Branching of O-mannose glycans regulates node of Ranvier organization and saltatory conduction

Abstract

The myelin sheath of axons is organized into domain structures with nodes of Ranvier that facilitate saltatory conduction. Here, we show that a brain-specific glycosyltransferase, MGAT5B that catalyzes β1,6-GlcNAc branching of an O-mannose (Man) glycan, is required for node of Ranvier integrity. Mgat5b knockout (KO) mice displayed broadening of nodes in brain white matter. Consistently, electrophysiological analysis demonstrated a significant delay and variable axonal conduction in Mgat5b KO mice, indicating the importance of branched O-Man glycans in node morphology and functions. Biochemical and glycoproteomic analyses demonstrated that MGAT5B modifies the glycans of a key node-organizing glycoprotein, neurofascin 186 (NF186), and that interaction between NF186 and Contactin 1 is negatively regulated by branched O-Man glycans. Finally, neuron-specific restoration of MGAT5B in KO mice rescued these nodal defects, indicating a cell-autonomous role of MGAT5B in node organization. Our findings highlight a glycan-mediated mechanism for the maintenance of node structure and function.

Analysis of knockout mice for a glycosyltransferase designated as MGAT5B, branching of O-mannose glycans in brain is required for organization of the nodes of Ranvier in neurons, fast conductivity, and motor ability.

Introduction

A large number of mammalian proteins are post-translationally modified with glycans which play crucial roles in modulating protein functions and behaviors, including enzyme activity, subcellular localization, and turnover¹. In the secretory pathway, glycans on proteins are biosynthesized by various glycosyltransferases and glycosidases in a stepwise manner, during the transport of nascent polypeptides through the endoplasmic reticulum (ER) and the Golgi apparatus². These enzymes act cooperatively and competitively on glycoprotein substrates to shape particular glycan profiles on proteins.

Glycans on proteins are classified into N- and O-glycans based on the amino acids to which they are attached: N-glycans are linked to Asn residues, and O-glycans are attached to Ser or Thr^2,3. Among several types of O-glycans, this study focuses on O-Man glycans, that are known to play essential roles in muscle tissues⁴. Loss of a specific O-Man glycan structure selectively attached to the specific sites in α-dystroglycan (αDG) is causal for muscular dystrophy⁵. However, the functions of other O-Man glycans remain poorly understood. Notably, O-Man glycans are abundant in the mammalian brain, accounting for 10–30% of total brain O-glycans^6–8. This suggests that O-Man glycans play crucial roles in the brain.

Brain O-Man glycans often bear a β1,6-linked GlcNAc branch, which is synthesized by a brain-specific glycosyltransferase MGAT5B (also known as GnT-IX) (Fig. 1a), that was cloned as a homolog of N-glycan branching enzyme MGAT5^9,10. Although the physiological functions of MGAT5B remain elusive, its involvement in pathology has been reported. Mgat5b KO (Mgat5b^−/−) mice showed improved pathology in a chemically induced demyelination model¹¹. This phenotype was suggested to be mediated through suppressed activation of reactive astrocytes in Mgat5b KO mouse brain, and protein tyrosine phosphatase receptor type Z (PTPRZ) seems to be a key glycoprotein modified by MGAT5B in astrocytes. Furthermore, glycosylation of PTPRZ by MGAT5B was also suggested to be involved in glioma progression, and knockdown of either MGAT5B or PTPRZ resulted in reduced glioma growth, implying that MGAT5B promotes glioma progression¹². Despite these findings, physiological functions of MGAT5B remain elusive. This enzyme is predominantly expressed in the brain⁹, particularly in neurons¹³. Although low levels of expression are also detected in oligodendrocyte precursor cells (OPCs) (Brain RNA-Seq database)¹⁴, the robust expression in neurons suggests its primary role in fundamental neurophysiological processes. Therefore, in this study, we focused on O-mannosylated glycoproteins expressed in neurons to elucidate physiological roles of MGAT5B.

Fig. 1. Elongation of nodes of Ranvier in the Mgat5b KO mouse brain.

a Schematic model of the GlcNAc transfer reaction catalyzed by MGAT5B (GnT-IX). b Representative images of co-immunostaining for Caspr (green) and Nav1.6 (red) in white matter from 10-week-old WT and Mgat5b KO mice. Scale bars: 20 μm (leftmost and third images from the left); 5 μm (second and fourth images from the left). c Schematic illustration showing the four distinct white matter regions of the brain analyzed (i–iv). Created in BioRender. Kizuka, Y. (2026) https://BioRender.com/toa6hhs. d Schematic illustration of the region measured as the width of node of Ranvier. Violin plots illustrating the distribution of node of Ranvier widths measured in the four white matter regions. Each plot represents pooled data from three individual mice. Dashed lines in each plot indicate the median value (n = 60 nodes in 3 mice, **p < 0.01, ***p < 0.001; Wilcoxon rank-sum test.). e–g 12- to 14-week-old WT and Mgat5b KO mice were subjected to open field (e), Y-maze (f), and rotarod tests (g). Lines in each plot indicate the mean value (n = 11 WT and 13 KO mice for open field and Y-maze tests, and n = 14 WT and 16 KO mice for rotarod test, *p < 0.05; unpaired t-test).

Given that Mgat5b KO mice exhibited accelerated re-myelination in the disease model¹¹, we reasoned that MGAT5B may be involved in physiological myelination-related processes by modifying O-mannosylated proteins. To date, only a limited number of glycoproteins in the brain have been identified to be modified with O-Man glycans, including αDG¹⁵, PTPRZ¹⁶, CD24¹⁷, and NF186¹⁸. Among them, NF186 is a neuronal protein localized to nodes of Ranvier. These are specialized gaps between adjacent myelin sheaths on axons that allow for clustering of sodium channels and saltatory conduction of action potentials^19,20. Mice lacking NF186 exhibited disorganized nodes of Ranvier and dispersed sodium channels, which result in early lethality²¹. This demonstrates that NF186 plays a crucial role in the formation of nodes of Ranvier and is essential for saltatory conduction. Previous glycoproteomic analyses have revealed that NF186 carries branched O-Man glycans at multiple sites¹⁸, raising a possibility that its node-organizing function is regulated by MGAT5B. Here, we show that Mgat5b KO mice displayed altered node of Ranvier morphology and slower axonal conduction than wild-type (WT) mice. Our findings underscore the importance of O-Man glycans in neurophysiology, providing insights into the mechanism of glycan-mediated regulation of saltatory conduction.

Results

Nodes of Ranvier are expanded in Mgat5b KO mice

NF186 is specifically localized at nodes of Ranvier and is critically involved in forming the organized nodal structure^19,21. Branched O-Man glycans were also detected on NF186¹⁸, strongly suggesting that NF186 is a substrate glycoprotein of MGAT5B. We therefore hypothesized that loss of MGAT5B might lead to alteration in functions of NF186, thereby affecting node of Ranvier morphology. To test this hypothesis, brain sections from WT and Mgat5b KO mice were co-immunostained for Nav1.6 and Caspr, markers for node and its flanking paranode, respectively²², and morphology of these nodal structures was examined (Fig. 1b). Nodes of Ranvier widths in the four white matter regions (Fig. 1c) were compared between WT and Mgat5b KO mice, and we found that the nodal widths in Mgat5b KO mice were significantly increased compared with those in WT mice (Fig. 1d). These results indicate that the loss of MGAT5B results in a consistent enlargement of the nodes of Ranvier in myelinated axons throughout the cerebrum.

To evaluate whether the structural alterations in the nodes of Ranvier impact the behaviors of Mgat5b KO mice, we performed open field, Y-maze, and rotarod tests (Fig. 1e–g). Among these assessments, we observed a significant impairment, specifically in rotarod performance in Mgat5b KO mice compared to WT mice (Fig. 1g). As deficits in rotarod performance are known to be associated with abnormal axonal conduction between motor cortices²³, these results suggest that the morphological changes in the nodes of Ranvier in Mgat5b KO mice lead to functional deficits in motor coordination.

