Tracking the time course of structural plasticity in motor learning using DWI: skill learning vs adaptation.

YEFFAL, ABRAHAM; AMARO, EDSON; DOYON, JULIEN; LERNER, GONZALO; HIDALGO-MARQUES, MARCIA RENATA; JOVICICH, JORGE; JACOBACCI, FLORENCIA; BORÉ, ARNAUD; ARMONY, JORGE; DELLA-MAGGIORE, VALERIA

Roma

Congreso; 25th Annual Meeting of the Organization for Human Brain Mapping; 2019

Introduction:Structural remodeling induced by motor learning is a rapid, dynamic process associated withsynaptogenesis and enlarged astrocytic volume (Fu and Zuo, 2011; Xiao et al., 2016; Kleim et. al.,2007). In the last years, Diffusion-Weighted Imaging (DWI) has gained popularity as a non-invasive tool to quantify neuroplasticity in the cortex. Sagi and collaborators have demonstratedthat learning-specific reductions in mean diffusivity (MD) reflect astrocytic hypertrophy, and therefore can be used as a reliable marker of plasticity in humans (Sagi et. al., 2012). Here, we used MD maps to track the time course of structural plasticity in two motor tasks tapping on different neural substrates: motor sequence learning and visuomotor adaptation. In order to contrast LTP-like plasticity with structural changes induced by memory consolidation, subjectswere scanned before as well as 30 minutes and 24 hours post-training.Methods:We trained 21 healthy subjects (11 female, age 23.6±3.1 years) in two well-characterized motor tasks: motor sequence learning (MSL) and visuomotor adaptation (VMA). In MSL, subjects practiced a 5-element finger sequence in a self-paced manner. In VMA, subjects made ballistic movements to different targets displayed on a screen. They learned to adjust their movements to an imposed visual rotation by recalibrating their motor commands to compensate the error introduced by the perturbation. Subjects practiced each task until they reached asymptotic performance (~15 min. of practice for MSL and ~25 min. for VMA). In each one of the three sessions, DWI images (Siemens TRIO 3T, multi-band accelerated EPI pulse sequence, multi-band factor=2, 2×2×2 mm , 30 gradient directions, b-value = 1000 s/mm , TR=5208 ms, TE=89 ms, FOV=240x240 mm ) were acquired. Prior to the tensor and MD estimation, DWI images were corrected for geometric distortions, head motion, eddy currents, and b-vector correction. MDmaps were non-linearly transformed to MNI152 T1 template using a pipeline that minimized reproducibility errors (Jacobacci et al., 2018). Longitudinal MD changes associated with motor learning were statistically assessed using the threshold-free cluster enhancement (TFCE) approach, implemented in the Sandwich Estimator (SwE) toolbox for accurate modeling of longitudinal and repeated measures neuroimaging data (Guillaume et al., 2014). We ran two analyses aimed at distinguishing structural changes that were common from those that weredifferent between the motor learning tasks.Results:A marked reduction in MD over the lateral posterior parietal cortex 30 min. post-training, that persisted at 24 hours, distinguished MSL from VMA. This finding was in contrast with short-lasting changes occurring for both motor learning tasks. Specifically, MSL and VMA showed a reduction of MD in the right lateral cerebellum 30 min. post-learning that returned to baseline at24 hours. Surprisingly, MD in the left hippocampus followed the exact same temporal pattern. These results were specific to motor learning since they were not present in a control condition (CTL, go-nogo task).Conclusions:Although there is increasing literature pointing to a function of the hippocampus in the acquisition and/or consolidation of MSL when learned explicitly (Albouy et. al., 2013; Dohring et. al., 2017), there is no evidence for a role of the hippocampus in VMA. The transfer of rapidly encoded information from the hippocampus to long-term storage sites in the neocortex has been shown to play a key role in the consolidation of declarative and spatial memories (Khodagholy et al., 2017). However, recent work has highlighted the involvement of the hippocampus in the stabilization of non-hippocampal memories (Sawangjit et al., 2018; Schapiro et al., 2018). Our results shed lighton the time course of structural plasticity elicited by motor learning and add to the current debate challenging the traditional role of the hippocampus in explicit memory encoding.