The molecular, temporal and region-specific requirements of the beta isoform of Calcium/Calmodulin-dependent protein kinase type 2 (CAMK2B) in mouse locomotion

Genetic approaches using temporal and brain region-specific restricted gene deletions have provided a wealth of insight in the brain regions and temporal aspects underlying spatial and associative learning. However, for locomotion such extensive studies are still scarce. Previous studies demonstrated that Camk2b–/– mice, which lack the β isoform of Calcium/Calmodulin-dependent protein kinase 2 (CAMK2B), show very severe locomotion deficits. However, where these locomotion deficits originate is unknown. Here we made use of novel Camk2b mutants (Camk2bf/f and Camk2bT287A), to explore the molecular, temporal and brain region-specific requirements of CAMK2B for locomotion. At the molecular level we found that normal locomotion requires Calcium/Calmodulin mediated activation of CAMK2B, but CAMK2B autonomous activity is largely dispensable. At a systems level, we found that global deletion of Camk2b in the adult mouse causes only mild locomotion deficits, suggesting that the severe locomotion deficits of Camk2b–/– mice are largely of developmental origin. However, early onset deletion of Camk2b in cerebellum, striatum or forebrain did not recapitulate the locomotion deficits, suggesting that these deficits cannot be attributed to a single brain area. Taken together, these results provide the first insights into the molecular, temporal and region-specific role of CAMK2B in locomotion.


The role of Calcium/Calmodulin-dependent and autonomous activity of CAMK2B in locomotion.
We have previously shown that Camk2b -/mice (in which exon 11 was deleted) show severe locomotion deficits on the rotarod and the balance beam 4 . Since CAMK2 can have both structural and enzymatic functions in the brain, we first set out to understand which molecular aspects of CAMK2B are important for normal locomotion.
The Camk2b A303R mutation interferes with the Calcium/Calmodulin dependent activity, as well as the autonomous (Calcium/Calmodulin independent) activity of CAMK2B. To specifically investigate the role of autonomous activity in locomotion, we generated an autophosphorylation-deficient CAMK2B mouse mutant in which Threonine 287 is substituted by an Alanine, thus blocking autonomous CAMK2B activity (Camk2b T287A/ T287A ) (Fig. 1b,c). Protein quantification using western blot indeed revealed a near complete absence of T287-phosphorylated CAMK2B in Camk2b T287A/T287A mice without changes in T286-phosphorylated CAMK2A or total levels of both CAMK2A and CAMK2B ( Fig. 1d; for quantification see Table 1).
Temporal involvement of CAMK2B in locomotion. Expression of CAMK2B starts during early development (E12.5) 11 , hence the motor deficits seen in the Camk2b -/mice could very well be due to a crucial role for CAMK2B during development. To assess temporal contribution of CAMK2B in the severe locomotion deficits, we generated a novel floxed allele of Camk2b, with LoxP sites around exon 2 (containing the catalytic site) of the Camk2b gene (Camk2b f/f ), which was crossed into the C57Bl/6 background (see Methods and Fig. 2a). To ensure that the locomotion deficits are also present in the C57Bl/6 background (the previous mutants were tested in an F2 129P2-C57Bl/6 hybrid background), we crossed the Camk2b f/f mice with a Cag-cre transgene, deleting exon 2 from germline (Camk2b Δex2/Δex2 ). Using immunocytochemistry and western blots, we confirmed the global deletion of CAMK2B in this line. As expected, Camk2b Δex2/Δex2 showed loss of CAMK2B without changes in the expression of CAMK2A (Fig. 2b,c; Table 1). Furthermore, Camk2b Δex2/Δex2 mice showed a severe locomotion deficit compared to their wildtype littermates (effect of genotype: F 1,14 = 48.01, p < 0.001, repeated measures ANOVA; Fig. 2d), which indicates that the motor deficits are present in both F2 129P2-C57Bl/6 hybrid mice 4 as well as congenic (16 backcrosses) C57Bl/6 mice.
