A New Mouse Model Related to SCA14 Carrying a Pseudosubstrate Domain Mutation in PKCγ Shows Perturbed Purkinje Cell Maturation and Ataxic Motor Behavior

Spinocerebellar ataxias (SCAs) are diseases characterized by cerebellar atrophy and loss of Purkinje neurons caused by mutations in diverse genes. In SCA14, the disease is caused by point mutations or small deletions in protein kinase C γ (PKCγ), a crucial signaling protein in Purkinje cells. It is still unclear whether increased or decreased PKCγ activity may be involved in the SCA14 pathogenesis. In this study, we present a new knock-in mouse model related to SCA14 with a point mutation in the pseudosubstrate domain, PKCγ-A24E, known to induce a constitutive PKCγ activation. In this protein conformation, the kinase domain of PKCγ is activated, but at the same time the protein is subject to dephosphorylation and protein degradation. As a result, we find a dramatic reduction of PKCγ protein expression in PKCγ-A24E mice of either sex. Despite this reduction, there is clear evidence for an increased PKC activity in Purkinje cells from PKCγ-A24E mice. Purkinje cells derived from PKCγ-A24E have short thickened dendrites typical for PKC activation. These mice also develop a marked ataxia and signs of Purkinje cell dysfunction making them an interesting new mouse model related to SCA. Recently, a similar mutation in a human patient was discovered and found to be associated with overt SCA14. RNA profiling of PKCγ-A24E mice showed a dysregulation of related signaling pathways, such as mGluR1 or mTOR. Our results show that the induction of PKCγ activation in Purkinje cells results in the SCA-like phenotype indicating PKC activation as one pathogenetic avenue leading to a SCA. SIGNIFICANCE STATEMENT Spinocerebellar ataxias (SCAs) are hereditary diseases affecting cerebellar Purkinje cells and are a one of neurodegenerative diseases. While mutation in several genes have been identified as causing SCAs, it is unclear how these mutations cause the disease phenotype. Mutations in PKCγ cause one subtype of SCAs, SCA14. In this study, we have generated a knock-in mouse with a mutation in the pseudosubstrate domain of PKCγ, which keeps PKCγ in the constitutive active open conformation. We show that this mutation leading to a constant activation of PKCγ results in a SCA-like phenotype in these mice. Our findings establish the constant activation of PKC signaling as one pathogenetic avenue leading to an SCA phenotype and a mechanism causing a neurodegenerative disease.


Significance Statement
Spinocerebellar ataxias (SCAs) are hereditary diseases affecting cerebellar Purkinje cells and are a one of neurodegenerative diseases. While mutation in several genes have been identified as causing SCAs, it is unclear how these mutations cause the disease phenotype. Mutations in PKCg cause one subtype of SCAs, SCA14. In this study, we have generated a knock-in mouse with a mutation in the pseudosubstrate domain of PKCg , which keeps PKCg in the constitutive active open conformation. We show that this mutation leading to a constant activation of PKCg results in a SCA-like phenotype in these mice. Our findings establish the constant activation of PKC signaling as one pathogenetic avenue leading to an SCA phenotype and a mechanism causing a neurodegenerative disease.
PKCg is a serine/threonine kinase dominantly expressed in cerebellar Purkinje cells and playing an important role for Purkinje cell functions (Chopra et al., 2018;Hirai, 2018). Increased PKC activity has a strong negative impact on Purkinje cell dendritic outgrowth and development in organotypic slice cultures (Metzger and Kapfhammer, 2000), whereas Purkinje cells from PKCg -deficient mice show no gross morphologic abnormalities (Chen et al., 2005). SCA14 is a dominantly inherited disease; therefore, gain of function or a dominant negative function rather than a loss of function of PKCg might cause SCA14.
To date, .40 missense mutations or deletions in PRKCG have been reported, and many mutations have been found in the cysteine-rich regulatory domain (C1A and C1B domain) while some other mutations have been found in the pseudosubstrate domain, the calcium binding C2 domain or the kinase domain (Adachi et al., 2008). The question is how all these mutations in PRKCG are linked to the disease phenotype. For some mutations, an increased PKCg kinase activity was shown pointing toward a gain of function phenotype (Verbeek et al., 2005;Adachi et al., 2008). In contrast, other SCA14 mutations, especially in the C1 domain, are functionally defective because of decreased binding to diacylglycerol pointing toward a loss of function phenotype (Verbeek et al., 2008). These findings suggest that pathology in SCA14 is not simply because of one single mechanism but rather the result of complex mechanisms involving dysregulation of PKCg (Shimobayashi and Kapfhammer, 2018;Wong et al., 2018).
We have previously created a transgenic mouse model expressing a kinase domain mutant PKCg with a constitutive activation. In this mouse model, we found subtle changes of the Purkinje cell dendritic tree and a mild ataxia (Ji et al., 2014;Trzesniewski et al., 2019). In these mice, there are still two normal alleles of PKCg present, making it difficult to compare the model to human disease. We have now created a new knock-in mouse model with a mutation in the pseudosubstrate domain. This autoinhibitory domain is crucial for regulating PKCg activity by preventing access of substrates to the kinase domain (Newton, 2018). The pseudosubstrate domain is only dissociated from the kinase domain after binding of diacylglycerol (Baffi et al., 2019), allowing PKCg substrates to access the kinase domain and become phosphorylated (Newton, 2018). This "open-active" conformation of PKCg is then subject to dephosphorylation and degradation. This activation cycle of PKCg is well controlled in Purkinje cells and critical for the regulation of dendritic development, synaptic function in LTD, and synapse formation Saito and Shirai, 2002). Mutations in the autoinhibitory pseudosubstrate domain reduce its affinity to the kinase domain and will increase kinase activity but also induce PKCg dephosphorylation and degradation (Pears et al., 1990).
In this study, we introduced a mutation in the pseudosubstrate sequence, which keeps PKCg in the constitutive active open conformation; and we generated a knock-in mouse carrying this mutation. Another pseudosubstrate domain mutation at the same A24 position (A24T) was recently identified in a human SCA14 patient (Chelban et al., 2018). We found that the A24E mutation indeed induced increased PKC activity but at the same time made PKCg very prone to degradation. Purkinje cells expressing the mutated PKCg showed compromised dendritic development; and in the corresponding knock-in mouse model, we observed a marked ataxia, altered Purkinje cell morphology, and abnormal climbing fiber (CF) termination. Gene expression profiling revealed alterations in related signaling pathways, such as Type 1 metabotropic glutamate receptor (mGluR1) or mTOR. Our results support the concept that the regulation of PKC activity is crucial for Purkinje cell function and one important contributor to the pathogenesis of SCA14 and other SCAs.
The PCR products sequence were confirmed by DNA sequencing (Microsynth).

