DGKγ Knock-Out Mice Show Impairments in Cerebellar Motor Coordination, LTD, and the Dendritic Development of Purkinje Cells through the Activation of PKCγ

Abstract Diacylglycerol kinase γ (DGKγ) regulates protein kinase C (PKC) activity by converting DG to phosphatidic acid (PA). DGKγ directly interacts with PKCγ and is phosphorylated by PKCγ, resulting in the upregulation of lipid kinase activity. PKC dysfunction impairs motor coordination, indicating that the regulation of PKC activity is important for motor coordination. DGKγ and PKC are abundantly expressed in cerebellar Purkinje cells. However, the physiological role of DGKγ has not been elucidated. Therefore, we developed DGKγ knock-out (KO) mice and tested their cerebellar motor coordination. In DGKγ KO mice, cerebellar motor coordination and long-term depression (LTD) were impaired, and the dendrites of Purkinje cells from DGKγ KO mice were significantly retracted. Interestingly, treatment with the cPKC inhibitor Gö6976 (Gö) rescued the dendritic retraction of primary cultured Purkinje cells from DGKγ KO mice. In contrast, treatment with the PKC activator 12-o-tetradecanoylphorbol 13-acetate (TPA) reduced morphologic alterations in the dendrites of Purkinje cells from wild-type (WT) mice. In addition, we confirmed the upregulation of PKCγ activity in the cerebellum of DGKγ KO mice and rescued impaired LTD in DGKγ KO mice with a PKCγ-specific inhibitor. Furthermore, impairment of motor coordination observed in DGKγ KO mice was rescued in tm1c mice with DGKγ reexpression induced by the FLP-flippase recognition target (FRT) recombination system. These results indicate that DGKγ is involved in cerebellar LTD and the dendritic development of Purkinje cells through the regulation of PKCγ activity, and thus contributes to cerebellar motor coordination.


Introduction
Diacylglycerol (DG) is produced by the hydrolyzis of phosphatidylinositol-4,5-bisphosphate (PIP 2 ) by phospholipase C (PLC) in response to neurotransmitters, growth factors and hormones. DG functions as a lipid messenger and activator, by binding to several enzymes including protein kinase C (PKC; Almena and Mérida, 2011). DG signaling is terminated by DG kinase (DGK), which phosphorylates DG to produce phosphatidic acid (PA; Sakane et al., 2007). In other words, DGK indirectly inhibits PKC activity by reducing the level of DG. PA is also an important lipid messenger that activates various enzymes including phosphatidylinositol-4-phosphate-5kinase (PI4P5K; Jones et al., 2000), Raf1-kinase (Ghosh et al., 1996), mammalian target of rapamycin (mTOR; Fang et al., 2001), and atypical PKC (Limatola et al., 1994). Therefore, DGK is thought to have important physiological roles in response to different stimuli.
To date, 10 mammalian subtypes of DGK have been identified (Topham and Prescott, 1999;Kanoh et al., 2002;van Blitterswijk and Houssa, 2000). Among them, at least 8 subtypes are expressed in the brain with subtypespecific expression patterns, suggesting subtype-specific functions in neurons Tu-Sekine, 2011, Shirai and. For example, DGKg is abundantly expressed in cerebellar Purkinje cells (Goto and Kondo, 1999;Adachi et al., 2005). However, its function in neurons is still unclear.
Similarly, PKC has 10 subtypes and many of them show subtype-specific expression in the brain. For example, PKCa and g are expressed in cerebellar Purkinje cells  and regulate Purkinje cell function, including synaptic plasticity and dendritic development (Hirai, 2018). The activation of PKC induces long-term depression (LTD) and single climbing fiber (CF) innervation to Purkinje cells, which are important for motor coordination (Ito, 2002;Watanabe and Kano, 2011). PKCa knock-down blocks LTD (Leitges et al., 2004), and PKCg deficiency and constitutively active mutant PKCg induce the impairment of cerebellar motor coordination, including the loss of LTD, abnormal dendritic morphology and multiple CF innervation Kano et al., 1995;Shuvaev et al., 2011). These reports indicate that the regulation of PKC activity is important for cerebellar motor coordination.
Interestingly, a recent study previously reported that DGKg directly interacts with PKCg and is phosphorylated by PKCg , resulting in the upregulation of DGKg activity and the subsequent attenuation of PKC activity (Yamaguchi et al., 2006). These facts suggest that DGKg also has important roles in cerebellar motor coordination. However, its physiological role in cerebellar motor coordination has not been elucidated. Therefore, we investigated the function of DGKg in cerebellar motor coordination and its molecular mechanism using DGKg knock-out (KO) mice. We found that DGKg contributes to motor coordination, the induction of cerebellar LTD and the dendritic development of Purkinje cells through the regulation of PKCg .

Mice
Wild-type (WT) mice (C57BL/6N) were purchased from Japan SLC, Inc. DGKg KO (tm1a) mice and CAG-Flpo transgenic mice were kindly provided by Dr. Ikawa and Dr. Fujihara. Mice were housed under a 14/10 h light/dark cycle with ad libitum food and water. All procedures using mice were performed according to the guidelines of the Institute Animal Care and Use Committee. Male six-to 12-week-old mice were used for the motor coordination tests and the electrophysiological experiments.

