Protein-4.1G-Mediated Membrane Trafficking Is Essential for Correct Rod Synaptic Location in the Retina and for Normal Visual Function

•Protein 4.1G is essential for the arrangement of correct retinal synapse location•4.1G contributes to membrane-trafficking system mediated by neuronal AP3•Retinal synaptic location is important for establishing normal visual acuity•Normal membrane trafficking is essential for synaptic integrity maintenance In vertebrate retinal development, the axonal terminals of retinal neurons make synaptic contacts within narrow fixed regions, and these locations are maintained thereafter. However, the mechanisms and biological logic of the organization of these fixed synapse locations are poorly understood. We show here that a membrane scaffold protein, 4.1G, is highly expressed in retinal photoreceptors and is essential for the arrangement of their correct synapse location. The 4.1G-deficient retina exhibits mislocalization of photoreceptor terminals, although their synaptic connections are normally formed. The 4.1G protein binds to the AP3B2 protein, which is involved in neuronal membrane trafficking, and promotes neurite extension in an AP3B2-dependent manner. 4.1G mutant mice showed visual acuity impairments in an optokinetic response, suggesting that correct synapse location is required for normal visual function. Taken together, the data in this study provide insight into the mechanism and importance of proper synapse location in neural circuit formation. In vertebrate retinal development, the axonal terminals of retinal neurons make synaptic contacts within narrow fixed regions, and these locations are maintained thereafter. However, the mechanisms and biological logic of the organization of these fixed synapse locations are poorly understood. We show here that a membrane scaffold protein, 4.1G, is highly expressed in retinal photoreceptors and is essential for the arrangement of their correct synapse location. The 4.1G-deficient retina exhibits mislocalization of photoreceptor terminals, although their synaptic connections are normally formed. The 4.1G protein binds to the AP3B2 protein, which is involved in neuronal membrane trafficking, and promotes neurite extension in an AP3B2-dependent manner. 4.1G mutant mice showed visual acuity impairments in an optokinetic response, suggesting that correct synapse location is required for normal visual function. Taken together, the data in this study provide insight into the mechanism and importance of proper synapse location in neural circuit formation. During development of the vertebrate central nervous system (CNS), enormous numbers of various neurons need to be precisely interconnected to produce elaborate neural functions. Formation, maintenance, and impairment of synaptic connections in the CNS have been extensively studied. In contrast, little attention has been paid to the “location” of synaptic connections. In the developing hippocampus, dentate granule cell axons form synaptic connections with pyramidal cell dendrites in the CA3 to form the stratum lucidum. In the developing cerebellum, Purkinje cell dendrites contact climbing fibers and parallel fibers in the molecular layer. In retinal development, synaptic contacts are made in fixed locations to form plexiform layers and are maintained under normal conditions. However, the molecular mechanisms and physiological significance of fixed synaptic connection locations are almost unknown. The vertebrate retina, a part of the CNS, transmits light information through the lateral geniculate nucleus in the thalamus to the primary visual cortex. Retinal neural circuits are assembled from five types of neurons and one type of glial cell, forming three distinct layers: the outer nuclear layer (ONL), the inner nuclear layer (INL), and the ganglion cell layer (GCL). Retinal axons and dendrites form synapses in two plexiform layers: the outer plexiform layer (OPL) and the inner plexiform layer (IPL), which separate the nuclear layers (Masland, 2004Masland R.H. Neuronal cell types.Curr. Biol. 2004; 14: R497-R500Abstract Full Text Full Text PDF PubMed Scopus (168) Google Scholar, Sanes and Zipursky, 2010Sanes J.R. Zipursky S.L. Design principles of insect and vertebrate visual systems.Neuron. 