Delay and temporal dispersion of axonal conduction in Mgat5b KO mice

Nodal width greater than approximately 1.0 μm is negatively correlated with axonal conduction²⁴. Therefore, we reasoned that axonal conduction would be slower in Mgat5b KO neurons than that in WT. To test this idea, we performed electrophysiological recordings. An adeno-associated virus (AAV) vector encoding EYFP and Channelrhodopsin-2 (ChR2) for optical depolarization was injected into layer 5 of the left motor cortex of WT and Mgat5b KO mice (Fig. 2a). Three weeks post-AAV injection, optical stimulation was delivered to the contralateral motor cortex, and the electric responses were recorded (Fig. 2b). Measuring such antidromic spikes without synaptic transmission allowed us to precisely evaluate the conduction properties of axons, compared with orthodromic measurements which include synaptic transmission. We observed that the latency was significantly longer in Mgat5b KO mice than in WT mice (Fig. 2c, d). Furthermore, simultaneous recordings of multiple action potentials revealed that latency was significantly variable in Mgat5b KO mice compared with that in WT mice (Fig. 2c, e). These findings indicate that neuronal conduction is delayed and abnormally variable in the motor cortex of Mgat5b KO mice. The increased latency variability in Mgat5b KO mice indicates that their nodal widths in the KO mice are heterogeneous compared with WT mice.

Fig. 2. MGAT5B deficiency leads to delayed and temporally dispersed axonal conduction in callosal projections between motor cortices.

a AAV-mediated expression of ChR2-EYFP. The schematic depiction (center) illustrates the injection of AAV1-Syn-ChR2(H134R)-EYFP into the left motor cortex. Representative fluorescence images show ChR2-EYFP expression ipsilaterally at the injection site (right panel) and contralaterally in callosal axon terminals (left panel). Scale bars, 1 mm. b Schematic depiction of the in vivo experimental setup. Optogenetic stimulation was applied to callosal axon terminals in the right motor cortex to evoke antidromic spikes, which were recorded from the corresponding somata in the left motor cortex using a 16-channel electrode. c Four representative traces from different channels show spikes evoked by a 1-ms blue light pulse (top panel). A corresponding raster plot illustrates the spike timings from all 16 recording channels (bottom panel). d Increased spike latency in Mgat5b KO mice. Violin plots depict the distribution of latencies for every recorded spike (spike level) and the mean latency for each animal (individual level). (n = 1806 spikes from 7 WT mice and n = 1533 spikes from 6 Mgat5b KO mice. Spike level: ***p < 0.01, Wilcoxon rank-sum test; Individual level: ***p < 0.001, unpaired t-test). e Increased temporal dispersion of spike latencies in Mgat5b KO mice. Violin plots show the distribution of the standard deviation and coefficient of variation of spike latencies (n = 146 recordings from 7 WT mice and n = 120 recordings from 6 Mgat5b KO mice. Recording level: ***p < 0.001, Wilcoxon rank-sum test; individual level: **p < 0.01, ***p < 0.001, unpaired t-test).

MGAT5B modifies O-Man glycans on NF186

Next, we explored the mechanism by which loss of O-Man glycans resulted in enlarged nodes and slower axonal conduction. We focused on the O-Man-modified glycoprotein NF186¹⁸, which is a critical node organizer²¹. To investigate whether MGAT5B modifies NF186 O-Man glycans, we first co-expressed MGAT5B and a soluble truncated form (extracellular domain) of Myc-His-tagged NF186 (NF186∆TM) for purification. NF186 is also modified with O-GalNAc type glycans¹⁸; therefore, to clearly detect the differences in O-Man glycans, we used HEK293 cells lacking C1GALT1²⁵, which is a responsible enzyme for O-GalNAc glycan extension²⁶, as host cells for expression of NF186∆TM (Fig. 3a). Purified NF186∆TM was further treated with PNGaseF to remove all N-glycans, and removal of N-glycans was confirmed by the loss of reactivity with concanavalin A (ConA) (Supplementary Fig. 1a) that recognizes N-glycans²⁷. PNGaseF-treated purified NF186∆TM was analyzed by western and lectin blotting. Blotting with Maackia amurensis lectin (MAM), Ricinus communis agglutinin I (RCA-I), and Griffonia simplicifolia lectin II (GSLII), which recognizes α2,3-sialic acid²⁸, β1,4-Gal²⁹, and terminal GlcNAc residues³⁰, respectively, showed an upward shift in the size of NF186∆TM in SDS-PAGE gel upon MGAT5B overexpression (Fig. 3b). In addition, an increase in GSL-II signal could be derived from the generation of β1,6-GlcNAc branch by MGAT5B. We also confirmed that N-glycan branching is impaired in Mgat5^−/− brain but unaltered in Mgat5b^−/− (Supplementary Fig. 1b), by lectin blotting with leukoagglutinating-phytohemagglutinin (L4-PHA)³¹ and Datura stramonium agglutinin (DSA)³², consistent with the previous reports^11,33, supporting the notion that the primary substrates of MGAT5B in brain are O-Man glycans but not N-glycans. We also confirmed that the band pattern of N-glycosylated Nav1.6 was not altered in Mgat5b^−/− brain (Supplementary Fig. 1c). These findings indicate that O-Man glycans of NF186 are modified by MGAT5B.

Fig. 3. Modification of NF186 glycan by MGAT5B.

a Myc-His-tagged NF186ΔTM was expressed in HEK293 C1GALT1 KO cells with or without co-expression of MGAT5B and purified from the culture medium. Purified NF186ΔTM was treated with PNGaseF and subjected to SDS-PAGE and CBB staining. b Purified NF186ΔTM was treated with PNGaseF and blotted with anti-Myc, MAM, RCA-I, and GSLII. Proteins in whole cell lysates were western blotted for MGAT5B. c NF186 glycopeptides detected by glycoproteomic analysis that were only found in either MGAT5B (−) or MGAT5B (+) samples. Deduced glycan structures are shown. d Signal intensities of the detected glycopeptides in LC-MS analysis are shown.

To directly show that MGAT5B modifies the O-Man glycans on NF186, we next performed mass spectrometry (MS)-based glycoproteomic analysis of NF186∆TM purified from HEK293 C1GALT1 KO cell medium and treated with PNGaseF. Using Byonic software, we detected 5 glycopeptides in NF186 that were only found in either MGAT5B (−) or MGAT5B (+) samples (Fig. 3c, d). The extracted ion chromatograms and the MS² data of the glycopeptides are shown in Supplementary Figs. 2 and 3, respectively. The glycans found in MGAT5B (−) samples were assigned to be linear O-Man glycans, while those in MGAT5B (+) samples were branched (Fig. 3c, d). Although the detailed quantitative comparison of each glycan structure on O-mannosylted glycopeptides between MGAT5B (−) and (+) was technically difficult, our data strongly suggests that MGAT5B generates O-Man glycan branches on NF186.

In-node localization of NF186 is altered in Mgat5b KO neurons

Based on the critical functions of NF186 for node organization²¹ and its glycan modification by MGAT5B, we hypothesized that loss of branching of O-Man glycans on NF186 in Mgat5b KO mice would disrupt NF186 function, thereby disrupting the node structure. We first examined the protein expression level of NF186 by western blotting and observed no significant difference in levels between WT and Mgat5b KO mouse brains (Fig. 4a). We also confirmed that the NF186 (Nfasc) mRNA level was unchanged in Mgat5b KO mice (Fig. 4b). In addition, we investigated the levels of major proteins involved in the formation of myelin, nodes and paranodes, including Caspr, a neuronal protein involved in axon-myelin interactions³⁴, Contactin 1, a neuronal protein involved in paranodes formation³⁴, and myelin-associated glycoprotein (MAG), a major glycoprotein in myelin³⁵. Western blotting revealed that the levels of all these proteins were comparable between WT and Mgat5b KO mice (Fig. 4c). These results indicate that the nodal enlargement in Mgat5b KO mice is not caused by changes in the level of NF186 or other node-organizing proteins.

