Neuroinflammation in the normal appearing white matter of multiple sclerosis brain causes abnormalities at the node of Ranvier

Changes to the structure of nodes of Ranvier in the normal-appearing white matter (NAWM) of MS brains are associated with chronic inflammation. We show that the paranodal domains in MS NAWM are longer on average than control, with Kv1.2 channels dislocated into the paranode. These pathological features are reproduced in a model of chronic meningeal inflammation generated by the injection of lentiviral vectors for the lymphotoxin-α (LTα) and interferon-γ (IFNγ) genes. We show that tumour necrosis factor (TNF), IFNγ and glutamate can provoke paranodal elongation in cerebellar slice cultures, which could be reversed by an NMDA blocker. When these changes were inserted into a computational model to simulate axonal conduction, a rapid decrease in velocity was observed, reaching conduction failure in small diameter axons. We suggest that glial cells activated by proinflammatory cytokines can produce high levels of glutamate, which triggers paranodal pathology, contributing to axonal damage and conduction deficits.

Introduction the mean paranodal length (figure 2F) and proportion of paranodes longer than 4µm (figure 2G). The 143 same quantification procedure was followed for assessing dislocation of the nodal Nav voltage-gated 144 channels into the PNJ , but no significant dislocation was identified (Supplementary Figure2). 145 Prolonged exposure of the rat cortex to pro-inflammatory cytokines can generate paranodal 146 disruption 147 To further study the relationship between chronic inflammation in the NAWM and PNJ pathology, we 148 used a novel rat model of chronic meningeal inflammation (James et al, 2020). In this model, In order to confirm if the effect of TNF/IFNγ was due to glutamate release and action, 5 slices were 232 treated with two doses of 100 ng/ml of TNF/IFNγ together with the non-competitive NMDA antagonist 233 MK-801 (0.6mM) (figure 6D). The paranodal length distribution of the slices was not significantly 234 different between the MK-801 treated and non-treated slices (figure 6E). The proportion of paranodes 235 longer than 4µm was greatly reduced in the MK-801 group compared to the cultures treated with the 236 cytokines alone (from 11.38% of the paranodes to 3.3%, figure 6E). 238 To systematically examine the consequences of paranodal disruption on AP propagation through the 239 use of computational modelling, a double cable core model was built and solved numerically in 240 NEURON (figure 7A). A double cable core-conductor circuitry was chosen to represent the biophysical 241 parameters of the axonal membrane, the peri-axonal space and the myelin sheath, separately 242 (Richardson et al., 2000;McIntyre et al., 2002). We simulated the functionality of an axon membrane 243 made up of 4 types of compartments: nodes, paranodes, juxtaparanodes and internodes. The nodes 244 clustered a high density of fast Nav channels (Nav1.6), persistent Nav channels and slow Kv channels 245 (figure 7A, purple). Immediately flanking the nodes, the paranodes were built as compartments with 246 no active conductances, and the sites where the myelin end loops connect with the axolemma (figure 247 7A, dark grey). Next to the paranodes, the juxtaparanodes contained fast Kv channels (figure 7A, 248 medium grey). Finally, the internodes were sections surrounded by myelin with a low density of ion 249 channels (figure 7 A, light grey). Seven fibre diameters were simulated: d fibre = 0.5, 0.8, 1.1, 1.3, 1.6, 250 1.8, 3.5 [µm]. These small-caliber diameters were chosen taking into consideration previous human 251 brain and macaque EM studies, which indicated that the average axon core diameter within the CNS is and Kalu (1979) from small diameter axons in the cat hind limb nerves V = 4.6 * dfibre).

259
In order to simulate paranodal disruption, the resistance of the paranodal and/or juxtaparanodal 260 compartments was decreased, and the juxtaparanodal Kv channels were dislocated. The resistance was 261 decreased by increasing the peri-axonal space of both compartments progressively. The observed 262 paranodal lengthening was represented in this model by an increment in peri-axonal space on the 263 assumption that if some myelin-end loops at the PNJ detached from the axolemma, these spaces will 264 be progressively larger (Bhat et al., 2001;Zonta et al., 2008). Furthermore, juxtaparanodal Kv channels 265 were dislocated to the paranode, and their conductance was increased proportionally to the increment 266 in peri-axonal space of the paranode. In summary, the following structural arrangements were

272
The conduction velocity decreased as the paranodal peri-axonal space increased in all the axon 273 diameters (figure 8A). Fitting the data to a logarithmic curve, the velocity of the axons decreased faster 274 in the smaller-diameter axons than the larger-diameter ones as the paranodal peri-axonal space was 275 increased (figure 8A). When the juxtaparanodal fast Kv channels were dislocated to the paranodal 276 compartment (figure 8B), the velocity also decreased faster in the small-diameter axons, although the 277 dislocation of the channels did not significantly alter this. When a progressive increment of the 278 paranodal and juxtaparanodal peri-axonal spaces was introduced (figure 8 C), the axon with a core 279 diameter of 0.4 µm failed to conduct when the paranodal and juxtaparanodal peri-axonal spaces were 280 increased by 500% (the paranodal peri-axonal space was 0.012µm and the juxtaparanodal 0.12µm).

