DCLK1 Phosphorylates the Microtubule-Associated Protein MAP7D1 to Promote Axon Elongation in Cortical Neurons

Abstract

Doublecortin-like kinase 1 (DCLK1) is a member of the neuronal microtubule-associated doublecortin (DCX) family and functions in multiple stages of neural development including radial migration and axon growth of cortical neurons. DCLK1 is suggested to play the roles in part through its protein kinase activity, yet the kinase substrates of DCLK1 remain largely unknown. Here we have identified MAP7D1 (microtubule-associated protein 7 domain containing 1) as a novel substrate of DCLK1 by using proteomic analysis. MAP7D1 is expressed in developing cortical neurons, and knockdown of MAP7D1 in layer 2/3 cortical neurons results in a significant impairment of callosal axon elongation, but not of radial migration, in corticogenesis. We have further defined the serine 315 (Ser 315) of MAP7D1 as a DCLK1-induced phosphorylation site and shown that overexpression of a phosphomimetic MAP7D1 mutant in which Ser 315 is substituted with glutamic acid (MAP7D1 S315E), but not wild-type MAP7D1, fully rescues the axon elongation defects in Dclk1 knockdown neurons. These data demonstrate that DCLK1 phosphorylates MAP7D1 on Ser 315 to facilitate axon elongation of cortical neurons.

INTRODUCTION

Doublecortin-like kinase 1 (DCLK1) is a serine/threonine protein kinase expressed abundantly in the embryonic and adult brain (Lin et al., 2000). DCLK1 is closely related to Doublecortin (DCX), which is originally identified as mutated in a human cortical developmental malformation caused by defective neuronal migration (des Portes et al., 1998; Gleeson et al., 1998). In mice, targeted disruption of either Dclk1 or Dcx causes no obvious cortical abnormalities, whereas disruption of both Dclk1 and Dcx results in severe cortical lamination defects resembling the pathological features observed in human patient with DCX mutations (Corbo et al., 2002; Deuel et al., 2006; Koizumi et al., 2006). In addition, Dclk1 knockout mice show a significant defect in development of the corpus callosum, whereas Dclk1;Dcx double knockout mice exhibit a complete loss of the commissural axonal tracks through the corpus callosum, hippocampal commissure, and anterior commissure (Deuel et al., 2006; Koizumi et al., 2006; Bielas et al., 2007). Thus, DCLK1 functions together with DCX to regulate neuronal migration and axon elongation during cortical development. Recent genome wide association studies suggest that genetic variants of DCLK1 are associated with cognitive traits, schizophrenia and attention deficit hyperactivity disorder (ADHD) (Le Hellard et al., 2009; Havik et al., 2012), supporting the significant role of DCLK1 in neuronal functions.

DCLK1 protein contains a tandem DCX repeat domain at its N-terminus that have an ~75% amino acid identity with the repeat domain in DCX. DCLK1 additionally contains a serine-threonine protein kinase domain at its C-terminus. Interestingly, in the C. elegans ortholog of DCLK1, designated Zyg-8, both the DCX domain and the kinase domain are required for spindle positioning during asymmetric cell division (Gonczy et al., 2001), suggesting that the kinase activity is required for its function. Previous studies including crystallography suggest that the DCX microtubule binding domains contribute to microtubule bundling and stabilization in the leading process of migrating neurons at the neck of the growth cones (Gleeson et al., 1999; Tanaka et al., 2004; Bielas et al., 2007). Similarly, the DCX domains of DCLK1 likely play roles in microtubule bundling and stabilization (Kim et al., 2003). By contrast, little is known about the role of the kinase domain of DCLK1. Considering the DCLK functions in neural development, co-expressed kinase substrates are poised to mediate axon formation and/or neuronal migration (Deuel et al., 2006; Koizumi et al., 2006). To date, in vitro kinase assay and two-hybrid screens identified Synapsin II and the transcription factor Jun dimerization protein-2 (JDP2) as potential targets of the zebrafish homolog of DCLK kinase (zDCLK) (Shimomura et al., 2010; Nagamine et al., 2014), but their physiological significance remains unclear.

In this study, we have performed an unbiased mass spectrometry analysis and identified MAP7D1 (microtubule-associated protein 7 domain containing 1) as a novel kinase substrate of DCLK1 in developing mouse brain. Knockdown of MAP7D1 in layer 2/3 cortical neurons impairs callosal axon elongation, but not radial migration during corticogenesis. Biochemical analysis of MAP7D1 reveals the serine 315 (Ser 315) as a target of DCLK1. Overexpression of the phosphomimetic MAP7D1 in which Ser 315 is substituted with glutamic acid (MAP7D1 S315E) fully rescues the axon elongation defects in Dclk1 knockdown neurons. Our findings thus demonstrate that MAP7D1 functions downstream of DCLK1 in the commissural axon elongation of cortical neurons.

METHODS

Animals

All animal procedures were approved by the Institutional Safety Committee on Recombinant DNA Experiments and the Animal Research Committee of Osaka Bioscience Institute and University of Tokyo. Pregnant ICR mice were purchased from Japan SLC Inc. DCLK1 knockout mice (Koizumi et al., 2006) were backcrossed to the C57Bl/6J lines (CLEA Japan) at least 8 times. Mice were housed in a temperature-controlled room with 12 h light/dark cycle.

Plasmids

A full-length human DCLK1 gene in Kpn I site of pcDNA3.1/myc-6xHis (Lin et al., 2000) was digested with BamH I -Pme I and subcloned into BamH I - Sma I site of pGEX4T-1 expression vector (GE Healthcare). pGEX4T-1/hDCLK1 was digested with Kpn I and self-ligated to obtain control GST-myc-6xHis expression vector. pGEX4T-1/hDCLK1 D511A was created by using QuickChange site-directed mutagenesis kit (Agilent Technologies). MAP7D1 cDNA was PCR amplified from MGC Mouse Map7d1 cDNA (clone ID: 5354497) and subcloned into pET28b(+) (Novagen), pEGFP-C (clonetech) and pCAG-IRES-tdTomato that was modified from pCAGIG (Addgene plasmid 11159). Each MAP7D1 single amino acid mutant and deletion mutant was created by using QuickChange site-directed mutagenesis kit (Agilent Technologies). Human DCX cDNA was subcloned into pET28a (+) (Taylor et al., 2000). pCAG-IRES-tdTomato pCAG-EmGFP-miRNA RNAi constructs for Map7d1, Dclk1, Dcx were constructed by subcloning the PCR amplified EmGFP-miRNA fragments from pcDNA6.2-GW/EmGFP-miR containing each target sequences (generated by using BLOCK-iT Pol II miR RNAi Expression Kit, Invitrogen) and pcDNA6.2-GW/EmGFP-miR-neg control plasmid (Invitrogen) into EcoR I – Sph I site of pCAGIG. Target sequences for Map7d1, Dclk1 and Dcx were designed with Block-iT RNAi Designer (Invitrogen). Map7d1#1; TCCTGCCAAGCAAGATGTAAA, Map7d1#2; GAGCAGCATTGTGGATCGTCT, Dclk1#1 (target sequence in the kinase domain); GTGGACTTTCCATCTCCGTAT, Dclk1#2 (target sequence in the kinase domain); TTCTGCTAAGGAGCTCATCAA, Dclk1#3 (target sequence in the DCX domains); AGGTTCGATTCTACAGAAATG, Dcx; TCAA GTGACCAACAAGGCTAT.