Generation of transgenic mice
Animal experiments were conducted in accordance with the EU Directive 2010/63/EU for animal experiments and were reviewed and permitted by Swiss authorities. All experiments were done for both male and female mice. The point mutations (c. 71C . A; p. Ala 24 Glu, c. 78G . A; p. Arg 26 Arg) in the PRKCG gene (chromosome 7, 1.93 cM) were introduced into FVB background mice. Protospacer Adjacent Motif (PAM) sequence, 78G . A mutation that does not change amino acid was introduced to prevent the donor DNA from being a suitable target for Cas9 cleavage. The knock-in mice using Cas9/CRISPR engineering system were generated at the Center of Transgenic Models, University of Basel, with the rapid oocyte injection method. Alt-R CRISPR-Cas9 crRNA was designed specific to exon 1 of PRKCG (T TGC AGA AAG GGG GCG CTG) upstream of PAM sequence; Alt-R CRISPR-Cas9 tracrRNA and donor DNA (CAC CAG ATG AAG TCG  GTA CAG TGA CTG CAG AAG GTT GGC TGC TTG AAG AAA  CGA GCG GTG AAC TTG TGG CTC TTC ACC TCG TGG ACC  ACC TTC TGT CTC AGC TCC CCC TTT CTG CAA AAC AGG GGT  CGG GGT CCC CCC TCT GAG TCG CCT CCG CCA GGG CCC  AGA CCC GCC ATG) was obtained from Integrated DNA  Technologies. Crispr RNA, Cas9, and donor DNA were injected into FVB zygotes at the pronuclear site, and surviving embryos were transferred into pseudo-pregnant mothers. To identify founders, genotyping with genomic DNA samples from biopsies was performed by PCR. The primers for genotyping were as follows: forward primer, 59 TCC TTC CTA TCT CAG AGT CTG CG 39; and reverse primer, 59 GTT CCC AAG TCC CCT CCT TTT CC 39 (Microsynth). Then, the mutations at 71C . A and 78G . A in the PRKCG were confirmed by DNA sequencing. The fragment for sequencing was obtained by PCR with genomic DNA samples and primers as mentioned above. The confirmed point mutation knock-in founders were crossed with FVB (Janvier labs) mice to obtain Wt, heterozygous (Het), and homozygous (Homo) PKCg -A24E mice.
Real-time qPCR RNA was purified from the cerebellum of control and PKCg -A24E mice at different ages using the RNeasy Mini-kit (QIAGEN) following the instructions of the manufacturer; 1 mg of total RNA was used for reverse transcription reaction with SuperScript IV Reverse Transcriptase (Takara). Real-time qPCR was performed on a StepOne real-time PCR system (Applied Biosystems) using the SYBR Green master mix (Applied Biosystems). The following primers and reaction conditions were used: mouse PRKCG forward primer, 59-CAAAACAGAAGAC AAAGACC-39; mouse PRKCG reverse primer, 59-GGCCTTGAGT AGCTCTGAGACA-39; GAPDH forward primer, 59-AACTTTGGC ATTGTGGAAGG-39; and GAPDH reverse primer, 59-ACACATTGG GGGTAGGAACA-39.
Reaction conditions are as follows: 1 cycle of (10 min at 95°C), 40 cycles of (15 s at 95°C and 60 s at 65°C), and 1 cycle of (15 s at 95°C, 30 s at 72°C and 15 s at 95°C).
Reactions were quantified by the relative standard curve system and the cycle threshold method using the SDS2.2 software (Applied Biosystems). A relative quantitation value for each sample from the triplicates of that sample was calculated for each gene. The data were analyzed with GraphPad Prism software.