2010; 66: 15-36Abstract Full Text Full Text PDF PubMed Scopus (383) Google Scholar). In the OPL, photoreceptor terminals connect with bipolar and horizontal cell processes to form the first synapses in the visual pathway (Figure 1A). The light-evoked signals are segregated into ON or OFF pathways that are mediated by ON or OFF bipolar cells, respectively. ON-bipolar cells, which depolarize upon light stimulation, develop axons which terminate in the inner half of the IPL. Conversely, OFF-bipolar cells hyperpolarize in response to light signals and extend axons that terminate in the outer half of the IPL (Ghosh et al., 2004Ghosh K.K. Bujan S. Haverkamp S. Feigenspan A. Wässle H. Types of bipolar cells in the mouse retina.J. Comp. Neurol. 2004; 469: 70-82Crossref PubMed Scopus (316) Google Scholar). Horizontal cells transmit a negative feedback signal to the photoreceptor terminals (Wang et al., 2014Wang T.M. Holzhausen L.C. Kramer R.H. Imaging an optogenetic pH sensor reveals that protons mediate lateral inhibition in the retina.Nat. Neurosci. 2014; 17: 262-268Crossref PubMed Scopus (70) Google Scholar). In the ONL, rod photoreceptor cell bodies are broadly distributed throughout the ONL, whereas cone photoreceptor cell bodies are arrayed on the outer side of the retina. Photoreceptor axon terminals are located at relatively restricted positions in the OPL, indicating that photoreceptor axon terminal location is regulated in a genetically fixed manner, although the distance between each cell body and synaptic terminal varies among photoreceptor cells. Normal aging disrupts this retinal layer integrity (Eliasieh et al., 2007Eliasieh K. Liets L.C. Chalupa L.M. Cellular reorganization in the human retina during normal aging.Invest. Ophthalmol. Vis. Sci. 2007; 48: 2824-2830Crossref PubMed Scopus (81) Google Scholar, Liets et al., 2006Liets L.C. Eliasieh K. van der List D.A. Chalupa L.M. Dendrites of rod bipolar cells sprout in normal aging retina.Proc. Natl. Acad. Sci. USA. 2006; 103: 12156-12160Crossref PubMed Scopus (81) Google Scholar, Samuel et al., 2011Samuel M.A. Zhang Y. Meister M. Sanes J.R. Age-related alterations in neurons of the mouse retina.J. Neurosci. 2011; 31: 16033-16044Crossref PubMed Scopus (112) Google Scholar). Despite the molecular mechanisms for triad synapse formation having been extensively studied, the molecular basis and functional importance of OPL formation at a fixed laminar location throughout the retina during development remains unexplored. Protein 4.1 family genes, 4.1R (Epb4.1), 4.1N (Epb4.1l1), 4.1G (Epb4.1l2), and 4.1B (Epb4.1l3) in vertebrates, encode membrane-cytoskeletal proteins, which anchor a transmembrane protein to the actin-spectrin cytoskeleton. In rat neurons, 4.1N associates with AMPA glutamate receptor subunits, GluR1 and GluR4, to regulate the surface expression of AMPA receptors (Coleman et al., 2003Coleman S.K. Cai C. Mottershead D.G. Haapalahti J.P. Keinänen K. Surface expression of GluR-D AMPA receptor is dependent on an interaction between its C-terminal domain and a 4.1 protein.J. Neurosci. 2003; 23: 798-806Crossref PubMed Google Scholar, Shen et al., 2000Shen L. Liang F. Walensky L.D. Huganir R.L. Regulation of AMPA receptor GluR1 subunit surface expression by a 4. 