Antibodies

Polyclonal antibodies for MAP7D1 N-terminal or C-terminal were generated by immunizing rabbits with GST-MAP7D1-N (a.a. 1–71) or GST-MAP7D1-C (a.a. 807–846) recombinant fusion proteins purified from bacteria (>90% purity) (Pocono Rabbit Farm & Laboratory). Rabbit polyclonal and goat polyclonal anti-GFP (GeneTex), rabbit polyclonal anti-RFP (Rockland), sheep polyclonal anti-human DCLK1 C-terminus (R&D Systems), rabbit polyclonal anti-human DCLK1 N-terminus (Lin et al., 2000), rabbit monoclonal anti-β-actin (13E5, Cell Signaling) were used for immunohistochemistry, immunoprecipitation and immunoblotting. An affinity-purified polyclonal antibody for phospho-MAP7D1 Ser315 was generated by immunizing rabbits with the phosphopeptide FLARSR[p-S]AVTLPRN (21st Century Biochemicals, Inc).

Preparation of Recombinant Protein

Recombinant proteins were expressed using BL21 (DE3) Escherichia coli. (Novagen). GST-myc-6xHis, GST-DCLK1-myc-6xHis and GST-DCLK1 D511A (kinase-inactive)-myc-6xHis recombinant proteins were affinity-purified by using GSTrap4B (GE Healthcare). A DCLK1 kinase-inactive Asp 511-to-alanine substituted mutant is equivalent to the D527A mutant used by Burgess et. al. (Burgess and Reiner, 2002). 6xHis-T7 tag-MAP7D1–6xHis recombinant proteins were purified by using His GraviTrap (GE Healthcare). These purified recombinant proteins were buffer exchanged to phosphate-buffer saline using PD-10 (GE Healthcare) column for further studies.

GST Affinity-Column Chromatography and 2D-LC-MS/MS

Affinity column purification was performed as described (Fukata et al., 1997). Briefly, mouse embryonic day 17 (E17) brains were homogenized with buffer A (20 mM Tris/HCl (pH 7.4), 1 mM dithiothreitol, 1 mM EDTA with phosphatase inhibitors and protease inhibitors) and centrifuged at 100,000g for 1 h at 4 °C and collected supernatant fraction. Mouse E17 brain cytosol was then loaded on GST-myc-6xHis or GST-DCLK1-myc-6xHis bound to glutathione beads. The columns were washed and bound fraction was eluted with 0.5 M NaCl in buffer A. The eluted fractions were trypsinized and subjected to shotgun multidimensional protein identification technology (MudPIT) analysis (Chen et al., 2006)

In Vitro Kinase Assay

In vitro kinase assays were performed as previously described (Silverman et al., 1999). GST-DCLK1-myc-Hisx6 or GST-DCLK1 D511A-myc-Hisx6 was incubated with Hisx6-MAP7D1-Hisx6 or other recombinant proteins in a buffer containing 50 mM HEPES, pH 7.5, 10 mM magnesium acetate, 10 μg/mL leupeptin, 10 μg/mL aprotinin, 50 mM β-glycerophosphate, 100 mM orthovanadate, 20 mM [γ-³²P] ATP (20 cpm/fmol) at 30°C. The reactions were terminated by adding 2× SDS-PAGE sample buffer and boiled for 5 min. The reaction mixtures were separated with SDS-PAGE. The gel was stained with Coomassie Brilliant Blue and dried onto Whatman 3MM paper with Bio-Rad gel dryer. The dried gel was exposed to BioMax MR autoradiography film (Kodak) at −80°C to detect the phosphoproteins.

Identification of the Phosphorylation Site by Mass Spectrometry

Hisx6-MAP7D1-Hisx6 was phosphorylated in vitro and separated by SDS-PAGE and fixed and stained with Coomassie blue. The band was excised from the gel and destained followed by in gel digestion with Trypsin. Extracted peptides were applied on the time-of flight mass spectrometer (Applied Biosystems) and the MALDI-TOF MS spectra were obtained.

Western Blot Analysis

Brain tissues or cells were collected and lysed with IP buffer (50 mM Tris pH 7.4 150 mM NaCl, 1 mM EDTA, 1% TritonX-100) containing Phos-stop phosphatase inhibitor cocktail (Roche) and Complete Protease Inhibitor cocktail (Roche Diagnostics). Lysates were centrifuged at 20,000g for 20 min at 4°C, and the supernatants were mixed with 4x SDS-PAGE sample buffer and boiled for 5 min. The protein samples were separated by 8% or 7.5% acrylamide SDS-PAGE gel and transferred to PVDF membrane followed by immunoblotting with the indicated antibodies.

Primary Cortical Cell Culture

Primary cell cultures were prepared from the cerebral cortex of ICR mouse or DCLK1 knockout mouse embryos at E15. The cerebral cortices were washed with cold HBSS (Invitrogen) and then digested with papain (Worthington) for 15 min at 37°C. The tissues were then triturated with fire-polished glass pipettes and filtered with 70 μm cell strainer (Falcon). The dissociated cells were pelleted, resuspendend and plated with neurobasal medium (Invitrogen) containing 5% fetal bovine serum (Biowest), B-27 supplement (Invitrogen), GlutaMAX (Gibco), and penicillin/streptomycin (Gibco). The culture medium was changed to neurobasal medium containing B-27 supplement, Gluta-MAX, and penicillin/streptomycin after 1 h and maintained in 37°C in 5% CO₂. For immunostaining, the dissociated cortical neurons were grown on the glass coverslips (Matsunami) pre-coated with poly-l-lysine (PLL) (100 μg/mL). For the MAP7D1 immunoprecipitation from DCLK1 knockout cortical neurons, cortical cells prepared from individual embryo were plated on PLL-coated 100 mm dishes and harvested at days in vitro (DIV) 6 days with IP buffer followed by immunoprecipitation experiments.