Organotypic slice cultures
Slice cultures were prepared as described previously (Gugger et al., 2012). Briefly, mice were decapitated at postnatal day 8 (P8), their brains were aseptically removed, and the cerebellum was dissected in ice-cold preparation medium: MEM, 1% Glutamax (Invitrogen), pH 7.3. Sagittal sections (350 mm thickness) were cut on a McIllwain tissue chopper under aseptic conditions. Slices were separated, transferred onto permeable membranes (Millicell-CM, Millipore), and incubated on a layer of neurobasal medium: Neurobasal A medium (Invitrogen) supplemented with B27 supplement (Invitrogen) and Glutamax (Invitrogen), pH 7.3, in a humidified atmosphere with 5% CO 2 at 37°C. The medium was changed every 2-3 d; 300 nM PMA (Tocris Bioscience) was added for PKC activation 24 h before slices were fixed; 10 mM Gö6983 (Tocris Bioscience) was added to the medium at each medium change for PKC inhibition, starting at 3 DIV (DIV3). Slices were kept in culture for a total of 7 d and analyzed with Western blot and immunohistochemical staining. To monitor protein degradation, slices were treated with 50 mg/ ml cycloheximide (Sigma Millipore) or 10 mM MG132 (Sigma Millipore) or both at DIV7, and protein was extracted at DIV8.

Histology and immunohistochemistry
Immunohistochemistry was performed as described previously (Shimobayashi et al., 2016). For the analysis of the cerebellar sections, Wt and PKCg -A24E mice were killed by perfusion with 4% PFA; then the cerebellum was removed, fixed with 4% PFA for 1 d at 4°C, and cryoprotected in 30% sucrose for 1 d at 4°C. Cerebella were frozen in isopentane on dry ice and embedded in optimal cutting temperature compound, and parasagittal cryosections of 20 mm thickness were cut on a Leica Microsystems CM1900 cryostat and used for immunohistochemistry. Organotypic slice cultures were fixed at DIV7 in 4% PFA overnight at 4°C. All reagents were diluted in 100 mM PB, pH 7.3. Cryosections or organotypic slices were incubated in blocking solution (0.5% Triton X-100, 3% normal goat serum [Invitrogen]) for 1 h to permeabilize the tissue and block nonspecific antigen binding. Primary antibodies were added in blocking solution and incubated overnight at 4°C with the following antibodies: mouse anti-calbindin D28K (1:1000, Swant; #300), guinea pig anti-vesicular glutamate transporter 2 (vGlut2) (1:1000, Millipore; AB2251-I), rabbit anti-PKCg (1:1000, Santa Cruz Biotechnology; sc-211). The staining was visualized with AlexaFluor-568 goat anti-rabbit (1:500, Invitrogen; A11011) and AlexaFluor-488 goat anti-mouse (1:500, Invitrogen; A11001). Stained slices or sections were mounted on glass slides using Mowiol. Cryosections or organotypic slices were viewed on an Olympus AX-70 microscope equipped with a Spot digital camera. Recorded images were adjusted for brightness and contrast with Photoshop image processing software.

Dissociated cerebellar cultures
Dissociated cerebellar cultures were prepared from mice essentially as described previously (Shimobayashi and Kapfhammer, 2018). After starting the culture, half of medium were changed twice a week. For PKC activation or inhibition assay, 15 nM PMA (Tocris Bioscience) for PKC activation or 5 mM Gö6983 (Tocris Bioscience) for PKC inhibition was added to the medium at each change starting at DIV7 or DIV4. For plasmid transfection, L7-based expression vectors were constructed using the primers mentioned above.
Immunohistochemistry of dissociated cerebellar cells After 14-18 d, cells were fixed in 4% PFA for 1 h at 4°C. All reagents were diluted in 100 mM PB, pH 7.3. Cells were incubated in blocking solution (0.5% Triton X-100, 3% normal goat serum [Invitrogen]) for 30 min at room temperature. Two different primary antibodies were simultaneously added to the cells in fresh blocking solution and incubated for 30 min at room temperature. After washing in PB, secondary antibodies were added to the slices in PB containing 0.1% Triton X-100 for 30 min at room temperature. For the analysis of vector expression in Purkinje cells, mouse anti-Calbindin D-28K (1:1000, Swant; #300) and polyclonal rabbit anti-GFP (1:1000, Abcam; ab6556) were used as primary antibodies, and AlexaFluor-568 goat anti-rabbit (1:1000, Invitrogen; A11011) and AlexaFluor-488 goat anti-mouse (1:1000, Invitrogen; A11001) were used as secondary antibodies to visualize Purkinje cells (Ji et al., 2014). Stained cells were viewed on an Olympus AX-70 microscope equipped with a Spot digital camera or were acquired with confocal microscopy (Carl Zeiss, LSM710) equipped with solid state lasers using a LD Plan-Neofluar 40Â objective (Carl Zeiss). Recorded images were adjusted for brightness and contrast with Photoshop image processing software.