After having shown that our novel Camk2b mutant indeed recapitulates the severe locomotion deficit described before, we continued determining the developmental component of CAMK2B in the locomotion deficit. Therefore Camk2b f/f mice were crossed with CAG-Cre ER , giving us Tamoxifen-dependent temporal control over gene deletion. As expected, [12][13][14] week old Camk2b f/f ;CAG-Cre ER mice showed loss of CAMK2B 4 weeks after 4 daily consecutive injections with Tamoxifen and showed no changes in levels of CAMK2A (Fig. 2b, We tested 8-10 week old mice prior to Tamoxifen-induced deletion on the rotarod and found no difference between genotypes (effect of genotype: F 1,38 = 0.53, p = 0.47, repeated measure ANOVA), indicating that before Tamoxifen mediated gene deletion, the Camk2b f/f ;CAG-Cre ER mice do not have a locomotion deficit (Fig. 2e, left). When tested 28 days after Tamoxifen-induced deletion, Camk2b f/f ;CAG-Cre ER mice showed a clear locomotion deficit on the accelerating rotarod (effect of genotype: F 1,38 = 24.25, p < 0.001, repeated measures ANOVA; Fig. 2e, right), however, this phenotype was not as severe as upon germ-line deletion. To make sure that the milder phenotype was not caused by the initial rotarod testing before gene deletion, we induced Camk2b gene deletion in a naïve cohort of 8-10 week old mice, and tested the mice 4 weeks after gene deletion. Although these Camk2b f/f ; CAG-Cre ER mice showed no improvement of locomotion over time compared to their Camk2b f/f littermates without expression of CRE (overall effect of genotype: F 1,14 = 4.01, p = 0.06; effect of time: F 1,14 = 6.48, p < 0.001; interaction: F 1,14 = 3.02, p < 0.05 Fig. 2f), performance at the first two trials was indistinguishable from control mice, which is markedly different from Camk2b Δex2/Δex2 mice (Fig. 2d) Taken together, deletion of CAMK2B in adulthood resulted in a much milder locomotion deficit compared to germline deletion of CAMK2B, indicating a significant developmental origin for the locomotion deficits seen in Camk2b Δex2/Δex2 mice.
Brain areas contributing to the CAMK2B dependent rotarod deficits. Lesion studies in rodents have indicated that several brain areas are involved in locomotion, the most important being the cerebellum, Figure 1. A crucial role for Calcium/Calmodulin-dependent activity, but not autonomous activity of CAMK2B in locomotion. Locomotion was tested using the accelerating rotarod. (a) Camk2b A303R/A303R mice (n = 6) show a severe impairment in locomotion compared to Camk2b +/+ control littermates (n = 14). (b) Schematic picture for the generation of the Camk2b T287A mutants. Exons are depicted as black boxes. The asterisk in exon 11 indicates the mutation at Thr287, where a new TspRI restriction site was introduced. The LoxP sites flanking the neomycin gene are depicted as triangles. The Diphtheria Toxin cassette (DTA) was cloned in the construct for positive selection. Recombined depicts the mutant Camk2b T287A locus after homologous recombination. Δ NEO depicts the mutant Camk2b T287A locus after Cre recombination. (c) Sequence of Thr287 in exon 11 showing the specific mutation made to induce Thr287Ala and introducing the TspRI restriction site used for genotyping. (d) Western blots probed with a phospho-specific antibody against ph-T287 reveal no detectable Thr287 phosphorylation in the hippocampus of Camk2b T287A/T287A mice. Note that the Thr287Ala mutation has no effect on ph-Thr286, CAMK2A, and CAMK2B protein levels. Actin levels are shown as loading control. (e) Camk2b T287A/T287A mice (n = 14) show no impairment in locomotion compared to Camk2b +/+ control littermates (n = 9). Error bars indicate SEM. striatum and motor cortex (for review see 16 ). To our knowledge, the specific contribution of each of these brain areas to rotarod motor behaviour has not been systematically assessed through genetic lesions. CAMK2B is expressed in all these brain regions, therefore we used our conditional mutant to assess which brain area, if not all, is responsible for the severe rotarod phenotype seen in Camk2b Δex2/Δex2 mice. Knowing that loss of CAMK2B in the cerebellum reverses plasticity at the parallel fiber-Purkinje cell synapse, we started with assessing the importance of the cerebellum in the rotarod phenotype. Camk2b f/f mice were crossed with two different cerebellum-specific Cre-lines: L7-cre (specific for cerebellar Purkinje cells) and Gabaa6-cre (specific for cerebellar granule cells). Both Cre-lines express the CRE protein within the first postnatal week 17,18 .
