NCAN is a major CSPG produced by developing cortical neurons. (a) Detection of the neoepitope of CSPGs after digestion with chondroitinase ABC (Chase). (b) Immunoblotting of the neoepitope from undigested (Chase-) and digested (Chase +) cerebral cortex lysates prepared at E18.5. The arrowhead indicates the major CSPG at 130 kDa. (c) List of peptide fragments identified from the 130-kDa band. Positions indicate the amino acid number in full-length NCAN. (d) Domain structure of mouse NCAN. Ig: immunoglobulin-like domain, Link: hyaluronan-binding link module, EGF: epidermal growth factor-like repeat, CLD: C-type lectin domain, CS: chondroitin sulfate chain. (e, f) NCAN expression in the developing cerebral cortex from E13.5 to postnatal day (P) 42. The arrowheads indicate the full-length and N-terminal fragment of NCAN. Values in (f) are normalized to GAPDH and represented relative to P2. N = 3 for each point. Mean ± SD. (g) Quantitative RT-PCR analysis of Ncan mRNA in the developing cerebral cortex. Values are normalized to Gapdh and represented relative to E13.5. N = 3 for each point. Mean ± SD. (h) Immunoblot analysis of NCAN in the cultured medium of primary cultured cortical neurons 1, 3, and 5 days after plating. (i) Experimental model for in utero labeling. Radial glial cells in the VZ were labeled with GFP on E14.5. GFP-positive cells were isolated 0.6–4 days later. (j) Expression of Ncan mRNA in GFP-labeled cells isolated on the indicated days after in utero electroporation. Values are normalized to Gapdh and represented relative to 0.6 days. N = 3 for each point. Mean ± SD.

Neuron-derived NCAN forms a pericellular matrix with HA. (a) In situ hybridization analysis of Ncan mRNA on the E16.5 cerebral cortex. (b) Localization of NCAN protein (green) in the E16.5 cerebral cortex. Nuclei were counterstained with DAPI (blue). (c) Distribution of HA (green) visualized by the biotinylated HA-binding protein (b-HABP) in the E16.5 cerebral cortex. (d) High magnification views of a Tuj-1-positive primary cultured cortical neuron (green) 5 days after plating. White arrows indicate the co-localization of NCAN (magenta) and HA (cyan). Orthogonal projections in the X-Z and Y-Z planes taken along the white lines showed the localization of NCAN and HA at the adhesion sites between the neuron and culture substrate. (e) Schema of the pull-down assay for analyzing binding between endogenous HA and its interactors. (f) Immunoblotting of the input, precipitate (P), and supernatant (S) with an anti-NCAN antibody. NCAN was precipitated with HA by adding b-HABP (b-HABP +). NCAN was not precipitated without b-HABP (b-HABP-) or after digestion with hyaluronidase (HAase +). (g) Co-precipitation of NCAN with exogenously added biotinylated HA (b-HA +) or endogenous HA (b-HABP +). Scale bars represent 200 μm (a, c) and 5 μm (d).

Screening of the interacting partners of NCAN. (a) Schematic of the pull-down assay to identify NCAN-interacting partners. (b) Co-precipitation of TNC with the full-length and C-terminal half of NCAN. Interactions between NCAN and TNC disappeared following the addition of EDTA but were not affected by Chase digestion. The N-terminal half of NCAN did not bind to TNC. (c, e) TNC expression in the developing cerebral cortex from E13.5 to postnatal day (P) 42. Values in (e) are normalized to GAPDH and represented relative to E18.5. The same brain lysate as in Figure. 1e were analyzed. The values for GAPDH were calculated based on the data presented in Figure 1e. N = 3 for each point. Mean ± SD. (d) Quantitative RT-PCR analysis of Tnc mRNA in the developing cerebral cortex. Values are normalized to Gapdh and represented relative to E13.5. N = 3 for each point. Mean ± SD. (f, g) In situ hybridization analysis of Tnc mRNA (f) and immunohistochemical localization of TNC protein (g) in the E16.5 cerebral cortex. The scale bar represents 200 μm.

List of proteins identified from GFP-NCAN and negative control resins.