Western blotting
Western blotting was performed as described previously (Kano et al., 2014). Briefly, the samples were subjected to 10% SDS-PAGE, followed by blotting onto a poly-vinylidene difluoride membrane (Millipore). Nonspecific binding sites were blocked by incubation with 5% skim milk in 0.01 M PBS containing 0.03% Triton X-100 (PBS-T) for 1 h. The membrane was incubated with the appropriate antibody for 1 h at room temperature. After washing with PBS-T, the membrane was incubated with peroxidaselabeled anti-rabbit or anti-mouse IgG for 30 min. After three rinses with PBS-T, the immunoreactive bands were visualized using ImmunoStar (Wako). The densities of the bands were analyzed by ImageJ. To detect phosphorylated proteins, we used 5% BSA instead of skim milk for blocking and 0.01 M TBS containing 0.03% Tween 20 (TBS-T) instead of PBS-T.

Immunohistochemistry
For immunohistochemical analysis of mouse brains, mice were perfused with 0.9% NaCl through the left ventricle followed by 4% paraformaldehyde (PFA). The brains were removed and immersed for 48 h in 4% PFA at 4°C. The brains were immersed in a 30% sucrose solution for at least 2 d at 4°C. The brains were sliced with a cryostat (Leica CM1850) to produce 20-mm-thick coronal and parasagittal sections. Immunohistochemistry was performed as described previously (Ueda et al., 2013). The sections were incubated in 0.03% H 2 O 2 in PBS (-) for 1 h and permeabilized with 0.3% Triton X-100 in PBS (-) for 30 min. Subsequently, the sections were washed with 0.1% Tween 20 in PBS (-) and incubated in 0.1% Tween 20 in PBS (-) supplemented with goat serum as a blocking reagent for 1 h. Next, the sections were incubated with an appropriate antibody for 72 h at 4°C. The bound antibodies were visualized using EnVision 1 System-HRP (Agilent Technology).

Rotarod test
The rotarod apparatus (MK-630B single lane rotarod, Muromachi Kikai) consisted of a rod (30 mm in diameter, 90 mm wide) flanked by two large round plates (40 cm in diameter). The speed of rotation was increased from four to 40 rotations per minute over 5 min. We recorded the latency for the mice to fall from the rod. The test was performed three times daily for 2 d.

Footprint test
The hind paws of the mice were coated with nontoxic ink, and the mice were allowed to walk through a tunnel (30 cm long, 16 cm wide, and 17 cm high). The footprint patterns were analyzed for stride length. The test was performed three times.

Beam test
Mice were trained to traverse an elevated metallic beam (70 cm long, 10 mm in diameter, 60 cm high). They were placed at one end of the beam and an enclosed escape box was placed at the other end. Each hind paw slip was recorded and counted. The test was performed five times daily for 2 d.

Electrophysiology
Mice were sacrificed by cervical dislocation under anesthesia with diethyl ether. The cerebellum was excised, and parasagittal cerebellar slices (250 mm thick) were prepared from the vermis (Edwards et al., 1989;Kakizawa et al., 2000Kakizawa et al., , 2012. Whole-cell recordings were obtained from visually identified Purkinje cells under an upright microscope (BX51WI, Olympus) using a 40Â water-immersion objective at room temperature (23-25°C). The resistance of the patch pipettes was 2.0-3.5 MV when filled with an intracellular solution composed of the following: 130 mM K-gluconate, 10 mM KCl, 10 mM NaCl, 1 mM EGTA, 4 mM ATP-Mg, 0.4 mM GTP-Na, and 10 mM HEPES (pH 7.3; adjusted with KOH). The standard bathing solution was composed of the following: 125 mM NaCl, 2.5 mM KCl, 2 mM CaCl 2 , 1 mM MgSO 4 , 1.25 mM NaH 2 PO 4 , 26 mM NaHCO 3 , and 20 mM glucose, bubbled with 95% O 2 and 5% CO 2 . Bicuculline (10 mM) was always added to block the IPSCs. Scu (100 mM) was added to the extracellular solution for 1 h prior to the stimulation. For focal stimulation of parallel fibers (PFs), a stimulation pipette (with a tip 5-10 mm in diameter) was filled with standard bathing solution and used to apply square pulses (0.1 ms in duration, 0-20 V in amplitude) to the middle portion of the molecular layer. The intensity of each stimulus was adjusted to evoke PF-excitatory postsynaptic currents (EPSCs) with amplitudes of 60-120 pA. The ionic current was recorded from Purkinje cells using a patchclamp amplifier (EPC-9, HEKA) at a holding potential of À90 or À80 mV, after compensating for liquid junction potential. The signals were filtered at 2 kHz and digitized at 20 kHz.
On-line data acquisition and off-line data analyses were performed using PULSE (HEKA) software. For the LTD experiments, the intensity of the stimulus was adjusted to evoke PF-EPSCs whose initial amplitudes were between 60 and 120 pA. After obtaining a stable initial recording for at least 10 min, conjunctive stimulus (CJS) was applied to induce LTD. The CJS protocol is composed of 300 single PF stimuli in conjunction with depolarizing pulse (À80 to 0 mV for 50 ms) in conjunction with single PF stimuli (at 40 ms after the onset of the pulse) at 1 Hz. The test stimulus was applied to PFs every 10 s. The amplitude of PF-EPSCs was averaged every 60 s, and normalized to the mean value observed 10 min before the LTD-inducing stimulus. A 100 ms, À5 mV hyperpolarizing test pulse preceded each PF stimulus to monitor the series resistance and input resistance of Purkinje cells throughout the experiment. Data were discarded if the resistance changed by .10% (Kakizawa et al., 2012). Data were also discarded when the slope of PF-EPSC amplitude averaged every minute during the initial recording for 10 min was larger than 2% or when the amplitude did not become stable within 20 min after the onset of whole-cell configuration.
For the analysis of membrane capacitance of Purkinje cells, hyperpolarizing current pulse (À80 to À90 mV, for 500 ms) was applied, and the resulting capacitive current was recorded. Membrane capacitances were then calculated according to the model described previously (Llano et al., 1991), which distinguishes two regions of Purkinje cells: region 1 representing the soma and the main proximal dendrites, and region 2 representing the dendritic tree. C1 represents the membrane capacitance of region 1, and C2 represent the capacitance of the dendritic tree of Purkinje cells.
To examine number of CFs innervating the recorded Purkinje cell, we systematically moved the stimulation pipette in the granule cell layer and tried stimulation at 10 different sites. The stimulus intensity was gradually increased at each stimulation site. When a CF was stimulated, an EPSC was elicited in an all-or-none fashion. In some Purkinje cells, more than one discrete CF-EPSC could be elicited when the stimulus intensity was increased or when the stimulating electrode was moved to a different site. The number of CFs innervating the Purkinje cell was estimated by counting the number of discrete CF-EPSC steps elicited in that cell Kakizawa et al., 2000Kakizawa et al., , 2003.