For neuronal polarity analysis, dissociated cortical neurons from E15 mouse cortex were resuspended in OPTIMEM (Invitrogen) and electroporated with knockdown constructs using NEPA21 (NEPAGENE). The transfected cells were plated on PLL-coated 12-well plate with Neurobasal medium containing 5% FBS, B-27 supplement, GlutaMAX and penicillin/streptomycin and changed to the medium without FBS after 1 h. The cortical cells were fixed after 48 h and immunostained using anti-EGFP antibody and the fluorescence images were obtained with fluorescence microscope (BZ-9000, Keyence, Osaka, Japan) using a 20x objective. Cortical neurons extending a single process that was at least 20 μm longer than the other remaining processes were scored as polarized neurons. For examination of Map7d1 miRNA and Dclk1 miRNA-induced effects for dendrite formation, cortical neurons were plated at 3.0 × 105 cells per well in PLL-coated 12-well dishes and transfected with knockdown constructs using calcium phosphate method at DIV6. Neurons were fixed in 4% PFA at DIV10 and immunostained with mouse anti-MAP2 monoclonal antibody and anti-MAP7D1 or anti-DCLK1 C-terminal antibody to verify the knockdown, and the fluorescence images were obtained with fluorescence microscope (BZ-9000, Keyence, Osaka, Japan) using a 20x objective. Quantification of total dendrite length and Sholl analysis were performed using NIH Image J software with the NeuronJ plugin (Meijering et al., 2004) and the Sholl Analysis plugin.

Immunofluorescence Analysis

Culture neurons were fixed with 4% paraformaldehyde in PBS at 37°C for 10 min. The fixed neurons were then rinsed with PBS, incubated with blocking buffer (1% bovine serum albumin, 0.1% Triton X-100 in PBS) and stained using anti-MAP7D1, sheep polyclonal anti-DCLK1 antibodies. Immunofluorescence images were obtained using a fluorescence microscope (BZ-9000, Keyence, Osaka, Japan) or a confocal fluorescence microscope (TCS-SP8, Leica).

Immunoprecipitation Analysis

For immunoprecipitation of endogenous MAP7D1 protein, whole brains from P1 mice or cortical cultures at DIV 6 were homogenized and lysed in IP buffer with phosphatase inhibitors and protease inhibitors. Lysates were centrifuged at 20,000g for 20 min at 4°C and supernatants were pre-incubated with Protein G sepharose (GE Healthcare). Anti-MAP7D1 antibody- protein G beads were added to the supernatants pre-incubated with Protein G and incubated for 1 h at 4°C. The beads were then spin-downed and washed twice with IP buffer with phosphatase inhibitors and protease inhibitors, and washed additional three times with IP buffer. The proteins bound to the beads were eluted by boiling with 2x SDS sample buffer, separated by 7.5% acrylamide SDS-PAGE gel and transferred to PVDF membrane followed by immunoblotting. For the evaluation of the Ser 315 phospho-specific MAP7D1 antibody, endogenous MAP7D1 was immunoprecipitated from mouse brains using IP buffer without phosphatase inihibitors and the protein-beads complex was treated with or without lambda protein phosphatase (NEB) for 30 min at 30°C. Alternatively, EGFP-MAP7D1 wild-type or S315A were immunoprecipitated from transfected Neuro-2a cell lysates using IP buffer and the protein-beads complex were subjected to in vitro kinase reactions with recombinant GST-DCLK1 proteins.

Radial Migration and Callosal Axon Elongation Assay Using in Utero Electroporation

Pregnant ICR mice were purchased from Japan SLC Inc. miRNA-mediated knockdown constructs (pCAG-EmGFP-miRNA) or overexpression constructs (pCAG-IRES-tdTomato) were injected into the lateral ventricle of embryo at E14 (for radial migration assay) or E15 (for callosal axon elongation assay) and electroporated using NEPA21 electroporator with tweezers-type electrode (CUY650P5) (NEPAGENE) (Tabata and Nakajima, 2001). For in vivo rescue experiment, Dclk1 miRNA#1 knockdown construct and MAP7D1 overexpression constructs were mixed at a molar ratio of 5: 1 and injected into the lateral ventricle of embryo at E15. Surgeries were performed at 4 pm - 7 pm. After 4 days (for radial migration assay) or 7 days (P3) and 18 days (P14) (for callosal axon elongation assay), embryos or pups were perfused and the dissociated brains were post-fixed overnight with 4% paraformaldehyde in PBS. The fixed brains were then subjected to graded dehydration in 10% (W/V), 20% (W/V), and 30% (W/V) sucrose in PBS for cryoprotection at 4°C. Coronal brain sections (50 μm thick) were prepared using freezing-sliding microtome (Yamato, REM-710). The free-floating sections were processed for immunohistochemistry to GFP or RFP to increase the signals by blocking with 3% donkey serum and 0.1% Trioton X-100 in PBS. Nuclei were stained with TO-PRO-3 or DAPI.

Analyses of the Radial Migration

The fluorescence images were obtained with Leica TCS SP-8 confocal microscope using a 20x objective. To quantify the neuronal migration of cortical neurons, we chose the brains that have similar population of cells labeled by GFP in the cortex, and chose the dorsolateral area for measuring the number of the cells. Measuring was performed by counting the GFP-positive cells manually in the upper cortical plate (uCP), the lower cortical plate (loCP) and the intermediate zone (IZ). Statistical significances were assessed by One-way ANOVA followed by the Bonferroni post-hoc test using SYSTAT13 software (HULINKS, Tokyo, Japan).

Analyses of the Callosal Axon Elongation

The fluorescence images were obtained with Keyence BZ-9000 fluorescence microscope using a 10x objective and by jointing the images. To quantify the callosal axon elongation, we chose the 20 longest GFP-positive or tdTomato-positive axons from each slice of brain and measured the absolute distance of axon tips to the midline. We chose the brains that have similar population of cells in the somato-sensory cortex labeled by GFP or tdTomato. Images were converted to black/white using Adobe Photoshop software (Adobe Systems, Inc., San Jose, CA) and the distances from the axon tips to the midline were measured by using NIH Image J software. Statistical significances were assessed by non-parametric Kruskal-Wallis test with Conover-Inman test using SYSTAT13 software (HULINKS, Tokyo, Japan).