Alphafold2 prediction of the NCAN-TNC complex. (a) Domain structure of mouse (m) TNC. EGF: epidermal growth factor-like repeat, FNIII: fibronectin type-III domain. (b) Five predicted complex models of the mNCAN and mTNC were generated using AlphaFold2 multimer implemented in ColabFold, and the best-predicted complex was shown. (c, d). Sequence alignments of CLD of mNCAN and rat (r) ACAN (c) and FNIII-3-5 of mTNC and rTNR (d). Circles under alignments indicate the key residues of the rACAN-rTNR complex. Triangles over alignments indicate the residues in the interface of mNCAN-mTNC. Green and magenta circles/triangles are the residues involved in the interaction between L4 loop of CLD and βC, F, G strands of FNIII-4 and β6, 7 strands of CLD and CC’ loop of FNIII-4, respectively. Black circles/triangles indicate residues involved in the interaction between CLD and FNIII-4-5 linker region/FG loop of FNIII-5. (e) Model for forming the ternary complex of NCAN, HA, and TNC.

HA, NCAN, and TNC form the ternary complex in the developing cerebral cortex. (a) Triple staining of the E17.5 mouse cerebral cortex with HA (cyan), NCAN (magenta), and TNC (green). (b) High magnification images show the co-localization of the three components in the upper part of the SP/IZ but not in the CP or VZ. (c, d) Localization of TNC (c, magenta), NCAN (c, d, cyan), and HA (d, magenta) around GFP-labeled bipolar neurons (green) in the upper SP/IZ at E17.5, three days after in utero labeling. Orthogonal views taken along the white dashed lines showed the contact between the ternary complex and the surface of bipolar neurons, as indicated by the arrows. Scale bars represent 100 μm (a), 5 μm (b), and 2 μm (c, d).

Defects in NCAN and TNC retards neuronal migration. (a) Low magnification images of FT-labeled cells (black or green) in WT and DKO mouse coronal sections at E16.5. Nuclei were counterstained with DAPI (blue). (b, c) Radial distribution of FT-labeled cells (green) in the lateral (b) and medial (c) cortices of WT and DKO mice at E16.5. A quantitative analysis of migration profiles across the cortex is shown on the right. The cerebral cortex is divided into five equal areas (bins 1–5) from the pia to the ventricle, and the proportion of FT-labeled cells in each bin was calculated. The nuclei staining images (blue) on the left illustrate the boundaries of the cortical layers. N = 19–20 mice per group. Mean ± SD; *p < 0.05; Student’s t-test. Scale bars represent 200 μm (a) and 50 μm (b, c).

Single deletion of NCAN or TNC results in mild abnormalities in neuronal migration. (a) Radial distribution of FT-labeled cells (green) in the lateral cortices of WT, NCAN KO, and TNC KO mice at E16.5. A quantitative analysis of migration profiles across the cortex is shown on the right. The nuclei staining images (blue) on the left illustrate the boundaries of the cortical layers. N = 20-21 mice per group. Mean ± SD; *p < 0.05 vs. WT; Dunnett’s test. (b, c) Localization of TNC in WT and NCAN KO mice at E16.5 (b). Localization of NCAN in WT and TNC KO mice at E16.5 (c). The normalized fluorescence intensity profiles of TNC and NCAN are shown on the right of the images. The maximum intensity value for WT mice was set to 1. N = 8-13 mice for each group. Mean ± SD (shaded area). (d) In utero electroporation of Turbo-RFP alone or with GFP-fused full-length NCAN (GFP-NCAN) into the NCAN KO brain at E16.5. (e) Immunostaining of TNC (magenta) around Turbo-RFP-positive neurons (cyan) 2 days after electroporation. (f) Staining intensity of TNC around the control and GFP-NCAN-expressing neuron. N = 20 images per group. Mean ± SD; *p < 0.05; Student’s t-test. (g) High magnification of the boxed region in (e). The arrows indicate the juxtaposed localization of TNC (magenta) and GFP-NCAN (green). Scale bars represent 50 µm (a, b, c), 5 µm (e), and 2 µm (g).

