ADP Ribosylation Factor 4 (Arf4) Regulates Radial Migration through N-Cadherin Trafficking during Cerebral Cortical Development

Abstract During the development of the cerebral cortex, N-cadherin plays a crucial role in facilitating radial migration by enabling cell-to-cell adhesion between migrating neurons and radial glial fibers or Cajar–Reztius cells. ADP ribosylation factor 4 (Arf4) and Arf5, which belong to the Class II Arf small GTPase subfamily, control membrane trafficking in the endocytic and secretory pathways. However, their specific contribution to cerebral cortex development remains unclear. In this study, we sought to investigate the functional involvement of Class II Arfs in radial migration during the layer formation of the cerebral cortex using mouse embryos and pups. Our findings indicate that knock-down of Arf4, but not Arf5, resulted in the stalling of transfected neurons with disorientation of the Golgi in the upper intermediate zone (IZ) and reduction in the migration speed in both the IZ and cortical plate (CP). Migrating neurons with Arf4 knock-down exhibited cytoplasmic accumulation of N-cadherin, along with disturbed organelle morphology and distribution. Furthermore, supplementation of exogenous N-cadherin partially rescued the migration defect caused by Arf4 knock-down. In conclusion, our results suggest that Arf4 plays a crucial role in regulating radial migration via N-cadherin trafficking during cerebral cortical development.


Introduction
The six-layered neocortex is a unique feature of mammals, responsible for higher brain functions such as cognition, sensory perception, emotion, learning, and memory.
During radial migration, neurons undergo dynamic changes in their cell shapes and migrate in a specific direction by extending processes and sensing environmental cues through cell-cell and cell-extracellular matrix adhesions (Kawauchi, 2015;Peyre et al., 2015;Martinez-Garay, 2020).N-cadherin, a calcium-dependent adhesion molecule of the classical cadherin family, mediates almost every step of radial migration through cell-cell adhesion between migrating neurons and radial glial fibers or Cajar-Reztius cells (Gil-Sanz et al., 2013;Martinez-Garay et al., 2016).Accumulating evidence has revealed that membrane trafficking of N-cadherin plays a crucial role in radial migration by regulating the spatiotemporal expression of surface N-cadherin in migrating neurons (Kawauchi et al., 2010;Wu et al., 2016;Hor and Goh, 2018).However, our understanding of the molecular mechanisms by which membrane trafficking controls N-cadherin-dependent radial migration remains incomplete.
The ADP ribosylation factor (Arf) small GTPases are crucial for regulating membrane trafficking and maintaining organelle integrity.Arfs function as molecular switches that cycle between GDP-bound and GTP-bound states, and their conversion to the GTP-bound state by Sec7 domain-containing guanine nucleotide exchange factors (GEFs) induces conformational changes, allowing them to recruit effector proteins and to activate lipid-modifying enzymes, thereby facilitating various steps of membrane trafficking (D' Souza-Schorey and Chavrier, 2006;Donaldson and Jackson, 2011).The canonical Arf family comprises six members (Arf1-6) in mammals (Kahn et al., 2006), which can be structurally divided into three classes: Class I (Arf1-3), Class II (Arf4-5), and Class III (Arf6).Concerning the roles of Arfs in cortical development, mutations in human genes for ARF1 and its GEF, ARFGEF2, have been linked to periventricular nodular heterotopia, suggesting the involvement of Arf1 in cortical radial migration (Sheen et al., 2004;Gana et al., 2022).Arf6 also regulates multipolar migration, multipolar-to-bipolar transition in the IZ and N-cadherin recycling in migrating neurons in rodents (Falace et al., 2014;Hara et al., 2016).However, the functional roles of Class II Arfs in cortical development are not yet fully understood, as they have been considered supplementary or redundant to Arf1 because of their high sequence similarity and overlapping localization to the Golgi.However, recent evidence suggests that Arf4 has unique functions in cellular processes, such as the transport of rhodopsin in photoreceptors and Notch components in differentiating keratinocytes (Deretic et al., 2005;Ezratty et al., 2016).Genetic deletion of Arf4 in mice results in mid-gestational lethality, likely because of growth retardation by dysfunction of the visceral endoderm (Follit et al., 2014).In terms of neuronal functions, heterozygous deletion of Arf4 in mice results in impaired dentate gyrus-dependent pattern separation with reduced spine density in the dentate gyrus (Jain et al., 2012), whereas Arf4 1/À /Arf5 À/À mice exhibit essential tremor-like behaviors with impaired targeting of Nav1.6 to the axon initial segment in cerebellar Purkinje cells (Hosoi et al., 2019).
To elucidate the role of Class II Arfs in cortical development, we first examined the cellular and subcellular localization of Class II Arfs in the developing cerebral cortex by immunohistological analyses using isoform-specific antibodies.We then examined the effect of Arf4 and Arf5 knockdown on cortical layer formation, cell morphology and Golgi orientation in the migrating neurons in the IZ, and their migration speed in both the IZ and CP.Furthermore, we compared N-cadherin subcellular localization and the morphology of cell organelles between Arf4-knock-down and control migrating neurons.Our results provide the first evidence for specific roles of Arf4 in cortical radial migration through Ncadherin trafficking.

