Auditory experience refines cortico-basal ganglia inputs to motor cortex via remapping of single axons during vocal learning in zebra finches

Abstract

Experience-dependent changes in neural connectivity underlie developmental learning and result in life-long changes in behavior. In songbirds axons from the cortical region LMAN(core) (core region of lateral magnocellular nucleus of anterior nidopallium) convey the output of a basal ganglia circuit necessary for song learning to vocal motor cortex [robust nucleus of the arcopallium (RA)]. This axonal projection undergoes remodeling during the sensitive period for learning to achieve topographic organization. To examine how auditory experience instructs the development of connectivity in this pathway, we compared the morphology of individual LMAN(core)→RA axon arbors in normal juvenile songbirds to those raised in white noise. The spatial extent of axon arbors decreased during the first week of vocal learning, even in the absence of normal auditory experience. During the second week of vocal learning axon arbors of normal birds showed a loss of branches and varicosities; in contrast, experience-deprived birds showed no reduction in branches or varicosities and maintained some arbors in the wrong topographic location. Thus both experience-independent and experience-dependent processes are necessary to establish topographic organization in juvenile birds, which may allow birds to modify their vocal output in a directed manner and match their vocalizations to a tutor song. Many LMAN(core) axons of juvenile birds, but not adults, extended branches into dorsal arcopallium (Ad), a region adjacent to RA that is part of a parallel basal ganglia pathway also necessary for vocal learning. This transient projection provides a point of integration between the two basal ganglia pathways, suggesting that these branches convey corollary discharge signals as birds are actively engaged in learning.

Fig. 1.

A: lateral magnocellular nucleus of the anterior nidopallium (LMAN) consists of core and shell regions. The core region (C, black) forms a basal ganglia loop from area X of the medial striatum (area X) to DLM and back to LMAN_core, which projects to vocal motor cortex [robust nucleus of arcopallium (RA)]. The shell region (S, gray) forms a parallel basal ganglia loop from medial striatum (M Str) to DLM back to LMAN_shell, which projects to dorsal arcopallium (Ad). Although both afferent and efferent connections of LMAN are refined during vocal learning, large-scale topographic organizational changes are limited to the pathway from LMAN_core to RA. High vocal center (HVC) is not shown for clarity; like LMAN_core, HVC neurons send a major projection to RA. DLM, medial dorsolateral nucleus of the thalamus (DL contains neurons in the dorsolateral portion of DLM that project to LMAN_core, whereas VM contains neurons in the ventromedial portion of DLM that project to LMAN_shell); dNCL, dorsal region of the caudolateral nidopallium. B: schematic coronal sections (medial is left). The axonal projection from LMAN_core to RA is topographically organized from the medial-lateral axis within LMAN_core to the dorsal-ventral axis within RA. M, medial; I, intermediate; L, lateral. Modeled after Iyengar and Bottjer (2002b).

Fig. 2.

Top: biotinylated dextran amine (BDA)-labeled injection site in LMAN_core; injections completely filled several neurons within a limited area. Bottom: axon arbors were well filled; endings (e) and varicosities (v) could be clearly identified at ×1,000 magnification.

Fig. 3.

Location of all BDA-labeled cells for each injection site in relation to the Nissl-stained outline of LMAN_core. Injections were made bilaterally (see materials and methods), but all LMAN_core outlines are shown with medial to the left for consistency of viewing. wn, White noise. Scale bars, 100 μm.

Fig. 4.

Axon arbor reconstructions from LMAN_core neurons in 20-day, 27-day, and 35-day birds. Gray outlines show the Nissl-stained border of RA and the location of Ad. One axon branch in the 20-day bird projects through Ad to the most ventral portion of the arcopallium, Av. Schematic on right shows location of injection sites within LMAN_core. Numbers correspond to injection sites depicted in Fig. 3. Scale bars, 100 μm.

Fig. 5.