Golgi-Cox staining
Mice from P14 to 1 year old were used for the Golgi-Cox study. The FD Rapid GolgiStain Kit (FD Neuro Technologies) was used for Golgi staining. Mice were killed and perfused with 4% PFA. The cerebellum was collected in 4 ml of impregnation Solutions A and B according to the instructions. After 3 weeks of impregnation, the cerebellum was placed in Solution C from the FD Rapid GolgiStain kit and stored at room temperature for 3 d. Then, cerebella were frozen in isopentane on dry ice and kept in À80°C until use. Sections were cut at 100 mm using a Leica Microsystems CM1900 cryostat, collected and mounted on gelatincoated slides using Solution C, and dried at room temperature overnight in the dark place. The next day, the staining was developed using Solutions D and E and distilled water at a 1:1:2 concentrations. The sections were then dehydrated in increasing alcohol concentrations and coverslipped using Eukitt mounting medium (Sigma Millipore). Slides were viewed on bright field microscope and images were taken using a Spot digital camera.

Experimental design and statistical analysis
The age of the mice for each experiment is shown in Table 1. Statistical comparisons were made using the GraphPad Prism 8.3.1 software package (GraphPad Software). The quantification of Purkinje cell dendritic tree size was done as previously described (Gugger et al., 2012). Purkinje cells that had a dendritic tree isolated from its surroundings were selected for analysis. Cells were photographed with a digital camera. An image analysis program (ImageJ, https://imagej.nih.gov/ij/) was used to trace the outline of the Purkinje cell dendritic trees yielding the area covered by the dendritic tree. More than 20 cells were acquired from at least three independent experiments were analyzed. The statistical significance was assessed by nonparametric Mann-Whitney's test. CIs were 95%, statistical significance when p , 0.05. Graphical data are represented as the mean 6 SEM.
To quantify the protein expression in Western blots, the immunoreactivity of each sample was normalized to the actinb signal, and the ratio was evaluated and then normalized to Wt control using LI-COR software (LI-COR Biosciences). Protein or phospho-protein expression differences between each sample were analyzed using the two-tailed Mann-Whitney test.
Statistical significance of the Footprint pattern test, the Rotarod test, and the Walking beam test was analyzed using ANOVA models, and Wt and PKCg -A24E mice groups were analyzed by two-way ANOVA test with Bonferroni correction.
Behavioral testing. Rotarod. To measure motor function, mice were placed on an accelerating rotarod (Rotamex-5, Columbus Instruments), and the speed of rotation was increased from 2 to 52 revolutions per minute over 4 min. The latency to fall from the rotarod was recorded. Data were collected for 5 trials per day after a training period of 5 trials per day for 4-5 d. The two-way ANOVA test with Bonferroni correction was conducted to examine the main effect of genotype on each day.
Walking beam test. In this test, the mice had to traverse an 80-cmlong and 8-mm-wide wooden bar. Every slip of a hind paw was recorded and counted. After a 1 d training period, data were collected for 10 trials on 4 or 5 different days. The statistical significance was assessed by twoway ANOVA test with Bonferroni correction.
Footprint pattern test. The footprint patterns were evaluated at 9month-old Wt and PKCg -A24E mice groups. Mouse paws were painted with nontoxic ink, and mice were placed at one end of a dark tunnel. Mice walked through the tunnel, and their footprints were analyzed for the width and length of each step. The statistical significance was assessed by two-way ANOVA test with Bonferroni correction.
Total RNA extraction and total RNA sequence The cerebellum was isolated, and quickly slices of 500 mm thickness were made. Purkinje layer and molecular layer were microdissected and harvested in RNAlater. Cerebellum lysates from 3 mice in each group were collected for total RNA isolation, following the instructions of the manufacturers of the Trizol and RNeasy Lipid Tissue Mini Kit (QIAGEN). Samples were eluted in 15 ml of RNase-free H 2 O, quantified using a Nanodrop ND-1000 (Thermo Fisher Scientific) spectrophotometer. RNA qualities were checked using Agarose Gel Electrophoresis and Bioanalyzer 2100 (Agilent Technologies). All samples were adjusted as 100 ng/ml and only used when RNA integrity number was ! 7.1. Library preparation and sequencing were performed at the Quantitative Genomics Facility of the University of Basel and the ETH Zurich at Basel.