Fig. 4. Levels of NF186 and related proteins in the Mgat5b KO mouse brain.

a Western blotting of NF186 in brain homogenates from 10- to 30-week-old WT and Mgat5b KO mice. Samples were probed with antibodies against NF186 and GAPDH. Signal intensity for NF186 was quantified and normalized against that of GAPDH, and the values relative to WT are shown (n = 6 mice; mean ± S.D.; n.s. not significant; unpaired t-test). b Real-time PCR analysis of the Nfasc mRNA levels in the brains of 10-week-old WT and Mgat5b KO mice. The expression levels were normalized against those of 18S rRNA, and the values relative to WT are shown (n = 5 mice; mean ± S.D.; n.s. not significant; unpaired t-test). c Western blotting of myelin-related proteins in brain homogenates from 10- to 30-week-old WT and Mgat5b KO mice. Samples were probed with antibodies against Caspr, Contactin 1, and MAG. Signal intensities were quantified, and the values relative to WT are shown (n = 6 mice; mean ± S.D.; n.s. not significant; unpaired t-test). d Immunofluorescence staining of brain white matter (region (iv) in Fig. 1c) in mouse brain sections with antibodies against Caspr (red) and NF186 (green). Scale bar: 5 μm. Created in BioRender. Kizuka, Y. (2026) https://BioRender.com/toa6hhs. e Schematic illustration of the region measured as the width of the NF186-positive region. Violin plots quantifying the width of the NF186-positive region within the white matter. Measurements were taken from brain sections stained as described in (a). (n = 200 nodes in 10 WT or 6 KO mice, ***p < 0.001; unpaired t-test).

We next analyzed intranodal localization of NF186 by co-immunostaining of NF186 and Caspr in white matter. In WT mice, NF186 was detected in the nodes flanked by Caspr-positive paranodes (Fig. 4d). In Mgat5b KO mice, NF186 was also localized in the nodes, suggesting that the fundamental node structure is maintained (Fig. 4d). However, measurement of the widths of the NF186-positive region revealed that it was narrower in Mgat5b KO mice than in WT mice (Fig. 4e). These findings indicate that the absence of MGAT5B leads to the altered localization of NF186 in the node of Ranvier.

Branched O-Man glycan negatively regulates the interaction between NF186 and Contactin 1

The above results suggest that loss of glycan modification of NF186 in Mgat5b KO mice resulted in altered NF186 function. To reveal the mechanism of this effect, we next focused on protein complexes involving NF186, because the lack of one of its interacting partners (e.g., Brevican and the Na⁺ channel β1 subunit) causes the nodes of Ranvier to become disorganized, resulting in decreased conduction velocity^36,37. To search for NF186-interacting proteins, we immunoprecipitated NF186 from mouse brain homogenates and performed a proteomic analysis of the co-precipitating proteins. Two specific bands (Band 1 and Band 2) observed in the NF186-immunoprecipitation (IP) sample but not in the negative control with normal rabbit IgG (Fig. 5a) were subjected to proteomic analysis. Among four identified proteins (Table 1), we focused on Contactin 1 because it has a large extracellular region, is known to be localized closely to nodes and paranodes, and plays a critical role in neuron–myelin interactions³⁸.

Fig. 5. Interaction between NF186 and Contactin 1.

a Immunoprecipitation of NF186 from brain homogenates of WT mice. Co-precipitated proteins were separated by SDS-PAGE and visualized by silver staining. Normal rabbit IgG was used as a negative control for immunoprecipitation. Bands 1 and 2, specifically present in the NF186-IP samples, were excised for proteomic analysis. The proteins identified by MS are listed in Table 1. b Immunoprecipitation of NF186 from mouse brain homogenates, followed by SDS-PAGE and immunoblotting for Contactin 1 and NF186. The intensity of the co-immunoprecipitated Contactin 1 signal was normalized against that of the immunoprecipitated NF186 and further by that of Contactin 1 in the brain lysate sample. These normalized values from Mgat5b KO mice were divided by those from paired WT mice (n = 12 mice; mean ± S.D.; *p < 0.05; unpaired t-test). c HEK293 cells were transfected with empty vector or expression plasmids encoding CNTN1-Fc, NF186ΔTM, and MGAT5B. CNTN1-Fc was pulled down from the conditioned medium using Protein G-beads. The culture medium before pulldown and the pulldown samples were subjected to western blotting to detect NF186ΔTM co-precipitated with CNTN1-Fc. The signal intensity of co-immunoprecipitated NF186 was normalized against that of the NF186 in Input and further normalized against that of pulled down CNTN1. Expression of MGAT5B was confirmed by western blotting of cell lysates (n = 5 cultures; mean ± S.D.; ***p < 0.001; unpaired t-test). d A docking model for NF155-Contactin 1 with branched O-Man glycans. The model was built based on the crystal structure of the NF155-Contacitn 1 complex (PDB: 7OL4)³⁹. Branched O-Man glycans were attached using CHARMM-GUI Glycan Reader&Modeler. Pink, NF155; Light blue, Contactin 1.

Table 1.

Proteomic analysis of proteins co-precipitated with NF186 from mouse brain

Sample Name		Protein	Gene symbol	MW	Score	Peptide	Coverage	Accession	Note*
Band 1	1	Contactin 1	Cntn1	113,388	144	10	10	P12960	P12960
	2	Sodium/potassium-transporting ATPase subunit alpha-3	Atp1a3	111,692	76	5	5	Q6PIC6	Q6PIC6
Band 2	1	Guanine nucleotide-binding protein G(o) subunit alpha	Gnao1	40,085	105	9	24	P18872	P18872
	2	Fructose-bisphosphate aldolase A	Aldoa	39,356	36	2	7	P05064	P05064

Note* : Accession in Mascot Search Results

To examine whether the interaction between NF186 and Contactin 1 is affected by loss of branched O-Man glycans, NF186 was immunoprecipitated from WT and Mgat5b KO mouse brain homogenates, and the amount of co-precipitated Contactin 1 was evaluated. The level of Contactin 1 co-precipitated with NF186 was increased in Mgat5b KO mice compared with that in WT mice (Fig. 5b). To further confirm this finding, we performed similar co-pulldown experiments using cultured cells. NF186ΔTM and Fc-tagged Contactin1 (CNTN1-Fc) were exogenously expressed in HEK293 C1GALT1 KO cells with or without co-expression of MGAT5B. CNTN1-Fc was precipitated from the conditioned media using Protein G-beads, and the levels of co-precipitated NF186ΔTM were analyzed by western blotting. Overexpression of MGAT5B resulted in a significant reduction in the amount of NF186ΔTM associated with Contactin 1 (Fig. 5c), indicating that the branched O-Man glycans produced by MGAT5B have a negative impact on the interaction between NF186 and Contactin 1.

To gain structural insights into how the branched O-Man glycans on NF186 impact its interaction with Contactin 1, we constructed a model of the complex using the experimentally solved co-crystal structure of the Contactin 1 extracellular domain and NF155 (PDB ID: 7OL4)³⁹, a splicing variant of NF186, with O-Man glycan. NF155 and NF186 share the identical Ig domains, and NF186 was reported to carry branched O-Man glycans at multiple sites, including T92¹⁸. Our glycoproteomics analysis (Fig. 3c) also identified another modification site T203 in this crystalized region. Therefore, we modeled O-Man glycans at T92 and T203 of NF155 in the NF155-Contacin 1 complex to assess their potential effects on the NF186-Contactin 1 complex (Fig. 5d). The modeled O-Man glycan at T92 was located within the inner cavity of the horseshoe-shaped Ig domain of NF155 in close proximity to several N-glycans (N316- and N420-glycans of NF155 and N208-glycan of Contactin 1) (Fig. 5d). As these N-glycans contribute to stabilization of the complex³⁹, our model raised the possibility that linear and less bulky T92 O-Man glycans stabilize the complex by interacting with these N-glycans. In addition, the T203 O-Man glycan is located closely to the interface with Contactin-1, and extension of the branched glycan likely causes a steric hindrance. These predictions suggest that the bulky branched O-Man glycans lead to destabilization of the NF186-Contactin 1 complex.