RESULTS

Identification of DCLK1-Interacting Proteins by Proteomic Analysis

To screen potential substrates of DCLK1, we first aimed to isolate a comprehensive collection of proteins that interact with DCLK1 using affinity column chromatography followed by a proteomic analysis. Using wild-type mouse E17 brain cytosolic fractions, proteins bound to the affinity columns were separated and subjected to silver staining on an SDS-PAGE gel [Fig. 1(A)]. Numerous proteins were detected in the eluted fraction from the affinity GST-DCLK1 column, but not as much from GST column [Fig. 1(A)]. To identify proteins interacting specifically with DCLK1, the eluted fractions were subjected to shotgun multidimensional protein identification technology (MudPIT) analysis (Chen et al., 2006). We identified ~100 proteins that specifically associated with GST-DCLK1 compared to GST (Supporting information Table S1). Notably, we identified proteins previously reported to interact with DCLK1, such as 14-3-3 epsilon (Ballif et al., 2006), DCX (Koizumi et al., 2006), and tubulins (Burgess and Reiner, 2000; Lin et al., 2000; Kim et al., 2003). These data suggest that our proteomic approach functions efficiently to identify DCLK1-interacting proteins, with the potential to identify kinase substrates of DCLK1.

Figure 1.

Isolation of DCLK1-interacting proteins. (A) Mouse brain cytosol fraction was loaded onto a glutathione-Sepharose column coated with either GST-myc-Hisx6 or GST-DCLK1-myc-Hisx6. The bound proteins were eluted by 500 mM NaCl and analyzed by SDS-PAGE followed by silver staining. Protein bands specifically observed in the eluted fraction from the affinity GST-DCLK1 column are indicated (arrowheads). (B) Recombinant Hisx6-DCX-Hisx6, Hisx6-MAP7D1-Hisx6 were incubated with recombinant GST-DCLK1-myc-Hisx6 or GST-DCLK1 D511A-myc-Hisx6 (kinase dead mutant) in the presence of [γ-32P] ATP. The reactions were resolved by SDS-PAGE and analyzed by autoradiography (above) and Coomassie blue staining (below). p-DCLK1 indicates the autophosphorylated DCLK1. p-DCX and p-MAP7D1 indicates the phosphorylated DCX and phosphorylated MAP7D1 respectively. Note that proteolytic degradations of recombinant MAP7D1 proteins (labeled as “partial”) are also phosphorylated by DCLK1.

Identification of MAP7D1 As a Novel Kinase Substrate of DCLK1

Given that DCLK1 is associated with microtubules through the N-terminal DCX domain (Burgess and Reiner, 2000; Lin et al., 2000; Kim et al., 2003), we reasoned that DCLK1 might phosphorylate proteins associated with the microtubule cytoskeleton. We initially focused on three microtubule-associated proteins, MAP7D1 (microtubule-associated protein 7 domain containing protein 1), EML4 (echinoderm microtubule-associated protein like 4), and DCX among the DCLK1-interacting candidates (Supporting information Table S1). To examine whether DCLK1 was capable of phosphorylating these proteins, we prepared recombinant proteins and performed in vitro kinase assay with either GST-DCLK1 or a kinase-dead mutant GST-DCLK1 (D511A). We successfully obtained bacterial recombinant proteins of MAP7D1 and DCX for the analysis, but EML4 resisted isolation due to degradation. In the kinase assay, both MAP7D1 and DCX were phosphorylated upon incubation with GST-DCLK1 in the presence of [γ-³²P] ATP, but not by incubation with GST-DCLK1 D511A [Fig. 1(B)].

Next, to determine the phosphorylation sites of MAP7D1 by DCLK1, we performed the mass spectrometry analysis of MAP7D1 after incubation with GST-DCLK1 in the presence of ATP. This analysis detected six phosphorylated peptides [Ser 275, Ser 296, Ser 308, Ser 315, Ser 344, and Thr 455; 49% amino acid sequence coverage, Fig. 2(A–C)]. Notably, the phosphorylation stoichiometry of one residue in particular, Ser 315, was much higher (96%; 27 phosphopeptides out of 28 total peptides detected) compared to other phosphorylation sites [Fig. 2(A,B)], suggesting Ser 315 as a major phosphorylation site by DCLK1. To confirm this, we constructed a series of the serine/threonine-to-alanine substituted MAP7D1 mutants at the 6 residues and found that phosphorylated MAP7D1 was no longer detected in the S315A mutant following incubation with DCLK1 [Fig. 2(C)]. In addition, ³²P incorporation appeared to be reduced in S275A and S308A mutants [Fig. 2(C)]. A previous study on the substrate specificity of DCLK1 by using synthetic peptides indicated the primary selectivity for the phosphorylation site as arginine at the P-3 position and hydrophobic residues at the P-5 position (Shang et al., 2003). The Ser 315 site (LARSRSAVTLP) is consistent with this phosphorylation motif preference, and is also conserved among MAP7D1 orthologs in vertebrates [Fig. 2(D)]. Indeed, MAP7D1 phosphorylation on Ser 315 was previously detected by large-scale phosphoproteomics analysis in the mouse brain (Huttlin et al., 2010) although the responsible protein kinases were not identified. Taken together, these data suggest that DCLK1 associates with and is capable of phosphorylating MAP7D1 on Ser 315.

Figure 2.

MAP7D1 is phosphorylated by DCLK1 on Ser 315. (A) Table of the identified 6 phosphorylation sites within tryptic peptides from recombinant MAP7D1 protein incubated with GST-DCLK1 in the presence of ATP. The estimate of the stoichiometry of phosphorylation is based on the ratio of number of MS/MS spectra observed. The ratio and the number of phosphopeptides out of total peptides detected are shown. (B) MS3 spectra of phospho-SRSAVTLPR. Phospho-Ser 315 is indicated by Sp. Identified b-ions (N-terminal peptide fragments) and y-ions (C-terminal peptide fragments) are indicated. b-ions and y-ions demonstrated the presence of the phosphate group on Ser 315. (C) Recombinant MAP7D1 and its serine/threonine-alanine mutants were incubated with GST-DCLK1 in the presence of [γ-32P] ATP and reactions were resolved by SDS-PAGE and analyzed by autoradiography (above) and Coomassie blue staining (below). p-DCLK1 indicates the autophosphorylated DCLK1. p-MAP7D1 indicates the phosphorylated MAP7D1. The single mutation of S315A (Ser 315-Ala) resulted in almost no phosphorylation. The single mutation of S275A and S308A also showed mild reduction in phosphorylation. (D) Schematic representation of mouse MAP7D1 and MAP7 domain structure (bellow). They have two highly conserved regions (56% and 70% amino acid identity). Two coiled-coil domains are indicated as CC1 and CC2. Previously identified microtubule binding domain and KIF5B interacting domain of MAP7 are indicated (Faire et al., 1999; Metzger et al., 2012). Position of amino acids 1 – 71 and 807 – 846 are used for establishing antisera. Protein alignment shows that Ser 315 in mouse MAP7D1 is highly conserved in vertebrates (above). Previously predicted substrate specificity of DCLK1 for arginine at the P-3 position and hydrophobic (Hyd) residue at the P-5 position are indicated (Shang et al., 2003).