Phosphoproteomics
The cerebella were isolated and quickly frozen with liquid nitrogen. Protein was extracted with lysis buffer, including PhosSTOP (Roche Diagnostics); then protein concentration was measured with the BCA Protein Assay kit (Invitrogen); 1 mg of mouse tissue was lysed in 80 ml of 8 M urea, 0.1 M ammonium bicarbonate, phosphatase inhibitors (Sigma Millipore; P5726 and P0044) by sonication (Bioruptor, 10 cycles, 30 s on/off, Diagenode), and proteins were digested as described previously (PMID:27345528). Pewptide samples were enriched for phosphorylated peptides using Fe(III)-IMAC cartridges on an AssayMAP Bravo platform as recently described (PMID: 28107008). Phospho-enriched peptides were resuspended in 0.1% aqueous formic acid and subjected to LC-MS/MS analysis using an Orbitrap Fusion Lumos Mass Spectrometer fitted with an EASY-nLC 1200 (Thermo Fisher Scientific). LC-MS was performed by Alexander Schmidt in the Biocenter of the University of Basel, Switzerland.
Ingenuity pathway analysis (IPA) IPA is a web-based biological analysis tool for omics data, and genomic data. We applied RNA sequence data and phosphoproteomics data into QIAGEN's IPA software (https://www.qiagenbioinformatics.com/products/ ingenuity-pathway-analysis/) and performed a core analysis that indicates not only direct but also indirect relationships between genes and proteins. The IPA software shows the overlapping canonical pathways, upstream regulators, and affected genes/proteins networks. 4-and 53-week-old 6D 4-and 40-week-old 7, Beam walk 3-and 6-month-old 7, Rotarod 3-and 6-month-old 7, Footprint 9-month-old 8 5-and 7-week-old

Results
A24E mutant PKCc is unstable, aggregates, and is partially degraded by proteasomes There are now .40 different missense mutations or deletions which have been found in human SCA14 patients (Chelban et al., 2018;Wong et al., 2018) (Fig. 1A). We have previously reported that, within these mutations, PKCg -S361G has increased kinase activity resulting in abnormal Purkinje cell development (Ji et al., 2014;Shimobayashi and Kapfhammer, 2018). The PKCg -S361G mutation is located in the kinase domain and thus not subject to regulation of PKC activity within the cell. We now aimed to generate a constitutive active form of PKCg by a mutation in the regulatory parts of the molecule. From PKCa, it is known that the substitution of the conserved alanine residue in the pseudosubstrate site of PKCa with a charged glutamic acid residue (E25) causes a reduction in the affinity of this sequence on the catalytic domain leading to the activated conformation of the PKCa protein (Pears et al., 1990). The pseudosubstrate domain is well conserved and exactly the same amino acid sequence from amino acids 22-29 from PKCa is found in PKCg with the alanine residue being in position 24. We first introduced the single point mutation on alanine 24 changing it to glutamic acid, called PKCg -A24E. Then GFP-PKCg -A24E was transfected to HeLa cells or HEK293T cells to observe the protein expression. As shown in Figure 1B, GFP-PKCg -A24E had a strong tendency for aggregation compared with GFP-PKCg -Wt, and the amount of protein expression was markedly reduced (GFP-PKCg -Wt = 100% vs GFP-PKCg -A24E = 29.7%; Fig. 1C, at 0 min [Starting point]). In order to see the protein half-life and degradation, we applied the translation inhibitor cycloheximide. Twenty-four hours after cycloheximide treatment, GFP-PKCg -Wt protein was still present in normal amounts (96.68% of the original GFP-PKCg -Wt expression), whereas GFP-PKCg -A24E was reduced down to 41.27% of the original GFP-PKCg -A24E expression (Fig. 1C), indicating that the reduced amounts of GFP-PKCg -A24E protein are because of increased degradation. As it is known that PKC protein family is mainly degraded via ubiquitin proteasome pathway (Wang et al., 2016), we used the proteasome inhibitor MG132 to block the ubiquitin proteasome degradation. Twenty-four hours after MG132 treatment, reduced GFP-PKCg -A24E protein level was rescued (PKCg -Wt = 100.0%; PKCg -Wt 1 MG132 = 192.9%; PKCg -A24E = 32.26%; PKCg -A24E 1 MG132 = 114.6%; n = 6), indicating that GFP-PKCg -A24E is unstable in cells and at least partially targeted for degradation via the proteasome pathway (Fig. 1D).
Increased PKC kinase activity in PKCc-A24E mice despite reduced protein levels As the mutation in the pseudosubstrate domain is supposed to keep the protein in a constantly active open conformation, we studied whether the overall PKC activity in cerebellar slice cultures of PKCg -A24E mice was reduced or increased. We used a general anti-phospho-serine PKC substrate antibody to see whether the phosphorylation of PKC substrates was increased or not and found that both Het and Homo PKCg -A24E mice had increased PKC kinase activity compared with Wt littermate controls (Fig. 4A). We also studied the phosphorylation of a known PKCg target protein, the NMDAR, which is composed of a heterodimer of at least one NR1 and one NR2A-D subunit. The NR1 subunit can be phosphorylated by PKCg at Ser890 (Sánchez-Pérez and Felipo, 2005). Using a phospho-specific antibody, we found that phosphorylation at this site was markedly increased both in Het and Homo PKCg -A24E mice (Fig. 4B). MARCKS is also a known phosphorylation target of PKC in many cells. Surprisingly, phospho-MARCKS protein expression compared with total MARCKS expression was strongly decreased in the PKCg -A24E mouse. A similar reduction was found in Wt slice cultures treated with PMA (Wt = 100.4 6 4.37%, n = 4; Het = 61.92 6 15.25%, n = 4; Homo = 33.98 6 5.83%, p = 0.0009, n = 4; Wt 1 PMA = 42.54 6 15.4%, p = 0.0033, n = 4; Het 1 PMA = 34.58 6 9.39%, p = 0.0010, n = 4; Homo 1 PMA = 16.89 6 3.557%, p , 0.0001, n = 4), suggesting that phospho-MARCKS protein is more dephosphorylated by phosphatases under the PKCg activation condition. This dephosphorylation could be mostly prevented by the use of the phosphatase inhibitor okadaic acid (Fig. 4C).