Primary culture of mouse Purkinje cells
Dissociated cerebellar cultures were prepared as described previously (Seki et al., 2009). EXVIII embryos were removed from pregnant WT and DGKg KO mice anesthetized with Ravonal. The cerebellum was dissected and kept in ice-cold HBSS (Wako). Cerebellar cells were dissociated by using dissociation solutions from the Sumitomo Nerve-Cell Culture System according to the manufacturer's instructions. Dissociated cerebellar cells were suspended in neuronal culture medium from the Sumitomo Nerve-Cell Culture System, and plated at a concentration of 1 Â 10 5 cells/50 ml in the center of a multi-well glass bottom dish (Matsunami Glass) coated with polyethylenimine (Sigma-Aldrich), and cultured in a humidified atmosphere containing 5% CO 2 at 37°C. After the cells attached to the bottom (1-2 h later), 250 ml of culture medium was added to each well. The culture medium was supplemented with 100 pM T3. The cells were cultured for 21 d in vitro (DIV). Half of the medium was changed every 3-4 d. TPA (200 nM), GFX (200 nM), and Gö (1 mM) were added to the medium for 3 d.

Confocal microscopy
For immunofluorescent staining, primary cultured Purkinje cells at DIV21 were fixed with 4% PFA for 1 h at 4°C. After fixation, the cells were permeabilized with 0.3% Triton X-100 for 30 min and then blocked with 10% normal goat serum (NGS) for 2 h. The cells were incubated with anti-calbindin antibody for 16 h at 4°C and then visualized with Alexa 546-labeled goat anti-rabbit IgG, followed by observation under confocal microscopy.
The fluorescence of Alexa 546 was observed with a confocal laser scanning fluorescence microscope (FV-500 IX81 Olympus). The images were recorded as TIFF files. To count the number of neurites and branches and their lengths, the images were analyzed with Neurolucida and Neurolucida Explorer software (MBF Bioscience).

Measurement of PA in the crude synaptosomal membrane fraction
Crude synaptosomes were prepared as described previously (Gu et al., 2009). Briefly, the cerebellum was homogenized in sucrose buffer: 320 mM sucrose, 20 mM Tris-HCl, 1 mM EGTA, 1 mM EDTA, 1 mM MgCl 2 , 1 mM PMSF, 20 ng/ml leupeptin, 1 mM NaF, 1 mM Na 3 VO 4 and 1Â phosphatase inhibitor cocktail solution II; pH 7.4) with a Potter-Elvehjem homogenizer. After centrifugation at 800 Â g for 5 min, the supernatant was centrifuged at 10,000 Â g for 10 min. The pellet (crude synaptosome fraction) was homogenized in sucrose buffer containing 1% Triton X-100 and 300 mM NaCl and centrifuged at 16,000 Â g for 30 min. The supernatant (S2) and pellet (P2) were obtained, and the P2 fraction was dissolved in 1% SDS. PA was measured as described previously (Morita et al., 2009). Equal amounts of crude synaptosomal membrane fraction (P2) and 1 M NaCl were mixed, and 20 ml of 1% perchloric acid (PCA), 450 ml of chloroform/methanol (1 : 2), 150 ml of chloroform and 150 ml of 1% PCA were added to the samples in order. After centrifugation at 10,000 rpm for 2 min, the lower layer was harvested and dried. The extracts were dissolved in 70 ml of 1% Triton X-100. Then, 30 ml of the solution and 120 ml of reagent buffer A1 (50 mM Tris-HCl, 50 mM NaCl, and 10 kU/ml LPL; pH 7.4) were mixed and reacted for 1 h at 37°C. After incubation for 3 min at 96°C, the mixture was centrifuged at 10,000 rpm for 5 min at room temperature. A total of 50 ml of the supernatant and 50 ml of reagent buffer A2 (50 mM Tris-HCl, 50 mM NaCl, 0.2% Triton X-100, 5 U/ml HRP, and 0.3 mM Amplex Red) were mixed and reacted for 30 min at room temperature. Then, the amount of PA was measured using a Wallac 1420 ARVOsx (PerkinElmer).

