Microglia-neuron interaction at nodes of Ranvier depends on neuronal activity through potassium release and contributes to remyelination

Abstract

Microglia, the resident immune cells of the central nervous system, are key players in healthy brain homeostasis and plasticity. In neurological diseases, such as Multiple Sclerosis, activated microglia either promote tissue damage or favor neuroprotection and myelin regeneration. The mechanisms for microglia-neuron communication remain largely unkown. Here, we identify nodes of Ranvier as a direct site of interaction between microglia and axons, in both mouse and human tissues. Using dynamic imaging, we highlight the preferential interaction of microglial processes with nodes of Ranvier along myelinated fibers. We show that microglia-node interaction is modulated by neuronal activity and associated potassium release, with THIK-1 ensuring their microglial read-out. Altered axonal K⁺ flux following demyelination impairs the switch towards a pro-regenerative microglia phenotype and decreases remyelination rate. Taken together, these findings identify the node of Ranvier as a major site for microglia-neuron interaction, that may participate in microglia-neuron communication mediating pro-remyelinating effect of microglia after myelin injury.

Fig. 1. Microglial cells contact nodes of Ranvier in mouse central nervous system.

A In adult mouse nervous tissue, microglial cells (Iba1, green) contact nodes of Ranvier (AnkyrinG, indicated as AnkG in the figures, red) in grey (i, cortex and ii, cerebellum) and white matter (iii, corpus callosum and iv, dorsal spinal cord, cortico-spinal tractus). Ai–iv 3D reconstruction of boxed area in i to iv respectively. Bi–ii′ Transmission electron micrographs showing microglial processes (Iba1⁺, green) contacting directly the nodal axolemma (pink) in adult mouse cerebellum. ii′ Higher magnification of the micrograph (ii), showing the interaction between the microglial process, the axolemma and first paranodal loop. C, D Immunofluorescent stainings of adult mouse dorsal spinal cord showing nodes of Ranvier (AnkyrinG, red) contacted by both a microglial cell (Iba1, green) and an astrocyte (GFAP, white) in control condition (ctrl) (C) and a microglial cell (Iba1, green) and an oligodendrocyte progenitor cell (PDGFRα, white) in remyelinating condition (LPC) (D). 3D reconstructions correspond to the boxed area. Scale bars: A, C, D 2D: 10 μm; 3D: A 1 μm, C, D 2 μm; Bi–ii 500 nm, Biii 200 nm. A, C, D: n = 4 animals, B: n = 3 animals.

Fig. 2. Microglial cells contact nodes of Ranvier in human central nervous system.

A Immunohistostainings of post-mortem human hemispheric white matter tissue (healthy donor) showing microglial cells (blue, resident microglia, TMEM119) contacting nodes of Ranvier (Na_V, brown). B, C Immunofluorescent stainings of post-mortem human hemispheric white matter tissue (healthy donor) showing microglial cells (green; B, resident microglia, TMEM119, and C, homeostatic microglia, P2Y12R) contacting nodes of Ranvier (Na_V, red). Bii, Cii 3D reconstructions correspond to the boxed area in Bi and Ci respectively. Scale bars: (2D) A, B, C 10 μm; (3D), B, C 2 μm. A: n = 6 samples, B, C: n = 5 samples.

Fig. 3. Microglial cell contacts at nodes of Ranvier are stable in vivo and increase in remyelination.

A LPC injection in mouse spinal cord (cortico-spinal tractus). B–G Contacts are observed between nodes of Ranvier (Na_V, red) and microglia (Iba1, green) in control (without injection) (B), Sham 7 and 11 days post-NaCl injection (Sham DPI 7 and 11) (C, F, respectively), in perilesional tissue 7 days post LPC injection (P-Lesion DPI 7) (D) and in remyelinating tissue 11 days post-LPC injection (rem DPI 11) (G). E Corresponding quantifications of the percentage of nodes contacted. B–G: n = 4 animals per condition; 4–6 areas per animal, 42 nodes minimum per area). H Nfasc186mCherry colocalizes with Nav at the node of Ranvier in Thy1-Nfasc186mCherry mouse dorsal spinal cord (n = 3 animals). I Glass-window system above dorsal spinal cord for 2-Photon live-imaging. J Images from a 3-h movie (Movie 1) from a CX3CR1-GFP/Thy1-Nfasc186mCherry mouse shows a stable interaction between a microglial cell (green) and a node of Ranvier (red) (arrow head). scale bar: 10 µm. K Percentage of time in contact in 1-h movies, with one acquisition every 10 min. L Longest sequence of consecutive timepoints with contact in 1-h movies. Each dot is a microglia-node pair. The number of node-microglia pairs imaged is indicated on each bar. J–L n = 6–11 animals per condition. Scale bars: (B–D, F, G, J) 10 µm. E ANOVA with post hoc Tukey test; K, L Two-sided Kruskal–Wallis test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant. Bars and error bars represent the mean ± s.e.m. For detailed statistics, see Supplementary Table.