Altered morphology of Purkinje cells from PKCc-A24E mice in vitro
We studied the dendritic morphology of Purkinje cells from PKCg -A24E mice. In organotypic slice culture, the dendritic expansion of Purkinje cells from PKCg -A24E mice was severely impaired and the cells developed only short thickened dendrites similar to those found after PMA treatment (Ji et al., 2014). This was most evident in Homo PKCg -A24E mice, but also noticeable in Het PKCg -A24E mice (Fig. 5A). Statistical analysis showed that the size of Purkinje cells from Homo PKCg -A24E mice (62.1 6 5.23%, p , 0.0001) is significantly smaller than that of Wt (100 6 5.64%). Purkinje cell size of Het is also smaller than that of Wt (Het, 84.2 6 5.40%, p = 0.0718), but the difference did not reach statistical significance (values are shown as percentage of Wt [100%] and are the mean 6 SEM of 20 Purkinje cells).
With footprint gait analysis, we calculated the step width. A representative trace image of the walking pattern of 9month-old Wt, Het, and Homo PKCg -A24E mice is shown in Figure 7D, highlighting the ataxic gait with increased step width of Het and Homo PKCg -A24E mice. The difference in the step width was significant for Het and Homo PKCg -A24E mice (Wt = 1.30 6 0.058 cm, n = 12; Het = 2.60 6 0.154 cm, n = 12; Homo = 2.95 6 0.047 cm, n = 12; Fig. 7D,E). The footprint pattern in both Het and Homo PKCg -A24E mice showed clear signs of an ataxic gait.
We have also profiled protein phosphorylation by using phospho-proteomics, which might provide valuable insight into the mutant PKCg signaling. The abundance of 105 phosphopeptides was significantly upregulated while 69 phosphopeptides were significantly downregulated in Homo PKCg -A24E mice (p , 0.05) ( Table 2; Extended Data Fig. 8-3). The volcano plot data show proteins with increased phosphorylation in Homo PKCg -A24E mice to the right side and with less phosphorylation to the left side (Fig. 8D). The data show that Rmdn3 (also known as PTPIP51) and Homer3 are highly phosphorylated in Homo PKCg -A24E mice ( Fig. 8D; Extended Data Fig. 8-3). Rmdn3 was shown to be expressed in the Purkinje cell soma and dendrites (Koch et al., 2009) and is implied in calcium handling and interactions between mitochondria and the endoplasmic reticulum (Fecher et al., 2019). In the RNA sequence data, both RMDN3 and HOMER3 genes are also upregulated in PKCg -A24E mice (Fig. 8E). Western blotting confirmed the abundant expression of Rmdn3 and Homer3 in the cerebellum, with a slight upregulation in PKCg -A24E (Fig. 8F). IPA network analysis shows that Dlgap1-Homer3-Shank1/3-Syne1-Pclo signaling is more phosphorylated in PKCg -A24E mice, suggesting that postsynaptic scaffolding proteins might also be targets of PKCg phosphorylation.

