Planar Cell Polarity Protein Vangl2 Interacts With Protein Ap2m1 to Regulate Dendritic Branching in Cortical Neurons

Van Gogh-like 2 (Vangl2) is a mammalian homolog of Drosophila core planar cell polarity (PCP) protein Vang/Strabismus, which organizes asymmetric cell axes for developmental proliferation, fate determination, and polarized movements in multiple tissues, including neurons. While the PCP pathway has an essential role for dendrite and dendritic spine formation, the molecular mechanism remains to be claried. To investigate the mechanism of Vangl2-related neuronal development, we screened for proteins that interact with the Vangl2 cytosolic N-terminus from postnatal day 9 mouse brains using a yeast two-hybrid system. From 61 genes, we identied adaptor-related protein complex 2, mu 1 subunit (Ap2m1) as the Vangl2 N-terminal binding protein. Intriguingly, however, the pull-down assay demonstrated that Vangl2 interacted with Ap2m1 not only at its N-terminus but also at the C-terminal Prickle binding domain. Furthermore, we veried that the downregulation of Ap2m1 in the developing cortical neurons reduced the dendritic branching similar to what occurs in a knockdown of Vangl2. From these results, we suggest that the membrane internalization regulated by the PCP pathway is required for the developmental morphological change in neurons.


Introduction
Van Gogh (Vang), also known as Strabismus (Stbm), was originally identi ed in Drosophila as a core planar cell polarity (PCP) protein, which when mutated, causes considerable misorientation of organized epithelial structures (Torban et al., 2004;Tissir & Go net, 2013). Vang/Stbm is evolutionarily conserved, and can be seen in several species including insects, sh, and mammals. Mice have two Vang/Stbm family members: Van Gogh-like protein (Vangl) 1 and Vangl2, whose mRNAs are expressed in the developing and adult nervous system (Tissir & Go net, 2006). Mutations in VANGL genes have been identi ed in sporadic and familial cases of neural tube defects in humans, and similar defects have been found in the looptail (Lp) mouse mutant that has mutations in Vangl2.
Vangl2 is a membrane protein comprising four transmembrane domains and two intracellular domains, one each at the amino (N)-and carboxyl (C)-terminals. The C-terminus has multiple domains including the PDZ-binding motif, the Prickle binding domain, and some missense mutations that cause the Lp mutant (Torban et al., 2004). Indeed, two independent Lp mutations in Vangl2 have been shown to impair interactions with Dvl proteins, which suggests that Vangl2-Dvl-mediated signaling underlies the neural tube defect (Torban et al., 2004). In the nervous system, Vangl2 regulates commissural axon growth-cone guidance by antagonizing Dvl-mediated signaling (Shafer et al., 2011). Additionally, the Vangl2 PDZbinding motif is tightly associated with postsynaptic density (PSD)-95 protein. They form a protein complex with NMDA receptors, Prickle2, and N-cadherin, which regulate the clustering of postsynaptic molecules and dendritic spine formation (Yoshioka et al., 2013;Nagaoka, Ohashi, et al., 2014;Nagaoka et al., 2015;Nagaoka & Kishi, 2016). Intriguingly, the Vangl2 C-terminal deletion mutant was shown to enhance dendritic branching, while the N-terminus deletion mutant reduced both spine density and dendritic branching (Hagiwara et al., 2014), which indicates that Vangl2 has an essential role in modulating the constitution of neuronal morphology.
Vangl2 plays critical roles in early neural development and axon/dendritic branching. However, despite these broad-ranging functions, the functional domains and proteins that it interacts with (other than PCP proteins) have not been well studied. Unlike the well-characterized domains of the C-terminal region, which include a region that interacts with PCP partner proteins (Bailly et al., 2018), the roles of the Nterminal region remain poorly characterized. Here, we identi ed proteins that interact with the N-terminus of Vangl2 (Vangl2N) using yeast two-hybrid screening. Following the pull-down assay, we found proteins that interact with Vangl2: adaptor-related protein complex 2, mu1 subunit (Ap2m1), eukaryotic translation elongation factor 1 α1 (Eef1a1), and Ras/Rap GTPase-activating protein SynGAP (SynGAP1). Ap2m1 was found to be the most likely to interact with Vangl2N, binding to both the N-terminus and the Cterminal Prickle binding domain. Remarkably, knockdown (KD) of Ap2m1 resulting in less branching of cortical neurons, similar to what happens in the KD of Vangl2. Because the AP2 protein regulates membrane internalization via clathrin-mediated endocytosis, the Vangl2-involved PCP signal coordinates the temporal and spatial regulation of neuronal morphology.

