Chronic demyelination and myelin repair after spinal cord injury in mice: A potential link for glutamatergic axon activity

Abstract

Our prior work examining endogenous repair after spinal cord injury (SCI) in mice revealed that large numbers of new oligodendrocytes (OLs) are generated in the injured spinal cord, with peak oligodendrogenesis between 4 and 7 weeks post-injury (wpi). We also detected new myelin formation over 2 months post-injury (mpi). Our current work significantly extends these results, including quantification of new myelin through 6 mpi and concomitant examination of indices of demyelination. We also examined electrophysiological changes during peak oligogenesis and a potential mechanism driving OL progenitor cell (OPC) contact with axons. Results reveal peak in remyelination occurs during the 3rd mpi, and that myelin generation continues for at least 6 mpi. Further, motor evoked potentials significantly increased during peak remyelination, suggesting enhanced axon potential conduction. Interestingly, two indices of demyelination, nodal protein spreading and Nav1.2 upregulation, were also present chronically after SCI. Nav1.2 was expressed through 10 wpi and nodal protein disorganization was detectable throughout 6 mpi suggesting chronic demyelination, which was confirmed with EM. Thus, demyelination may continue chronically, which could trigger the long-term remyelination response. To examine a potential mechanism that may initiate post-injury myelination, we show that OPC processes contact glutamatergic axons in the injured spinal cord in an activity-dependent manner. Notably, these OPC/axon contacts were increased 2-fold when axons were activated chemogenetically, revealing a potential therapeutic target to enhance post-SCI myelin repair. Collectively, results show the surprisingly dynamic nature of the injured spinal cord over time and that the tissue may be amenable to treatments targeting chronic demyelination.

Keywords: DREADDs; demyelination; glutamate; myelin repair; spinal cord injury.

FIGURE 1.. Remyelination continues for 6m post-injury with peak new myelin formed during the 3^rd month after SCI.

A. Schematic of experimental design. Adult PDGFRα-CreER^T2: Tau-mGFP reporter mice were given 4 doses of tamoxifen at the beginning of the 2^nd, 3^rd, or 6^th month post-injury (mpi), perfused 1 month later, and new (mGFP+) myelin was quantified in spinal cord sections. B. Confocal images of GFP in naïve and injured spinal cord white matter at 2mpi, 3mpi, and 6mpi. Insets show high-power single-channel and merged confocal images of GFP, neurofilament-heavy (NF), and GFP/NF co-localization. Boxes in insets at top of images indicate the region of image collection. Scale bar = 25 μm. C. Quantification of remyelinated axons (GFP+/NF+) divided by total axons (NF+) in naïve and injured reporter mice examined 2mpi (n=11), 3mpi (n=5), and 6mpi (n=7) at 2 mm, 1 mm, and 0.5 mm rostral and caudal to the injury epicenter. In naïve mice, only ~3% of axons were wrapped in new myelin. After SCI, remyelination increased and peaked during the 3^rd mpi when >20% of axons became wrapped with new myelin. During the 6^th mpi, new myelin continued to be produced on spared axons throughout the extent of the tissue. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as mean + SEM. All three groups were significantly different from their respective controls at each distance. *p<.05.

FIGURE 2.. New OLs and myelin accumulate after SCI.