Experimental animals
All experimental procedures in this study were approved by the Animal Experimental and Ethics Committee of The Kitasato University School of Medicine (#2018-138, #2019-138, and #2020-138).Pregnant ICR (Institute of Cancer Research) mice were purchased from The Jackson Laboratory Japan.Mice were maintained in standardized pathogen-free conditions with 12/12 hour (h) light/ dark cycle at room temperature with at libitum access to food and water at the Center for Genetic Studies of Integrated Biological Functions of Kitasato University School of Medicine.Mouse embryos and pups of either sex were used in the experiments.

Cell culture and transfection
To evaluate the efficiency of shRNAs, primary cortical neurons were prepared from E14 mouse embryos, as described previously (Hara et al., 2016).Before plating, cortical neurons were transfected with shRNA plasmids and pCAGGS-EGFP by electroporation (Amaxa Nucleofector 2D, Lonza) according to the manufacturer's protocol.Three days after plating, neurons were subjected to immunoblotting with antibodies against Arf4, Arf5, and a-tubulin.

Antibodies
Antibodies used in this study were summarized in Table 1.An anti-STX16 antibody was raised by immunizing a rabbit with a keyhole limpet hemocyanin (KLH)conjugated 15-aa peptide (CSLDPEAAIGVTKRS), which corresponded to amino acids 61-74 of rat STX16.For characterization of the anti-STX16 antibody, the total lysate of adult mouse brains was prepared as previously (Sakagami et al., 2013).HEK293T cells were transfected with pCAGGS-FLAG-STX16 using polyethylenimine Max (Polyscicences) and harvested with 2Â SDS sample buffer (125 mM Tris-HCl [pH 6.8], 4% SDS, 20% glycerol, 1% sodium deoxycholate, and 10% 2-mercaptoethanol) at 1 d after transfection.After boiling for 5 min, the lysates of brains (10 mg) and HEK293T cells were to immunoblotting with the anti-STX16 antibody and anti-FLAG IgG.

Immunoblotting
Cortical neurons were harvested with a buffer consisting of 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 30 mM MgCl 2 , 1% Triton X-100, 10% glycerol, and a cocktail of protease inhibitor (Roche), and then dissolved with 2Â SDS sample buffer.After boiling at 95°C for 5 min, 10 mg of lysates were electrophoretically separated on SDSpolyacrylamide gels and transferred onto polyvinylidene difluoride (PVDF) membranes.The blots were incubated with antibodies against Arf4, Arf5, or a-tubulin.After incubation with horseradish peroxidase-linked speciesspecific secondary antibody (Table 1), immunoreactive bands were detected using the ECL-Plus Western Blotting Detection kit (Thermo Fisher Scientific) and an image analyzer (GE HealthcareImager 680, Cytiba).The optical density of each immunoreactive band was quantified from three independent blots using Fiji, an open-source image processing software (Schindelin et al., 2012;RRID: SCR_ 002285).