MAP7D1 Is Expressed in Developing Mouse Brain

MAP7D1 is a member of the MAP7 family and is closely related to MAP7/E-MAP-115 (hereafter MAP7) that was originally identified as a microtubule-associated protein in epithelial cells (Masson and Kreis, 1993). There are four MAP7 paralogs (MAP7, MAP7D1, MAP7D2, and MAP7D3) in mammals (Metzger et al., 2012), and of these, MAP7 was shown to function in spermatogenesis (Komada et al., 2000) and in nuclear positioning in myotubes (Metzger et al., 2012), but no information about expression patterns and physiological roles of the other MAP7 paralogs, including MAP7D1, is available.

We first assessed temporal expression patterns of MAP7D1 in developing mouse brain. To detect endogenous MAP7D1 proteins, we generated rabbit polyclonal antisera against recombinant GST-fusion proteins corresponding to the N-terminus [amino acid (a.a.) 1–71] and the C-terminus (a.a. 807–846) (see MAP7D1 domain structure in [Fig. 2(D)]), because these two regions show no significant similarities in amino acid sequences with other MAP7 family proteins. Both antisera recognized multiple bands at ~120 kDa on immunoblots of mouse P0 brain lysates [Supporting information Fig. S1, arrows]. Similar molecular weight bands were detected in the cell lysates prepared from Neuro-2a cells transfected with MAP7D1, but not in untransfected cells [Supporting information Fig. S1], indicating that the ~120 kDa bands correspond to MAP7D1 proteins.

Immunoblot analysis of mouse whole brain lysates revealed that MAP7D1 is expressed in developing brain during embryonic stages. MAP7D1 expression was already visible at E11 and slightly increased by P0, and then gradually decreased in the later postnatal stages [Fig. 3(A)]. Interestingly, this MAP7D1 expression pattern mirrored that of DCLK1, which itself was first detected at E11, became robust around P0 ~ 5, and gradually decreased in the later postnatal stages [Fig. 3(A)]. We next investigated MAP7D1 localization in developing mouse cortex. At P0, MAP7D1 was detected abundantly in the cortical plate and the ventricular zone [Fig. 3(B)]. These expression patterns were reproduced by the other anti-MAP7D1 antibody [Fig. 3(C)], confirming that the two anti-MAP7D1 antibodies recognize endogenous MAP7D1 in the brain. Double staining with anti-DCLK1 and -MAP7D1 antibodies indicated that MAP7D1 was co-localized with DCLK1 in the cortical plate but not in the subcortical axonal fiber tracts [Fig. 3(B)]. In the dissociated cortical cultures, MAP7D1 was localized to the growing axons as well as cell bodies near axons. Double-staining with anti-MAP7D1 and -DCLK1 antibodies indicated that MAP7D1 was co-localized with DCLK1 in the proximal region of axons as well as the cell body [Fig. 3(D) arrows], whereas no significant co-localization was observed in the distal part of axons including growth cones [Fig. 3(D) arrowheads]. Taken together, these observations indicate that MAP7D1 protein is expressed in developing cortical neurons and imply a potential interaction between MAP7D1 and DCLK1 during neural development.

Figure 3.

MAP7D1 is expressed in mouse brain during development. (A) Immunoblotting for MAP7D1 (pAb 22013), DCLK1 (N-terminal antibody) and β-actin in mouse brain lysates during development (20 μg total protein per lanes). (B, C) Immunohistochemistry of mouse coronal cortical sections at P0 with anti-MAP7D1 pAb 22011 or pAb 22013 (green) and anti-DCLK1 C-terminal antibody (magenta). CP, cortical plate, IZ, intermediate zone, VZ, ventricular zone. Scale bars: 50 μm. (D) Dissociated E15 mouse primary cortical neurons (DIV2) were immunostained with anti-MAP7D1 pAb 22011 (green) and anti-DCLK1 C-terminal antibody (magenta). Note that MAP7D1 is co-localized with DCLK1 in the proximal region of axons as well as the cell bodies (arrows), whereas no significant co-localization of MAP7D1 with DCLK1 is observed in the distal part of axons including growth cones (arrowheads). Scale bars: 10 μm.

MAP7D1 Is Required for Callosal Axon Elongation

DCLK1 is involved in multiple stages of brain development including neuronal migration and axon elongation (Deuel et al., 2006; Koizumi et al., 2006). As we found that MAP7D1 is predominantly expressed in the cortical plate and localized to growing axons [Fig. 3(C,D)], we first asked whether MAP7D1 might play a role in axon elongation in cortical neurons. To do this, we utilized the miRNA-based RNA interference (RNAi) to knockdown expression of endogenous MAP7D1. We chose the miRNA-based RNAi strategy because the shRNA-based RNAi was shown to have off-target effects (Baek et al., 2014). We designed two RNAi constructs targeting different cording sequences of MAP7D1 (designated as Map7d1 miRNA #1 and #2). We first confirmed that Map7d1 knockdown constructs efficiently silenced the expression of EGFP-MAP7D1 that was transiently expressed in Neuro-2a cells after 48 h (~75% and ~30% reduction for Map7d1 miRNA #1 and Map7d1 miRNA #2, respectively) [Supporting information Fig. S2(B)]. In addition, in E15 primary cortical neurons electroporated with Map7d1 miRNA constructs, we found substantial reduction of endogenous MAP7D1 proteins after 48 h based upon immunostaining with anti-MAP7D1 antiserum (~70% reduction for Map7d1 miRNA #1, ~25% reduction for Map7d1 miRNA #2, respectively) [Supporting information Fig. S2(C)].

