Mice lacking EFA6C/Psd2, a guanine nucleotide exchange factor for Arf6, exhibit lower Purkinje cell synaptic density but normal cerebellar motor functions

ADP ribosylation factor 6 (Arf6) is a small GTPase that regulates various neuronal events including formation of the axon, dendrites and dendritic spines, and synaptic plasticity through actin cytoskeleton remodeling and endosomal trafficking. EFA6C, also known as Psd2, is a guanine nucleotide exchange factor for Arf6 that is preferentially expressed in the cerebellar cortex of adult mice, particularly in Purkinje cells. However, the roles of EFA6C in cerebellar development and functions remain unknown. In this study, we generated global EFA6C knockout (KO) mice using the CRISPR/Cas9 system and investigated their cerebellar phenotypes by histological and behavioral analyses. Histological analyses revealed that EFA6C KO mice exhibited normal gross anatomy of the cerebellar cortex, in terms of the thickness and cellularity of each layer, morphology of Purkinje cells, and distribution patterns of parallel fibers, climbing fibers, and inhibitory synapses. Electron microscopic observation of the cerebellar molecular layer revealed that the density of asymmetric synapses of Purkinje cells was significantly lower in EFA6C KO mice compared with wild-type control mice. However, behavioral analyses using accelerating rotarod and horizontal optokinetic response tests failed to detect any differences in motor coordination, learning or adaptation between the control and EFA6C KO mice. These results suggest that EFA6C plays ancillary roles in cerebellar development and motor functions.


Introduction
ADP ribosylation factor 6 (Arf6) is a small GTPase that regulates actin cytoskeleton remodeling and vesicular transport between the plasma membrane and endosomes [1][2][3]. In the a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 mammalian brain, Arf6 mediates a variety of neuronal processes related to cell shape, motility and responsiveness, including the maintenance of structural integrity of the neuroepithelium [4], neuronal migration [5,6], formation of the axon [7] and dendrites [8], maturation and maintenance of dendritic spines [9][10][11], trans-regulation of oligodendrocyte differentiation [12], axonal transport [13], recycling of synaptic vesicles [14] and hippocampal long-term depression [15]. The diverse neuronal functions of Arf6 are facilitated by strict spatiotemporal regulation of the GDP/GTP cycle of Arf6 in neurons by two types of regulatory proteins: guanine nucleotide exchange factors (GEFs) that facilitate the exchange of GDP for GTP, and GTPase activating proteins (GAPs) that enhance GTP hydrolysis. Several GEFs were previously identified to activate Arf6, including the EFA6/PSD (exchange factor for Arf6/pleckstrin and Sec7 domain-containing protein), BRAG/IQSEC (brefeldin A-resistant Arf-GEF/IQ motif and Sec7 domain-containing protein) and cytohesin families [1,2] The EFA6/PSD family comprises EFA6A/PSD1 [16,17], EFA6B/PSD4 [18], EFA6C/PSD2 [19] and EFA6D/PSD3 [20], which are generated from distinct genes and function primarily as an Arf6-specific GEF [16,21,22]. They are structurally characterized by a conserved domain organization consisting of a central catalytic Sec7 domain, an adjacent pleckstrin homology (PH) domain responsible for interaction with the plasma membrane and F-actin [18,23], and a C-terminal region containing a coiled coil motif that mediates protein-protein interaction and GEF-independent actin cytoskeleton remodeling [16,17,24,25]. In the adult mouse brain, EFA6A, EFA6C and EFA6D are abundantly expressed with distinct expression patterns. EFA6A is expressed predominantly in the forebrain and localized to the plasma membrane, postsynaptic density and endosomes in the dendritic shaft and spines in hippocampal neurons [17,25,26]. The EFA6A-Arf6 pathway was shown to regulate dendritic formation [26,27], maturation and maintenance of dendritic spines [9,11], and directionality of axonal transport [28] in primary cultured neurons. EFA6D is widely expressed throughout the brain and is localized to various subcellular compartments in hippocampal neurons, including cell bodies, dendritic shafts and spines, axons and presynaptic terminals [20,29]. A single nucleotide polymorphism in the human EFA6D gene was reportedly associated with alcohol drinking behaviors and neuronal activity in the prefrontal cortex during the go/no-go executive control task, suggesting modulatory roles of EFA6D in addictive and cognitive behaviors [30]. On the other hand, EFA6C is unique in that it is expressed predominantly in the cerebellar cortex, particularly in Purkinje cells [19]. However, the physiological significance of individual EFA6 members in the brain is still unknown at the whole animal level.