Quantitative analysis
Colocalization coefficient of Arf4, Arf5, or N-cadherin with several organelle markers was analyzed using ZEN software (Carl Zeiss; RRID: SCR_013672).The contours of EGFP-labeled transfected migrating neurons in the upper IZ at E17 were outlined by segment line tool as region of interest (ROI), and colocalization coefficient within ROI was measured by colocalization tool after the threshold was automatically selected (Costes et al., 2004).Data were statistically analyzed using one-way ANOVA with post hoc Tukey-Kramer's test (Tables 2, 3).
The number of EGFP-labeled or mCherry-labeled transfected cells in each cortical zone was counted using Fiji software.Each cortical zone was identified by the following criteria: In E15 brain sections, the VZ and SVZ were visualized by the immunoreactivity for Sox2 and the VZ and SVZ were identified by nuclear shapes, with DAPI staining.In E17 brain sections, the VZ, IZ, and CP were identified by the nuclear density with DAPI staining.In neonatal brain sections, upper cortical plate (uCP), and deep cortical plate (dCP) were identified by the combination of the immunoreactivity for Cux1, a marker for upper cortical layers, and nuclear density by DAPI staining, and the IZ and VZ were further distinguished by DAPI staining as low-cell-density and high-cell-density zones, respectively.Data in each experimental condition were taken using two consecutive sections from four to five individual embryos or neonates in two pregnant mice per group.The percentage of EGFP-labeled or mCherry-labeled cells in each cortical zone was compared with that in the corresponding zone in control animals transfected with control shRNA or indicated plasmids.Data were statistically analyzed using two-way ANOVA with the post hoc Tukey-Kramer's test (Table 3), or Bonferroni test ( Table 3).
The number of cell processes was counted by marking cell processes extending from the cell body and a leading process of EGFP-labeled transfected cells in the upper IZ at E17 using counter tool of Fiji software with sequential images.Data were statistically analyzed using two-way ANOVA with the post hoc Tukey-Kramer's test (Table 3).
The analyses of organelle morphology were performed using Fiji software as follows: EGFP-labeled transfected cells in the IZ were selected by outlining their contours using polygon selection tools.The channel image for an organelle marker was duplicated into a new window, and immunoreactive puncta were extracted by setting thresholds to obtain data using the command "analyze particle."The ratios of total areas of GM130-immunoreactive, TGN38A-immunoreactive, early endosome antigen 1 (EEA1)-immunoreactive, STX12-immunoreactive, Rab11-immunoreactive, STX16immunoreactive, or VAMP4-immunoreactive puncta to those of cell soma were statistically analyzed using Kruskal-Wallis test followed by Dunn's multiple comparison test (Table 3).
The fluorescence intensity for N-cadherin in the cell body was obtained by subtracting that in the nucleus in transfected cells, and normalized by that of control shRNA-transfected cells.Data were statistically analyzed using Kruskal-Wallis test followed by Dunn's multiple comparison test (Table 3).
The contact index was defined as the ratio of the contact length of an EGFP-positive migrating neuron with BLBP-immunoreactive radial glial fibers to the total length of its cell body and leading process observed on a single image, as indicated in Figure 9C.Data were analyzed statistically using Mann-Whitney U test (Table 3).
The length of a leading process was measured by tracing a leading process of EGFP-labeled transfected cells in the IZ at E17 from the distal tip to the proximal base at the cell body using segmental line tools of ZEN software with stacked images.Data were analyzed statistically using Mann-Whitney U test (Table 3).Statistical analyses in this study were performed using the GraphPad Prism9.0 for Macintosh (GraphPad Software; RRID: SCR_002798).

