Deletion of the α subunit of the heterotrimeric Go protein impairs cerebellar cortical development in mice.

Go is a member of the pertussis toxin-sensitive Gi/o family. Despite its abundance in the central nervous system, the precise role of Go remains largely unknown compared to other G proteins. In the present study, we explored the functions of Go in the developing cerebellar cortex by deleting its gene, Gnao. We performed a histological analysis with cerebellar sections of adult mice by cresyl violet- and immunostaining. Global deletion of Gnao induced cerebellar hypoplasia, reduced arborization of Purkinje cell dendrites, and atrophied Purkinje cell dendritic spines and the terminal boutons of climbing fibers from the inferior olivary nucleus. These results indicate that Go-mediated signaling pathway regulates maturation of presynaptic parallel fibers from granule cells and climbing fibers during the cerebellar cortical development.

Introduction

When G-protein coupled receptors (GPCRs) bind their cognate ligands, their respective heterotrimeric GTP binding proteins (G-proteins) are activated, inducing the dissociation of Gα from Gβγ. Specificity of GPCR signal transduction is determined by the functions of the specific Gα subunits because all G-proteins share a pool of Gβγ subunits. Gα subunits are classified into four subfamilies: Gαs, Gαi/o, Gαq/11, and Gα12/13. So far, 16 different mammalian Gα subunits have been identified [1, 2]. Go belongs to the pertussis toxin-sensitive Gi/o family, the α subunits of which share 71% amino acid sequence homology. Gαi proteins are ubiquitously expressed and well-studied. They inhibit adenylyl cyclase (AC) directly and consequently reduce intracellular cyclic AMP (cAMP) levels [3]. Gnai (the gene encoding Gαi protein) knockout (Gnai) mice show several abnormalities, especially in their immune systems and blood clotting [4, 5]. In comparison, Gαo expression is restricted to the central nervous system (CNS), particularly in synapses, and to endocrine cells and cardiac myocytes [3, 6, 7]. Accordingly, Gnao (the gene encoding Gαo protein) knockout (Gnao) mice show severe neurological deficits including seizures, hyperactivity, and abnormal sexual behavior accompanied by early death [8, 9]. Mutations of Gnao gene in human patients induce early infantile epileptic encephalopathy (EIEE), which is the basis for the Gnao to be called EIEE17 [10, 11]. These findings suggest Gαo plays an important role in the CNS. Still, compared to Gi, the role of Go in the CNS is largely unknown. This is likely due to the fact that no enzymatic functions have yet been associated with Gαo and to the fact that the scarcity of Gnao mice due to their low birth and survival rates make Gαo difficult to study.

Gαo is abundantly found in the neurite tips and growth cones of in vitro cultured neurons where it stimulates neurite formation [7, 12, 13]. In Gnao mice, the olfactory nerve layer comprised of projecting axons of olfactory sensory neurons is atrophied, which may be associated with defective Gαo signals in neurite outgrowth [9]. Gαo is highly expressed in the cerebellar cortex [14], however, the function of Go in the cerebellum is mostly unknown except for the interaction with the Purkinje cell protein 2, Pcp2 (L7) [15] and Gi/o-coupled cannabinoid receptor 1 (CB1) [16]. Pcp2(L7) mice are almost normal, showing only mild hypoplasia accompanied by normal motor learning and enhanced motor function [17]. WIN55212–2, a CB1 receptor agonist predominantly activates Gαo among various G-proteins in the rat cerebellum [16].

The cerebellum is a well-defined neural system that can help clarify the correlation between the anatomical architecture of a neural network and its function [18]. The cerebellar cortex consists of three distinct layers: the molecular layer (ML), the Purkinje cell layer (PCL), and the granule cell layer (GCL). Purkinje cells (PCs) arranged in a single layer of the PCL, each sending a single, long axon to deep cerebellar nuclei (DCN) and are the only output neurons of the cerebellar cortex. Their complex dendrites extend through the ML where they receive presynaptic inputs from climbing fibers (CFs) projecting from the inferior olivary nucleus (ION) and parallel fibers (PFs) originating in granule cells (GCs). The PC dendrites are segregated into proximal and distal territories during the postnatal period [19]. Immature proximal PC dendrites are first innervated by multiple CFs during the initial 3 weeks of the postnatal period. During the subsequent 3 weeks, a single dominant CF moves toward the pial surface while the other redundant CFs are pruned [20]. Simultaneously, the immature PCs undergo a transformation from multi- to mono-planarity as they acquire the fan-like shape that is typical for mature PCs [21]. These morphogenetic processes are controlled cell-autonomously by intrinsic PC factors in the early phases of postnatal development and by extrinsic synaptic connections with ION-CFs in the later phases [22].