281
Meanwhile, the axons with dcore of 0.6 µm and 0.8µm also failed to conduct when the paranodal and 282 juxtaparanodal peri-axonal spaces were increased by 1000% (the paranodal peri-axonal space was 283 0.022 µm and the juxtaparanodal 0.22µm). Additionally, larger-diameter axons had a significant 284 velocity decay in the same conditions. For example, the velocity of the axon with a core diameter of 285 2.7 µm decreased 78.47 % (y = −0.209 * ln(x) + 1.6457, r 2 = 0.99). In the last structural arrangement, 286 the paranode and juxtaparanode peri-axonal spaces were increased and the Kv1 were dislocated (figure 287 8 D). In this condition, conduction failure occurred in the axons with a core diameter of 0.4, 0.6, and 288 0.8 µm and a decrease in axons with a core diameter of 2.7 µm. This data suggests that AP conduction 289 in the smaller-diameter axons might be more susceptible to paranodal and juxtaparanodal disruption 290 than larger-diameter axons.
In the previous four conditions, all 21 nodes of the axon were disrupted by increasing the peri-axonal 292 space widths and displacing the juxtaparanodal voltage-gated Kv channels. However, conduction failure 293 only occurred in the small-diameter axons (dcore of 0.4, 0.6 and 0.8 µm). Therefore, we also examined 294 the number of consecutive nodes needed for conduction failure at these diameters, and the velocity 295 decay after simulating conduction in an axon where patches of healthy nodes (figure 8 E, purple) were 296 interspersed with patches of disrupted nodes (figure 8E, orange), which is more likely to reflect the 297 pathological situation. In the axon with dcore= 0.4 µm, five consecutive nodes were needed for 298 conduction failure when the width of the paranodal and juxtaparanodal peri-axonal spaces were 299 increased to 0.012 and 0.12 µm respectively (figure 8E). We then explored if damaging less than five   IFNg, which indicate that proinflammatory cytokines could be the trigger for this pathology. Further investigations using primary microglial cultures and organotypic cerebellar slice cultures suggested that 322 glutamate release by microglia in response to stimulation with pro-inflammatory cytokines could 323 mediate these pathological changes at the paranodes. Accumulation of abnormal nodes of Ranvier 324 could be responsible for some of the generalised MS symptoms that cannot be attributed to focal 325 lesions.

326
The elongation of paranodal axo-glial junctions elongation, indicated by the paranodal axonal protein 327 Caspr1 or by its glial counterpart, Nf155, suggests that the glial and axonal proteins may have detached 328 from each other, leading to diffusion along the axolemma. Therefore, it could represent a partial or 329 complete detachment of the myelin end loops, and between the loops themselves, that could result in   In conclusion, our data strongly suggests that diffuse pathology in the NAWM, which includes 413 paranodal disruption, could be caused by the presence of local cytokine induced inflammation leading 414 to excess glutamate release from microglia. Such paranodal pathology, that cannot be attributed to 415 focal demyelinating lesions, would be expected to alter the efficiency and velocity of AP conduction, 416 adding to the overall neurological dysfunction in MS. This could also be relevant to white matter 417 changes seen in other neurodegenerative conditions in which chronic microglial activation is a feature.  ThermoFisher Scientific). One-two cerebellar slices were plated per insert and maintained in an 564 incubator (HeraCell Vios 160i, ThermoFisher) at 37 o C, 5% CO2 for 9-10 days, replacing half the medium 565 every other day to replace the growth factors. In order to check the integrity of the slices, the 566 macroscopic structure was checked with an inverted microscope (Olympus CKX53). They were also 567 stained with the cell integrity marker propidium iodide (PI, Molecular Probes). PI was added at a   The culture medium was replaced entirely when the cytokines or glutamate were added to the primary 585 microglia or to the tissues.   Safronov et al., 1993). From human nerve 713 electrophysiological studies, the single conductance of a channel was quantified to be between 7-10 714 pS (Scholz et al., 1993;Safronov et al., 1993;Reid et al., 1999). In this model, the single conductance of 715 a slow Kv channel was set to 8 pS. Thus, with a density of 110 channels /µm 2 and the maximum The leak conductance of the node was set to gl=0.007 S/cm 2 (Bostock and Rothwell, 1997) while the 729 leak conductance of the paranode was gl=0.0005 S/cm 2 , the juxtaparanode and internode 730 conductances were gl = 0.005 S/cm 2 (Chiu and Schwarz, 1987  The cause of death is that presented on the death certificates.    encoding LT-a and IFN-g genes were injected into the subarachnoid space in the midline of the brain.

1139
The white rectangles are a representative of the 10 selected ROIs at the corpus callosum, cingulum and