A. Schematic of experimental design. Tamoxifen was given 1 week prior to injury (−1wpi), then PDGFRα-CreERT2: Tau-mGFP reporter mice received a moderate (75 kD) contusion injury. EdU was given in drinking water from 1–14dpi to label dividing cells, then spinal cords were collected at 4wpi or 10wpi. B-D. Confocal z-stacks from (B) naïve, (C) 4wpi, and (D) 10wpi spinal cord white matter immunolabeled for GFP, EdU, and nuclei (DAPI). GFP+/EdU+ cells are indicated by white arrowheads and site of image collection shown in inset. Scale bar = 25 μm. E. Quantification of GFP+/EdU+ cells (OLs) from 2 mm rostral to 2 mm caudal from the injury epicenter in naïve (n=6), 4wpi (n=4), and 10wpi (n=4) tissue. The number of GFP+/EdU+ cells increased from an average 0–1 cell in naïve tissue to ~17 cells at ±0.5 mm from the epicenter at 4wpi and 10wpi. The number of GFP+/EdU+ cells did not change between 4–10wpi. F-G. The total number of OLs (CC1+/DAPI; F) and new OLs (GFP+/DAPI; G) increased significantly between 4–10wpi. Compared to naïve, total and new OLs increased 3-fold by 4wpi and 5-fold at 10wpi. H. Quantification of axons with new myelin (GFP+/NF+) divided by total axons (NF+) in naïve (n=6) and injured PDGFRα-CreER^T2: Tau-mGFP mice examined 4wpi (n=4) and 10wpi (n=4) at 2 mm, 1 mm, and 0.5 mm rostral and caudal to the injury epicenter. Axons with new myelin near the lesion increased from ~5% in naïve cords to ~10% by 4wpi and ~25% by 10wpi. At 10wpi, axon remyelination was significantly increased compared to naïve and 4wpi values. I-J. Single- and double-channel confocal z-stack images from 10wpi spinal cords immunolabeled for Caspr and GFP. GFP/Caspr co-localization indicates re-established nodes of Ranvier on GFP+ myelin. Solid white arrowheads (J) indicate node of Ranvier, and open white arrowheads indicate hemi-nodes. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as mean + SEM. *p<.05. Scale bars = (F) 8 μm and (G) 10 μm.

FIGURE 3.

Functional improvements persist chronically post-injury. A. Schematic of experimental timeline. Tamoxifen was given 1 week prior to injury (−1wpi), then PDGFRα-CreERT2: Tau-mGFP mice received a moderate T9 contusion injury (n=6). Motor evoked potentials (MEPs) were recorded at baseline (−2wpi), and 1, 4, and 10wpi. BMS and automated horizontal ladder were performed at baseline, 1 dpi, and weekly thereafter through 10wpi. B. BMS scores dropped to either 0 or 0.5 1dpi, and then increased to a score of 5 by 28dpi. BMS scores remained stable chronically and were not significantly different between 28–70dpi (4–10wpi). C. BMS subscores increased through 7wpi then plateaued; these changes were not significantly different at any time between 21–70dpi. D. The number of missed steps on the automated horizontal ladder increased after SCI, but remained stable at ~20 missed steps per pass. Minor changes in missed steps were not statistically significant. E. Representative MEP waveforms at baseline and after SCI. X-axis represents time (2 ms per division) and y-axis represents voltage (baseline, 1000 μV per division; SCI, 100 μV per division). F. MEP peak latency, defined as the moment of maximum upward response, significantly increased 10wpi compared to baseline. At 4wpi, MEP latency was significantly shorter than 1wpi and 10wpi. G. MEP amplitudes were significantly reduced at 1, 4, and 10wpi compared to baseline. MEP amplitude increased from an average of 125 μV at 1wpi, to 233 μV at 4wpi, and reached 400 μV by 10wpi, but this increase over time after SCI was not significantly different. H. MEP area under the curve (AUC) measured the voltage and duration of MEP responses. By 10wpi, MEP AUC significantly increased 2.5-fold compared to 1wpi. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as mean + SEM. *p<.05.

FIGURE 4.. Nodes of Ranvier disruption occurs acutely through chronically in spared white matter in the injured mouse spinal cord.