The cerebellum is a brain region that is engaged in motor coordination and learning with a uniform three-layered cortical structure consisting of the molecular layer, Purkinje cell layer and granular layer [31,32]. Purkinje cells are the sole output neurons of the cerebellar cortex, with an extremely elaborate dendritic tree studded with numerous dendritic spines, where asymmetric excitatory synapses are formed with parallel fibers from granule cells and climbing fibers from the inferior olive at the distal and proximal dendritic segments, respectively. Owing to their unique morphology and relatively simple synaptic connections [31,32], Purkinje cells have been extensively studied as a model for neuronal morphogenesis, synaptogenesis and synaptic plasticity.
To examine the physiological role of EFA6C in cerebellar development and functions, we generated mice lacking EFA6C using the CRISPR/Cas9 system and examined their cerebellar phenotypes by histological and behavioral analyses.
(MEXT) of Japan and approved by the Animal Experimentation and Ethics Committee of the Kitasato University School of Medicine (Permission number: 2017-146). All efforts were made to minimize animal distress and to reduce the number of animals used.
To check CRISPR/Cas9-mediated mutation of the EFA6C gene, genomic DNA was extracted from tail or ear tissues and subjected to PCR using primers to amplify the exon 1 region including the gRNA target sequence: forward (5'-ACCGACATGGATGAAGAGAA GCTC-3') and reverse (5'-TCCTCCTCTGCTGGTCCGCTTCTC-3'). The PCR amplicons were visualized by agarose gel electrophoresis, subcloned into pGEM-T Easy vectors (Promega, Madison, WI) and sequenced using the BigDye terminator Cycle Sequencing Kit v3.1 and ABI 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA). The male founder mice were crossed to wild-type C57BL/6J female mice to generate heterozygous offspring.
Adult mice (less than 5 per cage) and pregnant mice (1 per cage) were housed and maintained under barrier conditions in an air-conditioned room (about 20-24˚C) under 12-h light/dark cycle with free access to food and water. In all experiments, wild-type mice generated by intercrossing heterozygous male and female mice were used as control animals.

Histological analyses
Three male 10-12-week-old mice of each genotype were perfused through the vasculature via the left cardiac ventricle with 4% paraformaldehyde and 2% paraformaldehyde/2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2) for light and electron microscopic analyses, respectively.
Electron microscopic analyses were performed as previously described [39] with some modifications. Parasagittal sections (200 μm thick) of cerebella prepared with a microslicer (VT1000S; Leica, Nussloch, Germany) were post-fixed with 2% osmium tetroxide, dehydrated, and embedded in epoxy resin. Ultrathin sections of the cerebral cortex of lobules 4/5 were made at a thickness of 70 nm in the parasagittal plane on an ultra-microtome (Ultracut; Leica), and stained with 2% uranyl acetate and mixed lead solution. Thirty electron microscopic images of the middle one-third of the molecular layer with an area of 300 μm 2 were randomly taken from three mice of each genotype using an electron microscope (H-7650; Hitachi, Tokyo, Japan). Asymmetric synapses were identified as synaptic contacts between parallel fiber terminals, which contain synaptic vesicles at a lower packaging density than climbing fiber terminals, and the Purkinje cell spine heads, which are observed as round or oval profiles containing thick electron-dense postsynaptic specialization and smooth endoplasmic reticulum without mitochondria or microtubules [40]. The largest diameter of the spine head was measured as the spine head width. The length of thick electron-dense postsynaptic specialization in the Purkinje cell spine was measured as the postsynaptic density length. The measurement was done in a blinded manner using the Image J software (NIH). To determine the percentage of free spines of Purkinje cells, approximately 100 spines from each mouse (WT, 330 spines; KO, 337 spines) were examined with several sets of serial electron microscopic images from three wild-type and EFA6C KO mice as described previously [39].