Camk2b f/f ;L7-cre mice showed selective loss of CAMK2B in cerebellar Purkinje cells (Fig. 3a-c; Table 1) and Camk2b f/f ;Gabaa6-cre mice showed selective loss of CAMK2B in cerebellar granule cells. Surprisingly, Camk2b f/f ; L7-cre mice also showed a decrease in levels of CAMK2A in the cerebellum, whereas in Camk2b f/f ;Gabaa6-cre mice CAMK2A levels were unaffected (Fig. 3a,c; Table 1). Importantly, Camk2b f/f mice lacking CRE recombinase did not show any detectible reduction of CAMK2B protein compared to Camk2b +/+ mice, indicating that the insertion of the LoxP sites did not interfere with Camk2b gene expression (Table 1).
For the rotarod experiments with the L7-cre mice, we used all four genotypes obtained from crossings of heterozygous Camk2b f/+ (with and without cre) mice: Camk2b f/f ;L7-cre, Camk2b f/f , Camk2b +/+ and Camk2b +/+ ;L7-cre. Finding no difference in locomotion between the Camk2b +/+ and Camk2b f/f mice, we decided for subsequent experiments to cross homozygous Camk2b f/f mice (with and without cre), resulting in only Camk2b f/f and Camk2b f/f ;cre offspring. Surprisingly, neither the cerebellar Purkinje cell nor granule cell specific Camk2b mutants showed a rotarod deficit on the one-day training paradigm (Fig. 3d,f; for statistics see Table 2). When tested for 5 consecutive days however (to assess motor learning), Camk2b f/f ;L7-cre mice showed impaired motor learning ( Fig. 3e; Table 2), whereas the granule cell specific mutants still showed no hint of any deficit compared to their littermate control mice ( Fig. 3g; Table 2).
Since the loss of CAMK2B in cerebellar granule cells or Purkinje cells does not appear to contribute to the locomotion deficit as seen in Camk2b Δex2/Δex2 mice, we then focused on striatum and motor cortex. Camk2b f/f mice were crossed with Rgs9-cre (deletion specifically in most medium spiny projection neurons of the striatum 19 ) and Emx-cre (deletion in glutamatergic pyramidal neurons in cortex and hippocampus) transgenic mice to test respectively striatal and cortical involvement in the rotarod phenotype. Rgs9-cre is expressed from P8 onwards 20 and Emx-cre is expressed from E10.5 21 . Camk2b f/f ;Rgs9-cre mice showed selective loss of CAMK2B in striatum (Fig. 4a,b; Table 1) and Camk2b f/f ;Emx-cre mice showed selective loss of CAMK2B in glutamatergic pyramidal neurons in hippocampus and cortex with no changes in levels of CAMK2A in both mutants (Fig. 4a,b; Table 1). Importantly, none of the specific lines showed notable off-target deletion in other brain areas (data not shown).
Notably, neither the striatal nor forebrain-specific Camk2b mutants showed a rotarod deficit on the one-day training paradigm (Fig. 4c,e; for statistics see Table 2). When tested for 5 consecutive days these region-specific mutants still showed no deficits compared to their controls (Fig. 4d,f; Table 2). Interestingly, the Camk2b f/f ; Emx-cre mice even showed significantly enhanced locomotion ( Fig. 4f; Table 2).   Taken together these results show that selective deletion of CAMK2B in brain areas supporting locomotion, is not sufficient to recapitulate the locomotion deficits observed in either the Camk2b Δex2/Δex2 or the Camk2b f/f ; CAG-Cre ER mice. This suggests that multiple brain areas are responsible for locomotion deficits observed in Camk2b Δex2/Δex2 mice. The only mouse line showing a small but significant effect on motor learning (5 day paradigm), but not motor performance (1 day paradigm), was the Purkinje cell specific knockout (Camk2b f/f ;L7-cre). However, this effect is still marginal compared to Camk2b Δex2/Δex2 mice.