Next, to investigate whether MAP7D1 is involved in axon elongation in vivo, we introduced the miRNA expression constructs via in utero electroporation in neural progenitor cells at E15 (Tabata and Nakajima, 2001) and labeled cortical layer 2/3 neurons that project their axons to the contralateral hemisphere through the corpus callosum [Fig. 4(A)]. In the control groups at P3, a large number of growing axons was observed around the contralateral hippocampus [Fig. 4(B,B’)]. By contrast, when Map7d1 knockdown constructs were expressed, only a small percentage of axons reached a comparable region compared with control [Fig. 4(C,C’)]. To quantitate this effect, we chose 20 longest EGFP-positive axons from each slice of the brain and measured the absolute distance to the midline. We found that the absolute distance to the midline for the MAP7D1-depleted axons was significantly shorter than that of control axons (median distance to midline was 1358.6 μm in control knockdown, n = 80 axons from four independent animals; 727.2 μm in map7d1 knockdown, n = 100 axons from five independent animals) [Fig. 4(E)]. Notably, Map7d1 miRNA #1 showed severer defects compared to Map7d1 miRNA #2 (median distance was 926.8 μm, n = 100 axons from five independent animals), consistent with the down-regulation efficiency of two RNAi constructs. As expected, Dclk1 knockdown resulted in impairment of callosal axon elongation (median distance to midline was 682.1μm for Dclk1 miRNA #1; 936.2 μm for Dclk1 miRNA #2) [Fig. 4(D,D’, and E)], which was similar to the axon elongation defects in Dclk1 knockouts (Deuel et al., 2006; Koizumi et al., 2006). Taken together, these data indicate that MAP7D1 is required for axon elongation in developing cortical neurons.

Figure 4.

MAP7D1 is required for callosal axon elongation in corpus callosum. (A) Schematic illustration of the callosal axon elongation assay using in utero electroporation. (B-D) Coronal cortical sections from brains electroporated miRNA-knockdown constructs (Control scrambled miRNA (B and B’), Map7d1 miRNA #1 (C and C’), Dclk1 miRNA #1 (D and D’)) at E15 and analyzed at P3. Representative images of EmGFP fluorescence converted to black/white images are shown. Dashed line is the midline of brain. (B-D) are high magnifications of boxed regions in (B-D) respectively. Scale bars: left 500 μm, right 100 μm. (E) Quantification of the axonal absolute distance to the midline for each condition. 20 longest axons were chosen in the slice from each mice and absolute distances to the midline of the growing axonal tips were measured. Box-and-whisker plots indicate the median (line in the box), 25th and 75th percentiles (box), the data range (whiskers) and outliers (circles). Outliers are defined as data points greater than the 75 th percentile of all data points plus 1.5 times the interquartile range or lower than the 25 th percentile of all data points minus 1.5 times the interquartile range. n = 80 axons from 4 mice for control, n = 100 axons from 5 mice for Map7d1 miRNA #1, n = 100 axons from 5 mice for Map7d1 miRNA #2, n = 60 axons from 3 mice for Dclk1 miRNA #1 and n = 160 axons from 8 mice for Dclk1 miRNA #2 were measured. Statistical significances were assessed by Kruskal-Wallis test with Conover-Inman Test. ***p < 0.001.

Previous studies reported that genetic depletion of the serine/threonine protein kinase LKB1, a key regulator of neuronal polarity, causes a significant defect in callosal axon development (Barnes et al., 2007; Shelly et al., 2007). Thus, Map7d1 knockdown might affect neuronal polarity formation. To investigate this possibility, we examined the neuronal polarization in the dissociated cortical neurons expressing Map7d1 or Dclk1 knockdown construct at DIV2. We found no significant difference in the percentage of polarized neurons between Map7d1 or Dclk1 depletion and control [Supporting information Fig. S3]. We also examined whether MAP7D1 might regulate dendrite development in dissociated cortical neurons and found that neither Map7d1 nor Dclk1 knockdown significantly affected total dendrite length, dendrite brunching, and dendrite complexity [Supporting information Fig.S4 (A), (B), (C)]. It is thus likely that MAP7D1 and DCLK1 play critical roles in axon elongation rather than axon initiation and dendrite development.

Radial Migration Is Unaffected in Map7d1 Knockdown Cortical Neurons

We next examined whether MAP7D1 is required for radial migration of cortical neurons. We introduced the miRNA-mediated knockdown constructs into neural progenitor cells in the ventricular zone of E14 mouse brain and inspected the distribution of GFP-positive cells at E18 [Fig. 5]. Previous studies demonstrated that double knockout mice for DCLK1 and DCX showed severe defects in radial migration of cortical neurons, whereas no obvious migration defects were observed in single knockout mice of either DCLK1 or DCX (Corbo et al., 2002; Deuel et al., 2006; Koizumi et al., 2006). Consistently, knockdown of either Dclk1 or Dcx caused no significant defects in cortical migration [Fig. 5(A,C, and D)], whereas in the Dclk1/Dcx double miRNA-transfected brains, the proportion of GFP-positive cell reached the uCP was significantly reduced (50.5 ± 5.8%, n = 6 slices from 3 independent animals) [Fig. 5(G,H)] compared to control (77.0 ±5.4%, n = 6 slices from 3 independent animals) [Fig. 5(A,E, and H)]. In addition, no significant difference was observed in the distribution of GFP-positive cells in double knockdown of Dcx and Map7d1 [Fig. 5(F,H)] as well as Map7d1 knockdown alone [Fig. 5(B)]. These data suggest MAP7D1 might be less important for radial migration of cortical neurons compared to axon elongation and that axon elongation defects in Map7d1 knockdown neurons are unlikely to be secondary to migration defects of cortical neurons.

Figure 5.

Radial migration is not significantly affected in Map7d1 knockdown cortical neurons (A-G) Coronal cortical sections from brains electroporated miRNA-based knockdown constructs (Control scrambled miRNA (A and E), Map7d1 miRNA#1 (B), Dclk1 miRNA#3 (C), Dcx miRNA (D), Map7d1(#1)/Dcx double miRNA (F), Dclk1(#3)/Dcx double miRNA (G)) at E14 and analyzed at E18. Representative images of EmGFP fluorescence with DAPI for the dorsolateral area of the telencephalon are shown. Scale bar, 200 μm. (H) Quantification of the radial migration in control, Map7d1/Dcx knockdown and Dclk1/Dcx knockdown. The cortical plate was divided in half to uCP and loCP, and the proportion of GFP-positive cells in uCP, loCP and the IZ were measured. Data represents mean ± SEM. Each data was measured from n = 6 cortical slices from 3 mice. Statistical significances were assessed by One-way ANOVA followed by the Bonferroni post-hoc test; **p < 0.01, ***p < 0.001. NS, not significant.

MAP7D1 Phosphorylation on Ser 315 Is Critical for Callosal Axon Elongation

Given that DCLK1 associates with and phosphorylates MAP7D1 on Ser 315 and that both DCLK1 and MAP7D1 are required for callosal axon elongation in cortical neurons, we asked whether the MAP7D1 Ser 315 phosphorylation is required for axon elongation in developing cortical neurons. Firstly, we examined whether levels of MAP7D1 phosphorylation might be changed in Dclk1 knockout neurons. To do that, we generated a polyclonal antibody against phosphorylated MAP7D1 on Ser 315. This antibody specifically recognized endogenous MAP7D1 but not lambda protein phosphatase-treated MAP7D1 [Supporting information Fig. S5(A)]. Further, the antibody recognized the in vitro phosphorylated MAP7D1 but not the MAP7D1 S315A mutant, confirming that the antibody specifically recognizes phosphorylated MAP7D1 on Ser 315 [Supporting information Fig. S5(B)]. Immunoblots with the anti-phosphorylated MAPD1 antibody indicated that p-MAP7D1/MAP7D1 ratio was ~50% reduced in Dclk1 knockout cortical neurons compared to wild-type [Fig. 6(A)], suggesting that DCLK1 is responsible, at least in part, for MAP7D1 phosphorylation on Ser 315 in cortical neurons.