Transient disruption of the ternary complex by hyaluronidase injection. (a) Localization of HA, NCAN, and TNC in the cerebral cortex two days after intraventricular injection of PBS or hyaluronidase (HAase) at E14.5. The normalized fluorescence intensity profiles of HA, NCAN, and TNC are shown on the right of each image. The maximum intensity value for PBS-injected mice was set to 1. N = 5 mice for each group. Mean ± SD (shaded area). (b) Immunoblot analysis of NCAN and TNC in cerebral cortex lysates two days after intraventricular injection of PBS or HAase. The broadening of the NCAN band is due to the absence of chondroitinase ABC digestion. (c) The NCAN (left) and TNC (right) amounts were represented relative to the PBS-injected group. N = 4 mice per group. Mean ± SD; *p < 0.01; Student’s t-test. (d, e) Distribution of FT-labeled cells (green) in the lateral (d) and medial (e) cortices two days after injection of PBS or HAase at E14.5. N = 9–10 mice per group. Mean ± SD; *p < 0.05; Student’s t-test. Scale bars represent 50 µm.

Delayed multipolar-to-bipolar transition in DKO mice. (a) Schematic of the morphological analysis. (b, c) Migration of mCherry-labeled cells in the WT and DKO cerebral cortices 53 h after in utero labeling (b). High magnification images of mCherry-labeled cells in the IZ (c). The images below show the morphology of neurons in the upper IZ (U-IZ) and lower IZ (L-IZ). (d, e) The length-to-width ratio (e) and the major neurite angle (f) of mCherry-positive neurons in WT and DKO mice. N = 128–134 cells from 6 mice per group. *p < 0.05; Student’s t-test. (f) The proportion of bipolar neurons among mCherry-labeled neurons in the WT and DKO cortices. N = 6 mice per group. Mean ± SD; *p < 0.05; Student’s t-test. Scale bars represent 50 μm (b).

Morphological maturation of cortical neurons by TNC and NCAN.

(a) Experimental model for in utero cell labeling and the primary neuronal culture. (b) Morphological stages of primary cultured cortical neurons. (c) Representative images of GFP-labeled neurons cultured for 2 days on cover glasses coated with the control substrate, poly-L-ornithine (POL), and 10 μg/mL of TNC, HA, and NCAN. (d-f) The length of the longest neurite (d), total length of neurites (e), and number of neurites (f) of neurons cultured on the indicated substrate for 2 days. N = 55–80 cells per condition. *p < 0.05 vs Control (Con); Dunnett’s test. (g) The percentage of neurons with each morphological stage after culturing on the indicated substrate for 2 days. N = 4 wells per condition. Mean ± SD. *p < 0.05 vs Con for stage 1; Dunnett’s test. (h) Representative images of GFP-labeled neurons derived from WT and NCAN KO mice cultured for 2 days on cover glasses coated with POL. (i-k) The length of the longest neurite (i), total length of neurites (j), and number of neurites (k) of WT and NCAN KO neurons cultured for 2 days. N = 96–101 cells per condition. *p < 0.05; Student’s t-test. Scale bars represent 20 μm (c, h).

Location of GFP-positive cells 0.6, 2, and 4 days after in utero electroporation. Most GFP-positive cells are in the VZ on day 0.6. GFP-positive cells begin to migrate radially through the SP/IZ on day 2 and reach the CP on day 4. The scale bar represents 100 μm.

Pull-down assay of HA with recombinant GFP-fused NCAN. (a) Recombinant expression of GFP-fused full-length, N-terminal, and C-terminal half of NCAN. The culture medium of transfected HEK293 cells was digested with chondroitinase ABC (Chase) and analyzed by immunoblotting with an anti-GFP antibody. (b) Quantification of precipitated HA by the full-length, N-terminal, and C-terminal half of NCAN. A negative control experiment (NC) was conducted without NCAN. N = 3 for each point. Mean ± SD.

Alphafold2 prediction of NCAN-TNC complex. (a) Model confidence of NCAN-TNC complex. Each residue is colored by pLDDT score, and a high pLDDT score indicates high accuracy. (b) Crystal structure of rat (r) ACAN-TNR complex (PDB ID: 1TDQ). Each domain is colored the same as in Figure 4c. (c) Sequence alignments of fourteen FNIII of mTNC. Green letters indicate residues involved in the interaction between FNIII-4 and CLD of NCAN.