Discussion
Using several novel as well as previously described Camk2b mouse mutants, we dissected the molecular, temporal and systems requirements for CAMK2B in locomotion. At the molecular level we showed that even though the autonomous activity of CAMK2B in locomotion is dispensable, the ability of CAMK2B to bind Calcium/ Calmodulin is crucial for normal locomotion. At the developmental level we found that deletion of Camk2b in the adult mouse causes a much milder locomotion deficit compared to germ line Camk2b deletion. Finally, at a systems level we showed that early onset deletion of Camk2b in any of the major motor areas of the brain (cerebellum, striatum or forebrain) did not recapitulate the locomotion deficit of global Camk2b deletion. Taken together these results suggest that the locomotion deficits are mainly of a developmental origin, as well as a result of interplay between multiple regions (systems level).
We showed here that loss of autophosphorylation of CAMK2B at the T287 site does not significantly affect rotarod behaviour, although a small trend could be seen. In contrast, mutating the Calcium/Calmodulin binding site of CAMK2B, such that it cannot bind Calcium/Calmodulin and rendering CAMK2B bound to actin (Camk2b A303R/A303R ), led to a severe locomotion deficit. Whether the inability to bind Calcium/Calmodulin and becoming active, or the inability to be released from actin causes the deficits observed, remains to be investigated. CAMK2B forms heteromers with CAMK2A 22 and ratios of CAMK2A/CAMK2B can differ between cell types 22,23 . It has been shown for the hippocampus (in dissociated neuronal cultures and using Camk2b A303R/A303R mice) that CAMK2B kinase activity is not important for synaptic plasticity or hippocampal learning, but that CAMK2B is required for regulating the subcellular location of CAMK2A 5,9 . However, in Purkinje cells, CAMK2B not only regulates the subcellular localization of CAMK2A, but also plays an important enzymatic role in regulating the direction of plasticity at the parallel fiber-Purkinje cell synapse 4 . These different requirements likely reflect the differences in the CAMK2A/CAMK2B ratio in the different brain areas. Thus, when the relative expression level of CAMK2A is low, both the enzymatic and the non-enzymatic function of CAMK2B may be important, whereas in areas where CAMK2A is abundant, CAMK2B serves mainly to regulate the subcellular localization of CAMK2A. It would be interesting to investigate the CAMK2A/CAMK2B ratio in the specific motor areas of the brain to get more insight in which brain areas are involved in the severe locomotion deficits seen in the Camk2b -/and Camk2b A303R/A303R mice. Notably, Camk2b f/f ;L7-cre mice showed a significant decrease in CAMK2A in western blot when only CAMK2B was deleted. A similar trend could be found in the cerebellum of Camk2b f/f ;CAG-Cre ER and in Camk2b -/mice 4 . These results imply that specifically in Purkinje cells, which are the only cells in the cerebellum expressing CAMK2A, CAMK2B stabilizes CAMK2A, and that absence of CAMK2B therefore results in reduced levels of CAMK2A. However, the exact mechanism behind this regulation requires further research.
From a developmental point of view, we found that deleting most of Camk2b in the adult mice resulted in a significant locomotion impairment showing its necessity for normal motor behaviour. However, the phenotype was rather mild compared to the germ-line knock out mouse, indicating that acute deletion of Camk2b does not dramatically affect neuronal function. These results indicate that the severe locomotion deficits of Camk2b -/mice largely arise during development of the nervous system. Unfortunately, little is known about the role of both the CAMK2 isoforms in brain development. For CAMK2A, which starts to be expressed around P1, we recently showed that deletion of Camk2a in adulthood is as detrimental as a germline deletion of Camk2a for hippocampal learning and plasticity 12 , indicating an important post-developmental role. CAMK2B however, starts to be expressed already during early embryonic development (around E12.5) 11,24 . Therefore it is likely that CAMK2B plays a significant role during development. Previous studies looking at the developmental role of CAMK2B have shown an important role for CAMK2B in neurite extension and spine formation in acute knock-down experiments in neuronal cultures 25 . However, no morphological changes were found in the Camk2b -/mice 4,6 , nor in neuronal cultures obtained from Camk2b -/mice 5 , thus it is unclear how a germ-line deletion of Camk2b can be so profoundly affected.