Behavioral analyses
The rotarod test was performed as previously described with modification [41]. Briefly, the rotarod apparatus was programmed to accelerate from 4 rpm to 40 rpm over 4 min and then remain constant speed at 40 rpm for 1 min. Ten male mice aged 10-12 weeks from each genotype were placed on the rod rotating at 4 rpm and underwent three trials per day with a 30-min inter-trial interval for 5 consecutive days. The latency until mice fell off the rod or rotated all the way around the rod successively twice while clinging to the rod was recorded, and data from the three trials were averaged to give the daily score. One day before beginning Recombinant protein corresponding to amino acids 559-582 at the Cterminus of mouse VGulT2 guinea pig polyclonal the behavioral tests, mice were placed on the rod rotating at 4 rpm for 1 min to acclimate to walking on the rod. Five male mice aged at 10-12 weeks from each genotype were subjected to a horizontal optokinetic response test as described previously [42,43]. Briefly, mice were placed on a table with their heads fixed by a screw towards a checked-pattern screen, which was sinusoidally oscillated around the table by 15 degrees (peak-to-peak) at 0.33 Hz in light. The evoked eye movements were recorded by an infrared television camera system (Flea3 0.3 MP Mono; FLIR system, OR), and several cycles of the evoked eye movements without blinks and saccades were averaged every 10 min for 1 h, and the mean amplitudes were calculated by a modified Fourier analysis. The gain of the eye movement was calculated by dividing the peak-to-peak amplitude of eye movement by that of the screen oscillation at each time point.

Statistical analyses
Data from histological and behavioral analyses were statistically analyzed using Student's t-test and two-way repeated measures ANOVA, respectively. Significance was indicated by a p-value less than 0.05.

Generation and characterization of EFA6C KO mice
To investigate the physiological role of EFA6C in vivo, we generated EFA6C KO mice using the CRISPR/Cas9 system. The gRNA containing 20 nucleotides followed by a 3-nucleotides PAM sequence targeting exon 1 of the EFA6C gene (Fig 1A) was electroporated into fertilized zygotes together with the Cas9 protein. Among the ten founder mice carrying indels in the EFA6C gene, PCR and sequencing analyses revealed that six founder mice (5 males and 1 females) carried bi-allelic mutations around the target sequence, two mice (2 females) carried a mono-allelic mutation and the remaining two male mice were mosaic carrying more than two mutant alleles (S1 Fig). By crossing the male founder mice to wild-type C57BL/6J female mice, heterozygous offspring were generated, and a mouse line derived from the founder mice #10 (S1 Fig), in which 112 nucleotides had been deleted in the exon 1, was chosen for further investigation (Fig 1B and 1C). Homozygous mice grew normally without any abnormal neurological signs such as ataxic gait.