Experimental design and statistical analysis
All animal data were analyzed for male mice. All data are shown as the mean 6 SEM, and Student's t test, Dunnett's test, and Tukey's multiple comparisons test were used as appropriate to test statistical significance. Data were analyzed using Excel (Microsoft) and R version 3.5.1 (The R Foundation for Statistical Computing). The statistical analysis of multiple CFs was performed using the Kolmogorov-Smirnov test (Igor Pro, Hulkins). Differences were considered significant when p , 0.05.

Generation of DGKc KO mice
We produced homozygous tm1a mice (tm1a/tm1a) by mating tm1a heterozygous mice containing a promotordriven cassette (Fig. 1A). The genotype of the tm1a/tm1a mice was confirmed by PCR (Fig. 1C). However, we could not confirm that DGKg was not expressed in the brains of tm1a/tm1a mice, because the insertion occurred between exons 3 and four of the gene that encodes DGKg . Therefore, we examined DGKg expression in the brain by Western blotting and immunohistochemistry using an anti-DGKg antibody. Western blotting revealed that there was no expression of DGKg in the cerebrum or cerebellum of tm1a/tm1a mice (Fig. 1C). DGKg immunoreactivity was clearly observed in the hippocampus and cerebellar Purkinje cells and weakly and diffusely expressed in the granule cells of WT mice, but not in those of tm1a/tm1a mice (Fig. 1D,E). These results indicate that the DGKg protein was not expressed in the brains of tm1a/tm1a mice; thus, we concluded that these mice were DGKg KO mice.

Motor coordination of DGKc KO mice
DGKg is abundantly expressed in cerebellar Purkinje cells which are important for motor coordination (Goto and Kondo, 1999;Adachi et al., 2005). Thus, we tested the motor coordination of DGKg KO mice by the rotarod, footprint and beam tests. In the rotarod test, WT and DGKg KO mice showed steady improvements over trials, but the latency for DGKg KO mice to fall from the rod was significantly shorter than that of WT mice ( Fig. 2A; *p , 0.05, **p , 0.01; Student's t test). In the footprint test, WT and DGKg KO mice seemed to walk normally (Fig. 2B). However, a precise analysis of the hind footprint pattern revealed that the stride length of DGKg KO mice was significantly (;15%) shorter than that of WT mice ( Fig. 2C; p = 0.023; Student's t test). In the beam test, DGKg KO mice exhibited significantly more frequent slips than WT mice ( Fig. 2D; p = 0.033, Student's t test). These results clearly indicate that DGKg KO mice show impairments in motor coordination.

LTD in DGKc KO mice
Since cerebellar LTD is a basis of motor coordination, we examined cerebellar LTD in DGKg KO mice. LTD was induced by CJS, which is paired depolarization and parallel fiber (PF) stimulation at 1 Hz for 300 s (Kakizawa et al., 2012), and the amplitude of EPSCs at PF-to-Purkinje cell synapses (PF synapses) after CJS was decreased in WT mice. In contrast, DGKg KO mice showed impaired LTD (Fig. 3A,B; p = 0.0003, Student's t test); the average EPSC amplitude of DGKg KO mice 21-30 min after CJS were equal to baseline levels. On the other hand, no significant abnormalities were observed in the electrophysiological properties of PF synapses; there was no significant difference in the input-output relationship, that is, the slope of the EPSC amplitude (output)-stimulus intensity (input) curve, between WT and DGKg KO mice ( Fig. 3C; p . 0.05, Student's t test). Additionally no significant difference was observed in the paired-pulse ratio (PPR; Fig.  3D; p . 0.05, Student's t test), an index of the change in transmitter release probability at presynaptic terminals (Zucker and Regehr, 2002). Taken together, the data indicate that, although the basic electrophysiological properties of PF synapses were normal, cerebellar LTD was impaired in DGKg KO mice.

CF innervation to Purkinje cells in DGKc KO mice
CF innervation of Purkinje cells is also responsible for motor coordination. Single Purkinje cells are originally innervated by multiple CFs; however, they are innervated by a single CF as they grow due to pruning (Watanabe and Kano, 2011). A deficiency in the mGluR1-PKCg signaling cascade results in multiple CF innervation in mature Purkinje cells and subsequent impaired motor coordination (Kano et al., , 1998Offermanns et al., 1997). Therefore, we investigated the number of CFs innervating Purkinje cells in DGKg KO mice. Multiple CF innervation was not observed in DGKg KO mice ( Fig.  3E; p = 0.473, Kolmogorov-Smirnov test). These results indicate that DGKg deficiency did not affect CF innervation in Purkinje cells.