Screening of novel proteins that interact with Vangl2
To search for novel Vangl2 binding proteins, we conducted yeast two-hybrid screening with the N-terminal region (amino acid 1-114, Vangl2N) as bait, and mated it with a postnatal day 9 (P9) mouse brain cDNA library containing the Y187 yeast strain (Fig. 1A, B). The resulting diploid cells were screened on different stringency plates, and approximately 4.5 × 10 7 diploid cells were cultured. Among 213 positively identi ed clones, we identi ed 61 genes by sequencing and BLAST searches (Fig. 1B, Table 1). To investigate how the N-terminal region of Vangl2 regulates dendrites and/or dendritic spines formation, we selected 10 proteins from those that were identi ed, choosing them based on proteomic data from postsynaptic density (PSD) proteins in the P9 mouse cortex (Shao et al., 2017) (Table 1 indicated by *).
Among them, two housekeeping proteins, one nuclear protein, and two presynaptic proteins were excluded from further analysis. As for the remaining ve proteins, Ap2m1, ATPase Na + /K + transporting subunit β1 (Atp1b1), Eef1a1, kinesin family member 1A (Kif1a), and SynGAP1, we examined how they interacted with the N-terminal region of Vangl2. We performed pull-down assays using puri ed GST-Vangl2N and GST-Vangl2C (amino acid 242-521) with cell lysates from HEK293 cells that expressed amino-terminally FLAG-tagged protein. Cell lysates expressing FLAG-Ap2m1 or FLAG-SynGAP1 were mostly retained on GST-Vangl2N, with fewer retained on GST-Valng2C. Conversely, cell lysates expressing Eef1a1 interacted with both GST-Vangl2N and GST-Vangl2C, while those expressing Atp1b1 and Kif1a were retained the least (Fig. 1C, Supplementary Fig. 1, Table 2). Considering the amount of pulled down FLAG-tagged protein, we focused on Ap2m1 for further analysis.  was su cient for binding to the N-terminal region of Vangl2, we performed pull-down assays using puri ed GST-Ap2m1 (195-367aa) with cell lysates from HEK293 cells that expressed the amino-terminally HA-tagged Vangl2 deletion mutant. Cell lysates from HEK293 cells that expressed HA-Vangl2 or Vangl2 lacking the C-terminal region (Vangl2ΔC) were retained on GST-Ap2m1 (195-367aa) ( Fig. 2A, B). The interaction of the Ap2m1 fragment with Vangl2 was also observed in HEK293 cells co-transfected with EGFP-tagged Ap2m1 fragments and HA-tagged Vangl2 (Fig. 2C). These results indicated that the region of Ap2m1 that encodes the amino acids from 195 to 367 was su cient for the interaction with Vangl2.

The Functional Role Of Ap2m1 In Neuronal Dendritic Branching
From the yeast two-hybrid screening, we found an association between Vangl2 and the adaptor-related protein complex, AP2. To evaluate the role of AP2 on the morphology of neurons, an shRNA construct was designed to downregulate endogenous Ap2m1. The e ciency of Ap2m1 shRNA was rst con rmed by transfection in HEK293 cells (Fig. 4A, B, Supplementary Fig. 1). Next, we electroporated the Ap2m1 shRNA construct with EGFP into embryonic cortical neurons (E14. 5-15.5). At P21, we characterized the morphology of cortical pyramidal neurons identi ed with EGFP. The total length of apical and basal dendrites was signi cantly lower in Vangl2 and Ap2m1 KD neurons than that in control with scramble shRNA (Fig. 4C, D, Table 3). Conversely, spine density was higher in Ap2m1 KD neurons compared to the comtrol and Vanlg2 KD (Fig. 4E). The total number of tips was signi cantly lower in Vangl2 and Ap2m1 KD neurons than control; branches at the secondary dendrite were especially devoid of tips (Fig. 4F, G). These results suggest an association between the Vangl2 portion of PCP proteins and an adaptor protein for membrane internalization that might have a role in the dendritic remodeling in neurons, which can affect neuronal development and plasticity. Values are means ± SEM.