To assess the protein expression of EFA6C, cerebellar lysates from each genotype were immunoblotted with anti-EFA6C antibodies raised against the N-terminal 136 amino acids of mouse EFA6C (anti-EFA6C[N]) [19] (Fig 1D). In addition, the anti-EFA6C(C) antibody detected a band around 50 kDa without differences in the immunoreactive intensity among the three genotypes ( Fig  1D, WT, 1.0 ± 0.42; Hetero (+/-), 0.76 ± 0.10; KO, 0.74 ± 0.01, n = 3 of each genotype). No other bands corresponding to truncated EFA6C products were detectable in the cerebellar lysates of homozygous or heterozygous mice. Next, to examine the compensatory changes in the Arf6 pathway, the cerebellar lysates were immunoblotted with antibodies directed against Arf6 and other Arf6 GEFs such as EFA6A [16], EFA6D [20], BRAG1-3 [34,36,44] and cytohesin-2 [45]. There were no significant differences in the protein expression of Arf6 and other Arf6 GEFs in the cerebellar lysates between wild-type and EFA6C KO mice (Fig 1E and 1F). Further immunoblotting of cerebellar lysates with antibodies against calbindin and mGluR1a,  N]), EFA6C C-terminal region (EFA6C[C]), EFA6A, EFA6D, BRAG1-3, cytohesin-2 (Cyth2), Arf6, calbindin and mGluR1. The same blots were reprobed with anti-α-tubulin IgG. Note that immunoreactive bands for EFA6C were completely absent in the cerebellar lysates of homozygous mice, and there were no apparent differences in the protein expression of Arf6 GEFs, Arf6, calbindin and mGluR1 in the cerebellar lysates between the two genotypes. The positions and sizes (kDa) of molecular weight markers are indicated on the right. (F) Quantification of the protein expression of Arf6 GEFs, Arf6, calbindin and mGluR1. Relative optical densities for immunoreactive bands normalized to those for α-tubulin are shown as mean ± SEM of relative ratio to the wild-type control group set as 1.
Immunohistochemical analyses with the anti-EFA6C(N) antibody demonstrated that intense immunolabeling for EFA6C was detected in cell bodies and dendrites of Purkinje cells and the pinceau of basket cells in the wild-type cerebellar cortex (Fig 2A and 2B). In addition, faint but discrete immunolabeling was detected in the cerebellar glomeruli in the granular layer. In contrast, no immunolabeling was detectable in the cerebellar cortex of homozygous mice (Fig 2B"), further confirming the absence of EFA6C at the protein level in the cerebellum of EFA6C KO mice.

Histological assessment
Nissl staining of parasagittal cerebellar sections failed to detect differences in the gross anatomy of the cerebellar cortex between wild-type, heterozygous and homozygous mice in terms of the formation of folia, the three-layered structural organization of the cerebellar cortex, and the thickness and cellularity of each layer (Fig 2C-2C" and 2D-2D").
Immunofluorescence with an antibody against calbindin, a reliable marker to visualize the overall morphology of Purkinje cells [48,49], revealed that the Purkinje cells of both wild-type and homozygous mice developed extensive dendritic arbors that were indistinguishable ( Fig  2E and 2E'). Double immunofluorescence with antibodies against VGluT1 and VGluT2, which label axon terminals of parallel fibers from cerebellar granule cells and climbing fibers from inferior olivary neurons, respectively, revealed that there were no apparent qualitative differences between the two genotypes in the labeling pattern or intensity for VGluT1 and VGluT2 in the molecular layer (Fig 2F and 2F'), suggesting the normal development of the two major afferent fibers onto Purkinje cells. Further double immunofluorescence with antibodies against VGAT and PV, which label GABAergic axon terminals, and Purkinje cells and interneurons such as basket cells and stellate cells, respectively, did not detect any qualitative differences in the immunoreactive cell number or distribution pattern between the two genotypes (data not shown).
Next, we quantitatively analyzed the ultrastructural synaptic organization of the molecular layer by electron microscopy (Fig 3A and 3B). In the middle molecular layer where the dendritic spines of Purkinje cells mainly form synaptic contacts with parallel fiber terminals from granule cells, the density of asymmetric synapses was significantly lower in EFA6C KO mice compared with that in wild-type mice (Fig 3C, WT, 15.8 ± 1.4 / 100 μm 2 ; KO, 13.1 ± 0.5 / 100 μm 2 , n = 3 mice of each genotype, P = 0.032, Student's t-test). In contrast, the width of spine heads and the length of the postsynaptic density on asymmetric synapses of Purkinje cells were comparable between the two genotypes (Fig 3D and 3E). To further examine the incidence of free dendritic spines that contain a postsynaptic density but lack synaptic contact with the presynaptic terminal, we performed serial section electron microscopy of the molecular layer from the two genotypes (Fig 4A and 4B). The incidence of free dendritic spines was extremely low in both genotypes and the difference between the groups was not statistically significant (Fig 4C; WT, 0.58 ± 1.02%, n = 330 spines from 3 mice; KO, 0.58 ± 0.52%, n = 337 spines from 3 mice, P = 0.989, Student's t-test).