A. Schematic of experimental timeline. C57BL/6 mice were naïve (n=6) or received a T9 contusion and were sacrificed at 1wpi (n=6), 2.5wpi (n=6), 4wpi (n=6), 10wpi (n=5), and 26wpi (6mpi; n=7). B. Representative images of longitudinal sections from naïve and injured spinal cords from 1wpi to 26wpi showing normal and aberrant nodes of Ranvier. Nodes were identified by Kv1.2+ segments (green) flanking Caspr+ segments (red). Axon neurofilaments are labeled with blue. Ion channel and Caspr spreading are indicated by arrowheads. Inset in naïve image shows high-power example of proper nodal protein organization. JXP = juxtaparanode (Kv1.2), P = paranode (Caspr), N = node of Ranvier (Nav1.6 not labeled). Inset in 1wpi image shows improper nodal protein organization from spreading. Inset in 26wpi image shows high-power example of a short Kv1.2 profile. Scale bars = 10 μm. C. Quantification of nodal profile organization from 1–26wpi based on proper Kv1.2 and Caspr localization (Kv1.2, Caspr, Caspr, Kv1.2) and length. Nodes were counted in spared white matter 2 mm, 1 mm, and 0.5 mm rostral and caudal from the lesion epicenter (+, rostral; -, caudal). After SCI, the greatest loss of nodes in spared white matter occurred ±0.5 mm from the epicenter, with a 5-fold decrease in properly arranged nodal profiles compared to naïve tissue at acute and chronic time points. Distal to the lesion, the percent of proper nodes were half that in naïve white matter at acute time points and continued to decrease to a peak 3.5-fold lower than naïve at 26wpi. D. Quantification of aberrantly short Kv1.2 profiles (<5 μm) in white matter bordering SCI lesion sites. The number of short profiles rose progressively after SCI with significantly increased numbers peaking between 2.5 – 10w post-injury near the epicenter. Levels remained significantly elevated through 6mpi at all distances examined. E. Aberrantly long Kv1.2 profiles (>10 μm) were fewer and rose in a more protracted time course after SCI, with peak levels detected at 10wpi, which dropped to non-significant levels rostral to injury but remained elevated at and caudal to the injury site at 6mpi. F. Significant axon loss occurred in SCI sampled regions compared to those in naïve at all timepoints and regions but did not differ between SCI groups. Black asterisks represent significant differences relative to naïve; colored asterisks represent differences with group indicated by that color. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as box plots with median and bound by the 25^th and 75^th percentiles, whiskers set to min and max smallest to largest values. * all ranges from p < .05 – .001).

FIGURE 5.. Nav1.2 is upregulated in white matter after SCI.

A. Schematic of experimental timeline. C57BL/6 mice received a T9 contusion and were sacrificed at 1wpi (n=6), 2.5wpi (n=6), 4wpi (n=6), 10wpi (n=5), and 26wpi (6mpi; n=7). Naïve mice (n=6) served as controls. B. Representative confocal z-stacks of longitudinal sections from naïve and injured spinal cords from 1wpi to 26wpi showing changes in Nav1.2 expression over time in spared white matter bordering the pia. Scale bar =10 μm. C. Quantification of Nav1.2 proportional area/ NF area in naïve and SCI mice at 2 mm, 1 mm, and 0.5 mm rostral and caudal to the injury epicenter revealed Nav1.2 expression increased significantly after SCI. Nav1.2 labeling was highest caudal to the epicenter between 2.5wpi – 10wpi where percent area increased ~5-fold compared to 1wpi. By 26wpi, Nav1.2 expression had declined significantly compared to earlier time points. Nav1.2 area in naïve tissue was negligible. Asterisk color corresponds to significant differences between timepoints. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as box plots with median and bound by the 25^th and 75^th percentiles, whiskers set to min and max smallest to largest values. * all ranges from p < .05 – .001.

FIGURE 6.. Remyelinated and demyelinated axons are present 26 weeks post-injury.

A-B. Representative images of (A) transverse and (B) longitudinal semithin Epon embedded sections stained with toluidine blue from naïve and 26wpi spinal cords. Several demyelinated axons (black arrowheads) and remyelinated axons (pink arrowheads) are present in 26wpi spinal cord tissue. Naïve age-matched controls have thick myelinated axons with no indices of demyelination. Scale bar = 5 μm. C. Transmission electron microscopy confirms thin myelin sheaths wrap axons, indicative of remyelination after SCI. Numbers represent g-ratios, defined as axon diameter divided by axon + myelin diameter. Scale bar = 1 μm. A-C. Associated maps indicate region of image acquisition. D. Average g-ratios are significantly larger in 26wpi white matter compared to age-matched naïve. A larger g-ratio is indicative of reduced myelin thickness. E. Quantification of mature myelin (g < 0.75), remyelinated (g > 0.75), and unmyelinated (g = 1) axons in spared white matter from naïve (n=4) or 26wpi (n=3) mice. The percent of remyelinated axons increased from <1% in naïve tissue to 26% of total axons at 26wpi. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as box plots with median and bound by the 25^th and 75^th percentiles, whiskers set to min and max smallest to largest values. *p < .05 vs Naive.