Branched O-Man glycan has a neuron-autonomous role in maintaining the node structure

Finally, we investigated whether the role of branched O-Man glycans in node organization is cell autonomous. MGAT5B expression was restored in layer 5 neurons of the motor cortex in Mgat5b KO mice by injecting an AAV1 encoding MGAT5B-IRES-mScarlet under the Synapsin (Syn) promoter (Fig. 6a, left). We first confirmed the ability of this AAV to overexpress MGAT5B in neurons by infecting primary cultured cortical neurons isolated from WT mouse embryos with the same AAV, and verified that both MGAT5B and mScarlet were expressed in neurons as expected (Supplementary Fig. 4a, b). Three weeks after the AAVs encoding mScarlet or MGAT5B-IRES-mScarlet were injected into Mgat5b KO mouse brains, the widths of nodes of Ranvier in white matter were measured and compared between control (mScarlet) and MGAT5B-expressing (MGAT5B-IRES-mScarlet) mice (Fig. 6a, right). As a result, the enlarged nodes of Ranvier observed in Mgat5b KO neurons were rescued in MGAT5B-expressing neurons (Fig. 6b and Supplementary Fig. 4c). This indicates that Syn promoter-driven expression of MGAT5B in neurons is sufficient to re-organize the node structure in the Mgat5b KO mouse brain.

Fig. 6. Node of Ranvier length in Mgat5b KO mice with exogenous MGAT5B expression in neurons.

a Left: the schematic depiction illustrates the injection of AAV1-Syn-mScarlet or MGAT5B-IRES-mScarlet into the left motor cortex (region (iv) in Fig. 1c). Right: representative images of immunostaining for Caspr (green) in the white matter of 25-week-old Mgat5b KO mice injected with either control mScarlet (magenta) AAV or MGAT5B-IRES-mScalet AAV. Scale bars: 5 μm. b Violin plots showing the width of nodes of Ranvier in the white matter of mouse brains. Dashed lines in each plot indicate the median value (n = 149 nodes from 3 control mice and 149 nodes in 2 rescued mice, ****p < 0.0001; Wilcoxon rank-sum test.).

Discussion

In this study, we demonstrated that MGAT5B plays a critical role in the formation and maintenance of nodes of Ranvier under physiological conditions, thereby contributing to axonal conduction. As a possible mechanism of its involvement, we showed augmented interaction of NF186 with Contactin 1 and altered localization of NF186 within nodes of Ranvier in Mgat5b KO neurons (Fig. 7). In addition to previous studies describing pathological roles of MGAT5B in demyelination and glioma growth^11,12, our study reveals a physiological function of branched O-Man glycans in neurons.

Fig. 7. Model depicting regulation of saltatory conduction by O-Man glycan branching.

In WT neurons, a node organizer NF186 is modified with branched O-Man glycans, which negatively regulate the interaction of NF186 with Contactin 1. In Mgat5b KO neurons, branching of O-Man glycans on NF186 is lost, which enhances the interaction between NF186 and Contactin 1, leading to an enlarged node, thereby inducing delay and temporal dispersion of axonal conduction. Created in BioRender. Kizuka, Y. (2026) https://BioRender.com/toa6hhs.

We observed a significant delay and variable latency in the axonal conduction in the motor cortex of Mgat5b KO mice. This finding is relevant to the significantly impaired performance in the rotarod test (Fig. 1g). The previous studies have established that abnormalities in axonal conduction within callosal projections connecting the motor cortices result in deficits in rotarod performance²³. While significant differences were not observed in the open field or Y-maze test, the presence of enlarged nodes across various brain regions suggests that axonal conduction could be affected in multiple neural circuitry. The lack of behavioral phenotypes in certain tests might reflect the specific sensitivity of those paradigms or potential compensatory mechanisms in the associated circuitry. The abnormalities in axonal conduction identified in this study represent a loss of temporal precision, which has been previously demonstrated to be a fundamental requirement for motor learning^40,41. While we did not directly measure neuronal synchronization in this study, such abnormalities in axonal conduction are expected to be a factor that potentially disrupts the precise timing of neuronal firing across hemispheres. Therefore, we suggest that this disrupted temporal precision represents a key physiological mechanism underlying the behavioral deficits observed in Mgat5b KO mice.

Regarding a possible mechanism that can explain node enlargement in Mgat5b KO axons, aberrant formation of paranodal junction could occur. Previous studies showed that forming paranode critically serves as barriers to restrict the position of the nodal proteins³⁶ and that defective paranode formation results in nodal enlargement⁴². At the molecular level, Contactin 1 plays an essential role in paranode formation³⁸ by interacting with Caspr and oligodendroglial NF155^43,44. Our observation of an enhanced interaction between Contactin 1 and NF186 in Mgat5b KO mice, probably in the node region, indicates that reduced distribution of Contactin 1 in the paranodal region could contribute to the nodal enlargement in Mgat5b KO neurons (Fig. 7). Our results extend the “double barrier model” proposed by ref. ³⁶. by demonstrating that not only the presence of NF186 protein, but also its proper glycosylation state, is crucial for preventing the lateral diffusion of Nav1.6 within the nodal compartment.

We focused on the interaction between NF186 and Contactin 1 in neurons; however interaction between Contactin 1 and oligodendroglial NF155 has also been previously studied³⁹. NF155 and NF186 share most of their extracellular domains, including the threonine residues (T92 and T203) modified with O-Man, and MGAT5B is expressed in OPCs^45,46, raising the possibility that NF155 may also be modified by MGAT5B and that branched O-Man on NF155 may regulate paranode formation. Notably, MGAT5B expression is barely detectable in mature oligodendrocytes⁴⁵. This suggests that O-Man glycan structures in oligodendrocyte lineage cells change with differentiation, which might contribute to regulation of the NF155-Contactin 1 interaction. Analyzing the O-Man glycan structures in the oligodendrocyte lineage and assessing interactions between NF155 and Contactin 1 in Mgat5b KO mice are warranted in future studies.

Although O-Man glycans account for 10–0% of all O-glycans in the mammalian brain, depending on brain regions^6,8,47, their carrier proteins remain poorly characterized. The best-studied O-mannosylated glycoprotein is αDG whose unique O-Man glycan structure (core M3-type) is essential for skeletal muscle integrity^4,5. However, the absence of αDG barely alters the profile of brain O-Man glycans⁴⁸, indicating that other glycoproteins mainly carry O-mannosylated glycans in the brain. Although NF186, PTPRZ, its soluble isoform phosphacan, and CD24 can be modified with branched O-Man glycans^16,17, little is known about what other glycoproteins carry MGAT5B-modified branched O-Man structures. As specific probes are lacking for detecting O-Man glycans, technical development is needed for non-biased detection of O-mannosylated glycopeptides by glycoproteomic analysis. Moreover, the whole glycan structures of extended branched O-Man glycans for each carrier protein are also unclear. As our glycoproteomics analysis detected the fucosylated glycans on NF186 only in MGAT5B (+) sample, the branch formation by MGAT5B could be related to O-Man glycan extension. Consistent with this, we recently reported that MGAT5B-dependent formation of β1,6-branch acts as an efficient scaffold for keratan sulfate extension on O-Man glycans¹⁰. Solving the structures of whole O-Man glycans on each carrier protein in brain is an important next issue.