1N-linked actin cytoskeletal association.J. Neurosci. 2000; 20: 7932-7940Crossref PubMed Google Scholar), and 4.1N knockdown decreases the surface expression of GluR1 and long-term potentiation in the hippocampus (Lin et al., 2009Lin D.T. Makino Y. Sharma K. Hayashi T. Neve R. Takamiya K. Huganir R.L. Regulation of AMPA receptor extrasynaptic insertion by 4.1N, phosphorylation and palmitoylation.Nat. Neurosci. 2009; 12: 879-887Crossref PubMed Scopus (271) Google Scholar). On the other hand, another study reported that 4.1G and 4.1N double-mutant mice showed unaltered glutamatergic synaptic transmission and synaptic plasticity in the hippocampus (Yang et al., 2011Yang S. Weng H. Chen L. Guo X. Parra M. Conboy J. Debnath G. Lambert A.J. Peters L.L. Baines A.J. et al.Lack of protein 4.1G causes altered expression and localization of the cell adhesion molecule nectin-like 4 in testis and can cause male infertility.Mol. Cell. Biol. 2011; 31: 2276-2286Crossref PubMed Scopus (31) Google Scholar). Thus, the exact in vivo function of the 4.1 family genes in the CNS is still unclear. In the current study, we found that 4.1G is predominantly expressed in rod photoreceptor cells in the retina. 4.1G-null mutant mice showed an aberrant localization of photoreceptor terminals in the ONL. Our analysis showed that 4.1G functions at least partly through the AP3-mediated membrane-trafficking system. Although photoreceptor synapses normally connect with bipolar and horizontal cell terminals, optokinetic response (OKR) analysis demonstrated impaired visual responses in the 4.1G−/− mice. Our study implies that synapse location is functionally important in the neural circuit. In order to identify genes important for retinal photoreceptor synapse development, we performed a microarray analysis comparing the retinal gene-expression profiles of wild-type (WT) and Crx-Cre-driven Otx2 conditional knockout retinas, in which photoreceptor cell fate is converted to that of amacrine-like cells (Nishida et al., 2003Nishida A. Furukawa A. Koike C. Tano Y. Aizawa S. Matsuo I. Furukawa T. Otx2 homeobox gene controls retinal photoreceptor cell fate and pineal gland development.Nat. Neurosci. 2003; 6: 1255-1263Crossref PubMed Scopus (460) Google Scholar). We focused on genes that were annotated to encode membrane skeletal proteins, extracellular matrix proteins, and receptor-like transmembrane proteins. In this screen, we identified 4.1G, a gene that encodes an adaptor protein that connects membrane proteins with cytoskeletons. To examine the tissue distribution of mouse 4.1G, we performed northern blot analysis and found that 4.1G is highly expressed in the retina and testis (Figure 1B), whereas a previous work reported that 4.1G is ubiquitously expressed in various tissues (Parra et al., 1998Parra M. Gascard P. Walensky L.D. Snyder S.H. Mohandas N. Conboy J.G. Cloning and characterization of 4.1G (EPB41L2), a new member of the skeletal protein 4.1 (EPB41) gene family.Genomics. 1998; 49: 298-306Crossref PubMed Scopus (98) Google Scholar). We cloned three splicing variants of 4.1G cDNAs from mouse retinal RNAs and found that retinal 4.1G protein isoforms, which are produced from the three transcript variants, lack the spectrin-actin-binding (SAB) domain (Figure S1A; Cheng and Molday, 2013Cheng C.L. Molday R.S. Interaction of 4.1G and cGMP-gated channels in rod photoreceptor outer segments.J. Cell Sci. 2013; 126: 5725-5734Crossref PubMed Scopus (10) Google Scholar, Yang et al., 2011Yang S. Weng H. Chen L. Guo X. Parra M. Conboy J. Debnath G. Lambert A.J. Peters L.L. Baines A.J. et al.Lack of protein 4.1G causes altered expression and localization of the cell adhesion molecule nectin-like 4 in