Class II Arfs exhibit overlapping but distinct expression in the developing cerebral cortex
A previous in situ hybridization study has shown that Arf4 and Arf5 mRNAs are substantially expressed in the developing rat brain (Suzuki et al., 2001).To examine the expression of Arf4 and Arf5 proteins in developing cerebral cortices, we performed immunohistological analyses of the mouse cerebral cortex using specific antibodies against Arf4 and Arf5 (Hosoi et al., 2019).In the dorsal pallium at E17 (Fig. 1A,B), both proteins were expressed throughout the cerebral zones, including the VZ, IZ, and CP (Fig. 1A,B).We further performed double immunofluorescence using antibodies against Class II Arfs and microtubule-associated protein-2 (MAP2) for postmigratory neurons, neurofilament (NF) 165 for axons, the polysialylated neural adhesion molecule (PSA-NCAM) for immature neurons, and the brain lipid-binding protein (BLBP) for radial glia.Both Arf4 and Arf5 were expressed prominently in cell bodies and proximal processes of MAP2-positive postmigratory neurons in the CP (Fig. 1C,  D) and PSA-NCAM-positive migrating neurons in the IZ (Fig. 1G,H), and BLBP-positive radial glia in the VZ (Fig. 1I,J).In the IZ, intense immunofluorescence for Arf5, but not for Arf4, was observed in the axon bundle labeled by NF165, presumably corresponding to developing fibers projecting to subcortical regions (Fig. 1F, arrowheads).These results suggest that both Arf4 and Arf5 are widely expressed in the developing cerebral cortex.
To further examine the subcellular localization of Arf4 and Arf5, migrating neurons in the IZ at E17 were visualized by expressing EGFP using IUE at E14 and subjected to                   2, 3), suggesting that Class II Arfs mediate various steps of membrane trafficking in migrating neurons.
Furthermore, immunofluorescence staining with Tuj1 (Class III b -tubulin) and Cux1, differentiation markers for neurons and cortical Layers II-IV excitatory neurons, respectively, revealed that Arf4-knock-down cells in the IZ were immunoreactive for Tuj1 at E17 and Cux1 at P0 to the same extent as surrounding neurons and cortical Layer II/III neurons, respectively (Fig. 4D,E, arrows), suggesting that Arf4 knock-down did not affect neuronal differentiation.
Finally, we examined the distribution of Arf4 knockdown cells in the P10 cerebral cortex that had been electroporated with shArf4#1 and pCAGGS-EGFP at E14. Arf4-knock-down neurons were still observed in the lower cortical layer at P10 compared with the control, suggesting that Arf4 knock-down led to a permanent defect in radial migration (Fig. 4F).
We also examined whether knock-down of Arf4 or Arf5 regulates the transition from multipolar to bipolar morphology in neurons migrating in the IZ by classifying transfected cells as multipolar, round, and bipolar shapes.At E16, most of control migrating cells in the lower IZ exhibited multipolar morphology with multiple processes extending from the cell bodies in various directions, and transformed to bipolar morphology with a leading processes extending toward the pial direction at the upper IZ at E17 (Fig. 5H,J) Quantification of the cell morphology revealed that there were no significant differences in the proportion of cell shapes among the control, shArf4#1transfected, and shArf5-transfected migrating cells in the IZ at E16 and E17 (Fig. 5H,J; Table 3).However, it should be noted that Arf4-knock-down bipolar cells in the upper IZ at E17 possessed numerous filopodia-like, fine, short processes extending from their cell bodies and leading processes (Fig. 5I, arrows).Quantification revealed that Arf4#1-transfected cells possessed more short processes extending from the cell body than the control or shArf5-transfected cells (Fig. 5K; Table 3).These findings suggest that Arf4 regulates the Golgi polarization and cell morphology, although it is not involved in multipolar-to-bipolar morphologic transition.
Furthermore, time-lapse imaging of an organotypic brain slice culture from E17 embryos electroporated at E14 revealed that Arf4 knock-down significantly reduced the speed of multipolar migration in the lower IZ at E17, compared with that of the control (Fig. 6A,B; Table 3, Control: 8.8 6 2.2 mm/h, n ¼ 30 cells, 3 embryos from 2 pregnant mice; shArf4#1: 4.6 6 1.6 mm/h, n ¼ 30 cells, 3 embryos from 2 pregnant mice, p , 0.0001).These results suggest that Arf4 also regulates cell motility during multipolar migration in the IZ.
Arf4, but not Arf5, also regulates locomotion in the CP Because Arf4 knock-down reduced the proportion of cells that reached the upper CP at P0 (Fig. 4B), we examined the effect of Arf4 knock-down on radial migration behaviors in the CP using time-lapse imaging of an organotypic brain slice culture.Arf4 knock-down significantly reduced the speed of locomotion toward the pia, compared with that in the control (Fig. 6C,D; Table 3; Control: 16.1 6 4.6 mm/h, n ¼ 30 cells, 3 embryos from 2 pregnant mice; shArf4#1: 9.4 6 3.9 mm/h, n ¼ 30 cells, 3 embryos from 2 pregnant mice, p , 0.0001), suggesting that Arf4 regulates cell motility during locomotion in the CP as well as multipolar migration in the IZ.

Knock-down of Arf4 or Arf5 affects the morphology of the Golgi and endosomes in migrating neurons
We examined whether knock-down of Arf4 or Arf5 affected the morphology of organelles related to secretory and endocytic pathways, including the Golgi, TGN, and endosomes, in migrating neurons by immunostaining with antibodies against GM130, STX16, VAMP4, STX12, Rab11, and EEA1.The specificity of an anti-STX16 antibody was confirmed by immunoblotting in which the antibody detected an immunoreactive band of 45-48 kDa in the lysates of mouse brains and HEK293T cells transfected with FLAG epitope-tagged STX16 (Fig. 7A).The immunoreactivity of STX16 was detected in the entire E17 cerebral cortex (Fig. 7B) and migrating neurons visualized by mCherry (Fig. 7D), which was completely attenuated by preabsorption of the antibody with STX16 (Fig. 7C,E).Furthermore, the antibody labeled punctate structures partially overlapped and/or closely associated with TGN38A in migrating neurons visualized by EGFP in the upper IZ (Fig. 7F).These findings suggested the specificity of the newly generated anti-STX16 antibody.
Since N-cadherin mediates cell-cell adhesion between radially migrating neurons and radial glial fibers (Kawauchi et al., 2010;Martinez-Garay et al., 2016), we examined the effect of Arf4 knock-down on their interactions by calculating the contact index, which was defined by the ratio of contact length of EGFP-positive transfected migrating neurons with BLBP-immunoreactive radial glial fibers to the total length of their cell bodies and leading processes as shown in Figure 9C.Arf4 knock-down significantly decreased the contact index by 33%, compared with that of control shRNA (Fig. 9C; Table 3; Control: 0.6 6 0.3, n ¼ 43 cells; shArf4#1: 0.4 6 0.3, n ¼ 41 cells, p ¼ 0.0004).Furthermore, we examined the effect of Arf4 knock-down on the length of leading processes.However, there were no significant differences in leading process length between control and shArf4#1transfected neurons (Fig. 9D; Table 3; Control: 26.7 6 9.8 mm, n ¼ 126 cells; shArf4#1: 25.5 6 11.9 mm, n ¼ 178 cells, p ¼ 0.1696).These results suggest that Arf4 regulates N-cadherin-mediated interaction with radial glial fibers in migrating neurons.