Various types of Gα proteins are expressed in PCs such as Gαq, Gαo, Gαi2, and Gαz. In PC dendrites, Gαz is constantly increased over development, whereas Gαq and Gαo show different temporal peaks of gene expression patterns during development stages [23], which suggested that the specificity of Gαq and Gαo in the developmental changes of PCs. During the cerebellar development, Gαq-mediated signaling cascades [metabotropic glutamate receptor type 1 (mGluR1)-Gαq-phospholipase Cβ4 (PLCβ4)-protein kinase Cγ (PKCγ)] are well-studied. Deletion of each component evokes the similar phenotypes including multiple CF innervation, motor discoordination, and ataxia [24-26].

In this study, we investigate possible roles of Gαo in cerebellar cortical development using Gnao mice. These results showed that the defective Gαo signals prevent the proper development of presynaptic PFs and CFs as well as maturation of postsynaptic PC dendrites. Our results indicate that Gαo is essential for full PC differentiation and the refinement of presynaptic CFs during the postnatal development.

Methods

Mouse genetics

Gnao mice were generated by breeding Gnao+/− mice. Hemizygotes were created via the insertion of a neomycin selection cassette at Gnao exon 6 [8]. Tail-genomic DNA was used for genotyping by polymerase chain reaction (PCR) using the following primers that allowed us to verify the disruption of the Gnao gene: int6R1 (5′- ACC TGG CCT CCC TTG GGA ATA CAG − 3′) and ex6F1 (5′- CAG CGA TCT GAA CGC AAG AAG TGG − 3′) for wild type, and int6R1 and pol2R1 (5′- TGT GCT CTA GTA GCT TTA CGG AGC − 3′) for mutant. The number of animals used for each quantitative analysis is described in each figure legend. All experimental procedures were reviewed and approved by the Institutional Animal Research Ethics Committee at Ajou University Medical Center (Suwon, South Korea). All mutant mice were compared to the wild type littermate control mice.

Cerebellar neuron culture

PCs were cultivated according to previous reports [27, 28] with slight modifications. Briefly, the extracted cerebella from C57BL/6 N at postnatal day (P) 0 were minced with a surgical knife, washed with Hank’s balanced salt solution (HBSS; Sigma-Aldrich, #H2387) containing gentamicin (50 μg/ml, Sigma-Aldrich, #G1914), and incubated in Accumax solution (Sigma-Aldrich, #A7089) for 15 min at 37 °C in a 5% CO2 incubator. After washing with HBSS, the tissues were triturated by repetitive pipetting and the debris was removed by passing the suspension through a nylon mesh. After centrifugation of the cell suspension, the cell pellet was resuspended in plating medium containing 10% (vol/vol) fetal bovine serum (FBS; Hyclone, #SH30084.03) and 50 μg/ml gentamicin in Dulbecco’s modified Eagle medium/F12 (DMEM/F12; Gibco, #11330–032) and then plated onto coverslips coated with poly-D-lysine (PDL; 0.1 mg/ml, Sigma-Aldrich, #P6407) and laminin (10 μg/ml, Gibco, #23017–015). After 2 h, the cells were washed with HBSS and the medium was replaced with Neurobasal-A medium (Gibco, #10888–022) containing B27 (Gibco, #17504–044), Glutamax (Gibco, #35050–061), and Penicillin-Streptomycin (10,000 U/ml, Gibco, #151–40,122). To culture GCs, C57BL/6 N mice were sacrificed at P7 when external granule cells extensively proliferate [29, 30]. Cerebellar cells were isolated and purified as described [31].