Dendritic development of Purkinje cells in DGKc KO mice
Next, we investigated the dendritic morphology of Purkinje cells in DGKg KO mice, because the dendritic morphology of Purkinje cells is also important for motor coordination. Purkinje cells were visualized by filling them with fluorescent dye from the a recording pipette, and impairments in the dendritic development of Purkinje cells were revealed in DGKg KO mice (Fig. 4A); the number of branches and the total length of the dendrites in DGKg KO mice were significantly lower than those in WT mice, although there was no significant difference in the number of dendrites per cell (Fig. 4B). Sholl analysis demonstrated that there were fewer branches in distal dendrites at distances of 120, 140, and 150 mm from the soma in KO mice ( Fig. 4C; 120 mm: p = 0.045, 140 mm: p = 0.043, 150 mm: p = 0.006, Student's t test).
In some cases, a smaller dendritic region is reflected by decreased membrane capacitance (Kakizawa et al., 2003). Therefore, we subsequently examined the membrane capacitance of Purkinje cells. Although there was no significant difference in the membrane capacitance of the somata and proximal dendrites, the membrane capacitance of the distal dendritic region was significantly lower in DGKg KO mice than in WT mice ( Fig. 4D; soma 1 prox. dend.: p = 0.305, distal dendrite: p = 0.009, Student's t test). This result is consistent with the impaired morphology of Purkinje cells in DGKg KO mice. To confirm dendritic morphologic impairments in Purkinje cells from DGKg KO mice, we compared morphologic differences between primary cultured Purkinje cells from WT and DGKg KO mice. Primary cultured Purkinje cells prepared from EXVIII embryos were cultured for 21 DIV and stained with an anti-calbindin antibody (Fig. 5A). The number of branches and total length of dendrites of primary cultured Purkinje cells from DGKg KO mice were significantly decreased compared with those from WT mice (control (con) in Fig. 5B, center and right; WT con vs KO con, branches: p = 0.000013, length: p = 0.000056, Tukey's multiple comparison test). However, there was no difference in the number of dendrites per cell (con in Fig. 5B, left; p . 0.05, Tukey's multiple comparison test). These results confirm the impaired branching of Purkinje cells from DGKg KO mice.
To investigate the molecular mechanism underlying the morphologic impairment of primary cultured Purkinje cells from DGKg KO mice, we focused on PKC activity because PKC is involved in the dendritic morphology of Purkinje cells and its activation results in the retraction of dendrites and a reduction in the number of branches (Metzger and Kapfhammer, 2000); an increase in DG due to DGKg deficiency activates PKC in primary cultured Purkinje cells from DGKg KO mice. Therefore, we treated primary cultured Purkinje cells with a PKC activator TPA and the panPKC inhibitor GF109203X (GFX) for the final 3 d. These pharmacological treatments did not change the number of dendrites in Purkinje cells from either WT or DGKg KO mice (Fig. 5B, left; p . 0.05, Tukey's multiple comparison test). In contrast, TPA treatment reduced the number of branches and total length of dendrites of Purkinje cells from WT mice but not of Purkinje cells from DGKg KO mice (TPA in Fig. 5B, center and right; WT con vs WT TPA, branches: p = 0.0026, length: p = 0.0013; KO con vs KO TPA, branches: p = 1.00; length: p = 0.98, Tukey's multiple comparison test). On the other hand, the decreased number of branches and total length of dendrites of Purkinje cells from DGKg KO mice were rescued to WT levels by treatment by the PKC inhibitor GFX (GFX in Fig. 5B, center and right; KO con vs KO GFX, branches: p = 0.00,041, length: p = 0.0066, Tukey's multiple comparison test). These results indicate that PKC activation is key for the morphologic impairment of primary cultured Purkinje cells from KO mice. To further investigate the PKC subtypes involved in these phenomena, we used the cPKC inhibitor Gö6976. Similar to GFX treatment, Gö treatment rescued the impairments in Purkinje cells from DGKg KO mice (Gö in Fig. 5B, center and right; KO con vs KO Gö, branches: p = 0.042, length: p = 0.043, Tukey's multiple comparison test). These results suggest that DGKg contributes to the dendritic development of Purkinje cells via the regulation of cPKC activity.  's t test). Furthermore, we investigated whether PKC is actually activated in the cerebellum of DGKg KO mice by Western blotting using an anti-PKC substrate antibody that recognizes PKC phosphorylation sites (Fig. 6C). Quantification of the band intensity revealed that the phosphorylation of total PKC substrates was elevated in the cerebellum of DGKg KO mice ( Fig. 6C; p = 0.021, Student's t test), indicating that PKC was indeed activated in DGKg KO mice.
The upregulation of PKC activity is results from an increase in the DG level induced by DGKg deficiency. In other words, the PA level is reduced in DGKg KO mice. Therefore, we measured the level of PA in the P2 (crude synaptosomal membrane) fraction of DGKg KO mice. The level of PA in the P2 fraction of DGKg KO mice was significantly decreased compared with that of WT mice ( Fig.  6D; p = 0.033, Student's t test). These results suggest that activated PKCg causes abnormalities in the dendritic development of Purkinje cells and LTD, resulting in the impairment of motor coordination.