In conclusion, this study demonstrates that MGAT5B plays a critical role in the formation and maintenance of nodes of Ranvier, as well as in maintaining the temporal precision of axonal conduction. Our findings establish a functional link between nodal broadening and behavioral deficits in motor coordination, revealing a physiological function for branched O-Man glycans in the central nervous system. While the precise causal relationship between the specific glycosylation state of NF186, the enlargement of nodes, and the resulting abnormalities in axonal conduction remains to be further clarified, this study provides a foundation for understanding glycan-mediated axonal regulation. In addition, conditional KO of Mgat5b would more directly elucidate the temporal and cell type-specific role of MGAT5B for node organization. Continued investigation into the physiological and pathological roles of MGAT5B will not only deepen our understanding of O-Man glycan functions in the brain but also help elucidate glycan-mediated mechanisms of demyelinating diseases and brain tumors.

Methods

Antibodies and lectins

The following antibodies were used: anti-Caspr (rabbit, Cell Signaling Technology, 97736), anti-Caspr (mouse, abcam, ab252535), anti-Nav1.6 (mouse, NeuroMab, 75-026), anti-Nav1.6 (rabbit, Millipore, AB5580), anti-NF186 (rabbit, Cell Signaling Technology, 15034), anti-GAPDH (mouse, Merck Millipore, MAB374), anti-MAG (rabbit, Cell Signaling Technology, 9043), anti-Contactin 1 (goat, R&D Systems, AF904), anti-MGAT5B (rabbit, Novus Biologicals, NBP3-05134), anti-myc (mouse, clone 4A6; Millipore, 05-724), horseradish peroxidase (HRP)-conjugated anti-mouse IgG (Cytiva, NA931V), HRP-conjugated anti-rabbit IgG (Cytiva, NA934V), HRP-conjugated anti-goat IgG (Jackson ImmunoResearch Labolatories Inc., 111-005-003), HRP-conjugated anti-human IgG (proteintech, SA00001-17), Alexa546-conjugated anti-mouse IgG (ThermoFisher Scientific, A10036), Alexa488-conjugated anti-rabbit IgG (ThermoFisher Scientific, A21206), Cy5-conjugated anti-mouse IgG (ThermoFisher Scientific, A10524), Agarose-conjugated WGA (Funakoshi, AL-1023), MAM (J-chemical, J110), biotinylated RCA-I (Vector Laboratories, B-1085), biotinylated GSLII (Vector Laboratories, B-1215), biotinylated ConA (MGC Woodchem, J203), unconjugated L4-PHA (MGC Woodchem, J212), and unconjugated DSA (MGC Woodchem, J105). The MAM, L4-PHA, and DSA lectins used for blotting were conjugated with horseradish peroxidase (HRP) using the Peroxidase Labeling Kit–NH₂ (Dojindo, LK11), according to the manufacturer’s instructions.

Animal experiments

Mgat5b KO (Mgat5b^−/−) and Mgat5 KO (Mgat5^−/−) mice were generated in the previous studies^11,49. Mice with C57BL/6N genetic background were housed in four or fewer per cage and bred at 20–26 °C and 40–70% humidity, under the cycle of 12 h light/dark. All experiments followed the guideline for animal experiments of Gifu University and Nippon Medical School and were approved by the respective animal experiment committees (Gifu University: No. AG-P-C-20250025; Nippon Medical School: No. 2024-033). The animal experiments in this study did not have humane endpoints.

Immunofluorescence staining

WT and Mgat5b KO mice were anesthetized and perfused with PBS and subsequently with 10% Formalin Neutral Buffer Solution (Wako). Brains were taken and immersed in the same fixation solution at 4 °C overnight, followed by immersion in 30% sucrose/PBS at 4 °C for 3 days with daily buffer renewal. Brains were sectioned at 30 μm thickness with a microtome (REM-710, YAMATO KHOKI).

For staining Nav1.6 and Caspr, sections were incubated with 0.01% Proteinase K/PBS for 6 min at room temperature for antigen retrieval, and Alexa Fluor Tyramide SuperBoost Kit (ThermoFisher Scientific) was used according to the manufacturer’s protocol. The activity of endogenous peroxidase was quenched with 3% hydrogen peroxide at room temperature for 1 h. The sections were blocked with 10% goat serum at room temperature for 1 h and incubated with anti-Nav1.6 antibody at 4 °C overnight, followed by incubation with HRP-conjugated secondary antibody at room temperature for 1 h and subsequently with Tyramide solution for 5 min. The sections were incubated with anti-Caspr antibody at 4 °C overnight, followed by incubation with the secondary antibody at room temperature for 30 min. For staining NF186 and Caspr, the sections were heated at 95 °C for 20 min in a buffer (10 mM Tris, 1 mM EDTA, 1% Tween 20) for antigen retrieval. The sections were blocked with PBS containing 3% bovine serum albumin (BSA) and 0.1% Nonidet P-40 (NP-40) at room temperature for 30 min and incubated with the primary antibodies at 4 °C overnight, followed by incubation with the secondary antibodies at room temperature for 30 min. After staining, the sections were mounted by Prolong diamond antifade reagent (ThermoFisher Scientific), and fluorescence images were obtained using a BZX-800 (KEYENCE).

Image analysis in the white matter region of the brain sections was performed using ImageJ software. For measuring the node lengths, node (Nav1.6-positive regions) and paranodes (Caspr-positive regions) were selected, and the distances between the edges of paranodes were manually measured as the lengths of node. For quantification of NF186-positive area, the distances between the edges of NF186-positive area in the node flanked by Caspr-positive area were manually measured.

Behavioral tests

Mice were habituated to the animal facility 1 h before behavior test began. For rotarod test, in training phase, mice were placed on the rod (Ugo Basile S.R.L., Varese, Italy) moving at 4 rpm for 2 min, and this was repeated three times. In testing phase, mice were placed on the rod moving at 4 rpm and rod speed was increased linearly from 4 to 40 rpm over 5 min. The time until the mouse fell from the rod was measured, and the average was calculated based on three trials. The rotarod apparatus was cleaned with 70% ethanol after every test to remove odors and excrements. A break of at least 15 min was scheduled between each trial.

For Y-maze test, mice were placed in Y-maze apparatus (40 cm, in length, 10 cm in width, and 12 cm in height) and they were allowed to move freely for 8 min. Mouse behavior was recorded using a video camera from above and analyzed manually. Alternating behavior was defined as entering three consecutive different arms. Alternation ratio was calculated according to the following formula: Alternation (%) = ([Number of alternations]/[Total arm entries−2]) × 100. The Y-maze apparatus was cleaned with 70% ethanol after every test to remove odors and excrements.

For open-field test, mice were placed in open field (30 cm in length, 30 cm in width, and 15 cm in height) and they were allowed to move freely for 15 min. Mouse behavior was recorded using a video camera from above. The total distance moved in the arena and the time spent in the center (15 cm in length, 15 cm in width) were analyzed using EthoVision XT (Noldus, Wageningen, Netherlands). The open field equipment was cleaned with 70% ethanol after every test to remove odors and excrements.

The investigators performing the tests and analyzing the data were blinded to the mouse genotypes.

Electrophysiology

To investigate the effects of MGAT5B deficiency on axonal conduction, 20-week-old WT and Mgat5b KO mice were used. AAV1-Syn-channelrhodopsin (H134R)-EYFP (ChR2; 0.5 μL, 2.3 × 10¹³ vector genomes/ml) was injected into layer 5 of the left motor cortex (0.2 mm anterior and 1.0 mm lateral from the bregma, and 0.5 mm below the pia mater). Electrophysiological experiments were performed 2–3 weeks after AAV injection. In vivo recordings were conducted under 0.8–1% isoflurane anesthesia. Antidromic spikes were evoked by optogenetic stimulation of callosal axons in the right motor cortex. Stimulation consisted of 20 pulses of blue light (473 nm, Thorlabs, Newton, NJ; 1 ms duration, 1 Hz frequency) delivered via a fiber-optic cable (~10 mW at the tip) positioned over the cortical surface. Evoked antidromic spikes were recorded from the left motor cortex using a 16-channel linear electrode. Recorded spikes were identified as antidromic for further analysis based on three criteria: a reliable response (evoked by ≥10 stimuli), a short latency (< 30 ms from stimulus onset), and low temporal jitter (< 0.3 ms)^40,41. To evaluate the temporal dispersion of antidromic spike latencies, the standard deviation and coefficient of variation were calculated using spikes simultaneously detected on at least two channels of the 16-channel electrode. The final dataset was composed of 1806 antidromic spikes from 146 recordings in seven WT mice and 1533 antidromic spikes from 120 recordings in six Mgat5b KO mice.