testis and can cause male infertility.Mol. Cell. Biol. 2011; 31: 2276-2286Crossref PubMed Scopus (31) Google Scholar). The SAB domain is also absent in the 4.1G clones that we isolated from human retinal cDNAs (19 out of 21 clones) as well as the 4.1G cDNA cloned from human brain astrocytoma cells (Saito et al., 2005Saito M. Sugai M. Katsushima Y. Yanagisawa T. Sukegawa J. Nakahata N. Increase in cell-surface localization of parathyroid hormone receptor by cytoskeletal protein 4.1G.Biochem. J. 2005; 392: 75-81Crossref PubMed Scopus (29) Google Scholar). We named the 4.1G splicing variant clones based upon their length: 4.1G short (4.1G(S); 689 residues), 4.1G middle (4.1G(M); 794 residues), 4.1G long (4.1G(L); 918 residues), and 4.1G full length (4.1G(full); 988 residues). We raised an antibody against 4.1G, which recognizes all 4.1G isoforms, and performed western blot analysis on adult retinal homogenate. For a positive control, we used HEK293T cell homogenate transfected with Flag-tagged 4.1G(S), 4.1G(M), or 4.1G(L) plasmids. In the retinal lysate, we detected strong bands consistent with the molecular weights of 4.1G(M) and 4.1G(S), a very faint band for 4.1G(L), but undetected a 4.1G(full) band (Figures 1C and S1N). To observe 4.1G expression, we carried out in situ hybridization using developing and adult mouse retinas and brains. We observed that all protein 4.1 family genes were widely expressed in the developing and adult retinas (Figures S1B–S1E). Next, we immunostained the adult retina using an anti-4.1G antibody and observed strong 4.1G protein staining signals in a region between the outer limiting membrane and OPL, including cell bodies and axons in photoreceptors, but none were detected in the outer and inner segments (Figure 1D). We further immunostained dissociated retinal cells using the anti-4.1G antibody and found that 4.1G is predominantly expressed in rods and Müller glial cells (Figure S1J). In the adult brain, 4.1G was weakly expressed (Figures S1F–S1I). To investigate the in vivo role of 4.1G in retinal development, we generated 4.1G-null mice by targeted gene disruption (Figures S1K–S1M). In the 4.1G−/− retina, no retinal 4.1G isoforms were detected by western blot (Figures S1N and S1O). The immunostained 4.1G signal totally disappeared in the 4.1G−/− retina (Figure S1P), indicating that the 4.1G−/− mice are 4.1G-null mutants. Both 4.1G+/− and 4.1G−/− mice were viable and showed no gross morphological abnormalities, whereas 4.1G−/− male mice are infertile as reported previously (Wozny et al., 2009Wozny C. Breustedt J. Wolk F. Varoqueaux F. Boretius S. Zivkovic A.R. Neeb A. Frahm J. Schmitz D. Brose N. Ivanovic A. The function of glutamatergic synapses is not perturbed by severe knockdown of 4.1N and 4.1G expression.J. Cell Sci. 2009; 122: 735-744Crossref PubMed Scopus (19) Google Scholar). We next immunostained PSD95 and pikachurin in retinal sections at 2 months old (2M) to label photoreceptor synaptic terminals. The 4.1G−/− retina displayed abnormally scattered rod synapse spherules in the ONL, which are normally arrayed in the OPL in the WT retina (Figure 1E). We further carried out in vivo single-cell labeling by subretinal injection of LIA murine retroviruses, which express human placental alkaline phosphatase to define the shape of a single rod cell, into retinal progenitor cells at postnatal day 0 (P0). We observed that rod axonal length was shortened and rod cell terminals did not reach the OPL in most rod cells in the 4.1G−/− retina (Figure 1F, arrow heads). Furthermore, the OPL in the 4.1G−/− retina was very thin and irregular (Figure 1F). The OPL length in 4.1G−/− (8.3 ± 0.5 μm; SD) retinas at 1M was significantly thinner than that in WT retinas (13.9 ± 1.8 