Effect of a PKCc inhibitor on impaired LTD in DGKc KO mice
It is known that PKCa but not PKCg is important for LTD. However, we found that there was no change in the activation of PKCa and that PKCg was activated in the cerebellum of DGKg KO mice in the basal state. To investigate the involvement of abnormal PKCg activation in impaired LTD in DGKg KO mice, we used a PKCg inhibitor. Scutellarin (Scu) is a flavonoid that is a major active ingredient of Erigeron breviscapus Hand. Mazz., a plant used in Chinese herbal medicine. A recent study reported that Scu inhibits the translocation of PKCg from the cytoplasm to the plasma membrane which is a hallmark of PKC activation, but has no effect on the translocation of PKCa (Xu et al., 2007;Su et al., 2012), suggesting that Scu specifically inhibits PKCg activity but not PKCa activity. Therefore, we tried to confirm its inhibitory effect on the autophosphorylation of PKCg and PKCa. Stimulating COS-7 cells overexpressing PKCg -GFP or PKCa-GFP with TPA induced the autophosphorylation of both PKCg and PKCa ( Fig. 7A; PKCg : p = 0.047 PKCa: p = 0.0021, Dunnett's test). However, pretreating the cells with Scu significantly inhibited the autophosphorylation of PKCg but not the autophosphorylation of PKCa in a dose-dependent manner, but not the autophosphorylation of PKCa (( Fig. 7A; PKCg , 20 mM: p = 0.99, 50 mM: p = 0.21, 100 mM: p = 0.47; PKCa, 20 mM: p = 0.046, 50 mM: p = 0.027, 100 mM: p = 0.0012, Dunnett's test). Then, we also investigated the inhibitory effect of Scu on PKCg activity in cerebellar slices from DGKg KO mice. Scu (100 mM) normalized the upregulated PKCg activity seen in DGKg KO mice to WT levels, but it had no significant effect on PKCa activity ( Fig. 7B; PKCg , KO: p = 0.0059, 20 mM: p = 0.0025, Dunnett's test). These results suggest that Scu can be used as a PKCg inhibitor. Finally, to investigate the effect of a PKCg inhibitor on impaired LTD in DGKg KO mice, we added Scu to the extracellular solution when we measured LTD. As expected, Scu rescued the impairment of LTD in DGKg KO mice (Fig. 7C,D; WT-KO: p = 0.00,002; KO-KO1Scu: p = 0.00,017, Tukey's multiple comparison test). These results indicate that the impaired LTD in DGKg KO mice is dependent on PKCg activation in the basal state.

Rescue effect of DGKc on the impairment of motor coordination
Finally, we performed a rescue experiment using the FLP-FRT recombination system. As shown in Figure 8A, the cassette was inserted into DGKg KO mice (tm1a/ tm1a), and we eliminated the cassette by mating DGKg KO mice with FLP mice. Recombination was confirmed by PCR (Fig. 8B), and the recovery of DGKg in the cerebrum and cerebellum of tm1c/tm1c (1c) mice was confirmed by Western blotting (Fig. 8C). The rotarod and beam tests showed that the motor coordination of the 1c mice was normal (Fig. 8D,E; pp , 0.05, ppp , 0.01; Tukey's multiple comparison test), suggesting the importance of DGKg for motor coordination.