Discussion
Vangl2 is a core component of PCP proteins, which are involved in cell-signaling that acts to divide and move cells along the tissue axis. Vangl2 comprises four transmembrane regions with both terminals exposed to the cytosol ( . Therefore, in this report, we searched for molecules that interact with Vangl2N to better understand the molecular mechanism underlying neural dendritic development. From the Y2H screening, we obtained 61 candidate genes including pre-and post-synaptic proteins (Table 1) and con rmed the Vangl2N speci c binding to adaptor protein Ap2m1 (Fig. 1, 2, 3). Furthermore, shRNA KD of Ap2m1 and KD of Vangl2N resulted in less dendritic branching, although spine density increased after Ap2m1 KD (Fig. 4).
From these results, we suggest that dendritic development associated with the clathrin-mediated endocytosis could be directed by the PCP pathway. C-terminus have been investigated in terms of neuronal morphology. However, even though the Nterminus has an essential role in neuronal dendritic development, interaction at the N-terminal region has not been well demonstrated (Hagiwara et al., 2014).
In mice, cortical layer 2/3 neurons have immature dendrites at P3, and apical and basal dendrites become longer at P5 and more complex by P15 (Hoshiba et al., 2016). Because Vangl2 expression gradually decreases during this dendritic morphogenesis during the rst 1-2 weeks (Yoshioka et al., 2013), we screened molecules interacting with the N-terminal region of Vangl2 using P9 mouse brain cDNA, and then identi ed Ap2m1 as a binding partner. In the pull-down assay using GST fused Vangl2 and Ap2m1expressing HEK293T lysates, Ap2m1 strongly bound to Vangl2N and weakly bound to Vangl2C (Fig. 1C and 3B). Consistent with these data, in the pull-down assay using GST fused Ap2m1 and Vangl2expressing HEK293T lysates, Ap2m1 bound to not only Vangl2ΔC but also Vangl2ΔN (Fig. 2B). These results suggested Ap2m1 preferentially bound to the N-terminus of Vangl2, although it bound to both the N-and C-terminal regions. A previous study showed that Ap1m1, an AP-1 subunit, binds to the C-terminal YYXXF motif of Vangl2, which is required for Vangl2 transport from the trans-Golgi network with the GTPbinding protein, Arfrp1 (Guo et al., 2013). In the present study, we showed that compared with GST-Vangl2-4 (corresponding to PKBD), Ap2m1 preferably interacted with the Vangl2 C-terminal, and hardly interacted with GST-Vangl2-3, which contains a YYXXF motif (279-283aa) (Fig. 3B). Therefore, the functional role of the interaction between AP-2 and Vangl2C might differ from that between AP-1 and , the Wnt/PCP pathway is essential for the formation of the dendrite, dendritic spine, and excitatory synapse. In this study, we showed that Ap2m1 KD also decreased dendritic length, the total number of tips, and the branch number in cortical neurons and the Vangl2 KD (Fig. 4). Clathrinmediated endocytosis (CME) regulated dendritic growth by mediating the internalization of receptors. Indeed, KD of the clathrin assembly protein CALM reduced dendrite length and the overall complexity of the neurites in cultured hippocampal neurons (Bushlin et al., 2008). KD of AP2b1, a subunit of the AP-2 complex, lowered the total number of dendritic tips in rat hippocampal neurons (Koscielny et al., 2018).
Consistent with these results, the direct interaction between Vangl2 and Ap2m1 suggests that the Wnt/PCP pathway and CME functionally coordinate the development of dendritic morphology.