Figure 6.

MAP7D1 phosphorylation on Ser 315 is critical for callosal axon elongation. (A) MAP7D1 Ser 315 phosphorylation is reduced in Dclk1 knockout neurons. Endogenous MAP7D1 was immunoprecipitated from cortical neurons at DIV6 dissociated from wild-type, Dclk1 heterozygous and Dclk1 knockout mouse followed by western blot analysis. Immunoblots using anti-MAP7D1 antibody (pAb22011) and anti-phospho Ser 315 MAP7D1 antibody (left). Quantification of phosphorylation level evaluated by the ratio of phosphorylated MAP7D1 to total MAP7D1 in wild-type Dclk1 heterozygous and Dclk1 knockout neurons (right). (B-D) Coronal cortical sections from brains electroporated with Dclk1 miRNA#1 knockdown construct together with pCAG-IRES-tdTomato (B and B’), pCAG-MAP7D1 wild-type-IRES-tdTomato (C and C’), and pCAG-MAP7D1 S315E-IRES-tdTomato (D and D’) at E15 and analyzed at P3. Representative images of tdTomato fluorescence converted to black/white images are shown. Dashed line is the midline of brain. (B’- D’) are high magnifications of boxed regions in (B–D) respectively. Scale bars: left 500 μm, right 100 μm. (E) Quantification of the axonal absolute distance to the midline. 20 longest axons were chosen in the slice from each mice and absolute distances to the midline of the growing axonal tips were measured. The box-and-whisker plots represent same as [Fig. 4(E)]. n = 140 axons from seven mice for control, n = 140 axons from seven mice for wild-type MAP7D1 and n = 160 axons from eight mice for MAP7D1 S315E were measured. Statistical significances were assessed by Kruskal-Wallis test with Conover-Inman Test. ***p < 0.001.

Next, to examine that DCLK1 regulates the callosal axon elongation through MAP7D1 phosphorylation, we asked whether a phosphomimetic MAP7D1 (MAP7D1 S315E) could rescue callosal axon defects in Dclk1 knockdown neurons. We used in utero electroporation to deliver Dclk1 knockdown construct together with MAP7D1 expression vectors which cocistronically express MAP7D1 and tandemTomato (pCAG-MAP7D1 WT-IRES-tdTomato or pCAG-MAP7D1 S315E-IRES-tdTomato, mixed at a molar ratio of 5: 1) to mouse cortical layer 2/3 neurons at E15 and observed tdTomato-positive axons at P3. We found that axon elongation defects in Dclk1 knockdown neurons was fully rescued by MAP7D1 S315E, but not by wild-type MAP7D1 [Fig. 6(B,B’,C,C’ and D,D’)]. Indeed, the absolute distance to the midline of the MAP7D1 S315E-expressing axons was comparable to that of wild-type animals (median distance for MAP7D1 S315E was 1494.0 μm, n = 160 axons from 8 independent animals, compared to [Fig. 4(B, B’, and E)]) [Fig. 6(E)]. These data strongly support the idea that DCLK1 promotes the callosal axon elongation, at least in part, through MAP7D1 phosphorylation.

We next examined whether overexpression of the unphosphorylated mutant of MAP7D1 on S315 (MAP7D1 S315A) might affect callosal axon elongation. Overexpression of MAP7D1 S315A in cortical layer 2/3 neurons resulted in a significant decrease in callosal axon length compared to control, whereas no significant difference in axon elongation was observed in neurons with wild-type MAP7D1 overexpression [Fig. 7(A,A’,B,B’, and C,C’)]. The absolute distance to midline of the MAP7D1 315A-overexpressing axons was decreased by ~50% compared to that of control (median distance was 740.0 μm in MAP7D1 S315A overexpressing neurons, n = 300 axons from 15 independent animals; 1406.1 μm in control neurons, n = 220 axons from 11 independent animals; 1594.6 μm in wild-type MAP7D1 overexpressing neurons, n = 180 axons from 9 independent animals) [Fig. 7(D)]. These data indicate that MAP7D1 S315A exerts dominant negative effects on callosal axon elongation and further support the idea that MAP7D1 phosphorylation on Ser 315 is crucial for axon elongation.

Figure 7.

Overexpression of a non-phosphorylated mutant of MAP7D1 on S315 shows dominant negative effects on callosal axon elongation. (A-C) Coronal cortical sections from brains electroporated with expression constructs (pCAG-IRES-tdTomato (A and A’), pCAG-MAP7D1 wild-type-IRES-tdTomato (B and B’), pCAG-MAP7D1 S315A-IRES-tdTomato (C and C’)) at E15 and analyzed at P3. Representative images of tdTomato fluorescence converted to black/white images are shown. Dashed line is the midline of brain. (A’-C’) are high magnifications of boxed regions in (A-C), respectively. Scale bars: left 500 μm, right 100 μm. (D) Quantification of the axonal absolute distance to the midline. 20 longest axons were chosen in the slice from each mice and absolute distances to the midline of the growing axonal tips were measured. The box-and-whisker plots represent same as [Fig. 4(E)]. n = 220 axons from 11 mice for control, n = 180 axons from 5 mice for wild-type MAP7D1, n = 300 axons from 15 mice for MAP7D1 S315A were measured. Statistical significances were assessed by Kruskal-Wallis test with Conover-Inman Test. ***p < 0.001. NS, not significant.

DISCUSSION

In this study, we have identified MAP7D1 as a novel kinase substrate of DCLK1 in cortical neurons. Starting from an unbiased affinity purification approach from murine brain, we identified MAP7D1 in tight association with DCLK1. MAP7D1 is co-expressed with DCLK1 in developing neurons of the cortical plate, and Map7d1 knockdown impaired callosal axon elongation in layer 2/3 cortical neurons, which is similar to the defects in Dclk1 knockdown neurons. Finally, the phosphomimetic MAP7D1 (MAP7D1 S315E), but not wild-type MAP7D1, fully rescued the impaired callosal axon elongation in Dclk1 knockdown neurons. These data together indicate that MAP7D1 is a major downstream target of DCLK1 in axon elongation in vivo.