Behavioral analyses
Finally, to examine whether motor coordination and learning were impaired in EFA6C KO mice, mice were subjected to two types of cerebellum-related behavioral paradigms, the accelerating rotarod test (Fig 5A) and the hOKR test (Fig 5C). In the accelerating rotarod test, the latency to falling off the rod consistently increased over 5 consecutive days of three daily trials, without any significant difference between the two genotypes ( Fig 4B, P = 0.8173, two-way repeated measures ANOVA). In the hOKR, wild-type and EFA6C KO mice displayed adaptation of eye movements tracking the horizontal oscillations of a checked-pattern screen at a comparable level during the 60-min training (Fig 5D and 5E; P = 0.8926, two-way repeated measures ANOVA). These results suggest that EFA6C KO mice exhibit normal cerebellumrelated motor coordination and learning in these two tests.

Discussion
Previous studies using primary cultured neurons implicated the EFA6-Arf6 pathway in neuronal processes related to the formation of neural circuits, such as the formation of dendrites [26,27] and dendritic spines [9] and axonal transport [28]. We have previously shown that three members of the EFA6 family, EFA6A, EFA6C and EFA6D, are abundantly expressed in the adult mouse brain with distinct spatiotemporal patterns [17,19,20,25,26,29]. However, the functional significance of individual EFA6 members is unknown at the whole animal level. In this study, we focused on EFA6C, an Arf6 GEF that is enriched in the cerebellum [19], and reported the cerebellar phenotypes of EFA6C KO mice for the first time.
The only significant phenotype found in the cerebellum of EFA6C KO mice was the reduced density of asymmetric synapses on dendritic spines of Purkinje cells, which could be caused by several structural abnormalities in the molecular layer. The present serial section electron microscopic analysis revealed that the incidence of free dendritic spines lacking synaptic contacts was comparably low in both genotypes, excluding the possibility that an increase in free spines lead to the reduced synaptic density as previously reported in mice lacking GluD2 [39] and Cbln1 [50]. Therefore, it is likely that the reduced synaptic density in EFA6C KO mice may be attributable to other deficits in cerebellar structures such as the arborization of dendrites, and the formation and maintenance of dendritic spines of Purkinje cells. Because . Thirty images of the middle one-third of the molecular layer with an area of 300 μm 2 were taken randomly from three male mice of each genotype, and examined in a blinded manner. Note the significantly lower density of asymmetric synapses in EFA6C KO mice compared with that in wild-type mice. Data are shown as mean ± SEM from three mice of each genotype. � p < 0.05; ns, not significant; Student's t-test. Scale bar, 1 μm.