In situ hybridization

The template DNA for in vitro transcription-Gnao exons 4 and 5 (nt 767–914, NM_010308.3)-was chosen for the preparation of RNA probes to avoid any regions of homology with Gnai2. The cDNA encompassing exons 4 and 5 (148 bp) was amplified with a forward primer (5′-CTT TGG GCG TGG AGT ATG GTG-3′) and a reverse primer (5′-CTC CTG GAT CCC CGA GTC GCC C-3′). The PCR product was then sub-cloned into the pGEM-T easy plasmid (Promega, #A1360). This plasmid was linearized with Nco I (Roche, #10835315001) or Sal I (Roche, #10348783001) for preparing the antisense- or sense-stranded RNAs, respectively. The antisense RNA was in vitro transcribed using SP6 RNA polymerase and a UTP-11-fluorescein isothiocyanate (FITC) labeling kit (Roche, #11685619910). Cryostat brain sections (40 μm-thick) were dried at room temperature (23 ± 2 °C) for 30 min and fixed in 10% neutral buffered formalin (NBF; BBC biochemical, #0151) for 10 min. After washing with diethylpyrocarbonate (Sigma-Aldrich, #D5758)-treated phosphate buffered saline (DEPC-PBS), the sections were acetylated with 0.1 M triethylamine (TEA, pH 8.0; Sigma-Aldrich, #T0886) and 2.5% acetic anhydride (Sigma-Aldrich, #A6404) in DEPC-PBS for 10 min at room temperature. The sections were then dehydrated and defatted with a 5-min chloroform (Sigma-Aldrich, #36919) treatment. Then, the sections were rinsed with ethanol, air dried, and stored at − 70 °C until used. The sections were pre-warmed at 55 °C for 15 min and pretreated at 37 °C for 2 h in hybridization solutions (Sigma-Aldrich, #H7782) supplemented with 0.2 g/ml dextran sulfate (Sigma-Aldrich, #D8906), 50% (vol/vol) formamide (Sigma-Aldrich, #F9037), 1 mg/ml Herring’s sperm DNA (Sigma-Aldrich, #D7290), and 1 mg/ml Ribonucleic acid from torula yeast (Sigma-Aldrich, #R6625) in 1.5× saline-sodium citrate (SSC; Invitrogen, #AM9763). Hybridization was performed in the presence of 0.1 μg/μl FITC-labeled RNA probe for 40 h at 37 °C. Then, unbound the probe was removed by sequential washing with 2× SSC for 5 min at room temp, 1× SSC for 1 min at room temp, 0.5× SSC containing 0.1% sodium dodecyl sulfate (SDS; Sigma-Aldrich, #L4390) for 20 min at 55 °C, 0.1× SSC containing 0.1% SDS for 20 min at 55 °C, 0.1× SSC containing 0.1% SDS for 20 min at room temp twice, and finally Tris-buffered saline (TBS, pH 7.5) without SDS for 5 min at room temperature 3 times. The sections were counterstained with bisbenzamide (Hoechst 33258; Invitrogen, #H3569) and mounted using Fluoromount-G® mounting solution (Southern Biotech, #0100–01). Images were acquired using an LSM710 confocal microscope (Carl Zeiss). Sense probes were generated with T7 RNA polymerase and used to verify Gnao-specific signals.

Immunofluorescence analysis

Mice were deeply anesthetized by i.p. injection of 2,2,2-tribromoethanol (0.02 ml/g, Sigma-Aldrich, #T48402) and perfused transcardially with 10% NBF. Extracted brains were incubated overnight in 10% NBF at 4 °C for post-fixation. For frozen sections, the brain was placed in 30% sucrose (Sigma-Aldrich, #S7903) in 0.1 M phosphate buffer (pH 7.4) at room temp for 48 h and embedded in O.C.T compound (Tissue-Tek, Sakura Finetek, #4583) to cut into 30 μm-thick slices. For paraffin sections, the brain was embedded in paraffin (Merck, #1.15161.2504) following standard procedures and cut into 7 μm-thick slices. To perform the staining, epitopes were unmasked by microwave (Daewoo Electronics, South Korea)-heating in 10 mM sodium citrate (pH 6.0; Sigma-Aldrich, #S4641) buffer including 0.05% (vol/vol) Tween 20 (Anatrace, #T1003) at 95 °C for 15 min. The samples were incubated with blocking solution [10% (vol/vol) normal goat serum (Gibco, # 16210–072), 1% bovine serum albumin (BSA; Sigma-Aldrich, #A2153) and 0.1% (vol/vol) Triton X-100 (Sigma-Aldrich, #T8787) in PBS (T-PBS)] for 1 h at room temp. The sections were incubated with primary antibodies overnight at 4 °C. The antibodies used were specific for Gαo (1:200, rabbit, Santa Cruz, #SC-387), Calbindin-D28K (Calb, 1:100, mouse, Sigma-Aldrich, #C9848 and 1:200, rabbit, Swant, #CB38a), Pcp2 (1:200, mouse, Santa Cruz, #SC-137064), vesicular glutamate transporter 1 (vGluT1, 1:100, rabbit, Invitrogen, #48–2400), vesicular glutamate transporter 2 (vGluT2, 1:100, mouse, Millipore, #MAB5504), and Tubulin β-III (Tubb3; 1:500, mouse, Biolegend, #801201). After washing with T-PBS, sections were incubated with secondary antibodies conjugated with Alexa 488 or 568 (Invitrogen) for 1 h at room temperature. The sections were counterstained with bisbenzamide and mounted as described above.