Materials And Methods
The use of animals was approved by the Institutional Committee for the Care and Use of Experimental Animals at the University of Yamanashi (protocol #A25-33, #A30-21). All experiments were conducted according to the recommendations in the Guidelines for Proper Conduct of Animal Experiments of the Science Council of Japan (2006). And this study was carried out in compliance with the AARIVE (Abunak Research: Reporting of In Vivo experiments) guidelines.
Yeast two-hybrid screening We screened a cDNA library from P9 ICR-mouse forebrain for Vangl2-interacting proteins using the Matchmaker Gold Yeast Two-Hybrid System (Clontech) according to the manufacturer's instructions.
Brie y, random-primed cDNA was synthesized using polyA + RNA taken from the P9 ICR-mouse forebrain using the SMART system (Clontech). The cDNA and pGADT7-Rec vector were co-transformed into the Y187 yeast strain (Clontech), which was plated on SD-lacking leucine (SD/-Leu) plates. Transformants were used for each screening. The Vangl2 amino acid residues 1-114 and 252-521 were subcloned into a pGBKT7 DNA-BD vector to yield pGBKT7-Vangl2 (1-114) and pGBKT7-Vangl2 (252-521), which were transformed into a Y2H Gold yeast strain that was plated on SD/-Typ plates. The positive clone was then used to mate with the Y187 transformants and selected on SD/-Leu/-Trp plates supplemented with X-a-Gal and aureobasidin A (SD/-Leu/-Trp/X/A). The positive clones were subsequently selected on SD/-Leu/-Trp/-Ade/-His/X/A plates. The yeast plasmid was transformed into E. coli and isolated DNA was sequenced.

Construction of expression vectors
The entire coding sequences of mouse, Ap2m1, Atp1b1, Eef1a1, Kif1a, and SynGAP1 were cloned into a pCAII-FLAG vector. The DNA fragment encoding amino acid residues 195-367 of Ap2m1 was ampli ed by PCR from the pCAII-FLAG-Ap2m1 vector and cloned into pGEX-4T-1 (GE Healthcare) and pCAII-EGFP

In utero electroporation
All experiments were performed in accordance with relevant guidelines including AARIVE. In utero electroporation was performed as described previously (Saito & Nakatsuji, 2001;Tabata & Nakajima, 2001). Brie y, pregnant ICR mice at embryonic day (E)14.5 or E15.5 were anesthetized with 10% pentobarbital solution (0.1mL/10g body weight), and the uterine horns were exposed. Approximately 1-2 ml of DNA solutions (0.5 mg/ml pCAII-EGFP with 1 mg/ml pSUPER-Scramble, pSUPER-Vangl2 (Hagiwara et al., 2014), or pSUPER-Ap2m1) containing 0.01% fast green was injected into the lateral ventricles of embryos using pulled borosilicate glass capillaries (B120F-4; World Precision Instruments). The head of an embryo in the uterus was pinched with a forceps-type electrode (CUY650P5; NEPA Gene) and ve square electric pulses (33 V, 50 ms) at intervals of 950 ms were delivered using an electroporator (CUY21E; NEPA Gene). After electroporation, the embryos were returned to the abdominal cavity to allow continuous development.

Histological analysis
Under deep anesthesia, mice were xed transcardially with 4% PFA in PBS (pH 7.4) at P21. Coronal sections (50 mm) were embedded and images were acquired with confocal laser microscopy (FV-1200, Olympus). Apical and basal dendrites of the cortical layer 2/3 pyramidal neurons were randomly sampled and analyzed using Amira 5.5 software (FEI).

Statistical methods
Statistical signi cance was evaluated by Student's t-test or one-way ANOVA followed by post hoc Tukey's test. Statistical signi cance was assumed when p < 0.05.

Declarations
Screening of Vangl2 interacting partners using yeast two-hybrid. (A) Structure of the Vangl2 protein includes 4 transmembrane regions (black boxes, 111-126, 148-166, 183-202, 218-237), and a PDZ binding motif (TSV, 519-521). To detect novel Vangl2 interacting-partners, the N-terminal (1-114) region of Vangl2 was incorporated as the bait plasmid. (B) The bait culture of Vangl2N was combined with prey plasmids of the P9 mouse DNA library and then plated. The positive clones were identi ed via blue color staining. (C) To con rm the screening results of the yeast two-hybrid test, binding of Vangl2N and Cterminal (242-521) proteins were investigated by pull-down assay via western blotting using anti-FLAG antibody. From the ve post-synaptically localized proteins, Ap2m1 and synGAP1 bound strongly to the N-terminal region rather than the C-terminal region of Vangl2. The full-length western blots were presented in Supplementary Fig.1.  However, binding of Ap2m1 was reduced by the deletion of the N-terminal and the C-terminal deletion. The full-length western blots were presented in Supplementary Fig.1. (C) Colocalization of Vangl2 and