Locomotion deficits can be caused by dysfunction of several different brain areas (as reviewed by 16 ). Using lesion studies it was shown that the cerebellum and thalamus are important in rotarod behaviour [26][27][28] , but lesioning the dorsal striatum did not cause motor impairments on the rotarod 29 . As for the motor cortex, some lesion studies have been performed, but the effect of the lesions was not assessed by the rotarod task. However, a recent pharmacological study showed that mGluR5 inhibition in primary motor cortex affected rotarod performance, which was not observed upon mGluR5 inhibition in dorsolateral striatum 30 . In contrast, deletion of the Nmdar1 gene from P8 onwards using Rgs9-cre mice in the striatum resulted in performance and learning deficits on the mice (n = 8) show a significant impairment in locomotion compared to wildtype littermates (n = 8). (e) [8][9][10] week old Camk2b f/f ;CAG-Cre ER mice (n = 13) and their Camk2b f/f control littermates (n = 17) were trained on the rotarod before (Day 1) and 4 weeks after Tamoxifen injections (Day 28). Before deletion, both genotypes performed equally. After deletion, Camk2b f/f ;CAG-Cre ER mice showed a significant impairment of locomotion compared to Camk2b f/f control littermates as shown in the 5 trials given 28 days after the first injection.   20 . Taken together, these studies all indicate that, depending on the lesion or genetic change, several specific brain areas could potentially be responsible for deficits in motor performance as measured with the rotarod.
Surprisingly, restricted deletion of CAMK2B in the major motor areas of the brain did not reveal a specific brain area or cell type to be responsible for the severe rotarod phenotype observed in the Camk2b -/mice. In these groups we not only assessed motor performance, but also motor learning by testing the mice over 5 consecutive days. Even though there are small variations in performance and learning in the control groups between the different experiments, the effects of Camk2b deletion on locomotion was only marginal compared to the profound phenotypes seen in Camk2b -/-, Camk2b Δex2/Δex2 and Camk2b A303R/A303R mice. There are three possible explanations for our findings. First, onset of Camk2b gene deletion in our mutants is likely to be too late. Three of the cre-lines used in this study start cre expression after onset of CAMK2B expression (L7-cre starts expression at P1 18 , Gabaa6-cre starts expression at P5-P7 17 , and Rgs9-cre expression is seen as early as P8 20 ). Thus, these lines start off with normal embryonic CAMK2B expression. Since we do not know the critical time window for CAMK2B in rotarod behaviour, it could be that in these mice CAMK2B is deleted too late in these separate brain regions or cell types to cause a significant effect on rotarod behaviour, comparable to Camk2b f/f ;CAG-Cre ER mice, where adult deletion resulted only in a mild phenotype when CAMK2B was deleted throughout the brain. Only the Emx-cre line, which expresses cre from E10.5 21 , induces deletion of Camk2b before onset of CAMK2B expression (E12.5). Interestingly, deletion in this cre-line resulted in enhanced, instead of impaired performance. Second, Camk2b deletion in thalamus, brainstem and the deep cerebellar nuclei is only obtained in the Camk2b Δex2/Δex2 mice. Since it has been shown that these areas are involved in locomotion 16 , there is a possibility that the motor deficits arise from those areas. Third, the severity of the locomotion deficits observed in the Camk2b -/mice 4 could indicate a distributed role for CAMK2B in several brain areas and cell types, each one of them contributing a little to the deficit. For example, we cannot exclude that deleting Camk2b in Purkinje-cells and cerebellar granule cells simultaneously would have an effect on locomotion. Additionally, even though the different cre-lines used in this study were very effective at deleting Camk2b throughout multiple brain areas and cell types (as judged by immunuhistochemstry and Western blot), we cannot exclude the possibility that a small number of cells did not undergo deletion, and prevented the lack of a severe phenotype.
Taken together we conclude that the nature of severe the locomotion deficits observed in the Camk2b -/is predominantly of developmental origin and cannot be attributed to one specific motor area or cell type in the brain. Moreover we conclude that the Calcium/Calmodulin-dependent activation of CAMK2B is essential for normal locomotion, but that the autonomous activity of CAMK2B is largely dispensable for normal locomotion. These findings are in sharp contrast to the role of CAMK2A in spatial learning, where loss of CAMK2A in adult mice recapitulates the phenotype of Camk2a -/mice, where selective loss of CAMK2A in the hippocampus causes spatial learning deficits, and where loss of autonomous CAMK2A activity causes severe spatial learning deficits 12,31 .