For immunocytochemistry, the coverslips with live cells attached were fixed with 10% NBF for 10 min at room temperature. After washing, the coverslips were incubated with blocking solution to block non-specific signals for 1 h at room temp and then incubated with primary antibodies overnight at 4 °C. The samples were incubated in the presence of secondary antibodies and mounted as described above. All fluorescence images were acquired using an LSM710 confocal microscope (Carl Zeiss) or a slide scanner Axio-Scan.Z1 slide scanner (Carl Zeiss). Isotype-specific IgG or normal serum were used instead of the primary antibody to validate the specificity of the immunoassays.

Western blot analysis

Approximately 100 mg of brain tissue was homogenized in 1 ml RIPA buffer (50 mM Tris-Cl, pH 8.0, 1% (vol/vol) IGEPAL® CA-630 (Sigma-Aldrich, #I8896), 0.1% SDS, 0.5% sodium deoxycholate (Sigma-Aldrich, #D6750), 150 mM Sodium chloride (Affymetrix/USB, #21618). The homogenate was centrifuged at 14,000 x g for 10 min at 4 °C and the supernatant was used for western blot analysis with anti- Gαo (1:1000, Santa Cruz), anti- Gαi1/2/3 (1:1000, Santa Cruz, #SC-26761), anti- Gαq (1:500, Santa Cruz, #SC-136181) and anti-Pcp2 (1:100, Santa Cruz, #SC-137064) as previously described [32]. The specific immunoreactivity was visualized using a secondary antibody conjugated with horseradish peroxidase (1:5000, Zymed) and an ECL kit (Pierce, #32106).

Cerebellar surface area measurement

The cerebella of mice older than P21 were considered mature. 1 mm-thick midsagittal mouse brain sections were obtained using a mouse brain matrix (ASI-Instruments, RBM-2000C) and used to make paraffin blocks. Paraffin-embedded cerebellar sections (7 μm-thick) were de-paraffinized via standard procedures and stained with cresyl violet (Sigma-Aldrich, #C5042) or nuclear fast red solution (Sigma-Aldrich, #N3020) for 1–5 min at room temperature. The sections were mounted using Cytoseal™ XYL (Thermo Scientific, #8312–4) and scanned with Aperio Scanscope XT scanner (Aperio Technologies). Cresyl violet-stained images of the whole cerebellar sections and the GCL were obtained, converted to black-and-white, and used to measure the occupancy of the GCL and ML via the ImageJ 1.50 program (NIH) [33]. The area covered by the ML was measured by subtracting the area covered by the GCL and white matter from the area of the whole cerebellum.

Measurement of PC dendrites and CF boutons

To measure the thickness of PC dendrites, paraffin sections (7 μm-thick) from Gnao mice were stained with an anti-Calbindin antibody. Fluorescent images were taken with an LSM710 confocal microscope using a 63X-oil immersion objective. 50 μm-long dendrites were selected from the confocal images. The thickness of each PC dendritic trunk was measured at a distance of one cell-body diameter from the cell body as described [34].

To measure spine length and density, all types of spines (i.e., protrusions, stubby, filopodia, and mushroom-type spines) were counted within a 10 μm stretch of distal dendrite in Gnao mice and their wild-type littermates, Gnao. According to the typical criteria for classifying dendritic spines, only spines separated from the dendritic shaft by at least 0.5 μm were counted [35]. vGluT2-positive CF boutons were counted on 50 μm stretches of the main dendritic trunks from Gnao mice and Gnao mice. The size of the vGluT2-positive fluorescent boutons was measured in a 200 μm-wide area of the ML from lobule II/III of each genotype. All quantifications for these results were performed semi-automatically using the ImageJ 1.50 software package (NIH).

Statistical analysis

Statistical analyses were performed to determine any significant differences between two groups by two-tailed unpaired t-tests using the GraphPad Prism 8 (GraphPad Software Inc.) or SigmaPlot 12.0 software (Systat Software, Inc.). All quantitative values are presented as means ± Standard error of mean (SEM). For all tests, p < 0.05 is considered significant and p values are presented in the results and/or figure legends.