MAP7D1 belongs to the MAP7 family of microtubule-associated proteins, originally identified as a MAP predominantly expressed in epithelial cells (Masson and Kreis, 1993). Four MAP7 paralogs, MAP7, MAP7D1, MAP7D2 and MAP7D3, are encoded in the mammalian genome, and phylogenetic analysis suggests that MAP7D1 is the most conserved with MAP7 (Metzger et al., 2012). Murine MAP7 was previously suggested to function in spermatogenesis presumably through organizing the microtubule structures (Komada et al., 2000). MAP7 binds microtubules through the N-terminal domain, which is highly conserved within the MAP7 family including MAP7D1 (Masson and Kreis, 1993). It is thus likely that MAP7D1 binds to microtubules through its N-terminal domain. Indeed, double staining with anti-MAP7D1 and anti-neuronal tubulin antibodies revealed that MAP7D1 is typically co-localized with microtubules in cultured cortical neurons. Previous studies proposed that other MAPs, such as MAP1b and Tau, function cooperatively in axon development through bundling microtubules (Takei et al., 2000). In growing axons, Tau is localized in the soma and entire length of axons (Kosik and Finch, 1987), whereas MAP1b is localized in the distal part of axons and the soma (Black et al., 1994). In contrast, our data suggested that MAP7D1 is predominantly localized in the proximal region of axons and the soma. Thus, MAP7D1 may play roles distinct from those of MAP1b and Tau in axonal development. It will be important to understand how MAP7D1 cooperates with other MAPs to regulate tempo-spatially axonal development in cortical neurons.

DCLK1 plays multiple distinct roles in neuronal development including axon elongation and migration (Deuel et al., 2006; Koizumi et al., 2006). In contrast to the significant defects in the commissural axon elongation in cortical neurons by MAP7D1 knockdown, no obvious migration defects were observed by Map7d1 knockdown or by Map7d1/Dcx double knockdown in cortical neurons. These data suggest that MAP7D1 phosphorylation by DCLK1 is probably more relevant to its role in axon elongation, compared with its role in neuronal migration. Our immunostaining data indicated that MAP7D1 is concentrated in the proximal region of growing axons where DCLK1 co-localized, whereas DCLK1 is localized at the growth cones as well as the proximal region of growing axons. It is thus most likely that DCLK1 phosphorylates MAP7D1 to promote axon elongation in the proximal region of growing axons.

We propose that DCLK1 promotes axon elongation of cortical neurons by phosphorylating MAP7D1 on Ser 315. First, our biochemical and immunochemical data indicate that DCLK1 associates with MAP7D1 in cortical neurons. Second, in vitro kinase assay followed by mass spectrometry demonstrated that DCLK1 directly phosphorylates MAP7D1 on Ser 315. Third, MAP7D1 S315A mutant was no longer phosphorylated by DCLK1. This notion is further supported by the evidence that the level of phosphorylated MAP7D1 was ~50% reduced in Dclk1 knockout neurons compared to wild-type neurons. Fourth, the phosphomimetic MAP7D1 (MAP7D1 S315E), but not wild-type MAP7D1, fully rescued axon elongation defects in Dclk1 knockdown neurons. Lastly, overexpression of the unphosphorylated MAP7D1 on Ser315 (MAP7D1 S315A), but not wild-type MAP7D1, in cortical neurons caused axonal elongation defects.

How does MAP7D1 phosphorylation promote axon elongation? One possible scenario is that Ser 315 phosphorylation may be involved in regulation of interactions between MAP7D1 and microtubules. Live imaging of GFP-fused MAP7 protein suggested that the MAP7-microtubules interaction is reversal with a higher turnover rate (t^1/2 = 4 sec) compared to other MAPs such as MAP1B and Tau, and that the turnover is halted by treatment with staurosporine, an inhibitor of protein kinases (Bulinski et al., 2001; Faire et al., 1999). Indeed, hyperphosphorylated MAP7 likely decreased its binding activity to microtubules (Masson and Kreis, 1995). These reports suggest that the microtubules-binding activity of MAP7 is regulated by phosphorylation as reported in other MAPs (Conde and Caceres, 2009). Given that Ser 315 is positioned inside of the potential microtubule-binding domain in MAP7D1 (Faire et al., 1999), Ser 315 phosphorylation may accelerate turnover rate of the MAP7D1-microtuble interaction, which in turn promotes axon elongation.

Alternatively, MAP7D1 phosphorylation might facilitate axonal transport in cortical neurons. Recent reports suggested that both MAP7 and Ensconsin (Ens), the Drosophila homolog of MAP7, interact with KIF5B/Khc (conventional kinesin heavy chain, kinesin-1 motor protein) to facilitate the kinesin-microtubule interaction and kinesin-1 activation (Sung et al., 2008; Metzger et al., 2012; Barlan et al., 2013). In growing axons, KIF5 selectively accumulates and transports the axonal components for axon elongation to the distal part of axons (Jacobson et al., 2006; Arimura and Kaibuchi, 2007). Consistently, KIF5 knockdown in hippocampal neurons impaired axonal elongation (Ferreira et al., 1992). Therefore, MAP7D1 phosphorylation may promote callosal axon elongation through KIF5-mediated axonal transport. Interestingly, recent studies suggest that DCLK1 regulates KIF1A-mediated axonal and dendritic transports in cultured hippocampal neurons (Liu et al., 2012; Lipka et al., 2016). It is thus feasible that DCLK1 regulates axonal and dendritic transports through different KIFs. Further studies are required to understand the mechanisms of how DCLK1-MAP7D1 pathway function in axon elongation.

Interestingly, we found that both Map7d1 miRNA and Dclk1 miRNA-expressed axons seemed to project into contralateral hemisphere by P14, which is similar to those in control neurons [Supporting information Fig. S6]. These data suggest that Map7d1 or Dclk1 knockdown results in a significant delay, rather than arrest, in callosal axon elongation. Further studies are required to examine whether the temporal delay of axon elongation by disruption of DCLK1-MAP7D1 pathway affects the final neural network connections and functions.

In summary, we have identified MAP7D1 as a novel substrate of DCLK1 and provided evidence that DCLK1 promotes axon elongation through phosphorylating MAP7D1 on Ser 315 in cortical neurons. Given that DCLK1 is likely associated with cognitive traits, schizophrenia, and attention deficit hyperactivity disorder (ADHD) (Le Hellard et al., 2009; Havik et al., 2012), it is possible that MAP7D1 dysfunction might be involved in mental diseases.

Supplementary Material

Additional Supporting Information may be found in the online version of this article.