Fig. 4. Microglia process dynamics are modified by nodal structure vicinity in myelinated and remyelinating cultured slices.

A Microglial cells (green) initially contacting an internode or a node (red) in myelinated slices (control), or a node in a remyelinating slice (rem). Arrowheads show the initial contact position (filled: contact, empty: no contact). (control: internode: n = 32 contacts from 16 animals, node: n = 9 contacts from 8 animals; rem: n = 6 contacts from 6 animals). B, C Dynamics of microglial processes contacting an internode vs a node in myelinated slices (internode: n = 32 contacts from 16 animals, node: n = 9 contacts from 8 animals). D Trajectories (with t0 position as reference, distance in µm) of microglial process tips in myelinated slices, initially contacting a node (nodal contact) or not (wo contact). wo contact: n = 14, nodal contact: n = 7, from 7 color coded animals. Type II Wald χ² (two-sided analysis), p = 1.679e-8 (for quantification, see Fig. S4F). E Instantaneous process velocity in myelinated slices. wo contact: 280 measures from 14 trajectories in 7 animals, nodal contact: 140 measures from 7 trajectories in 7 color coded animals. F, G Dynamics of microglial processes contacting a node in control (node ctrl) or remyelinating slices (node rem). ctrl: n = 9 contacts from 8 animals, rem: n = 6 contacts from 6 animals. H Trajectories (with t0 position as a reference, distance in µm) of microglial process tips in remyelinating slices, initially contacting a node or not. wo contact: n = 12, nodal contact: n = 6, in 6 color coded animals. Type II Wald χ² test (two-sided analysis), p = 2.2e-16 (for quantification, see Fig. S4H). I Instantaneous process velocity in remyelinating slices. wo contact: 240 measures from 12 trajectories in 6 animals, nodal contact: 120 measures from 6 trajectories in 6 color coded animals. Scale bars: A 10 µm. B, C, F, G Two-sided Mann–Whitney test, E, I Type II Wald χ² test (two-sided analysis). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant; bars and error bars represent the mean ± s.e.m. For detailed statistics, see Supplementary Table.

Fig. 5. Neuronal activity modulates microglia-node of Ranvier interaction.

A, D Experimental designs. B, C In myelinated CX3CR1^GFP/+ cerebellar organotypic slices, microglia (GFP, green) contacts with nodes (Na_V, red) are reduced following electrical activity inhibition with tetrodotoxin (TTX). E, F They are conversely increased following electrical activity activation using apamin (Apa). Arrowheads show the nodes of Ranvier contacted by microglia. C, F Percentage of nodes of Ranvier contacted by microglial cells in control (ctrl) vs 1-h treated slices from the same animal (B–C) TTX (500 nM, n = 6 animals), E, F Apamin (500 nM, n = 10 animals). Scale bars: B, E 10 µm. C Two-sided Paired t-test; F Two-sided Wilcoxon matched pairs test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant; bars and error bars represent the mean ± s.e.m. For detailed statistics, see Supplementary Table.

Fig. 6. Microglia preferential contact with nodes depends on potassium fluxes.