Results

Gαo is required for cerebellar cortex development

The anatomical development of the cerebellar lobules is completed around P21 [36]. Although the survival rates of Gnao mice is extremely low, we were able to analyze the cerebella of the few Gnao mice that survived to P180. To avoid lobular differences, we used sagittal sections of the vermal region in all experiments. Cresyl violet staining showed that the lack of Gαo does not alter the overall lobulation of the cerebellar cortex; the only significant difference we observed in Gnao mice was a 62% reduction in the depth of the intercrural fissure between lobules VI and VII [37] (0.25 ± 0.02 mm in Gnao mice; 0.10 ± 0.04 mm in Gnao mice; n = 3 mice for each genotype≥P25; *p < 0.05; arrows in Fig. 1a-b, and g). The overall size of the cerebellum was also reduced, with the cerebellar surface area in the midsagittal sections of the vermis being 25% smaller in Gnao mice compared to Gnao mice (5.9 ± 0.39 mm2 in Gnao mice; 4.4 ± 0.23 mm2 in Gnao mice; n = 4 mice for each genotype≥P25; *p < 0.05; Fig. 1h and Additional file 1). This hypoplasia was evident in the ML (Fig. 1c-d), the thickness of which in each folium was reduced by 11% in Gnao mice (138.6 ± 2.27 μm in Gnao mice; 123.7 ± 2.16 μm in Gnao mice; n = 4 mice for each genotype≥P25; ***p < 0.001; Fig. 1i). Since the thickness and shape of the GCL varies among the folia, we compared the total area occupied by the GCL in Gnao mice to that of their wild-type littermates (Additional file 1). The GCL was reduced to an extent similar to that of the ML (2.4 ± 0.14 mm2 in Gnao mice; 1.9 ± 0.01 mm2 in Gnao mice; n = 3 mice for each genotype≥P25; *p < 0.05; Fig. 1j), and a total number of GCs were reduced by 27% in Gnao mice compared to their wild-type littermates (155.0*103 ± 2.67*103 cells in Gnao mice; 113.7*103 ± 5.77*103 cells in Gnao mice; n = 3 mice for each genotype≥P25; **p < 0.005; Fig. 1k). Thus, we did not find a significant difference in the occupancy ratios of the ML to the GCL in Gnao mice compared to their wild-type littermates (n = 4 for each genotype≥P25; Fig. 1l).

Fig. 1

Gαo is required for cerebellar cortex development. a-b Overall cerebellum size of Gnao mice is smaller than that of their wild-type Gnaolittermates. Both Gnao and Gnao mice show the ten typical cerebellar lobules, but the depth of the folia in lobules VI-VII is reduced in Gnao mice (arrows). c-d Cresyl violet staining shows the normal arrangement of the three cortical layers. The ML is thinner in Gnao mice than Gnao mice. e-f Gαo is abundant in the ML and GCL of Gnao, but undetectable in the ML and GCL of Gnao mice. The asterisks in e highlight the absence of Gαo in PC soma. Note the difference of ML thickness. Scale bars, 500 μm in a-b, 50 μm in c-f. g The intercrural fissure between lobules VI and VII is reduced in Gnao mice. h Cerebellar areas in mid-sagittal sections are smaller in Gnao mice. i ML thickness is reduced in Gnao mice (n = 119 sections from 4 Gnao+/+ mice and 122 sections from 4 Gnao−/− mice). j GCL occupancy is 20% lower in Gnao mice than Gnao mice. k The number of total GCs in the GCL area is reduced by 27% in Gnao mice. l Occupancy ratios for the GCL (43.4 ± 0.73 in Gnao mice and 44.8 ± 0.96 in Gnao mice; p = 0.300) and for the ML (45.3 ± 0.97 in Gnao mice and 43.1 ± 1.33 in Gnao mice; p = 0.248) are similar when the smaller size of the cerebellum in Gnao mice is taken into account. m upper. Genotyping shows the mutant (Gnao−) and the wild type (Gnao+) alleles as 275 and 224 bp, respectively. Lower. Western blot analysis shows the absence of Gαo, but not Gαi. Data are means ± SEM; *p < 0.05, **p < 0.005, ***p < 0.001 vs. Gnao. N.S.; No significant difference, STD; standard marker

Immunostaining with an anti-Gαo antibody revealed a broad distribution of Gαo protein in the ML and GCL (Fig. 1e), but not in the PC somata within the PCL (asterisks in Fig. 1e). We verified the specificity of the anti-Gαo antibody by confirming an absence of immunoreactivity in Gnao mice (Fig. 1f) as well by western analysis (Fig. 1m).