Preparation of brain homogenate

WT, Mgat5 KO, and Mgat5b KO mouse brains were homogenized in 50 mM Tris-HCl (pH7.4) containing 150 mM NaCl and protease inhibitor cocktail using Potter homogenizer. The homogenates were used for subsequent experiments.

Western blotting, lectin blotting, Coomassie brilliant blue (CBB), and silver staining

The protein concentrations of samples were measured using Pierce BCA Protein Assay Kit (ThermoFisher), and the proteins were dissolved in Laemmli sample buffer and denatured at 95 °C for 5 min. Proteins were resolved by 5–20% SDS-PAGE. For western blotting, proteins in the gel were transferred to nitrocellulose membranes or PVDF membranes using a semi-dry blotter. The membranes were blocked with TBS containing 5% skim milk and 0.1% Tween 20 and then incubated with a primary antibody, followed by an HRP-conjugated secondary antibody. For blotting with HRP lectins, membranes were blocked with 1% BSA in TBS-T (TBS containing 0.1% Tween 20) overnight at 4 °C, followed by incubation at room temperature for 30 min with HRP-conjugated lectins that were diluted with 1% BSA in TBS-T. For blotting with biotinylated lectins, membranes were blocked with TBS-T for 30 min, followed by overnight incubation at 4 °C with biotinylated lectins diluted with TBS-T. After washing with TBS-T, the membranes were incubated with Streptavidin-HRP in VECTASTAIN ABC Standard kit (Vector Laboratories) (1:400 dilution in TBS-T) at room temperature for 1 h. Protein bands were detected with Western Lightning Plus-ECL (PerkinElmer Life Sciences) or SuperSignal West Femto Maximum Sensitivity substrate (ThermoFisher Scientific) using FUSION-SOLO 7 s EDGE (Vilber Lourmat). For CBB and silver staining, proteins in the gels were stained using GelCode Blue Safe Protein Stain (Thermo Fisher) and Silver Stain MS Kit (FUJIFILM), respectively, according to the manufacturer’s protocol.

Immunoprecipitation

The brain homogenates were solubilized in 500 μl of the lysis buffer (TBS, 60 mM Octyl-β-D-glucoside) containing protease inhibitor cocktail by sonication. After ultracentrifugation at 100,000 × g for 15 min at 4 °C, 0.5 mg of Dynabeads Protein G (ThermoFisher Scientific) was added to the supernatant, followed by gentle rotation for 1 h at 4 °C to capture the proteins which non-specifically bound to the beads. After removing the beads, antibody and Dynabeads Protein G were added to the supernatant, followed by gentle rotation at 4 °C overnight. The beads were washed with the buffer (TBS, 6 mM Octyl-β-D-glucoside) three times, and the proteins bound to the beads were eluted by boiling in Laemmli sample buffer.

Co-pulldown of NF186∆TM and Contactin 1-Fc

For co-pulldown experiments, Fc-tagged Contactin 1 (CNTN1-Fc) and NF186∆TM were co-expressed in previously established HEK293 C1GALT1 KO cells²⁵. After 4 h of transfection, the culture media were replaced with Opti-MEM I, followed by incubation for 44 h. The conditioned media were collected and centrifuged to remove cellular debris (950 × g for 5 min at 4 °C). The supernatants were incubated with 1 mg of Dynabeads Protein G (Thermo Fisher Scientific) that had been equilibrated with binding buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, and 1× Protease Inhibitor Cocktail Set V [EDTA-free] (Wako)). The mixtures were rotated for 3 h at 4 °C. The beads were washed three times with wash buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 1 mM EDTA), and proteins bound to the beads were eluted by adding Laemmli sample buffer to the beads and heating at 95 °C for 5 min. The supernatants were collected and used as samples for SDS–PAGE.

Glycoproteomics

To purify recombinant NF186∆TM, HEK293 C1GALT1 KO cells were transiently co-transfected with the plasmids encoding NF186∆TM and full-length MGAT5B using poly-ethyleneimine MAX (Polyscience) when confluence reached 70–80%. After 4 h, the medium was replaced with Opti-MEM I (Gibco), followed by 2 day-culture. Soluble myc-his-tagged NF186∆TM was purified through a Ni²⁺-column. The eluates containing 10 μg of NF186∆TM were suspended in 500 μL of a reducing solution containing 250 mM Tris-HCl (pH 8.5), 6 M guanidine hydrochloride, 2 mM EDTA, and 11 mg of dithiothreitol (DTT), and incubated at 50 °C for 1 h. The reduced proteins were alkylated with 20 mg of iodoacetamide (IAA) by incubating at room temperature for 1 h in the dark. The reaction mixture was passed through a NAP-5 column (Cytiva) equilibrated with 50 mM NH₄HCO₃ to remove salts and excess IAA. The desalted solution was evaporated to dryness. The dried residue was dissolved with 150 μL of 50 mM NH₄HCO₃ containing 0.5 μg trypsin (Thermo Fisher Scientific), and then incubated at 37 °C for 16 h. After boiling, 1 μg of endoprotease Glu-C (Thermo Fisher Scientific) dissolved in 20 μL of 50 mM NH₄HCO₃ was added, and the mixture was incubated at 37 °C for 16 h. After boiling, 2U of PNGaseF (Roche) dissolved in 2 μL of 100 mM NaH₂PO₄ (pH 7.3) was added, and the mixture was incubated at 37 °C for 16 h. After boiling, the solution was evaporated to dryness.

The sample was reconstituted with 15 μl of 0.1% (v/v) formic acid in water, centrifuged (13,200 × g × 5 min at 4 °C), and 1 μl of the supernatant was used for LC-MS analysis. The peptides and O-glycosylated peptides were separated on an ODS column (Nano HPLC Capillary Column; 3 μm C18, 120 mm × 75 μm ID; Nikkyo Technos) at a flow rate of 300 nL/min using nano-liquid chromatography system (Vanquish Neo; Thermo Fisher Scientific) by the direct injection mode, which does not involve trap elution with a trap column. The mobile phases for separation were solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). The column was eluted using a sequence of isocratic and linear gradient elution: 0–1 min, 4.8% (v/v) solvent B; 1–61 min, 4.8–32.8% (v/v) solvent B; 61–66 min, 32.8–40% (v/v) solvent B; 66–70 min, 40–72% (v/v) solvent B; keep to 72% solvent B for 5 min; and then re-equilibrate with 4.8% (v/v) solvent B using combined control mode (maximum flow rate; 2 μL/min, maximum pressure; 200 bar). The eluate from the separation column was continuously introduced into a nanoESI probe (Thermo Fisher Scientific) and analyzed by MS and two types of MS/MS (Orbitrap Eclipse; Thermo Fisher Scientific). In the nanoESI probe setting, the spray voltage for positive ion was 1800 V, the capillary temperature was 275 °C, and internal mass calibration was set EASY-IC (Run start). MS data were acquired in positive ion mode over the mass range m/z 500 to m/z 2000 using the Orbitrap (resolution; 60,000, maximum injection time; 100 ms, AGC target; 200% (8e5), RF lens; 30%). Data-dependent MS/MS was performed using a top-speed method (cycle time 3 s), selecting the most intense precursor ions with charge states z = 2–8. Dynamic exclusion was enabled (exclude duration; 15 s after 1 time, intensity threshold; 2.5e4) to minimize repeated selection of the same precursor. For initial fragmentation, selected precursor ions were isolated in the quadrupole with an isolation window of 1.6 m/z and subjected to higher-energy collisional dissociation (HCD). The normalized collision energy (NCE) was set to stepped 20–30–40 to promote glycan oxonium ion formation while retaining peptide backbone information. HCD MS/MS spectra were acquired in positive ion mode with a scan range mode defined with a first mass of m/z 120 using the Orbitrap (resolution; 15,000, maximum injection time; 22 ms, AGC target; 1000% (5e5)).