μm; SD; p < 0.01). These results show that 4.1G is required for proper location of the rod synaptic terminal and for OPL formation. To examine the integrity of synaptic terminals in horizontal and bipolar cell dendrites, which form synaptic contacts with photoreceptors, we immunostained retinal sections with cell-type-specific markers for horizontal cells (CALB1), ON-rod bipolar cells (PKCα), and OFF-cone bipolar cells (PKARIIβ for type 3b and calsenilin for type 4) together with photoreceptor synapse markers (CtBP2 or pikachurin; Sato et al., 2008Sato S. Omori Y. Katoh K. Kondo M. Kanagawa M. Miyata K. Funabiki K. Koyasu T. Kajimura N. Miyoshi T. et al.Pikachurin, a dystroglycan ligand, is essential for photoreceptor ribbon synapse formation.Nat. Neurosci. 2008; 11: 923-931Crossref PubMed Scopus (228) Google Scholar, Wässle et al., 2009Wässle H. Puller C. Müller F. Haverkamp S. Cone contacts, mosaics, and territories of bipolar cells in the mouse retina.J. Neurosci. 2009; 29: 106-117Crossref PubMed Scopus (321) Google Scholar). In the 4.1G−/− retina at 2M, we observed aberrantly sprouting neurites from horizontal and bipolar cells that contact photoreceptor synapse termini at ectopic locations in the ONL (Figures 2A–2D). In contrast, the cone photoreceptor synapse location was unaffected in the 4.1G−/− retina (asterisks in Figures 2C and 2D), indicating that only rod termini are aberrantly located. To analyze the ultrastructural differences of photoreceptor ribbon synapse between WT and 4.1G−/− mouse retinas at 2M, we carried out a conventional electron microscopy analysis. We did not find any substantial difference in the triad structure of ribbon synapses between WT and 4.1G−/− retina, although ribbon synapses in the 4.1G−/− retina were formed around cell bodies in the ONL (Figure 2E). To investigate whether the abnormal location of rod terminals is due to a developmental defect or degeneration, we examined developing WT and 4.1G−/− retinas by focusing on horizontal cells. In mouse development, horizontal cells begin to connect with rod terminals around P9 to shape the dyad structure. Then, bipolar cells begin to contact rod terminals at around P10 to form the triad structure. Rod ribbon synapse formation is completed by P14 or P15 (Sherry et al., 2003Sherry D.M. Wang M.M. Bates J. Frishman L.J. Expression of vesicular glutamate transporter 1 in the mouse retina reveals temporal ordering in development of rod vs. cone and ON vs. OFF circuits.J. Comp. Neurol. 2003; 465: 480-498Crossref PubMed Scopus (165) Google Scholar). We developmentally examined horizontal cells from P3 through P14 by immunostaining using an anti-calbindin antibody (Figure 3A). No obvious change was observed between the WT and 4.1G−/− retinas at P3 or P6. It should be noted that horizontal cells began to aberrantly sprout their termini into the ONL between P9 and P14 in the 4.1G−/− retina (Figures 3A and S2A–S2B’). In addition, at P12, we observed the aberrant sprouting of bipolar cell dendritic terminals into the ONL of the 4.1G−/− retina (Figures S2C–S2D’). These observations indicate that ectopic synapse location in the 4.1G−/− retina is due to the developmental defect. To further examine aberrant neurite extension in the developing 4.1G−/− retina, we immunostained retinal sections from WT and 4.1G−/− retinas using antibodies against pikachurin and CtBP2 at P12 when synapse formation between photoreceptor and bipolar cells is occurring (Figures 3B and 3C; Takada et al., 2004Takada Y. Fariss R.N. Tanikawa A. Zeng Y. Carper D. Bush R. Sieving P.A. A retinal neuronal developmental wave of retinoschisin expression begins in ganglion cells during layer formation.Invest. Ophthalmol. Vis. Sci. 2004; 45: 3302-3312Crossref PubMed