Gαo is targeted to the synaptic membranes of cerebellar neurons

The apparent absence of Gαo-immunoreactivity we observed in PC somata within the PCL was not in accordance with previous reports of Gαo expression in PCs (Table 1). To determine the precise location of Gαo protein, we performed an immunostaining experiment to detect Gαo and Pcp2 in the adult cerebellum. Pcp2 is a well-known PC-specific marker that interacts with Gαo [15]. We observed Gαo protein in the membrane compartment of PC dendrites but not in the cytosolic compartment of PC soma (Fig. 1e, asterisk), whereas we observed Pcp2 staining both in PC dendrites and soma (Fig. 2a). Importantly, Gαo was co-localized with Pcp2 in the dendritic spines (Fig. 2a). Such co-localization was evident in the sprouting dendritic tips and in pseudopodial protrusions emerging from the apical pole of early phase PCs in P7 (Fig. 2b). An in situ hybridization experiment revealed the presence of Gnao mRNA in the perikaryon of PCs, GCs, and interneurons of the ML (arrowheads in Fig. 2c1, c2, and c3, respectively). Gαo protein expression was more evident upon in vitro dissociation in cultured PCs (Fig. 2d). Unlike in the in vivo immunostaining experiment, we observed Gαo protein both in PC soma (asterisk) and dendrites (arrows in Fig. 2d). This immunoreactivity showed an overlap with that of anti-Calb, another PC-specific marker (Fig. 2d). These results indicate Gαo protein is highly mobile in the membrane compartment after being synthesized in the PC soma. This leads to the apparent absence of Gαo protein in the PCL in vivo (Table 1). Further studies with immune-electron microscopy with the Gnao and Gnao mice may clarify this issue. In dissociated cultures of P7 cerebellum in which external granule layer (EGL) cells highly proliferate, we found most cells are Tubb3- and Gαo-positive (Fig. 2e). Considering that GCs greatly outnumber interneurons by 414:1 [40, 41], it is very likely most Tubb3- and Gαo-positive cells in Fig. 2e are GCs although Gnao mRNA is expressed both in GCs and interneurons in the ML (Fig. 2c). Thus, the reduced size of the ML indicates a reduced total number of GCs in the absence of Gαo (Fig. 1k).

Table 1

Comparison of the localization of Gαo protein and Gnao mRNA in the cerebellar cortex. The localization of Gαo protein and Gnao mRNA is compartmentalized in ML, PCL and GCL. Gαo protein and Gnao mRNA were assessed by immunohistochemistry and in situ hybridization, respectively

	Species	ML	PCL	GCL	Ref.
Gαo protein	Mouse	+	–	+	Figure 1e
		ND	+	ND	[23]
	Rat	+	–	ND	[14]
Gnao mRNA	Mouse	–	+	+	Figure 2b Allen brain atlasa
	Rat	–	+	+	[38, 39]

ahttp://mouse.brain-map.org/experiment/show/507. ND Not Determined

Fig. 2

Gαo is highly targeted to the synaptic terminal regions of cerebellar neurons in wild-type mice. a Gαo-immunoreactivity appears at high levels throughout PC dendrites, especially in Pcp2-positive spines (Arrows in a1-a3, magnified views of a box shown in a). Scale bars, 50 μm in a, 10 μm in a1-a3. b PCs start to differentiate with many protrusions and neurites in a monolayer. Gαo is co-localized with Calb, another PC-specific marker at every PC boundary (yellow, arrows in b2 and b3) except in PC soma devoid of Gαo (asterisks in b2). Scale bars, 20 μm in b-b3. c Strong Gnao mRNA signals appear in the perinuclear regions of PC soma (c1), GCs (c2), and ML interneurons (c3). Scale bars, 20 μm in c, 10 μm in c1-c3. d In vitro culture of PCs shows Gαo and calbindin-D28K (Calb) expression at 21 days in vitro (DIV). The cultured PCs show significant Gαo signal in the soma (asterisk, d2) and in the dendrites (arrows in d1-d2). e In GCs at DIV7, Gαo protein appears in the soma (asterisks, e2) and in the neurites (arrows in e1-e2) of Tubb3-positive GCs. Scale bars, 20 μm in d-e. Bisbenzimide (Hoechst 33342) staining (blue) was used to visualize cell nuclei in a-e

