Author(s): LEHMANN, N., AYE, N., KAUFMANN, J., HEINZE, H.J., DÜZEL, E., ZIEGLER, G., TAUBERT, M., Institution: OTTO VON GUERICKE UNIVERSITY MAGDEBURG, Country: GERMANY, Abstract-ID: 1792

INTRODUCTION:
The mechanisms subserving motor learning in the intact human brain are not fully understood. We recently reported that learning a dynamic balancing task (DBT) over a 4-week period in young healthy adults (n = 24) leads to a behaviorally relevant increase in the complexity of cortical neurite orientation in motor brain areas [1], consistent with the idea of synaptic remodeling [2]. However, since a comprehensive understanding of motor learning can only be achieved by studying the interaction of distributed cortical and subcortical areas [2], we will focus here on learning-induced changes of the microstructural composition of the fiber tracts that interconnect hub regions of the motor network.
METHODS:
We used a powerful within-subjects design with a 4-week life-as-usual control period immediately followed by an equally long period of DBT learning, as described in [1]. A unique set of quantitative MRI (qMRI) contrasts weighted towards diffusion, relaxation times and magnetization transfer were measured at baseline as well as before and after the learning period. Advanced biophysical models of tissue microstructure were fitted to these data, resulting in parameter maps sensitive to features such as tissue density, neurite orientation and density, myelin and iron [3,4]. White matter fiber tracts were virtually dissected using tractography and neural network-based bundle recognition [5], and qMRI parameter maps were projected onto these bundles. Finally, along-tract statistics of latent microstructural changes over time [6] were performed.
RESULTS:
Non-overlapping 95% bootstrap CIs [6] of factor scores over time suggest significant learning-induced plasticity in major association (anterior thalamic radiation, superior longitudinal fascicle), commissural (corpus callosum) and projection tracts (corticospinal tract, thalamocortical and corticostriatal fibers). Importantly, based on factor loadings on the trajectory of latent change over time, we were able to infer three main drivers of neuroplastic change: process 1 is dominated by myelin-sensitive metrics, process 2 is dominated by neurite density and dispersion-sensitive metrics, process 3 is dominated by iron-sensitive metrics.
CONCLUSION:
Here we have characterized the multifaceted microstructural changes in white matter in response to motor learning with unprecedented biological specificity. Consistent with theoretical predictions based on animal studies [2], we demonstrate the presence of neuronal, extra-neuronal and myelin-related changes also in the human brain. In perspective, we anticipate our study to be a starting point towards a more comprehensive understanding of the mechanisms subserving motor learning in physiological and pathological aging or in movement disorders.
[1] Lehmann et al., J Neurosci, 2023
[2] Zatorre et al., Nat Neurosci, 2012
[3] Zhang et al., Neuroimage, 2012
[4] Weiskopf et al., Front Neurosci, 2013
[5] Wasserthal et al., Med Image Anal, 2019
[6] Madssen et al., PLoS Comput Biol, 2021