A Experimental design. B In myelinated CX3CR1^GFP/+ cerebellar organotypic slices, microglia (GFP, green) contacts with nodes (Na_V, red) are reduced following potassium channel inhibition by tetraethylammonium (TEA). Arrowheads indicate the nodes of Ranvier contacted. C Percentage of nodes of Ranvier contacted in control condition or following 1-h TEA treatment (B, C, 30 mM, n = 6 animals); the mean values per animal are individually plotted. D Microglial cell (green) initially contacting a node (red) in a myelinated slice treated with TEA. Arrowheads show the initial contact position (filled: contact, empty: no contact). E, F Dynamics of microglial tips contacting an internode vs a node in myelinated slices treated with TEA (D–F, internode: n = 37 contacts from 12 animals, node: n = 10 contacts from 8 animals). G Trajectories (with t0 position as reference, distance in µm) of microglial process tips whether they were initially contacting a node (nodal contact) or not (wo contact) at t0 (wo contact: n = 14; nodal contact: n = 7; from 7 color coded movies from 6 animals). H Distance between the process tip and its position at t0 for each timepoint in myelinated slices treated with TEA (wo contact: 280 measures from 14 trajectories in 6 animals, nodal contact: 140 measures from 7 trajectories from 7 color coded movies in 6 animals). I Instantaneous process velocity in myelinated slices (wo initial contact: 280 measures from 14 trajectories in 7 animals, initial nodal contact: 140 measures from 7 trajectories in 7 color coded animals). J Schematic of the experimental design. K Microglia (GFP, green) contacts at nodes (Na_V, red) are importanlty reduced following THIK-1 inhibition by tetrapentylammonium (TPA). L Percentage of nodes of Ranvier contacted by microglial cells in control condition or following 1-h TPA treatment (K, L, 50 μM, n = 6 animals); the mean values per animal are individually plotted. Scale bars: (B, D, K) 10 μm. C, L Two-sided Paired t-test; E, F Two-sided Mann–Whitney test; H, I Type II Wald χ² test (two-sided analysis). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant; bars and error bars represent the mean ± s.e.m. For detailed statistics, see Supplementary Table.

Fig. 7. Altered K⁺ fluxes following demyelination leads to a reduced number of pro-regenerative microglia and impairs remyelination ex vivo.

A In remyelinating CX3CR1^GFP/+ cerebellar organotypic slices, microglial (GFP, green) contacts with nodes (Na_V, red) are importantly reduced following potassium channels inhibition by TEA and TPA treatments. Arrowheads show the nodal structures (filled: contacted by microglia; empty: not contacted). B, C Percentage of nodal structures contacted by microglial cells in control remyelinating condition or following 1-h TEA (A, B, 30 mM, n = 6 animals) or TPA treatment (A, C, 50 μM, n = 6 animals); the mean values per animal are shown as dots. D–H In remyelinating C57bl6/J cerebellar organotypic slices, the number of microglial cells expressing IGF1 is decreased following potassium channel inhibition by TEA (D) or THIK-1 inhibition by TPA treatment (H). E–I Percentage of IGF1⁺ microglial cells at remyelination onset, with no treatment (ctrl) or with TEA (D, E, 2 h, 30 mM, n = 6 animals) or TPA treatment (H, I, 2 h, 50 μM, p = 2.195e-5, n = 6 animals), the mean values per animal are shown as dots and paired with the corresponding control. F, J In remyelinating C57bl6/J cerebellar organotypic slices, remyelination is reduced following TEA (F) or TPA (J) treatment. G, K Percentage of axonal area remyelinated in LPC-demyelinated slices without (ctrl) or with TEA (F, G, 2 h, 30 mM, n = 7 animals) or TPA treatment (J, K, 2 h, 50 μM, n = 7 animals). Scale bars: A 10 μm; D, H 20 μm; F, J 100 μm. B, C, E, G, I, K Two-sided Paired t-test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant; bars and error bars represent the mean ± s.e.m. For detailed statistics, see Supplementary Table.

Fig. 8. Altered K⁺ flux read-out by microglia following demyelination leads to a reduced number of pro-regenerative microglia and impairs remyelination in vivo.

A Osmotic pumps were placed 9 days post LPC-injection to deliver NaCl 9‰ (Ctrl) or 50 μM TPA at the lesion site and the spinal cords were collected 2 days later. B In remyelinating dorsal spinal cord lesion, the number of microglial cells expressing IGF1 is decreased following TPA treatment. C Percentage of IGF1⁺ microglial cells at 11 DPI in the remyelinating lesion, in control (Ctrl) or TPA condition (B–C, Ctrl: n = 5 animals, TPA: n = 6 animals). D Remyelination is reduced following TPA delivery at the lesion. E Percentage of the lesion devoided of myelin at 11 DPI, in control or TPA condition (D–E, Ctrl: n = 5 animals, TPA: n = 6 animals). Scale bars: B 10 μm, D 50 μm. C, E Two-sided Mann–Whitney tests. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant. Bars and error bars represent the mean ± s.e.m. For detailed statistics, see Supplementary Table.

Fig. 9. Model for microglial interaction at nodes of Ranvier.

Microglial contact with nodes of Ranvier depends on the tissular context. The interaction is in particular reduced in perilesional area at the peak of demyelination, and increased in remyelination compared to healthy myelinated tissue. Neuronal activity and nodal K⁺ efflux promote the interaction, with microglial read-out depending on THIK-1. Altering K⁺ efflux or THIK-1 activity in remyelination correlates with more pro-inflammatory microglia and decreased remyelination.