Materials and Methods
Animals. Camk2b A303R/A303R and Camk2b T287A/T287A mice were tested in a 129P2-C57Bl/6OlaHsd F2 hybrid background. Camk2b Δex2/Δex2 mice were backcrossed > 16 times into C57Bl/6JOlaHsd background. All conditional mice used in this study were backcrossed 10-12 times into the C57BL/6JOlaHsd background and crossed with cre lines maintained in the C57BL/6JOlaHsd background. Mice were genotyped when they were 7-10 days old, and re-genotyped after the mice were sacrificed. Genotyping records were obtained and kept by a technician not involved in the experimental design, performance and analysis. All mice were tested between 2-3 months of age, except the Camk2b f/f ;Rgs9-cre group, which was tested between 3-5 months of age. All mice were kept grouphoused in IVC cages (Sealsafe 1145 T, Tecniplast) with bedding material (Lignocel BK 8/15 from Rettenmayer) on a 12/12 h light/dark cycle in 21 °C (± 1 °C), humidity at 40-70% and with food pellets (801727CRM(P) from Special Dietary Service) and water available ad libitum. For all experiments mutants were compared to WT or cre-negative homozygous floxed littermates. All groups were matched for age and sex and all experiments were  Generation of Camk2b A303R/A303R and Camk2b T287A/T287A mice. Generation of the Camk2b A303R/ A303R mice has been described previously 5 . The Camk2b T287A targeting construct to generate Camk2b T287A/T287A mice was generated as follows. The Camk2b genomic sequence (ENSMUSG00000057897) was obtained from a public database (Ensembl) and used to design the primers for the targeting constructs. PCR fragments encompassing exon 6-11 using 5′ primer: 5′-GGTACCTGAGGAAGGTGCCAGCTCTGTCCC-3′ and 3′ primer: 5 ′ -GTC GAC CAGGGTAGTCAC GGT TGTC C-3′ (5.3 Kb; exon denot at ion according to ENSMUST00000019133) and exon 11-12 using 5′ primer: 5′-GCGGCCGCCTGTTAAAGGAATGGTTCTC -3′ and 3′ primer: 5′-ATGCATCTAAAAGGCAGGCAGGATGATCTGC-3′ (6 Kb) were amplified using High Fidelity Taq Polymerase (Roche) on ES cell genomic DNA and cloned on either site of a PGK-Neomycin selection cassette. All exons were verified by sequencing. Site directed mutagenesis was used to introduce the Thr287Ala point mutation, which induced a TspRI restriction site (Fig. 1). For counter selection, a gene encoding Diphtheria toxin chain A (DTA) was inserted at the 5′ end of the targeting construct. The targeting construct was linearized and electroporated into E14 ES cells (derived from 129P2 mice). Cells were cultured in BRL cell conditioned medium in the presence of Leukaemia inhibitory factor (LIF). After selection with G418 (200 μ g/ml), targeted clones were identified by PCR (long-range PCR from Neomycin resistance gene to the region flanking the targeted sequence). A clone with normal karyotype was injected into blastocysts of C57Bl/6 mice. Male chimeras were crossed with female C57Bl/6 mice (Harlan). The resulting F1 heterozygous mice (in the 129P2-C57Bl/6 background) were subsequently inter-crossed to obtain F2 129P2-C57Bl/6 hybrid mice to generate homozygous mutants and wild-type littermate controls.