Discussion
DGKg and PKCg are abundantly expressed in cerebellar Purkinje cells, and the functional correlation between these kinases is important for cerebellar motor coordination and the dendritic development of Purkinje cells. In this study, we investigated the function of DGKg in cerebellar motor coordination and related molecular mechanism using newly developed DGKg KO mice. Cerebellar motor coordination was impaired in DGKg KO mice most likely due to the loss of LTD and the impairment in the dendritic development of Purkinje cells. Our results suggest that DGKg contributes to cerebellar motor coordination through the regulation of LTD and the dendritic development of Purkinje cells through the regulation of PKCg activity.
To clarify the mechanism underlying the impairment of LTD in DGKg KO mice, we investigated the activity of PKCa because it is responsible for inducing LTD through the phosphorylation of GluR2 S880 (Leitges et al., 2004). However, PKCa activation in the cerebellum was not detected. We also checked the phosphorylation of the S880 residue of GluR2, and found that the phosphorylation of GluR2 in the cerebellum of DGKg KO mice was not changed compared with that in the cerebellum of WT mice (data not shown), confirming that PKCa was not upregulated in the cerebellum in the basal state. On the other hand, PKCg was activated in the cerebellum of DGKg KO mice. Why only PKCg was upregulated when both PKCg and PKCa can be activated by DG and Ca 21 . This may have been because PKCg but not PKCa can be activated by DG at a basal intracellular Ca 21 concentration (Kohout et al., 2002;Ananthanarayanan et al., 2003). Another possibility is the mutual regulation of DGKg and PKCg ; DGKg and PKCg directly interact, and the phosphorylation of DGKg by PKCg upregulates DGKg activity (Yamaguchi et al., 2006). In addition, DGKg accelerates the retranslocation of PKCg from the membrane to the cytoplasm (Shirai et al., 2000), indicating that DGKg and PKCg mutually regulate each other's activity and that DGKg negatively regulates PKCg activity under basal conditions. There have been several reports on the mutual regulation of other DGKs and PKCs (Topham et al., 1998;Imai et al., 2004;Van Baal et al., 2005) and the importance of these regulator mechanisms for synaptic plasticity (Lee et al., 2016). For example, DGKz directly interacts with PKCa and lowers PKCa activity by reducing DG levels in the basal state (Luo et al., 2003). An increase in the DG levels activates PKCa, which phosphorylates DGKz , resulting in the dissociation of PKCa and DGKz . The spatial regulation of PKCa activity by DGKz contributes to cerebellar LTD (Lee et al., 2015). Therefore, PKCa in the cerebellum of DGKg KO mice may be inhibited by its interaction with DGKz . Furthermore, PKCs have different sensitivities and dependencies on DG species (Kamiya et al., 2016), and each DGK utilizes different DG species (Sakane et al., 2018). Taken together, DGKg deficiency may upregulate PKCg but not PKCa. However, the upregulation of both PKCa and PKCg is likely induced by an increase of DG levels through DGKg deficiency. Indeed, the PKCg inhibitor Scu rescued impaired cerebellar LTD in DGKg KO mice. These results suggest that PKCg upregulation in the basal state is involved in the impairment of LTD in DGKg KO mice and that DGKg is responsible for lowering PKCg activity at LTD stimulation.
In this case, a new question of how upregulated PKCg impairs cerebellar LTD has arisen. Transient receptor potential canonical channel 3 (TRPC3) may be a key molecule. TRPC3 is abundantly expressed in Purkinje cells, and PKCg negatively regulates TRPC3 activity and extracellular Ca 21 influx (Adachi et al., 2008). TRPC3 KO mice show dysfunction of motor coordination, and the inhibition of TRPC3 impairs cerebellar LTD (Hartmann et al., 2008;Kim, 2013). PKCg negatively regulates TRPC3 activity and extracellular Ca 21 influx through TRPC3 (Adachi et al., 2008). Based on these facts, we speculate that the inactivation of TRPC3 by upregulated PKCg may inhibit Ca 21 influx, resulting in PKCa inactivation during LTD stimulation. LTD stimulation increases the DG levels and C, Phosphorylation of the PKC substrate was normalized to the PKCg expression level. The ratio of the phosphorylation of PKCg , PKCa, and PKC substrate to the expression levels of PKCg and PKCa to WT was plotted (PKCg : n = 3; PKCa: n = 3; PKC substrate: n = 4); pp , 0.05, followed by Student's t test. D, The amount of PA in the P2 (crude synaptosomal membrane) fraction of WT and DGKg KO mice was measured by an enzymatic method (n = 4); pp , 0.05, followed by Student's t test. Data are expressed as mean 6 SEM. the intracellular Ca 21 concentration through the mGluR1 cascade and depolarization (Ito, 2002); however, the latter but not the former controls PKCa activation during the induction of cerebellar LTD (Tsuruno and Hirano, 2007). In other words, the Ca 21 influx into Purkinje cells in DGKg KO mice is not enough to activate PKCa and release PKCa and DGKz . This speculation does not deny the involvement of PKCa in cerebellar LTD. However, to confirm this hypothesis, additional experiments are needed.
Our results also showed that impairments in the dendritic development of Purkinje cells in DGKg KO mice and the cPKC inhibitor Gö6976 normalized the dendrite retraction of Purkinje cells in DGKg KO mice to WT levels, suggesting that cPKC activity mainly functions in the dendritic development in Purkinje cells. In particular, the activity of PKCg is likely involved in the impairment of Purkinje cell branching in DGKg KO mice because the activation of PKCg was confirmed. Indeed, the dendrites of Purkinje cells in PKCg KO mice are enlarged, whereas those in PKCa KO mice are not changed compared with those in WT mice (Schrenk et al., 2002;Gundlfinger et al., 2003). Furthermore, DGKg activity was necessary for the development of neurites in Neuro-2A cells (data not shown). Taken together, these results suggest that DGKg contributes to the dendritic development of Purkinje cells through the regulation of PKCg activity. In addition, Figure 7. Rescue of impaired LTD in DGKg KO mice by a PKCg inhibitor. A, COS-7 cells overexpressing PKCg -GFP or PKCa-GFP were cultured for 48 h. The cells were treated with TPA (100 nM) for 30 min after preincubation with or without scutellarin (Scu; 20, 50, or 