A glycan-triggered EThcD workflow was implemented. HCD MS/MS spectra were continuously monitored for diagnostic glycan oxonium ions (m/z138.0545 (HexNAc-fragment), 204.0867 (HexNAc), 366.1396 (Hex-HexNAc)). When at least one predefined oxonium ion was observed, the corresponding precursor ion was automatically re-isolated and subjected to electron-transfer/higher-energy collision dissociation (EThcD). For EThcD, precursor ions were isolated with the same quadrupole window and reacted with fluoranthene reagent anions using an ETD reaction time optimized for the charge state. Supplemental HCD activation was applied with a low NCE (25%) to enhance fragment ion coverage. EThcD MS/MS spectra were acquired in positive ion mode with a scan range mode defined with a first mass of m/z 120 using the Orbitrap (resolution; 15000, maximum injection time; 150 ms, AGC target; 1000% (5e5)). Qual Browser and Free Style (ver. 1.8) included in Xcalibur software ver. 4.7. (Thermo Fisher Scientific) was used to show the base peak chromatogram (BPC), extracted ion chromatogram (EIC), and to calculate the peak intensity of monoisotope of the corresponding glycoform in the EIC. The structure of the glycopeptides was identified using MS/MS spectral data with the Byonic software ver. 5.9.5. (Protein Metrix).

Proteomics

After immunoprecipitation with anti-NF186 antibody and silver staining, two bands (Band 1 and 2 in Fig. 5a) from WT mouse samples were excised from the gel. Subsequent proteomic identification was carried out at JPROteomics (Sendai, Japan). The gel pieces were destained with 50 mM NH₄HCO₃ in 50% CH₃CN, dehydrated with CH₃CN, and evaporated. The samples were reduced in 100 mM DTT in 100 mM NH₄HCO₃ at 56 °C for 30 min. After removing the supernatant, the sample was alkylated in 100 mM iodoacetamide in 100 mM NH₄HCO₃ at 37 °C for 30 min. The supernatant was removed, and the samples were washed first with 100 mM NH₄HCO₃ and then with CH₃CN and evaporated. The proteins were digested in the gel with trypsin in 50 mM Tris-HCl, pH 7.5, at 35 °C for 17 h. The samples were desalted using a ZipTipC18 column (Millipore), and the peptides were eluted with 3 μl of 70% CH₃CN containing 0.1% TFA. The samples were diluted with 20 μl of 2% CH₃CN containing 0.1% formic acid and analyzed using a Bio NanoLC (KYA Technologies) linked to a QSTAR XL qTOF mass spectrometer (Applied Biosystems). The peptides were separated using an HiQ sil C18W-3 column (0.1 × 50 mm, KYA Technologies) and a four-step linear gradient (10 min at 2% CH₃CN, 0.1% formic acid; 30 min at 2–41% CH₃CN, 0.1% formic acid; 10 min at 41–80% CH₃CN, 0.1% formic acid; 20 min at 80% CH₃CN, 0.1% formic acid) at a flow rate of 200 nl/min. The MS spectra were obtained in the positive ion mode (ion spray voltage: 1.8 kV). The assignment of the MS/MS data to tryptic peptides was performed using ABSciex Analyst QS software version 1.1 (Applied Biosystems/SCIEX).

Plasmid construction

Primers used in this study are listed in Supplemental Table S1. pcDNA6.2/human MGAT5B-FLAG was constructed as described previously⁵⁰. To construct pcDNA6/mouse NF186ΔTM-myc-HisA, cDNA encoding the whole extracellular domain of mouse NF186 was amplified by PCR using a C57BL/6 adult mouse brain cDNA library as the template. The amplified fragment was inserted into the EcoRI/XhoI sites of pcDNA6 myc-HisA using Gibson Assembly. To construct pEF-Fc/human Contactin 1, cDNA encoding Met1 to Ser980 of human Contactin 1 was amplified by PCR using pME-Puro-sHA-Contactin 1 constructed as described previously⁵¹ as the template. The amplified fragment was inserted into the SalI/SpeI sites of pEF-Fc using DNA Ligation kit ver2.1 (Takara). To construct pGP-AAV-syn-WPRE/hMGAT5B-IRES-mScarlet, we utilized pGP-AAV-syn-WPRE/human prion-IRES-mScarlet as the template. To first construct pGP-AAV-syn-WPRE/human prion-IRES-mScarlet, cDNA encoding prion was amplified by PCR from HCT116 cell cDNA library and inserted to the EcoRI/XhoI sites of pcDNA6 myc-HisA using Gibson Assembly. Then the prion cDNA (template: pcDNA6-mycHisA/hprion), the IRES sequence (template: pIRES2-ZsGreen1 [Takara Bio]), and the mScarlet cDNA (template: pAAV-CaMKIIa-rsChRmine-oScarlet-KV 2.1-WPRE [Addgene: #183523, a gift from Karl Deisseroth]) were amplified by PCR and inserted into the EcoRI site of pGP-AAV-syn-jGCaMP8f-WPRE (Addgene: #162376, a gift from GENIE Project) using Gibson Assembly. To generate pGP-AAV-syn-WPRE/mScarlet, the mScarlet fragment was amplified by PCR using pAAV-CaMKIIa-rsChRmine-oScarlet-KV 2.1-WPRE as a template and inserted into the EcoRI site of pGP-AAV-syn-jGCaMP8f-WPRE by Gibson Assembly. To construct pGP-AAV-syn-WPRE/hMGAT5B-IRES-mScarlet, the IRES-mScarlet sequence and the hMGAT5B coding sequence were amplified by PCR using pGP-AAV-syn-WPRE/human prion-IRES-mScarlet and pcDNA6.2/hMGAT5B-FLAG as a template, respectively. The two fragments were then simultaneously inserted into the EcoRI site of pGP-AAV-syn-jGCaMP8f-WPRE by Gibson Assembly.

AAV production

HEK293T cells were seeded onto six 15 cm dishes and cultured to achieve approximately 70% confluency. Cells were transfected with 53 μg of the AAV expression plasmid (pGP-WPRE/AAV1-Syn- mScarlet or pGP-WPRE/AAV1-Syn-MGAT5B-IRES-mScarlet), 53 μg of the serotype plasmid (pAAV2.1, Addgene: #112862, a gift from James M. Wilson), and 84 μg of the helper plasmid (rAAV2-retro helper, Addgene: #81070, a gift from Alla Karpova and David Schaffer) using 270 μl of 2 mg/ml polyethyleneimine-MAX (Polysciences). After 96 h of incubation, 5 M NaCl was added to the medium to a final concentration of 0.5 M. The mixture was gently agitated at room temperature for 30 min, and the culture medium was harvested for virus purification. Cellular debris was then removed by centrifugation at 3400 × g for 15 min at 4 °C. The resulting supernatant was filtered through a 0.2 μm bottle-top filter. The filtrate was concentrated to 15 ml using a VIVAFlow50 tangential flow filtration system (MWCO 100 kDa) (Sartorius), then further concentrated to 4 ml using an Amicon Ultra-15 centrifugal filter unit (MWCO 50 kDa) (Amicon). The concentrate was subsequently diluted with 5 ml of PBS and reconcentrated to 4 ml using Amicon Ultra-15. To degrade contaminating nucleic acids, 1,000 units of Benzonase (Sigma-Aldrich) was added, followed by incubation at 37 °C for 60 min. For AAV purification via ultracentrifugation, OptiPrep (SEW Serumwerk Bernburg AG) solutions (15, 25, 40, and 60%) were prepared by diluting OptiPrep with PBS containing 500 mM NaCl, 20 mM MgCl₂, and 50 mM KCl. To facilitate layer visualization, 150 μl of 0.04% phenol red was added only to the 40% OptiPrep solution. A discontinuous gradient was formed in a 13 ml Ultra-Clear centrifuge tube (BECKMAN COULTER) by carefully loading 2 ml of 60%, 2 ml of 40%, 2.5 ml of 25%, and 3.5 ml of 15% OptiPrep solutions. Finally, 2 ml of the virus-containing concentrate was loaded on top. Ultracentrifugation was performed at 230,000 × g for 3 h at 18 °C. The virus-rich 40% OptiPrep fraction was carefully collected and concentrated to 500 μl using an Amicon Ultra-15 (50 kDa) unit. The concentrate was washed twice with 5 ml of PBS and reconcentrated to 500 μl after each wash. The final purified AAV solution was stored for subsequent injection experiments.