Discussion
In this study, we investigated the impact of Class II Arfs on cortical radial migration using IUE.Our results demonstrated that knock-down of Arf4, but not Arf5, led to an accumulation of transfected neurons in the IZ and dCP with disturbance in the Golgi orientation in the lower IZ, cell-cell adhesions between migrating neurons and radial fibers in the upper IZ, and cell motility during multipolar migration in the IZ and locomotion in the CP.The stalling of shArf4-knock-down neurons in the IZ was rescued by coexpressing shRNA-resistant Arf4, but not Arf5, despite the high similarity (;90%) between the two proteins at the amino acid level (Volpicelli-Daley et al., 2005).These findings suggest that Arf4 has specific and nonredundant roles in radial migration.Since Arf4 was expressed in both radial glia and migrating neurons, we were unable to definitively conclude which cell type(s) (migrating neurons, radial glial cells, or both) is primarily responsible for the migration defects caused by Arf4 knock-down in this study.Our attempts to express dominant active or negative Arf4 mutants specifically in postmitotic migrating neurons under the control of the NeuroD promoter were unsuccessful because of the induction of apoptosis.However, we found that expression of shArf4 did not significantly affect cell cycle progression or the delamination of neural progenitor cells in the VZ.Furthermore, we failed to observe apparent morphologic abnormalities in radial glial fibers extending from the VZ to the pia (data not shown).Therefore, we believe that the migration defects caused by Arf4 knock-down primarily result from Arf4 dysfunction in migrating neurons, which should be confirmed in future studies by conditionally deleting the Arf4 gene in migrating cortical neurons using Arf4floxed mice.
To gain insights into the role of Class II Arfs in migrating neurons, we first conducted immunohistological analyses to examine the subcellular localization of Class II Arfs.We found that both Arf4 and Arf5 were present in various organelles, including the Golgi apparatus (GM130), trans-Golgi network (TGN38A), retrograde transport vesicles to the TGN (VAMP4), and recycling endosomes (STX12), indicating the involvement of Class II Arfs in multiple membrane trafficking pathways.Furthermore, we observed that knock-down of Arf4 and Arf5 had overlapping but distinct effects on organelle morphology and distribution in migrating neurons.Knock-down of either Arf4 or Arf5 affected the sizes of GM130-immunoreactive, STX16-immunoreactive, and TGN38A-immunoreactive structures.Since Class II Arfs regulate vesicular transport from the Golgi to ER and within the Golgi through the recruitment of COPI (Hamlin et al., 2014), AP1 (Lowery et al., 2013), and GGAs (Lowery et al., 2013), the enlargement of the Golgi likely resulted from an imbalance between the influx and efflux caused by Class II Arf knock-down.Additionally, in HeLa cells, simultaneous knock-down of Arf1 and Arf4 was shown to inhibit retrograde transport of TGN38/46 from endosomes to the TGN (Nakai et al., 2013).On the other hand, knock-down effects on the size of VAMP2-immunoreactive, STX12-immunoreactive, and Rab11-immunoreactive puncta were specific for Arf4.VAMP4 and STX16 are SNARE partners that localize on transporting vesicles and their target TGN membrane and regulate retrograde transport to the TGN (Laufman et al., 2011), whereas STX12 is a component of the SNARE complex that localizes primarily on recycling endosomes (Prekeris et al., 1998) and Rab11 is a critical small GTPase for the recycling pathway to the plasma membrane (Ullrich et al., 1996).Therefore, it is tempting to speculate that Arf4 plays a distinct role from Arf5 in radial migration by controlling the balance of membrane trafficking in and out of the TGN via retrograde transport vesicles to the TGN and recycling endosomes from the TGN to the plasma membrane.However, it is also possible that Arf4 regulates neuronal migration by regulating the secretory pathway in the Golgi apparatus in an Arf5-independent manner.Further studies are needed to clarify these mechanisms.
Concerning cargo proteins that Arf4 regulates in migrating neurons, we demonstrated that knock-down of Arf4 resulted in the accumulation of N-cadherin in STX16-positive, VAMP4-positive, and TGN38-positive structures in migrating neurons, suggesting that Arf4 controls trafficking of de novo synthesized or endocytosed N-cadherin around the TGN.N-cadherin is a critical cell adhesion molecule that regulates various processes of radial migration, including those involved in cell proliferation and neurogenesis of radial glial progenitor cells in the VZ (Gil-Sanz et al., 2014), glial-independent somal translocation of early-born neurons (Franco et al., 2011), multipolar migration and multipolar-to-bipolar transition in the IZ (Jossin and Cooper, 2011), locomotion along radial glial fibers (Kawauchi et al., 2010), and glia-independent terminal translocation of late-born neurons in the uCP.We demonstrated that knock-down of Arf4 disturbed the Golgi orientation, cell-cell contact of bipolar neurons in the upper IZ, and cell motility during multipolar migration and locomotion, which were largely consistent with the phenotypes caused by N-cadherin dysfunctions.Furthermore, coexpression of N-cadherin with shArf4 partially rescued the migration defect caused by Arf4 knock-down.Taken together, these results suggest that Arf4 plays an important role in regulating radial migration by mediating the trafficking of N-cadherin from the TGN to the plasma membrane.