Gαo is required for the maturation of PC dendrites

After immunostaining PFs in Gnao mice with an anti-vGluT1 antibody, we found a generalized reduction in the thickness of the ML, but no more specific abnormalities (Additional file 2). Immunostaining of Gnao mice with an anti-Calb antibody revealed no difference in the arrangement and number of PC somata in the PCL in Gnao mice (n = 4 for each genotype; Fig. 3a-b, and g). We estimated PC monoplanarity by counting the number of PCs with primary dendrites stretching longer than 30 μm. About 70.3 ± 6.20% of the wild-type PCs retained primary dendrites longer than 30 μm (Fig. 3c, h), which is consistent with a previous report that roughly 70–80% PCs are monoplanar after P25 [21]. This suggests our semi-quantitative approach is relevant and indirectly indicates planarity as was previously noted [34]. In comparison, Gnao mice show only 47.1 ± 4.85% monoplanar PCs, suggesting that roughly 33% PCs in Gnao mice do not acquire monoplanarity during the early postnatal period (n = 3 mice for each genotype≥P25; *p < 0.05; Fig. 3d, h). Importantly, we observed a reduction in the thickness of PC dendritic trunks by 25% (3.2 ± 0.06 μm in Gnao+/+ mice and 2.4 ± 0.06 μm in Gnao−/− mice; n = 3 mice for each genotype≥P25; ***p < 0.001; Fig. 3e-f, and i) and in the number of spines by 16% (1.01 ± 0.01 /μm in Gnao mice; 0.85 ± 0.01 /μm in Gnao mice; n = 3 mice for each genotype≥P180; ***p < 0.001; Fig. 3j) accompanied by an increase in average spine length by 9% (0.7 ± 0.01 μm in Gnao+/+ mice; 0.8 ± 0.02 μm in Gnao−/− mice; n = 3 mice for each genotype≥P180; ***p < 0.001; Fig. 3k). These data suggest Gαo deficiency disrupts the development of the monoplanarity of typical fan-shaped PCs, their dendritic arborization, and their spine formation without affecting their migration or alignment. Interestingly, when we stained Gnao mice and visualized the vGluT2-positive terminal boutons of CFs with vertical extensions along the main dendrite trunks of PCs (Fig. 4a-b), we found a 32% reduction in the number of CF boutons on PCs (23.3 ± 2.02 in Gnao mice; 16.0 ± 1.38 in Gnao mice; n = 4 mice for each genotype≥P25; *p < 0.05; Fig. 4c) and an 11% reduction in the average size of CF boutons (2.1 ± 0.05 μm2 in Gnao mice; 1.9 ± 0.06 μm2 in Gnao mice; n = 4 mice for each genotype≥P25; **p < 0.005; Fig. 4d). This atrophy of CF boutons and PC spines may indicate an essential role for Gαo in full functional maturation of synapses between presynaptic ION neurons and postsynaptic PCs.

Fig. 3

Gαo is required for PC maturation. a-b The thickness of the ML is reduced (Bars in a, b) without changing PCL alignment. Scale bar, 100 μm. c-d Calbindin-D28k (Calb) staining reveals whole PC dendritic trees throughout the ML of the adult cerebellum. Scale bar, 20 μm in c-d. e-f Images of dendritic spines detected with an anti-Calb antibody. Arrows indicate spines longer than 1.5 μm in Gnao mice. Scale bar, 5 μm in e-f. g The average number of PCs per 200 μm is similar between Gnao mice and Gnao mice (11.5 ± 0.23 in Gnao mice and 11.0 ± 0.30 in Gnao mice; p = 0.232 vs. Gnao). h 70.3 ± 6.20% of total PCs in Gnao mice and 47.1 ± 4.85% of total PCs in Gnao mice extend primary dendrites longer than 30 μm (n = 61 PCs from 3 Gnao mice and 76 PCs from 3 Gnao mice). i Dendritic trunk widths are decreased in Gnao mice (n = 41 primary dendrites from 3 Gnao mice and 39 primary dendrites from 3 Gnao mice). j Gnao mice exhibit a significant reduction of spine density (n = 123 segments of 122 distal dendrites from 3 Gnao mice and 130 segments of 106 distal dendrites from 3 Gnao mice). k Average spine length is longer in Gnao mice (n = 210 dendrite spines from 3 Gnao mice and 195 dendrite spines from 3 Gnao mice). Data are means ± SEM; *p < 0.05, ***p < 0.001 vs. Gnao. N.S.; No significant difference

Fig. 4

Gαo is required for the maturation of CF-PC synapses. a-b Gnao mice show atrophied vGluT2-positive morphology of CF synaptic boutons (open arrowheads in b1) and PC dendrites in the ML relative to Gnao mice (arrowheads in a1). Scale bar, 10 μm. c Gnao mice show fewer vGluT2-positive dendritic boutons (n = 9 main dendrites from 4 Gnao mice and n = 7 main dendrites from 4 Gnao mice). d Bouton size is smaller in Gnao mice (P25) (n = 655 vGluT2-positive boutons from 4 Gnao mice and 432 vGluT2-positive boutons from 4 Gnao mice). Data are means ± SEM; *p < 0.05, **p < 0.005 vs. Gnao

Discussion

The cerebellum is a useful model utilized for studying many aspects of neural development because of its typical cytoarchitecture and developmental program [42]. In this study, we show that Go plays a crucial role in the postnatal development of the cerebellar cortex. Specifically, Gnao mice exhibit hypoplasia of the cerebellum (Fig. 1 and Additional file 1) with overall defects in cerebellar cortex development. In Gnao mice, the thickness of the ML and GCL are significantly reduced (Fig. 1) without more specific abnormalities in vGluT1-expression (Additional file 2), the monoplanarity, dendritic arborization and spinogenesis of PCs are reduced (Fig. 3), and the maturation of CF boutons are atrophied (Fig. 4).

