Myelination- and immune-mediated MR-based brain network correlates

Abstract

Background: Multiple sclerosis (MS) is an autoimmune disease of the central nervous system (CNS), characterized by inflammatory and neurodegenerative processes. Despite demyelination being a hallmark of the disease, how it relates to neurodegeneration has still not been completely unraveled, and research is still ongoing into how these processes can be tracked non-invasively. Magnetic resonance imaging (MRI) derived brain network characteristics, which closely mirror disease processes and relate to functional impairment, recently became important variables for characterizing immune-mediated neurodegeneration; however, their histopathological basis remains unclear.

Methods: In order to determine the MRI-derived correlates of myelin dynamics and to test if brain network characteristics derived from diffusion tensor imaging reflect microstructural tissue reorganization, we took advantage of the cuprizone model of general demyelination in mice and performed longitudinal histological and imaging analyses with behavioral tests. By introducing cuprizone into the diet, we induced targeted and consistent demyelination of oligodendrocytes, over a period of 5 weeks. Subsequent myelin synthesis was enabled by reintroduction of normal food.

Results: Using specific immune-histological markers, we demonstrated that 2 weeks of cuprizone diet induced a 52% reduction of myelin content in the corpus callosum (CC) and a 35% reduction in the neocortex. An extended cuprizone diet increased myelin loss in the CC, while remyelination commenced in the neocortex. These histologically determined dynamics were reflected by MRI measurements from diffusion tensor imaging. Demyelination was associated with decreased fractional anisotropy (FA) values and increased modularity and clustering at the network level. MRI-derived modularization of the brain network and FA reduction in key anatomical regions, including the hippocampus, thalamus, and analyzed cortical areas, were closely related to impaired memory function and anxiety-like behavior.

Conclusion: Network-specific remyelination, shown by histology and MRI metrics, determined amelioration of functional performance and neuropsychiatric symptoms. Taken together, we illustrate the histological basis for the MRI-driven network responses to demyelination, where increased modularity leads to evolving damage and abnormal behavior in MS. Quantitative information about in vivo myelination processes is mirrored by diffusion-based imaging of microstructural integrity and network characteristics.

Keywords: Demyelination; MRI; Modularity; Network Dynamics; Remyelination.

Fig. 1

Experimental outline. Schematic representation of the study showing the 7 experimental groups coinciding with different time points before, during, and after cuprizone diet. MRI data was performed longitudinally at every time point in the same mouse cohort (continuous blue line). New cohorts of mice were used at each time points to assess locomotor-, anxiety-like behavior, and memory abilities. All mice underwent the same tests and some of them were used for histological evaluation at every time point. Additional cohorts were used to assess effects of Pavlovian conditioning paradigm only at the indicated time points

Fig. 2

Structural and anatomical white matter changes during de- and remyelination in the cuprizone model. Example immunohistochemical images of coronal mouse slices containing the corpus callosum (CC) in control conditions (1st column), at 6 weeks after starting the cuprizone diet (cupri 6 weeks—full demyelination, 2nd column), and at full remyelination 6 weeks after reintroduction of normal food (3rd column). On the right of each panel, bar graphs show quantifications of changes. Stained for: a myelin specific marker PLP; b astrocytic specific marker GFAP; c exemplary T2-weighted images obtained in living mice during a longitudinal MRI scan. Pictures show frontal part of mouse brain containing neocortex and CC from control, cupri 6 weeks and remy 6 weeks mice. Bar graph shows the ratio calculated between the intensity of the myelin signal observed in Cx (SI_Cx) and CC (SI_CC). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Fig. 3

Structural and anatomical gray matter changes during de- and remyelination in the cuprizone model. Example immunohistochemical images of coronal mouse slices containing the neocortex (Cx) in control conditions (1st column), at 6 weeks after starting the cuprizone diet (cupri 6 weeks—full demyelination, 2nd column), and at full remyelination 6 weeks after reintroduction of normal food (3rd column). On the right of each panel, bar graphs show quantifications of changes. Stained for: a myelin specific marker PLP; b astrocytic specific marker GFAP; c exemplary T2-weighted images obtained in living mice during a longitudinal MRI scan. Pictures show frontal part of mouse brain containing neocortex, hippocampus, and CC. Bar graph shows the ratio calculated between the intensity of the myelin signal observed in caudal regions of the Cx (SI_Cx) and CC (SI_CC). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Fig. 4

Behavioral correlates of general de- and remyelination in the cuprizone model. a Schematic representation of the time course of histopathological markers for structural impairment in the cuprizone model. b Bar graphs showing travelled distance and time spent in the periphery in the Open Field test. c Bar graphs show the results of the EPM test. d Bar graphs show the results of the auditory Pavlovian conditioning paradigm conducted using 2.5 kHz or 10 kHz, where the latter is the conditioning stimulus associated to the foot shock. e Bar graphs show NOR index calculated 15 min, 4 h, and 24 h. *p < 0.05; **p < 0.01; ***p < 0.001

Fig. 5

Structural dynamics of de- and remyelination. a Schematic representation of the methodological approach used to perform neuronal network analyses. Diffusion tensor images were used to calculate the fractional anisotropy (FA), and a connectivity matrix was built with the help of an anatomical mask. b Regional FA values in the thalamus (left upper panel), in the neocortex (left lower panel), and the hippocampus (right upper panel at all and CC (right lower panel) investigated at all-time points). c FA-based network values had increased clustering coefficients and modularity compared to control. Horizontal black lines represent significance of a given time point in comparison to control: p < 0.01. Dashed lines represent significance of a given time point in comparison to control: p < 0.001

Fig. 6

Correlation analyses. a Correlation plots between FA values in the cortex and number of GFAP positive cells found in the cortex. Three time points corresponding to onset of demyelination (cupri 2 weeks, left column), full demyelination (cupri 6 weeks, middle column), and full remyelination (remy 6 weeks, right column). b Correlation plots between FA values in the corpus callosum and PLP signal intensity (indication for myelination) in the corpus callosum. Three time points corresponding to the start of demyelination (cupri 2 weeks, left column), full demyelination (cupri 6 weeks, middle column), and full remyelination (remy 6 weeks, right column). c Correlation plots between clustering (indicator of network activity) and the NOR index. Three time points corresponding to the start of demyelination (cupri 2 weeks, left column), full demyelination (cupri 6 weeks, middle column), and full remyelination (remy 6 weeks, right column). r² and p values are given in Table 1