FIGURE 7.. Vglut2+ puncta increase in spared white matter axons after SCI and are sites of contact by NG2 cells.

A. Schematic of experimental timeline. C57BL/6 mice were naïve (n=6) or received a T9 contusion and were sacrificed at 1wpi (n=6), 2.5wpi (n=6), 4wpi (n=6), 10wpi (n=5), and 26wpi (n=7). B. Quantification of total Vglut2+ puncta in spared white matter per axon profile. Vglut2 increases continuously after SCI to a peak 6-fold greater than naïve at 4wpi and remains significantly elevated for 6mpi. C. The ratio of Vglut2+/NG2+ puncta per NG2 cell increased significantly at 2.5wpi, rose to 10-fold higher at 4wpi, and remained elevated for at least 26wpi. D. Immunohistochemical labeling of NG2 (green), Vglut2 (red), and nuclei (DAPI, blue) in naïve and injured spinal cord white matter, from 1wpi through 26wpi. White arrowheads indicate NG2/Vglut2 co-localization. Inset shows high-power example of NG2/Vglut2 co-localization. Scale bar = 10 μm. Counts (B-C) represent averages of 1 mm and 2 mm images. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as mean + SEM. *p < .05

FIGURE 8.. DREADD activation modulates neuronal activity in spinal cord gray matter.

A. Schematic of experimental design. 2 week-old Vglut2-Cre mice received either Gi- or Gq-DREADD injections at T4/5 and T7/8. Mice received a T3 complete crush at 8-weeks, followed by twice daily CNO or saline i.p. injections from 2–4wpi until sacrifice (Gi saline, n=3; Gi CNO, n=4; Gq saline, n=3; Gq CNO, n=3). B. Representative images of FosB labeling indicating neuronal activity in the upper (I-IV) and lower (V-X) lamina of T7 gray matter. C-D. Quantification of the number of FosB+ cells in (C) lamina I-IV and (D) lamina V-X. CNO-mediated DREADD excitation of Vglut2+ neurons increased the number of FosB+ cells in lamina I-IV and lamina V-X compared to saline-injected controls. The number of FosB+ cells in lamina V-X decreased 2-fold in CNO-injected Gi-DREADD mice compared to controls. Neuronal activity was unchanged in lamina I-IV in Gi-DREADD mice. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as mean + SEM with individual data points shown by circles. *p < .05

FIGURE 9.. NG2 cells preferentially contact active intraspinal glutamatergic axons.

A-B. Representative merged and mCherry single-channel images of injured spinal cord white matter from (A) Gi-DREADD and (B) Gq-DREADD injected mice showing NG2 cell contacts on (A) silenced or (B) activated axons. Examples of NG2/mCherry double-labeling are indicated by white arrowheads. Inset shows high-power example of NG2/mCherry co-localization. Scale bar = 10 μm. C-D. Quantification of the total number of NG2/mCherry contacts in the injured thoracic spinal cord (Gi saline, n=3; Gi CNO, n=4; Gq saline, n=3; Gq CNO, n=3). NG2 labeled OPC contacts (C) decreased 2-fold on silenced glutamatergic axons compared to control axons, and (D) increased 2-fold on CNO activated glutamatergic axons compared to non-activated axons. E. Quantification of mCherry percent area revealed that mCherry labeling was significantly lower in mice injected with Gi-DREADDs compared to Gq-DREADDs; however, mCherry labeling was not different between saline- and CNO-injected groups. Data were analyzed using mixed-effect model followed by Tukey post hoc test and are presented as mean + SEM with individual data points shown by circles. *p < .05.