Characterization of DKO mice. (a) Immunoblot analysis of TNC and NCAN from WT and DKO mouse cerebral cortex lysates. (b) Immunostaining of NCAN and TNC on coronal sections of WT and DKO mouse brains at E17.5. Scale bars represent 200 µm. (c) No significant difference in cerebrum weights between WT and DKO mice at E16.5. N = 5 mice per group. Mean ± SD; ns > 0.05; Student’s t-test.

Restored neuronal migration in DKO mice after three days of labeling. (a, b) Radial distribution of FT-labeled cells (green) in the lateral (a) and medial (b) cortices of WT and DKO mice at E17.5. A quantitative analysis of migration profiles across the cortex is shown on the right. The nuclei staining images (blue) on the left illustrate the boundaries of the cortical layers. N = 4 mice per group. Mean ± SD; ns > 0.05; Student’s t-test. Scale bars represent 50 µm.

(a) Immunostaining of NeuN-positive (green) and Ctip2-positive (magenta) neurons in the cerebral cortices of WT and DKO at 2 weeks of age. Nuclei were counterstained with DAPI (blue). (b, c) Distribution of NeuN-positive (b) or Ctip2-positive (c) neurons in the cerebral cortices of WT and DKO mice. N = 3 mice per group. Mean ± SD; ns > 0.05; Student’s t-test. The scale bar represents 200 µm (a).

Histochemical analysis of radial glial cells, intermediate progenitor cells, and the morphology of radial fibers in DKO mice. (a) Immunostaining of Pax6-positive radial glial cells (green) and Tbr2-positive intermediate progenitor cells (red) in the VZ of WT and DKO cerebral cortices at E16.5. Bar graphs show the numbers of Pax6- and Tbr2-positive cells per 150 × 150 µm2. N = 5 mice per group. Mean ± SD; ns > 0.05; Student’s t-test. (b) Immunostaining of nestin-positive radial fibers in WT and DKO cerebral cortices at E16.5. (c) High magnification images of mCherry-positive bipolar neurons and radial fibers in the WT and DKO at E16.5, 2 days after in utero labeling. Bipolar neurons in the SP/IZ-attached radial fiber, irrespective of genotypes. Scale bars represent 20 µm (a), 50 µm (b), and 5 µm (c).

Localization of HA in WT, DKO, NCAN KO, and TNC KO mice. (a) Distribution patterns of HA in the E16.5 cerebral cortices of WT and DKO mice. The fluorescence intensity profile is shown on the right. The maximum intensity value for WT mice was set to 1. N = 11 mice for each group. Mean ± SD (shaded area). (b) Comparison of HA staining in the E16.5 cerebral cortices of WT, NCAN KO, and TNC KO mice. The normalized fluorescence intensity profiles of HA are shown on the right of the image. The maximum intensity value for WT mice was set to 1. N = 7-9 mice for each group. Mean ± SD (shaded area). Scale bars represent 50 µm.

Morphological analysis of cortical neurons cultured for 3 days on cover glasses coated with HA, NCAN, and TNC. (a) Representative images of GFP-labeled neurons after culturing on the indicated substrate for 3 days. (b-d) The length of the longest neurite (b), total length of neurites (c), and number of neurites (d) of neurons cultured on the indicated substrate for 3 days. N = 55– 78 cells per condition. *p < 0.05 vs Control (Con); Dunnett’s test.

Neurite outgrowth analysis with anti-Tuj-1 antibody staining (a) Representative images of neurons stained with anti-Tuj-1 antibody after 3 days of culture on the indicated substrate. (b-d) The length of the longest neurite (b), total length of neurites (c), and number of neurites (d) of neurons. Neurites were analyzed using AutoNeuriteJ. N = 30–76 cells per condition. *p < 0.05 vs Control (Con); Dunnett’s test.

Neurite outgrowth analysis of WT and NCAN KO neurons with anti-Tuj-1 antibody staining (a) Representative images of WT and NCAN KO neurons stained with anti-Tuj-1 antibody after 3 days of culture. (b-d) The length of the longest neurite (b), total length of neurites (c), and number of neurites (d) of neurons. Neurites were analyzed using AutoNeuriteJ. N = 70–80 cells per condition. *p < 0.05; Student’s t-test.