The intensive remodeling and maturation of PC dendrites that occurs during the postnatal period is controlled by intrinsic factors early in development and by a careful orchestration of intrinsic and extrinsic factors later in development [22]. The abnormalities we observed in the PC dendrites and spines of Gnao mice may therefore be either a consequence of the lack of Go-mediated signaling in the PCs themselves (intrinsic factors) or a consequence of their defective formation of synapses with CF boutons (extrinsic interactions).

Intrinsic factors that regulate Ca2+ homeostasis, such as mGluR1 and its downstream signaling pathways, are critical in PCs for the selection of the winning CF and the pruning of redundant CFs that originate in the ION [43]. Mice lacking even a single component of the mGluR1-Gαq-PLCβ4-PKCγ signaling cascade in PCs show multiple CF effects. Although such mice show normal PC dendritic growth, they have motor defects that include ataxia and impaired motor learning [24–26, 44]. Activation of PLCβ occurs when Gβγ dimers are released from Gq and Gi/o, but the binding sites and activation kinetics differ between Gαq and Gβγ [45]. Western analyses indicated that the expression levels of Gαi and Gαq were not altered by Gnao-deletion (Fig. 1m and Additional file 3). Furthermore, the loss of Gαo does not increase free Gβγ dimers in Gnao mice [8]. Thus, the phenotypes of Gnao mice are more likely due to the loss of some of Gαo specific unique functions rather than being due to the activation of the Gβγ-PLCβ pathway.

Gi/o signals are known to mediate presynaptic inhibitory effects of many neurotransmitters on transmitter release in axon terminals [46]. Activation of Gi/o-coupled CB1 receptors decreases glutamate release in presynaptic PF terminals by suppressing Ca2+ entry via voltage-gated Ca2+ channels (N-, P/Q-, and R-types) or modulating vesicle-release related proteins [47-49]. Considering that a CB1 agonist predominantly activates Go in the cerebellum and Go exerts its inhibitory function on N- and P/Q type channels [16, 50], the Ca2+ influx in PF terminals is highly likely mediated by the Go signals. In PCs, Pcp2 stimulates GDP release from Gαo through its GoLoco domain and modulates the inhibitory functions of Go on P/Q-type (Cav2.1) channels [15, 17, 51]. Deletion of Cacna1a, the gene encoding the pore-forming α1 subunit of P/Q-type channels in the entire cerebellum or in PCs impairs the maturation of CFs [52, 53]. Pcp2 expression was not altered in our Gnao mice, while the terminal boutons of CFs were atrophied (Fig. 4 and Additional file 3). It would be of great interest to determine whether the functions of voltage-gated Ca2+ channels are altered in Gnao mice.

Gαo is highly enriched in the cerebellar ML, particularly at the plasma membranes of PC dendrites and in PFs of the ML. Gαo membrane targeting may be associated with lipid modifications such as myristoylation or palmitoylation [54]. The distinctive membrane compartmentalization of Gαo in PCs (Table 1) makes it well-suited to perform synapse-specific studies. Consistent with our data, Gαo appears at high levels in postsynaptic densities in vivo as well as in the neurite tips and growth cones of in vitro cultured neurons [7, 12, 13, 55]. Consistently, we show that Gαo is found in the sprouting dendrites of PCs in P7 cerebellum (Fig. 2b). It is noteworthy that migration of granule cells from EGL to GCL appear normal, suggesting that the Bergman’s glia functions properly.

In conclusion, our study suggests Go may perform critical roles in the postnatal development of the cerebellar cortex. Future studies cell type specific deletion of Gnao will reveal the correlation of intrinsic signaling pathways in cerebellar cortical neurons with PF-PC synapse and CF-PC synapse. Gabra6 (Δα6)- and Pcp2(L7)-driven Cre-expressing mice have been utilized for the study of GCs and PCs, respectively [56-58]. However, Gαo protein is detectable due to its high stability even after the Gnao gene is deleted by Cre enzymes in postnatal neurons (data not shown). Inducible Cre mouse lines such as Gli1-CreER may be an alternative strategy to induce concomitant deletion of the Gnao gene and the existing Gαo proteins in proliferating EGL cells [59].

The gross anatomy of the cerebellum is normal in Gnao−/− mice. (PDF 187 kb)

PFs in the ML visualized in vGluT1-stained cerebellum. (PDF 85 kb)

Expression of Pcp2 in the cerebellum of Gnao−/− mice. (PDF 380 kb)