We have previously reported that Arf6 regulates multipolar migration through N-cadherin (Hara et al., 2016).However, knock-down effects of Arf4 and Arf6 differ in the morphology of organelles and subcellular localization of N-cadherin in migrating neurons.Arf6 knock-down led to the cytoplasmic accumulation of STX12-positive recycling endosomes in migrating neurons and disrupted the recycling of N-cadherin to the cell surface in cultured cortical neurons (Hara et al., 2016).On the other hand, Arf4 knock-down altered the morphology and distribution of various organelles including the Golgi, TGN, retrograde transport vesicles, and recycling endosomes, and induced accumulation of N-cadherin on the TGN and surrounding vesicles.Therefore, it is suggested that Arf4 and Arf6 regulate distinct steps of N-cadherin trafficking to the plasma membrane in migrating neurons.
It should be noted that the migration defect caused by Arf4 knock-down is not completely rescued by the coexpression of N-cadherin.In addition, knock-down of Arf4 did not affect multipolar-to-bipolar morphologic transition or leading process length, which was inconsistent with the previous findings observed by disruption of N-cadherin functions (Kawauchi et al., 2010;Martinez-Garay et al., 2016).These findings suggest that that Arf4 may regulate radial migration by trafficking other cargo proteins with Ncadherin.For instance, the b -amyloid precursor protein (b -APP) could be an attractive candidate for cargo regulated by the Arf pathway in migrating neurons.b -APP is a Type I transmembrane glycoprotein associated with the pathogenesis of familial Alzheimer's disease and can function as an adhesion molecule that interacts with the APP family proteins and extracellular matrix proteins, such as heparan sulfate proteoglycans, laminin, collagen, and F-spondin (Narindrasorasak et al., 1991(Narindrasorasak et al., , 1992(Narindrasorasak et al., , 1995;;Hoe and Rebeck, 2008).Notably, APP knockdown was previously shown to inhibit neuronal migration into the CP (Young-Pearse et al., 2007), similar to the phenotype induced by Arf4 knock-down.Furthermore, trafficking of APP to the cell surface and its localization to the Golgi/TGN are regulated in an Arf-dependent manner through the interaction of APP with Munc18-interacting proteins (MINTs) and phosphotyrosine binding (PTB) domain-containing coat proteins (Hill et al., 2003).Because MINTs can interact directly with GTP-bound Arf4 and function as a downstream effector of Arf4 (Hill et al., 2003), it is tempting to speculate that Arf4 may regulate radial migration by trafficking APP to the cell surface through interaction with MINTs.Another possible mechanism is Arf4-mediated ciliary transport.Arf4 has been proposed to mediate the sorting and transport of ciliary proteins from the TGN to the primary cilium (Deretic et al., 2005).Additionally, Arf4 plays a role in the trafficking of Notch components, such as Notch2 and presenilin-2, to basal bodies and/or primary cilia to promote epidermal differentiation (Deretic et al., 2005).The role of primary cilia in radial migration is still debated, but previous studies have identified 30 ciliopathy-related genes that impact cerebral cortex development, with knock-down of 17 genes resulting in disturbance of distinct steps of radial migration, including a transient multipolar stage in the lower IZ, multipolar-to-bipolar transition in the upper IZ, and radial glia-guided locomotion in the CP (Guo et al., 2015).It is therefore plausible to hypothesize that Arf4 regulates radial migration by facilitating the ciliary transport of these ciliopathy-related gene products.However, further investigation is required to confirm this hypothesis.
Lastly, mutations in the human genes for ARFGEF2 and ARF1 have been associated with cortical malformations, including periventricular nodular heterotopia and microcephaly, indicating that the ARFGEF2-Arf1 pathway is critical for cerebral cortical development (Sheen et al., 2004;Gana et al., 2022).Notably, a recent study reported a crosstalk cascade between Class II Arfs and Arf1 in the TGN: GBF1, a GEF for Class II Arfs, activates Arf4 and Arf5 at the TGN, where the resultant GTP-bound Class II Arfs interact with and recruit ARFGEF1/2 (Lowery et al., 2013), thereby initiating Arf1-dependent protein sorting and vesicle budding at the TGN.Therefore, it is attractive to speculate that Arf4 functions upstream of ARFGEF2-Arf1 signaling in radial migration.Further elucidation of the mechanisms by which Arf4 regulates radial migration may provide additional clues to understand the role of the ARFGEF2-Arf1 pathway in pathogenesis of human cortical malformation.