Quantification of viral titer

The viral genome (vg) titer of purified AAV preparations was determined by quantitative PCR (qPCR). The plasmid containing the AAV2 inverted terminal repeat (ITR) sequence (pGP-WPRE/AAV-Syn-MGAT5B-IRES-mScarlet) was used to generate a standard curve. Viral samples were serially diluted and mixed with 2× SYBR Green Master Mix (Applied Biosystems), forward (5’-GGAACCCCTAGTGATGGAGTT-3’) and reverse (5’-CGGCCTCAGTGAGCGA-3’) primers. The qPCR reactions were performed under the following conditions: 95 °C for 3 min, followed by 40 cycles of 95 °C for 15 s and 58 °C for 30 s. Viral genome titers (vg/mL) were calculated from the Ct values obtained for the ITR amplification using the standard curve generated from the concentration-known plasmid.

Culture of mouse cortical primary neuron

Primary cortical neurons were isolated from embryonic day 18 (E18) mouse brain as follows. Embryonic brains were isolated in Neurobasal medium (Gibco), and the cerebellum and midbrain were removed. The cortical tissues were minced in cold Neurobasal medium, followed by incubation on ice for 5 min. The supernatant was carefully removed, and the pellet was resuspended in Hanks’ Balanced Salt Solution (HBSS). After incubation for 5 min on ice, the supernatant was removed. The tissues were then resuspended in 0.25% trypsin (in HBSS) and incubated at 37 °C for 15 min with gentle agitation. 250 μl of 10 mg/ml DNase I (Rosh) was added, followed by incubation at 37 °C for 1 min. The cell suspension was gently mixed by pipetting, centrifuged at 800 × g for 5 min at 4 °C. The pellet was resuspended in HBSS containing 250 μL of DNase I, and the suspension was incubated at 37 °C for 3 min. 15 mL of HBSS was added, and the cell suspension was centrifuged at 800 × g for 5 min at 4 °C, and the resulting pellet was resuspended in Neurobasal medium. The cell suspension was passed through a 100-μm cell strainer to obtain a single-cell suspension. Cells were plated onto poly-D-lysine-coated culture dishes at a density of 5 × 10⁵ cells/mL. The culture medium was composed of Neurobasal medium supplemented with 2% (v/v) B-27 (Gibco) and 0.5 mM glutamate. Cultures were maintained at 37 °C in a humidified 5% CO₂ incubator, and half of the medium was replaced every 5 days.

AAV infection of cultured neurons

At DIV5, the purified AAV solutions were directly added to the neuronal culture medium. Virus was added at the final concentration of 1 μL of the virus solutions (AAV1-mScarlet: 1.01 × 10¹² vg/mL, AAV1- MGAT5B-IRES-mScarlet: 5.98 × 10¹⁰ vg/mL) per 1 mL medium, followed by further incubation for 5 days. Half of the conditioned medium was replaced with the equal volume of the fresh complete culture medium.

Expression of mScarlet was examined by fluorescence microscopy (EVOS Floid Cell Imaging Station, Thermo Fisher) between DIV8 and DIV12 to verify transduction efficiency. At DIV14, neurons were harvested and analyzed by western blotting.

Injection of AAV into mouse brain

For rescue experiments, AAV1 encoding mScarlet or MGAT5B-IRES-mScarlet was injected into layer 5 of the left motor cortex (0.2 mm anterior and 1.0 mm lateral from the bregma, and 0.5 mm below the pia mater) of 20–30-week-old Mgat5b KO mice.

Structural model of NF186-Contactin 1 complex

The model of glycosylated Contactin 1 complexed with glycosylated Neurofascin-155 (NF155) was constructed based on the crystal structure (PDB ID: 7OL4)³⁹ of the Ig domains of mouse NF155 (24-623) in complex with the Ig domains of Contactin 1 (35-604). The 3D structure of this model was presented using PyMOL (Schrödinger, Inc.). Based on the previously identified site for O-mannosylation in NF186¹⁸ and the site identified in this study, Core M2 glycan was generated and attached to T92 and T203 of NF155 by the Glycan Reader & Modeler module of CHARMM-GUI⁵².

RNA extraction, reverse transcription, and real-time PCR

The total RNA was extracted from WT and Mgat5b KO mice brains by homogenization with TRI Reagent (Molecular Research Center) and reverse-transcribed using SuperScript IV First-Strand Synthesis System (ThermoFisher Scientific) with random hexamers. The target cDNAs were amplified using a TaqMan Gene Expression Master Mix (Applied Biosystems) and the primers and probes described below. The amplified cDNAs were detected using a CFX Connect Real-Time PCR Detection System (BioRad). The primers and probes were purchased from Applied Biosystems as follows: Mm00813912_m1 for Nfasc and Hs99999901_s1 for eukaryotic rRNA. The mRNA levels of Nfasc were normalized to those of the rRNA levels.

Statistics and reproducibility

Statistical analyses were performed using GraphPad Prism 8 software (GraphPad Software, Inc) and RStudio (https://docs.posit.co/ide/user/#rstudio-ide-oss-downloads). Unpaired t test was used for comparison between two groups (except Figs. 1d, 4e, and 6b), and Wilcoxon rank-sum test was used for Figs. 1d, 2d, e, 4e, and 6b. All the experimental numbers (n) in this study mean biological replicates. All experiments were performed at least two independent experiments. For quantification and statistical analyses, data were collected from at least three biological replicates as described in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Y.K. conceived the study and supervised the project. Y.K. and D.K. designed experiments and acquired funding. S.T. T.N performed biochemical experiments. T.I. and D.K. performed electrophysiological and rescue experiments. K.O. and M.N. performed glycoproteomics and proteomics analyses. H.F., M.S., and S.T. performed behavioral tests. S.O. and Y.Y. built structural models. M.H. validated the data and provided critical comments. S.T. and Y.K. wrote the original manuscript draft, and Y.K. and D.K. revised the manuscript.

Peer review

Peer review information

. Communications Biology thanks Robert Mealer, Pamela Stanley and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Benjamin Bessieres. A peer review file is available.

Funding

This work was partially supported by the FOREST program [JPMJFR2145 to D.K. and JPMJFR215Z to Y.K.] from JST, Grants-in-Aid for Scientific Research (B) [24K02222 to Y.K.], Grants-in-Aid for Scientific Research (C) [JP23K06302 to D.K.], the Core-to-Core Program [JPJSCCA202000007 to Y.K.] and J-PEAKS program [JPJS00420230009] from the Japan Society for the Promotion of Science (JSPS), an AMED-CREST grant [JP23gm1410011 to Y.K.] from the Japan Agency for Medical Research and Development (AMED), and the Human Glycome Atlas Project (HGA) from the Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT).

Data availability

Proteomic raw data and search files for protein identification have been deposited in the ProteomeXchange Consortium (announced ID: PXD069311) via the jPOST partner repository (announced ID: JPST004119). Glycoproteomic raw data have been deposited in GlycoPOST (announced ID: GPST000680). Data are publicly available at each database and accessible by using announced ID. All other data are contained within the manuscript. Uncropped blot/gel images are shown in Supplementary Fig. 5.

Code availability

This study did not develop or use any original code.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-10622-0.