Contrary to our hypothesis about the importance of the EFA6C-Arf6 pathway in cerebellar development and functions, the cerebellar phenotype in EFA6C KO mice was mild. The lack of significant phenotypes in these mice suggests that EFA6C plays ancillary roles in the cerebellum. Alternatively, it may reflect the existence of various cellular mechanisms to compensate for the ablation of the EFA6C-Arf6 pathway. First, Arf6 can be activated by several Sec7-containing Arf GEFs: EFA6A-D in the EFA6 family [16][17][18]20], BRAG1-3 in the BRAG family [34,36,44] and cytohesin 1-3 in the cytohesin family [45,51]. Indeed, Purkinje cells abundantly express EFA6D [29], BRAG2 (our unpublished data), BRAG3 [36] and cytohesin 1-3 [51,52] as well as EFA6C. Although the present immunoblot analyses failed to detect any compensatory changes in the protein expression of these Arf6 GEFs, it is still possible that their subcellular localization or GEF activities may be changed in Purkinje cells lacking EFA6C. Second, considering the essential nature of actin cytoskeleton remodeling and endosomal trafficking in neurons, there are likely to be crosstalk mechanisms between Arf6 and other small GTPase pathways in neurons. For example, the EFA6A-Arf6 pathway was shown to partly converge on Rac1 to regulate the formation of dendritic spines in cultured hippocampal neurons [9,11]. In addition, Rab35 was shown to promote nerve growth factor (NGF)-induced neurite outgrowth of PC12 neuroendocrine cells by regulating the activity of Arf6 through the interaction with centaurin-β2/ACAP2, an Arf6 GAP [53]. These findings suggest that Arf6 coordinates with other small GTPases such as Rac1 and Rab35 to regulate neurite formation. Therefore, it is conceivable that such crosstalk mechanisms between Arf6 and other small GTPases may compensate for the lack of EFA6C to maintain the morphology and function of Purkinje cells. Third, there are short alternative isoforms of EFA6A and EFA6D, termed EFA6As and EFA6D1/2s, respectively, which share the C-terminal region containing a PH domain and coiled coil motifs but lack a catalytic Sec7 domain [27,29]. EFA6As reportedly regulates the dendritic formation of cultured cortical neurons probably through Arf6-independent actin cytoskeleton remodeling [27]. In the present immunoblotting, the anti-EFA6C The OKR performance showing the lack of significant differences in the progression of the adaptation between the two genotypes. The gain was defined as the ratio of the peak-to-peak amplitude of eye movement to that of the screen oscillation. Data are shown as mean ± SEM. ns, no significance using two-way repeated measures ANOVA.
https://doi.org/10.1371/journal.pone.0216960.g005 antibody raised against the C-terminal region of human EFA6C detected an immunoreactive band of 50 kDa in addition to a band for authentic EFA6C. Although the nature of this band is currently unknown, it is possible that a short isoform of EFA6C may exist and function to compensate the lack of EFA6C. Finally, we cannot exclude the possibility that the cerebellar phenotypes may become obvious with aging. For example, mice lacking the prion protein gene showed cerebellar Purkinje cell degeneration, progressive ataxia and impaired motor coordination only after they aged 70-90 weeks [54]. Therefore, follow-up studies are necessary to examine whether EFA6C KO mice develop age-dependent cerebellar phenotypes.
Despite~17% reduction in the spine density of Purkinje cells, EFA6C KO mice did not exhibit behavioral abnormalities in the accelerating rotarod and hOKR tests. Since~60% reduction of Purkinje cell synapses is associated with impaired motor functions in mice lacking GluD2 [39], it is possible that the spine reduction in EFA6C KO mice is below the threshold for these motor phenotypes. While~18% reduction of Purkinje cell spines was reportedly associated with essential tremor in humans [55], neuronal circuits other than parallel fiber-Purkinje cell synapse could be also affected in these patients. Recent functional imaging studies with human subjects and behavioral analyses with genetically modified animals revealed that the cerebellum is engaged not only in motor coordination and learning, but also in higher brain functions such as cognition, perception and emotion [56]. Furthermore, increasing evidence indicates that disturbance in cerebellar development and functions contributes to the pathogenesis of neuropsychiatric disorders such as autism spectrum disorder [57,58]. Therefore, further comprehensive behavioral analyses are necessary to examine whether EFA6C KO mice exhibit other cerebellum-related behavioral abnormalities such as autistic-like behaviors.
Supporting information S1 Fig. Sequences analysis of the EFA6C mutant founder mice. The PCR amplicons from genomic DNAs purified from founder mice were subjected to sequencing. The PAM sequence and insertions are shown in red and blue, respectively. Note that two mice (founder #1 and #2) carried a mono-allelic mutation, six mice (founder #3-#8) carried bi-allelic mutations, and two mice (founder #9 and #10) were mosaic with three or four mutant alleles. (TIF)