Scopus (86) Google Scholar). The number of rod terminals labeled with CtBP2 and pikachurin significantly decreased in the 4.1G−/− retina compared with the number in the WT retina (Figure 3D). Moreover, the typical horseshoe-shaped ribbon morphology seen with CtBP2 was barely detected in the 4.1G−/− retina (Figures 3B and 3C). Then, we examined pikachurin distribution in 4.1G−/− and WT retinas. We found that the number of aberrant pikachurin aggregates in the ONL was significantly higher in the 4.1G−/− retina than in the WT retina (Figures 3E–3G, arrowheads). These observations strongly suggest that synaptic protein transport is compromised in the developing 4.1G−/− retina. Because 4.1G expression was detected in Müller glial cells as well as rods (Figure S1J), we immunostained Müller glial cells at P12. The total morphology, including horizontal extension of Müller glial cells, in the 4.1G−/− retina was unaltered compared with that in the WT retina (Figures S2E and S2F), suggesting that Müller glial cells are not the cause of abnormality in the 4.1G−/− retina. Retinal detachment causes abnormal photoreceptor synapse location as well as abnormal rhodopsin transport to the outer segment (OS) (Fisher et al., 2005Fisher S.K. Lewis G.P. Linberg K.A. Verardo M.R. Cellular remodeling in mammalian retina: results from studies of experimental retinal detachment.Prog. Retin. Eye Res. 2005; 24: 395-431Crossref PubMed Scopus (259) Google Scholar). Immunohistochemistry revealed normal localization of rhodopsin in the OS in the 4.1G−/− retina, excluding the possibility of retinal detachment in the 4.1G−/− retina. In addition, no abnormality in the IPL in the 4.1G−/− retina was observed (data not shown). These results suggest that the mislocalization of rod synapses in the 4.1G−/− retina is due to a photoreceptor cell-autonomous mechanism. In order to understand the molecular mechanisms underlying ectopic synapse formation in the 4.1G−/− retina, the spectrum of proteins interacting with 4.1G in the retina must be first identified. To this end, we carried out immunoprecipitation with an anti-4.1G antibody by using retinal lysates prepared from WT and 4.1G−/− retinas. The obtained immunoprecipitates were fractionated by SDS-PAGE (Figure S3A) and were analyzed using mass spectrometry to identify proteins that were immunoprecipitated specifically in the WT retinal lysates. We identified 16 candidate proteins as 4.1G-interacting proteins (Table S1). Among those candidates, adaptor protein complex 3 beta 2 subunit (AP3B2) and tubulin beta 3 (Tubb3) showed the highest WT-specific probability. We also utilized the DAVID functional annotation tools (Huang da et al., 2009Huang da W. Sherman B.T. Zheng X. Yang J. Imamichi T. Stephens R. Lempicki R.A. Extracting biological meaning from large gene lists with DAVID.Curr. Protoc. Bioinformatics. 2009; Chapter 13 (Unit 13.11)PubMed Google Scholar) to acquire functional annotations for the 16 candidate proteins and identified four “vesicle”-related proteins including AP3B2, SYT-1, HSPA8, and TRAPPC4. We, therefore, focused on AP3B2, which is a component of the neuronal AP3 complex that controls synaptic vesicle formation from endosomes in the membrane-trafficking pathway (Faúndez et al., 1998Faúndez V. Horng J.T. Kelly R.B. A function for the AP3 coat complex in synaptic vesicle formation from endosomes.Cell. 1998; 93: 423-432Abstract Full Text Full Text PDF PubMed Scopus (247) Google Scholar, Seong et al., 2005Seong E. Wainer B.H. Hughes E.D. Saunders T.L. Burmeister M. Faundez V. Genetic analysis of the neuronal and ubiquitous AP-3 adaptor complexes reveals divergent functions in brain.Mol. Biol. Cell. 2005; 16: 128-140Crossref PubMed Scopus (65) Google Scholar). To