Figure 1 .
Figure number Panel Comparison

Figure 3 .
Figure3.Characterization of shRNAs against Arf4 and Arf5.A, B, Immunoblotting.Primary cortical neurons were transfected with control or indicated shRNA vector and subjected to immunoblotting (IB) with antibodies against Arf4, Arf5, and a-tubulin (TubA).The graphs show the relative expression of Arf4 and Arf5 to that of TubA.C-H, Representative immunofluorescence images showing the effect of control shRNA (C, F), shArf4#1 (D, G), or shArf5 (E, H) on endogenous expression of Arf4 (C 1 -E 1 ) and Arf5 (F 1 -H 1 ) in migrating neurons in the IZ.Sections of the E17 cerebral cortices that had been electroporated with indicated shRNAs and mCherry at E14 were subjected to double immunofluorescence with antibodies against Arf4 (C 1 -E 1 ) or Arf5 (F 1 -H 1 ) and mCherry (C 2 -H 2 ).Note the decrease in endogenous expression of the respective Arf protein by transfecting with shArf4#1 without compensatory upregulation of the other.Asterisks indicate the nuclei of transfected cells.Data were presented as mean 6 SD and statistically analyzed by one-way ANOVA with post hoc Tukey-Kramer's test (***p , 0.0005) in A or unpaired Student's t tests in B (*p , 0.05, n.s., not significant).Scale bar: 10 mm in H 3 .

Figure 5 .
Figure 5. Knock-down of Arf4 disturbs the Golgi orientation, but not multipolar-to-bipolar morphologic transition in the IZ.A, B, Representative immunofluorescence images of the ventricular zone at E15 at 1 d after electroporation with indicated shRNA and EGFP.Note that ventricular cells transfected with shArf4#1 (A) or shArf5 (B) exhibited reduced endogenous expression of the respective Arf.Asterisks in A and B indicate the nuclei of transfected cells.C, Representative micrographs of E15 cerebral cortices electroporated with control, shArf4, or shArf5 plasmids plus pCAGGS-mCherry.The VZ, SVZ, and IZ were divided by the immunoreactivity for Sox2 and nuclear density.Yellow bars indicate border between the VZ and SVZ.The graph shows the percentage of transfected cells in each zone.D, Representative micrographs showing the effect of knock-down of Arf4 or Arf5 on BrdU incorporation in the VZ at E15.Embryos were electroporated with indicated shRNA plasmids plus pCAGGS-mCherry at E14, killed at E15 after BrdU administration, and immunostained with antibodies against mCherry (magenta) and BrdU (green).Arrows indicate the BrdU incorporation in transfected cells.Yellow bars indicate border between the VZ and SVZ.The graph shows the percentage of BrdU-incorporated cells in total transfected cells.E, Representative micrographs showing the effect of knock-down of Arf4 or Arf5

Figure 6 .
Figure 6.Knock-down of Arf4 reduces migration speed in the IZ and CP.A, C, Representative time-lapse images of transfected neurons migrating in the lower IZ (A) and CP (C) at E17.Embryos were electroporated with the indicated shRNA plus pCAGGS-FloxP-EGFP, pCAGGS-FloxP-EGFP-F, pCAGGS-FloxP-mCherry-NLS, and pCAGGS-Cre recombinase at E14, and brains were subjected to slice culture and time-lapse observation at E17. Arrowheads indicate the nuclei of transfected neurons.B, D, Quantification of the migration speed of neurons transfected with indicated shRNAs in the lower IZ (B) and CP (D).Note the reduction of the migration speed in shArf4-transfected cells in the lower IZ and CP, compared with that in control cells.Data were presented as mean 6 SD and statistically analyzed using Mann-Whitney U test (****p , 0.0001).Scale bars: 30 mm in A and C.

