Structural plasticity within highly specific neuronal populations identifies a unique parcellation of motor learning in the adult brain

Abstract

Cortical networks undergo adaptations during learning, including increases in dendritic complexity and spines. We hypothesized that structural elaborations during learning are restricted to discrete subsets of cells preferentially activated by, and relevant to, novel experience. Accordingly, we examined corticospinal motor neurons segregated on the basis of their distinct descending projection patterns, and their contribution to specific aspects of motor control during a forelimb skilled grasping task in adult rats. Learning-mediated structural adaptations, including extensive expansions of spine density and dendritic complexity, were restricted solely to neurons associated with control of distal forelimb musculature required for skilled grasping; neurons associated with control of proximal musculature were unchanged by the experience. We further found that distal forelimb-projecting and proximal forelimb-projecting neurons are intermingled within motor cortex, and that this distribution does not change as a function of skill acquisition. These findings indicate that representations of novel experience in the adult motor cortex are associated with selective structural expansion in networks of functionally related, active neurons that are distributed across a single cortical domain. These results identify a distinct parcellation of cortical resources in support of learning.

Fig. 1.

Retrograde tracing and filling of corticospinal neurons. (A) Red latex microspheres injected into the dorsal horn (arrow) of the C8 spinal cord segment. (B) Green latex microspheres injected into the dorsal horn (arrow) of the C4 spinal cord segment. (C) Modified coronal sections cut with 20° forward angle (red plane) compared with normal coronal sections. (D) Projection pattern of apical dendrites of corticospinal neurons (red. C8; green, C4). (E) Representative neuron labeled by retrogradely transported red latex microspheres in the cell body filled by Lucifer Yellow (green); note several single labeled neurons (red, at arrowheads) around the filled neuron. (Scale bar: 20 μm.) (F) Representative Lucifer Yellow–filled corticospinal neuron in layer V of the motor cortex. (Scale bar: 100 μm.) (G) Photographic montage illustrating retrogradely labeled layer V corticospinal neurons projecting either to C8 (red) or C4 (green) spinal segments. Tracers do not colocalize within individual layer V cells, indicating that corticospinal projections do not collateralize across the C4 and C8 segments. (Scale bar: 60 μm.)

Fig. 2.

Acquisition of forelimb skilled grasping. Performance on a forelimb skilled grasping task significantly improved over a 14-d training period (***P < 0.0001; repeated-measures ANOVA).

Fig. 3.

Training-induced plasticity of C8-projecting neurons in motor cortex. (A) Neurolucida reconstructions of representative corticospinal neurons projecting to C8 in either untrained (Left), partially active control (Center), or skilled grasping (Right) animals. (B) Neurolucida reconstructed segments of representative apical dendrites from untrained (Left), partially active control (Center), and skilled grasping (Right) subjects. (C) Representative Lucifer Yellow-filled segments of apical dendrites from untrained (Left) partially active control (Center), and skilled grasping (Right) rats, illustrating increased spine density after skilled motor training. (D) Higher magnification of images in C. (Scale bars: A, 100 μm; B and C, 20 μm; D, 5 μm.)

Fig. 4.

Skilled grasp training selectively enhances structural plasticity in C8-projecting layer V corticospinal motor neurons. (A) A 22.5 ± 2.3% increase in spine density occured on distal apical dendrites of C8-projecting corticospinal neurons in rats that underwent forelimb skilled grasping; active controls show intermediate changes. ***P < 0.0001; **P < 0.001. (B) In contrast, apical dendrites of C4-projecting neurons exhibit no change at all in spine density (P = 0.9). (C) Similarly, there are significant changes in the density of basilar spines in skilled grasp trained rats and partially active controls that differ significantly from those of untrained controls. **P < 0.001; *P < 0.01. (D) In contrast, basilar dendrites of C4-projecting neurons exhibit no change at all in spine density (P = 0.8). (E) Measures of dendritic complexity also changed as a function of skilled grasp training, including total basilar dendritic length (***P < 0.0001; **P < 0.001). (F) Total dendritic branch number (**P < 0.001), and (G) Sholl analysis (***P < 0.0001; **P < 0.001). (H) Estimates of total basilar spine number indicate that skilled grasp training leads to significantly greater increases compared with partially active controls and untrained subjects (***P < 0.0001; *P < 0.05).

Fig. 5.

Neuronal organization in motor cortex: Interspersion of C8- and C4-projecting corticospinal cells. (A) Spinal cord injections of fluorescent microspheres resulted in retrograde labeling of cells throughout the forelimb area of the primary motor cortex. Mapping of the distribution of retrogradely labeled neurons reveals that C8- and C4-projecting neurons, which control distal and proximal forelimb musculature, respectively, exhibit no significant difference in cell distribution (P = 0.3 ANOVA group × rostral/caudal location). C8 injections labeled 67% more layer V cells than did injections at the C4 level in the same animal (P = 0.004; paired t test). (B) Skilled grasp training did not alter the distribution or total number of retrogradely labeled cells among either C8- or C4-projecting neurons (C8 distribution: P = 0.4, repeated-measures ANOVA across medial- to-lateral locations; C8 number: P = 0.7, unpaired t test; C4 distribution: P = 0.6; C4 number: P = 0.9).