confirm the potential interaction between 4.1G and AP3B2, we performed immunoprecipitation assays using HEK293T cells (Figure 4A) and detected an interaction between 4.1G and AP3B2. We further searched for the domain responsible for 4.1G binding to AP3B2 (Figures 4B and 4C). The headpiece (HP) domain plus four-point-one, ezrin, radixin, and moesin (FERM) domain fragment showed the strongest interaction with AP3B2. Because the HP domain alone failed to bind AP3B2, the FERM domain is the main 4.1G module interacting with AP3B2 (Figures 4B and 4C). In addition, due to lack of a usable anti-AP3B2 antibody for immunostaining, we examined colocalization of 4.1G with AP3B2 using PC12 cells. We observed the colocalization of 4.1G with AP3B2 by immunostaining (Figures S3B–S3E). We also confirmed the interaction between 4.1G and TUBB3 by immunoprecipitation (Figure S3F). Because the transport of glutamate transporter VGLUT1 to synaptic vesicles in the AP3-defective (Ap3d1−/−) brain was impaired (Salazar et al., 2005Salazar G. Craige B. Love R. Kalman D. Faundez V. Vglut1 and ZnT3 co-targeting mechanisms regulate vesicular zinc stores in PC12 cells.J. Cell Sci. 2005; 118: 1911-1921Crossref PubMed Scopus (72) Google Scholar), we next examined the subcellular localization of VGLUT1 in the WT and 4.1G−/− retinas to investigate if the association between 4.1G and AP3B2 is important for retinal synaptic vesicular sorting. Whereas VGLUT1 was localized predominantly to rod synaptic terminals in the WT retina at 1M, in the 4.1G−/− retina, a significant amount of VGLUT1 remained in cell bodies in addition to that in synaptic terminals (Figures 4D and 4E). Our observation of the shortened axons, the abnormal transportation of synaptic vesicle proteins to rod spherules in the 4.1G−/− retina, and the interaction of 4.1G with AP3B2 prompted us to suppose that 4.1G enhances vesicle transportation and promotes neurite extension. To examine this aspect, we overexpressed 4.1G(M) and 4.1G(S) in PC12 cells, which endogenously express the neuronal AP3 complex (Blumstein et al., 2001Blumstein J. Faundez V. Nakatsu F. Saito T. Ohno H. Kelly R.B. The neuronal form of adaptor protein-3 is required for synaptic vesicle formation from endosomes.J. Neurosci. 2001; 21: 8034-8042Crossref PubMed Google Scholar), and measured neurite length after nerve growth factor (NGF) treatment (Figures 5A and 5B ). The average neurite length in each quartile population (determined by neurite length) was significantly longer in the cells overexpressing 4.1G(M) or 4.1G(S) than that of mock-transfected cells. To test if the neuronal AP3 complex mediates the neurite elongation function of 4.1G, we carried out knockdown experiments of Ap3b2 using two different small hairpin RNA (shRNA) expression plasmids against Ap3b2 under 4.1G-overexpression conditions (Figures 5C and 5D). Neurites elongated by 4.1G were significantly shortened in Ap3b2 knockdown cells (Figure 5D). These observations encouraged us to ask if neuronal AP3 is important for vesicle transport in the retina, so we examined the histology of the Ap3b2−/− retina (Seong et al., 2005Seong E. Wainer B.H. Hughes E.D. Saunders T.L. Burmeister M. Faundez V. Genetic analysis of the neuronal and ubiquitous AP-3 adaptor complexes reveals divergent functions in brain.Mol. Biol. Cell. 2005; 16: 128-140Crossref PubMed Scopus (65) Google Scholar). The adult Ap3b2−/− retina exhibited abnormal localization of VGLUT1 in rod cell bodies and a slightly reduced accumulation of VGLUT1 in the rod spherules compared with the WT retina (Figures 5E and 5F). Furthermore, we found an increase in the number of ectopic rod synaptic terminals in the ONL compared with that of the WT, although the abnormality was less severe