Generation of floxed Camk2b and Camk2b
Δex2/Δex2 mice. The floxed Camk2b targeting construct was generated as follows. The Camk2b genomic sequence (ENSMUSG00000057897) was obtained from a public database (Ensembl) and used to design the primers for the targeting constructs. PCR fragments encompassing intron 1 using 5′ primer: 5′-TTTGGTACCGCATTGTGGGCATCTATGAAG-3′ and 3′ primer: 5′-AAAGGATCCAGTCAGCTGGAATGAGACGTG-3′ (2.6 Kb; intron and exon denotation according to ENSMUST00000019133), exon 2 using 5′ primer: 5′-TTTATGCATGATAGACCTGGGTGTTACACAG-3′ and 3′ primer: 5′-AAAGTCGACCTTCAGGTCTGGGACAGAG-3′ (583 bp) and intron 2 using 5′ primer: 5′-TTTGCGGCCGCAGAAGTCCTCATATTGGGGAGG-3′ and 3′ primer: 5′-AAACCGCGGTGACTCCT AATGCAGAAGACACC-3′ (4 kb) were amplified using High Fidelity Taq Polymerase (Roche) on ES cell genomic DNA. The 5′ fragment containing part of intron 1 was inserted before the first LoxP site whereas exon 2 and part of intron 2 were cloned on either site of a PGK-Neomycin selection cassette, which is flanked by frt and LoxP sites (for the schematics see Fig. 2). Exon 2 and flanking intronic sequences were sequenced to verify that the absence of secondary mutations. For counter selection, a gene encoding Diphtheria toxin chain A (DTA) was inserted at the 5′ of the targeting construct. The targeting construct was linearized and electroporated into E14 ES cells (derived from 129P2 mice). Cells were cultured in BRL cell conditioned medium in the presence of Leukaemia inhibitory factor (LIF). After selection with G418 (200 μ g/ml), targeted clones were identified by PCR (long-range PCR from Neomycin resistance gene to the region flanking the targeted sequence). To delete the Neomycin resistance cassette and obtain the conditional construct, the correctly targeted clones were transiently transfected with an flp recombinase (Fig. 2). Finally, a clone with correct karyotype was injected into blastocysts of C57Bl/6 mice. Male chimeras were crossed with female C57BL/6JOlaHsd mice (Harlan). The resulting F1 heterozygous Camk2b f/+ mice, which were backcrossed 10-12 times with C57BL/6JOlaHsd mice before generating the brain region and temporally restricted mutants as described below. To obtain the Camk2b Δex2/Δex2 mice, the Camk2b f/+ mice were crossed with transgenic Cag-cre mice 32 . The Camk2b Δex2/+ heterozygous offspring were backcrossed >16 times in C57BL/6JOlaHsd to obtain a congenic line. Mice were trained with an inter-trial interval of 45 minutes. We used two different paradigms on the accelerating rotarod. For motor performance (which we define as locomotion throughout the text) we used 5 consecutive trials with naïve mice (1 day paradigm). To assess motor learning we took the average of the first two trials of the first day and continued training the mice for another 4 consecutive days with 2 trials per day (5 day paradigm).

Generation of Camk2b
Immunohistochemistry and immunofluorescence. Mice were anaesthetized with pentobarbital and perfused transcardially with PBS followed by freshly prepared 4% paraformaldehyde solution (PFA, Sigma). Brains were taken out after perfusion and post-fixed for 1.5 hours in PFA and afterwards kept in 30% sucrose solution overnight. Immunohistochemistry was performed on free-floating 40μ m thick sagittal cryostat sections. Sections were washed in PBS once and afterwards primary antibodies were added (anti-CAMK2B, 1:2000, #ab34703, Abcam) diluted in PBS containing 2% NHS, 0.5% Triton-X 100 and 150 mM bovine serum albumin (BSA) and kept at 4 °C overnight for 48 hours. Two days later sections were washed 3 times with PBS and then secondary antibodies were added (biotinylated goat anti-rabbit IgG antibody, Vector Laboratories, Burlinghame, CA; 1:200) diluted in PBS containing 2% NHS, 0.5% Triton-X 100 and 150 mM bovine serum albumin (BSA) for 1 to 2 hours on room temperature. For diaminobenzidine (DAB) staining, sections were processed using a standard avidin-biotin-immunoperoxidase complex method (ABC, Vector Laboratories, USA) and 0.05% DAB as the chromogen. Sections were mounted on slices using chrome(3)potassiumsulfatedodecahydrate and left to dry. The next day, slides were dehydrated in alcohol, cleared with xylene and covered using Permount (Fisher Scientific, USA). For immunofluorescence, the same CAMK2B was used as in immunohistochemistry (1:1000) and we used Cy3 rabbit (1:200) as a secondary antibody. After 1-2 hours incubation of the secondary antibody at room temperature sections were washed four times in PB (0.05 M) and mounted on slices using chrome (3) potassiumsulfatedodecahydrate and left to dry. Finally, sections were covered using Mowiol (Sigma-Aldrich).
Data analysis and statistics. All experiments and analyses were performed blind to genotype. All behavioural tests were analysed using a 2-WAY repeated measures ANOVA to determine the performance and learning of the genotypes, with genotype as the between subjects factor and the repeated measures as within subject factor. α was set at 0.05. All values represent average+ /− SEM. Group sizes for each genotype are depicted in the figure legends. All statistics were performed in Graphpad Prism. * p < 0.05, * * * p < 0.001.