100 mM) for 30 min. The lysates from the cells were subjected to Western blotting and probed with anti-phospho-PKCg , anti-PKCg , anti-phospho-PKCa, and anti-PKCa antibodies. Quantification of the autophosphorylation of PKCg and PKCa was performed by ImageJ. The phosphorylation levels of PKCg and PKCa were normalized to the PKCg and PKCa expression levels. The ratio of the phosphorylation of PKCg and PKCa to that in the control was plotted (PKCg -GFP: n = 6; PKCa-GFP: n = 6); pp , 0.05, ppp , 0.01 (vs control), followed by Dunnett's test. B, Acute cerebellar slices from WT and DGKg KO mice were incubated with or without Scu (20 or 100 mM) for 1 h. Lysates from the slices were subjected to Western blotting as described in A (PKCg : n = 3; PKCa: n = 3); pp , 0.05, ppp , 0.01 (vs WT), followed by Dunnett's test. C, Changes in the PF-EPSC amplitude before and after the CJS of PF (1 Hz, 300 s) with depolarization applied at time 0. A PKCg inhibitor Scu (100 mM) was added to the extracellular solution prior to stimulation. The PF-EPSC amplitude was normalized to the mean over 10 min before CJS (WT: n = 8; KO: n = 8; KO1Scu: n = 7). Sample traces immediately before (1) and 30 min after (2) CJS. D, Average PF-EPSC amplitude over the 21-to 30-min period after CJS; ppp , 0.01, followed by Tukey's multiple comparisons test. Data are expressed as mean 6 SEM.
PKCg regulates spinogenesis and the morphology of the spines of distal dendrites through the phosphorylation of CaMKIIb (Sugawara et al., 2017). Purkinje cells in DGKg KO mice showed retracted distal dendrites, but the basic electrophysiological properties of PFs were normal, indicating that DGKg KO mice retained basic excitatory circuitry. On the other hand, morphologic abnormalities in spines of Purkinje cells cause impairments in motor coordination (Sugawara et al., 2013). Accordingly, morphologic abnormalities in spines of Purkinje cells may also contribute to cerebellar motor dysfunction in DGKg KO mice, although further study is necessary.
Cerebellar motor coordination was impaired not only by cerebellar LTD and alterations in the dendritic morphology of Purkinje cells but also by the disruption of refined motor program transmission through multiple CF innervation. A single Purkinje cell originally exhibits multiple CF innervation; however, it exhibits mono CF innervation as it grows due to pruning (Watanabe and Kano, 2011). CFs transmit signals to Purkinje cells to refine the motor program to prevent incorrect motor programming, resulting in the suppression of PF-EPSCs involved in incorrect motor programming and the induction of cerebellar LTD (Welsh et al., 1995). Previous reports have indicated the importance of PKCg in CF innervation; multiple CF innervation has been detected in mature PKCg KO mice , and a constitutively active PKCg mutant (S119P) also causes multiple CF innervation (Shuvaev et al., 2011). However, we demonstrated that the CF innervation of Purkinje cells in DGKg KO mice was normal, although PKCg activation was detected in the cerebellum. This is not surprising because the loss of function of PKCg seems to be critical for CF innervation. Indeed, the membrane residence time of the PKCg mutant (S119P) is shorter due to a lack of binding with DG induced by the mutation in the C1 domain; that is, the PKCg mutant does not exhibit activity in the membrane (Adachi et al., 2008). Alternatively, PKCg activity in the CFs of DGKg KO mice may be normal because CFs do not express DGKg , even in WT mice.
Thus far, we have focused on PKC to understand the molecular mechanism underlying the impairment of Figure 8. Rescue of motor coordination in DGKg floxed (tm1c) mice. A, WT, tm1a, and tm1c alleles. B, Typical results of PCR genotyping. Bands at 975, 907, and 523 bp were expected for the tm1c, WT, and tm1a alleles, respectively. C, Cerebral and cerebellar lysates from WT, tm1a (1a), and tm1c (1c) mice were subjected to Western blotting and probed with anti-DGKg and anti-GAPDH antibodies. D, Motor coordination of WT, tm1a (1a), and tm1c (1c) mice was assessed by the accelerating rotarod test (WT: n = 14; 1a: n = 15; 1c: n = 15); pp , 0.05, ppp , 0.01, followed by Tukey's multiple comparisons test. E, Motor coordination of WT, tm1a (1a), and tm1c (1c) mice was assessed by the number of hind paw slips in the beam test (WT: n = 5; 1a: n = 5; 1c: n = 3); pp , 0.05, followed by Tukey's multiple comparisons test. Data are expressed as mean 6 SEM. cerebellar LTD and dendritic development of Purkinje cells in DGKg KO mice; these processes are activated by DG but we also studied PA. Our results showed that PA was decreased in the cerebellum of DGKg KO mice, suggesting that aberrant PA-mTOR signaling has some influence on impaired motor coordination in DGKg KO mice. PA, which is generated by DGK, activates mTOR (Avila-Flores et al., 2005), which is involved in neurodevelopment and neuropsychiatric disorders (Costa-Mattioli and Monteggia, 2013;Bockaert and Marin, 2015). mTOR forms two distinct complexes, namely, mTORC1 and mTORC2, which are distinguished by their specific components and sensitivity to rapamycin (Laplante and Sabatini, 2012). Recent studies have revealed that mTORC1 and mTORC2 share a few functions but exhibit many distinct functions in Purkinje cells (Thomanetz et al., 2013;Angliker et al., 2015). A deficiency in mTORC2 results in multiple CF innervation but does not affect cerebellar LTD. A deficiency in mTORC1 results in a progressive loss of Purkinje cells, while its involvement in cerebellar LTD remains unknown. In other words, the dendritic atrophy of Purkinje cells induced by a deficiency in mTORC1 or mTORC2 is similar to that observed in DGKg KO mice, although DGKg KO mice do not exhibit the multiple CF innervation seen in mTORC2 KO mice and the loss of Purkinje cells seen in mTORC1 KO mice. Therefore, it is possible that PA-mTOR signaling also interferes with the phenotype of DGKg KO mice.
In summary (Fig. 9), the increase in DG levels caused by DGKg deficiency aberrantly activates PKCg Activated PKCg impairs LTD and the dendritic development of Purkinje cells, resulting in motor dyscoordination in DGKg KO mice. We showed for the first time the importance of DGKg in cerebellar LTD and the dendritic development of Purkinje cells through the regulation of PKCg activity and its contribution to cerebellar motor coordination.