Discussion
In this study, we present a new knock-in mouse model related to SCA14 with a constitutive activation by a mutated pseudosubstrate domain, which no longer can bind to the kinase domain and keeps the PKCg protein in the open active conformation. SEM. Statistical significance of the values of Wt and PKCg -A24E mice groups was analyzed by two-way ANOVA test with Bonferroni correction: *p , 0.05; ***p , 0.001; ****p , 0.0001. D, Footprints of 9-month-old Wt and PKCg -A24E mice were evaluated for step width. E, Wt = 1.30 6 0.058 cm, n = 12; Het = 2.60 6 0.154 cm, n = 12; Homo = 2.95 6 0.047 cm, n = 12. Data are mean 6 SEM. Statistical significance of the values of Wt and Het or Homo PKCg -A24E mice groups was analyzed by two-way ANOVA test with Bonferroni correction: ****p , 0.0001. This leads to a dramatic increase of dephosphorylation and protein degradation and to a drastic reduction of protein expression in cerebellar Purkinje cells. Despite this reduction in protein levels, there is clear evidence for an increased PKC activity in Purkinje cells from PKCg -A24E mice with the typical morphology of short thickened dendrites, a marked ataxia, and signs of Purkinje cell dysfunction. A similar mutation in a human patient was associated with overt SCA14. Our results show that the introduction of a new mutation leading to a constant activation of PKCg results in an SCA-like phenotype in these mice, establishing PKC activation as one pathogenetic avenue leading to an SCA phenotype.
Increased PKC activity despite reduced protein levels of PKCc-A24E because of increased degradation We introduced mutations in the pseudosubstrate domain for abolishing self-inhibitory effect result in a constantly activated PKCg enzyme (Pears et al., 1990;Newton, 2018). This concept was confirmed by transfecting pseudosubstrate domain mutated PKCg in Purkinje cells, which showed a severe reduction of their dendritic tree consistent with a constitutive activation of PKCg . Together with the increased PKC activation, we found a reduced amount of mutated PKCg -A24E protein. Treatment with cycloheximide and qPCR studies confirmed that this reduction was because of increased degradation as expected for the protein in the "open active" conformation (Newton, 2018). In this conformation, the kinase is subject to dephosphorylation by protein phosphatases and to ubiquitination and degradation via the proteasomal pathway (Hansra et al., 1999). Our in vitro studies thus agree with the known regulation of PKC activation and degradation and show that the increased degradation of the open active protein in transfected cells did result in reduced protein levels. These findings are similar to those reported for the original A25E construct in PKCa (Pears et al., 1990).
Dramatic reduction of PKCc-A24E protein in the PKCc-A24E mouse model is compatible with increased PKC activity While the effect of pseudosubstrate mutations has been studied extensively with diverse PKC variants in cell culture assays, the PKCg -A24E mouse is to our knowledge the first mouse model of such a pseudosubstrate mutation. We were surprised to find an almost complete absence of PKCg -A24E protein in the cerebellum, but qPCR confirmed that mRNA levels were similar in all genotypes, suggesting that the reduction was because of increased degradation (Fig. 3). Despite this dramatic reduction of PKCg -A24E protein, there was clear evidence for increased PKC activity. An antibody against an epitope recognizing PKCmediated phosphorylation showed increased phosphorylation of target proteins both in Het and Homo PKCg -A24E mice, and phosphorylation of the known target protein P890-NMDAR was also increased. The mice demonstrate that only a very small amount of active PKCg is required for inducing increased PKC activity in a cell.

Increased constitutive PKC activity interferes with Purkinje cell development and function
The increased constitutive activity of PKCg -A24E has negative effects on Purkinje cells both in cell culture and in vivo. This is particularly evident in organotypic slice cultures and dissociated cerebellar cultures. In both cases, Purkinje cells only develop a small abnormal dendritic tree with thickened dendrites. In slice cultures, we could show that this effect is because of increased PKC activity because it can be rescued by the application of PKC inhibitor like Gö6983. When analyzed by Golgi staining, a mild reduction of the Purkinje cell dendritic tree size becomes evident. A similar difference has already been found in the PKCg -S361G transgenic mouse (Ji et al., 2014;Trzesniewski et al., 2019). Importantly, also CF innervation is reduced, particularly in the distal parts of the Purkinje cell dendrites. This finding goes together with the known importance of Purkinje cell PKCg activity for CF innervation  and corresponds to similar findings in the PKCg -S361G transgenic mouse (Trzesniewski et al., 2019). The PKCg -A24E mice have a marked ataxia, which is most evident with testing on the walking beam which some Homo PKCg -A24E can barely cross; and also Het PKCg -A24E mice have a clearly A, IPA showed that many ubiquinone oxidoreductase subunits and cytochrome c oxidase subunits are significantly upregulated in Het PKCg -A24E mice. y axis is log2 fold change. B, Red complexes are upregulated in Het PKCg -A24E mice, which locate in the mitochondrial membrane. C, Network analysis was performed by IPA of gene sets upregulated or downregulated in Het PKCg -A24E mice compared with Wt littermates. Green genes represent decreases. Red genes represent increases. Many genes related to Purkinje cell morphology are downregulated in Het PKCg -A24E mice. D, Volcano plot of phospho-proteomics analysis of Wt versus Homo PKCg -A24E mice (n = 3). Lysates from 7-week-old Wt and PKCg -A24E mice were subjected to phosphoproteomic analysis by mass spectrometry. Differentially enriched phosphopeptides are shown in the volcano plot. x axis is log2 fold change, and y axis is p value. Rmdn3 and Homer3 are among the proteins with a significantly increased phosphorylation in PKCg -A24E mice. E, RNA sequence data show that Rmdn3 and Homer3 are upregulated in PKCg -A24E mice. F, Western blot analysis of Rmdn3 and Homer3 from organotypic slice cultures. Homer3 protein expression (Wt = 100.0%, n = 4; Het = 149.7%, p = 0.0286, n = 4; Homo = 184.2%, p = 0.0286, n = 4) and Rnmd3 protein expression (Wt = 100.0%, n = 4; Het = 167.0%, p = 0.0286, n = 4; Homo = 206.5%, p = 0.0286, n = 4) are increased in PKCg -A24E mice. Statistical analysis using the two-tailed Mann-Whitney test showing increased protein expression in PKCg -A24E mice.
increased number of slips at both ages tested. The presence of the ataxia reflects a dysfunction of the neuronal cerebellar circuits controlling precision of movements and integration of vestibular information.
The PKCc-A24E mouse is a novel mouse model related to SCA14 The PKCg -A24E mouse shows an ataxic phenotype but no extensive loss of Purkinje cells (data not shown). A similar situation applies to the PKCg -S361G mouse (Ji et al., 2014). The presence of the ataxia in the absence of major Purkinje cell loss points to an important aspect of SCAs. While it is generally assumed that Purkinje cell loss is the major cause of the patients' problems, the evidence for this assumption is weak. Indeed, in mice a loss of 90% of Purkinje cells is required for the manifestation of overt motor behavioral deficits (Martin et al., 2003). In a recent study of SCA14 families, most patients had only mild to moderate atrophy of the cerebellum, making it doubtful that their ataxia can be explained exclusively by Purkinje cell loss. In the patient carrying the A24T mutation corresponding to the PKCg -A24E mouse, only a mild cerebellar atrophy was found.
In an SCA1 mouse model, the development of ataxia and Purkinje cell loss could be dissociated (Duvick et al., 2010), suggesting that Purkinje cell dysfunction is a crucial aspect for the development of the ataxic phenotype together with Purkinje cell loss. The increased PKC activity, the impairment of dendritic development, and the behavioral deficits make the Het PKCg -A24E mice a valid mouse model related to SCA14.