Figure 7 .
Figure 7. Characterization of an anti-STX16 antibody.A, Characterization of an anti-STX16 antibody.The lysates of the mouse brain and HEK293T cells transfected or untransfected with pCAGGS-FLAG-STX16 were subjected to immunoblotting with antibodies against anti-STX16 (left) or anti-FLAG (right).B-E, Representative micrographs of STX16 immunoreactivity in E17 cerebral cortices (B, C) and migrating neurons at a high magnification (D, E).Sections of E17 cerebral cortices that had been transfected pCAGGS-mCherry at E14 were immunostained with antibody against mCherry (D 1 , E 1 ) and STX16 (B, D 2 ) or antibody preabsorbed with STX16 (C, E 2 ).F, Subcellular localization of STX16 in migrating neurons.Sections of E17 cerebral cortices that had been electroporated with pCAGGS-mCherry were immunostained with antibodies against STX16 (F 1 , green), TGN38A (F 2 , magenta), and mCherry (F 4 , blue).Note the partial colocalization of STX16 with TGN38A at the juxtanuclear region.An asterisk indicates the nucleus of a transfected migrating neuron in the IZ.Scale bars; 200 mm in C, 10 mm in E and F.

Figure 8 .
Figure 8. Knock-down of Arf4 or Arf5 affects the morphology and distribution of the Golgi and endosomes in migrating neurons.A-G, Representative micrographs showing the morphology of organelles and endosomes in transfected neurons in the IZ at E17. Cerebral cortices electroporated with indicated shRNAs and pCAGGS-mCherry at E14 were fixed at E17 and subjected to double immunofluorescence staining with antibodies against mCherry (magenta) and marker proteins (green), including GM130 (A), STX16 (B), VAMP4 (C), STX12 (D), Rab11 (E), TGN38A (F), or EEA1 (G).The lower graphs show the quantification of the immunoreactive areas for markers normalized by the cell body area.Note that Arf4 knock-down specifically reduced the immunoreactive areas for VAMP4, STX12, and Rab11, whereas either Arf4 or Arf5 knock-down increased the immunoreactive areas for TGN38A and STX16.Broken lines indicate the contour of transfected cells visualized by mCherry immunofluorescence.Data were presented as mean 6 SEM and statistically analyzed using Kruskal-Wallis test followed by Dunn's multiple comparison test (*p , 0.05, ***p , 0.005, ****p , 0.0001, n.s., not significant).Scale bars: 10 mm in A-G.

Figure 9 .
Figure 9. Arf4 regulates neuronal migration through N-cadherin trafficking.A, Representative micrographs showing the effect of knock-down of Arf4 or Arf5 on the subcellular localization of N-cadherin in migrating neurons.Brains were electroporated with shArf4 or shArf5 plus mCherry, killed at E17, and subjected to immunostaining with antibodies against N-cadherin (Ncad; green) and mCherry (magenta).Arrows indicate intracellular N-cadherin in shArf4#1 transfected cells.The lower graph shows the relative immunofluorescence intensity of intracellular N-cadherin to that of the control.Note the cytoplasmic accumulation of N-cadherin in migrating neurons transfected with shArf4#1, but not shArf5, in the IZ at E17. B, Representative images of triple immunofluorescence staining of Arf4-knock-down neurons in the IZ with antibodies against N-cadherin (green), STX16 (magenta), TGN38A (magenta), or VAMP4 (magenta), and EGFP (blue).Boxed areas show the colocalization of N-cadherin with STX16, TGN38A, or VAMP4 in transfected neurons at a high magnification.Broken lines indicate the contour of transfected cells visualized by EGFP immunofluorescence.C, Representative images of bipolar neurons migrating along BLBP-immunoreactive radial fibers (magenta) in the upper IZ.Arrowheads indicate the contact of EGFP-positive transfected neurons with BLBP-positive radial fibers.The cartoon illustrates the contact index, which is calculated by dividing the length of the contact between an EGFP-positive migrating neuron and BLBPimmunoreactive radial glial fibers by the total length of the cell body and leading process of the transfected neuron observed on a single image.A graph shows the quantification of the contact index.Note that Arf4 knock-down significantly reduced the cell-cell adhesion between migrating neurons and radial glial fibers.D, Representative images of the morphology of bipolar neurons with a leading process in the upper IZ at E17. Arrowheads indicate the distal tip and proximal base of leading processes of transfected bipolar cells.A graph shows the effect of Arf4 knock-down on the length of leading processes.E, Representative micrographs continued showing the effect of coexpression of N-cadherin with shArf4 on the migration defect caused by Arf4 knock-down.Embryos were subjected to in utero electroporation with shArf4#1 or shArf4#1 plus pCAGGS-N-cadherin (CAG-Ncad) at E14 and killed at P0.The graph shows the quantification of the distribution of EGFP-positive cells in cortical zones.Data were presented as mean 6 SD and statistically analyzed using Kruskal-Wallis test followed by Dunn's multiple comparison test (A, ****p , 0.0001), Mann-Whitney U test (C, D, ***p , 0.005, n.s.not significant), or two-way ANOVA followed by Bonferroni's multiple comparison test (E, *p , 0.05, ***p , 0.001).Scale bars: 10 mm in A-C, 200 mm in D.

Table 1 :
List of antibodies used in this study

Table 3 :
Summary of statistical analyses