Time course of structural plasticity induced by different types of motor learning

JACOBACCI, F; HIDALGO-MARQUES, M; ARMONY, J; LERNER, G; TAVERNA CHAIM, K; DOYON, J; BORÉ, A; AMARO, E; DELLA-MAGGIORE, V

San Diego, California

Encuentro; Society for neuroscience (SFN) annual meeting; 2018

The nervous system translates new knowledge into long-lasting plasticchanges that lead to the formation of memories. To date, the timescale ofstructural remodelling that accompanies functional neuroplasticity is largelyunknown. Magnetic resonance neuroimaging provides us with non-invasivetools to explore structural neuroplasticity in humans. There is wide evidenceof changes in grey and white matter structure in the healthy human brainfollowing extensive motor training (< 1 week, Dragansky et al., 2004; Landiet al., 2011; Scholz et al., 2009; Sampaio-Baptista et al., 2014). Animalexperiments, however, suggest that structural remodelling due to motorlearning is a rapid, dynamic process that can be detected over much shortertimescales (1-3 hours; Fu and Zuo, 2011; Xiao et al., 2016). A recentDiffusion Weighted Imaging (DWI) study has shown that training on a spatialmemory task induces a decrease in mean diffusivity (MD) that correlateswith N-methyl-D-aspartate (NMDA) dependent astrocyte hypertrophy in the hippocampus (Sagi et. al., 2012; Assaf, 2017). Here, we used DWI toinvestigate the emergence of early cortical plasticity elicited by a shortsession of motor learning. To this end, we trained 21 healthy subjects [11female; age (mean ± SD): 23,6±3,1 years old] in two well-characterizedmotor tasks, tapping on different neural substrates: motor sequence learning(MSL) and visuomotor adaptation (VMA). DWI images (2×2×2 mm3, 30gradient directions, b-value = 1000 s/mm2, TR=5208 ms, TE=89 ms,FOV=240x240 mm2) were obtained before, 30 min and 24 hours aftertraining to asymptotic performance (~15 min for MSL and ~25 min for VMA).Prior to the tensor and MD estimation, DWI images were corrected forgeometric distortions (Andersson et al., 2003), head motion, eddy currentsand b-vector correction (Andersson and Sotiropoulos, 2015). MD maps werenon-linearly transformed to MNI152 T1 using ANTs (Avants et al., 2011), andlongitudinal MD changes were statistically assessed using the threshold-freecluster enhancement (TFCE) approach of Randomise (Smith and Nichols,2009; Smith et al., 2004). Learning the MSL task was associated with asignificant reduction of MD in the left hippocampus 30 minutes post-learningthat returns to baseline levels at 24 hrs. MSL also lowered MD in the frontoparietalnetwork 30 min post-learning, but this decrement persistedovernight. No significant changes in microstructure were found after learninga VMA task. Our results suggest that acquiring a new motor policy -as inMSL- may involve changes in synaptic efficacy that are not triggered whenlearning is associated with recalibration of an existing motor policy.