Changes in gene expression and phosphorylation in
PKCc-A24E mice The mRNA profiling surprisingly yielded more results from the Het mice compared with the Homo mice. Many molecules in the oxidative phosphorylation pathway and mitochondrial molecules were upregulated in Het PKCg -A24E mice (Fig. 8A,B; Extended Data Fig. 8-1A). Mitochondria trafficking into dendrites is essential for Purkinje cell dendritic outgrowth, and proper oxidative phosphorylation for energy production in mitochondria is very important for neuron activity. Indeed, mitochondrial dysfunction has been found in several neurodegenerative diseases, such as Alzheimer's disease (Friedland-Leuner et al., 2014), Parkinson's disease, Huntington's disease, and SCA1 (Stucki et al., 2016). In addition, the Rictor signaling pathway is strongly affected (Extended Data Fig. 8-1C). This pathway is well known to be of outstanding importance for Purkinje cell development and function (Angliker et al., 2015). In the Homo PKCg -A24E mice, we find dysregulated genes involved in outgrowth and pruning of neuronal processes, such as the Eph receptors and changes in glutamate receptors and calcium homeostasis. This fits well with the idea that, in the PKCg -A24E mice, the constitutive activation of PKCg mimics a state of very strong synaptic activation, making it crucial for the Purkinje cells to handle the calcium release associated with this activation and limit the "natural" receptor activation through glutamate. With these compensatory mechanisms, the Purkinje cell would be stable and can survive, but it would be functionally compromised, resulting in the ataxic phenotype. This concept is also supported by the outcome of the phosphoproteomics analysis. Some of the most strongly phosphorylated proteins (e.g., Homer3 and Rmdn3) are involved in the control of receptor signaling and calcium handling; another one (Dpysl3) might control process outgrowth. Further studies will be required to further elucidate the exact role of these dysregulated proteins.

PKCc signaling and SCAs
In this manuscript, we show that the constitutive activation of PKCg in the PKCg -A24E mouse model is sufficient to induce a pathology related to SCA14, and this finding is supported by the recent identification of a human patient with a mutation at the same position (Chelban et al., 2018). However, other mutations causing SCA14 in humans affect the regulation of PKCg differently or even are kinase dead (Shirafuji et al., 2019); and of course, most SCAs are caused by mutations in different genes. Nevertheless, slowly a picture is emerging that many of these mutations affect the mGluR1-PKCg -inositol-1,4,5-trisphosphate receptor Type 1-calcium release pathway and can cause disease regardless of stimulation or inhibition of this pathway (Shimobayashi and Kapfhammer, 2018), meaning that it is not so important in which direction the activity if this pathway is pushed, but rather that the dynamic regulation of the activity of this pathway is disturbed. Although their overall activity pattern may look rather normal at first glance, Purkinje cells in these cases would be dysfunctional and the ataxic phenotype will become the more evident the more challenging the task is and can be present without noticeable Purkinje cell loss. Of course, in other cases, the loss of Purkinje cells rather than their dysfunction may be the key to SCA pathology. In the cases in which Purkinje cell dysfunction rather than death is at the base of the deficits, a pharmacological correction of the regulation of the affected pathway may be a valid therapeutic option.