Reconstructions of LMAN_core axon arbors branching within RA. Gray outlines show the Nissl-stained border of RA; coronal view, medial is to the left. At each age 2 axons from 3 different LMAN_core injection sites are shown. Each color represents a single axon. Axon arbors frequently branched just before entering RA; for clarity, only the branches that enter RA are shown. Schematics on right show the location of each injection within LMAN_core; numbers correspond to injection sites shown in Fig. 3. Scale bars, 100 μm.

Fig. 6.

Spatial extent of LMAN_core axon arbors within RA decreased with age independent of normal auditory experience (all graphs show means + SE). Mean total length, volume, percentage of RA occupied, and dorsal-ventral extent of axon arbors decreased between 20 and 27 days. Measures of length and tangential extent are given in mm; volume is given in mm³. Axons of 35-day white noise (wn) birds did not differ from those of normal 35-day birds in any of these measures. *P < 0.05 compared with 20-day birds; ‡P < 0.005 compared with 20-day birds. See Table 1 for complete statistical comparisons.

Fig. 7.

Between 27 and 35 days the number of branches, number of branch orders, number of varicosities, and varicosity density decreased while mean branch length increased (all graphs show means + SE). White noise-reared 35-day birds had a significantly higher number of branches, number of varicosities, and varicosity density (number/mm) and a shorter mean branch length (mm) than normal 35-day birds. White noise birds were not significantly different from 20-day birds in any of these measures. *P < 0.05 compared with 20-day birds; ‡P < 0.005 compared with 20-day birds. See Table 1 for complete statistical comparisons.

Fig. 8.

LMAN_core axon arbors become better targeted to the correct topographic location with age. RA was divided into thirds in the dorsal-ventral direction (top left), and the percentage of total length, endings, and varicosities within each section were calculated as a function of the position of the injection site in LMAN_core (top right). Open circles represent the location and size of all injection sites. Bottom left: percentage of endings, length, and varicosities in the correct section of RA (means + SE). Bottom right: the volume encompassed by two LMAN_core axon arbors from the same injection site in combination was larger in white noise-reared birds than in normal birds. Box plots show the volume of all pairs of axon arbors across all injection sites in normal and white noise 35-day birds. Single dots represent outlier points.

Fig. 9.

Examples of axons with <50% of their varicosities in the correct section of RA in 20-day and white noise-reared 35-day birds. In each example the gray-shaded region highlights the middle section of RA to which axon arbors were matched based on the location of the injection site in LMAN_core. Some axons project to the wrong location in normal 20-day birds and in 35-day birds deprived of normal auditory experience. Top: the red axon in injection site 6 split into 5 branches just distal to the view shown here. Bottom: injection site 18 spanned both the lateral and lateral-intermediate sections of LMAN_core, so axon arbors could have been targeted to either the intermediate or the ventral section of RA. However, most of the orange axon was targeted to the dorsal section of RA, whereas the maroon axon arbor traversed RA dorso-ventrally and was less spatially refined, with varicosities spread throughout all 3 sections of RA. See Table 5 for quantitative data. Numbers correspond to injection sites shown in Fig. 3. One axon was reconstructed from injection sites 1, 3, and 19; 4 axons were reconstructed from injection sites 6 and 18 (only 3 are shown from injection site 6 for clarity). Scale bars, 100 μm.

Fig. 10.

LMAN_core→RA axon arbors of normal 35-day birds were topographically organized in the dorsal-ventral axis within RA, whereas some axon arbors of 35-day white noise birds projected to the wrong location. Top: location and size of injection sites within LMAN_core. The color of the axon arbors in the 2 bottom images correspond to the injection site of the same color. Middle: location within RA of axon arbors from each injection site in normal birds. Bottom: the axon arbors of 35-day white noise birds were less topographically organized. Axon arbors with asterisks have <50% of their varicosities in the correct section of RA. Numbers correspond to injection sites shown in Fig. 3.

Fig. 11.

Single axon arbors project to both RA and Ad in 35-day birds. Gray outlines show Nissl-stained border of RA and approximate location of Ad. Axon arbors are topographically organized within Ad; axon arbors from the most medial injection site project to the medial portion of Ad, while arbors from the most ventral site project to the lateral portion of Ad and arbors from an intermediate injection site are in between. Numbers correspond to injection sites shown in Fig. 3.