than that in the 4.1G−/− retina (Figures 2A, 2B, 5E, and 5G). These results suggest that 4.1G promotes neurite outgrowth by enhancing membrane trafficking mediated by the neuronal AP3 complex. To examine whether 4.1G deficiency affects physiological function in vivo, we first recorded electroretinogram (ERG) responses under the dark-adapted (scotopic) and the light-adapted (photopic) conditions of mice at P15 and 1M (Figures 6A–6D and S4A–S4L). In scotopic ERGs at P15, the amplitudes of a-waves were unaltered between WT and 4.1G−/− mice (Figures 6A and S4A). In contrast, the amplitudes of b-waves in scotopic ERGs at P15 were significantly reduced at stimulus intensities of −3.0 to 1.0 log cd-s/m−2 in 4.1G−/− mice (Figures 6A and S4B). As a result, the b/a wave ratio of the 4.1G−/− mice was significantly smaller than that of the WT mice (Figure S4I). The implicit time of b-waves in the 4.1G−/− mouse was significantly longer than that in WT mouse (Figure S4J). Interestingly, in retinal scotopic ERGs at 1M, there were no significant differences in both implicit time and amplitudes of a- and b-waves between the WT and 4.1G−/− mice (Figures 6C, S4E, S4F, S4I, and S4J). In photopic ERGs of the 4.1G−/− mice, there was no significant change either in the amplitudes or implicit times of a- and b-waves both at P15 and 1M (Figures 6B, 6D, S4C, S4D, S4G, S4H, S4K, and S4L). Moreover, we observed low-level rod ribbon synapse integrity in P15 4.1G−/− retinas that was recovered to normal by the adult stage (Figure S4M). These results showed that development of the rod synapse connection is delayed in the 4.1G−/− mice at P15, but the developmental delay is overcome by 1M. This fits well with the histological observations on WT and 4.1G−/− retinas at P12 and 2M (Figures 2B, 3B, and 3C). We assessed the visual acuity of WT and 4.1G−/− mice using OKRs induced by the rotation of a screen with various spatial frequencies and contrast sensitivities of black and white stripes (Figure 6E). WT mice had a spatial frequency threshold of 0.48 cycles/degree (n = 6). On the other hand, the spatial frequency threshold of the 4.1G−/− mice was significantly reduced to 0.22 cycles/degree (p < 0.01; n = 6) at 2M (Figure 6F). We next measured contrast sensitivity of WT and 4.1G−/− mice and observed a significant increase of contrast level in the OKRs of 4.1G−/− mice (55.5%; p < 0.01; n = 6) in comparison with that of WT mice (7.9%; n = 6; Figure 6G), indicating that contrast sensitivity was significantly impaired in 4.1G−/− mouse vision (Figure 6H). In addition, the decrease of contrast sensitivity was independent of spatial frequency (Figure 6I). These results indicate that visual acuity was impaired in 4.1G−/− mice. After 1 year of age (1Y), there is a significant correlation between mouse age and abnormal extension of bipolar dendrites (Liets et al., 2006Liets L.C. Eliasieh K. van der List D.A. Chalupa L.M. Dendrites of rod bipolar cells sprout in normal aging retina.Proc. Natl. Acad. Sci. USA. 2006; 103: 12156-12160Crossref PubMed Scopus (81) Google Scholar). To examine the effect of aging on the 4.1G−/− retina, we histologically evaluated retinal sections using toluidine blue O staining. In the 1Y 4.1G−/− retina, the ONL length was shorter by about 20% than that of the WT, although the ONL length was unaltered between WT and 4.1G−/− retinas both at P14 and 1M (Figures 7A and 7B ). We then immunostained retinal sections prepared from WT and 4.1G−/− retinas using cell-type-specific markers (Figures 7C–7F). Both the number of rod spherules in 1Y 4.1G−/− retinas labeled by CtBP2 and the number of ON bipolar cell dendritic tips labeled by TRPM1 (Koike et al., 2010Koike C. Obara T. Uriu Y. Numata T. Sanuki R. Miyata K. Koyasu T. U