Activity-dependent Organization of Topographic Neural Circuits

Abstract

Sensory information in the brain is organized into spatial representations, including retinotopic, somatotopic, and tonotopic maps, as well as ocular dominance columns. The spatial representation of sensory inputs is thought to be a fundamental organizational principle that is important for information processing. Topographic maps are plastic throughout an animal's life, reflecting changes in development and aging of brain circuitry, changes in the periphery and sensory input, and changes in circuitry, for instance in response to experience and learning. Here, we review mechanisms underlying the role of activity in the development, stability and plasticity of topographic maps, focusing on recent work suggesting that the spatial information in the visual field, and the resulting spatiotemporal patterns of activity, provide instructive cues that organize visual projections.

Keywords: Visual system; optic flow; spatiotemporal coding; synaptic plasticity; temporospatial coding; topographic map.

Figure 1.

Hebbian Models and Spike Timing Dependent Plasticity A. Schematic of Hebb’s statement “When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A’s efficiency, as one of the cells firing B, is increased.“ The red circle ‘A’ represents the presynaptic neuron and the blue triangle ‘B’ represents the postsynaptic neuron. B. Gunther Stent and von der Malsburg incorporated 2 additional concepts to the Hebb model: convergent coactive inputs are strengthened/maintained and uncorrelated convergent inputs are depressed/lost. Two presynaptic neurons, A and A’ in the red circles, converge on a postsynaptic neuron, B in the blue triangle. Convergent synaptic contacts would be maintained or lost depending on whether they were co-active or not. C. One form of the Hebbian STDP rule states that synaptic strength is enhanced or depressed depending on the temporal order and time windows during which correlated pre- and postsynaptic activity occurs. In this form of STDP, when presynaptic activity precedes postsynaptic activity by 20ms or less, synaptic strength is potentiated, resulting in LTP. By contrast when postsynaptic spikes precede presynaptic spikes, synaptic strength is decreased, resulting in LTD. There are other variations of temporal properties of STDP. Plasticity typically requires many repeated pairings of pre- and postsynaptic activity. Modified from Zhang et al, 1998.

Figure 2

Neural activity is required to establish the refined topographic retinotectal map in goldfish A, B. The positions of the recording electrodes in the tectal neuropil (A) and receptive fields of tectal cells (B) measured in the regenerated retinotectal projection of goldfish under control conditions (B, left) and in animals treated with TTX to block action potentials (middle) or exposed to 1 Hz strobe light, which synchronizes the activity of retinal ganglion cells (right). In the control, the receptive fields were organized topographically. In the TTX-treated group, the receptive fields were enlarged and overlapping. In animals exposed to strobe, the receptive fields were enlarged and disorganized. Modified from Schmidt, J.T., and Edwards, D.L. (1983) and Schmidt, J.T., and Buzzard, M. (1993).

Figure 3

Eye-specific segregation of retinotectal inputs requires neural activity in three-eyed Rana pipiens. A, B. Images of retinal projection patterns from the supernumerary eye in three-eyed tadpoles under control conditions or conditions in which retinal inputs were blocked by TTX for 4 weeks. The retinal projections were labeled with HRP and images were taken of flat-mounted tecta oriented with rostral up. Neuronal activity in the RGCs axons is required to maintain segregation of eye-specific inputs in the optic tectum. C, D. Drawings of individual retinotectal axons from control (C) and TTX-treated (D) animals show that blocking activity with TTX increases the size of the retinotectal axon arbor. Modified from Reh, T.A., and Constantine-Paton, M. (1985).

Figure 4.

Visual experience regulates the formation of ocular dominance columns and geniculocortical axon arbor structure. A. Ocular dominance columns in a monkey with normal visual experience (left) or monocular deprivation (right) revealed by autoradiographic detection of the labeling pattern, seen in white, following injection of radiolabeled amino acids into one eye. The right eye of the monocularly deprived monkey was closed at 2 weeks of age and ³H L-proline was injected into the left eye 18 months later. Note the expansion of the labeled ocular dominance columns corresponding to the open eye in white and decrease of the unlabeled dominance columns corresponding to the closed eye. Left and right panels of A are modified from Hubel, D. H. Wiesel, T. N. LeVay, S. (1977) and LeVay, S. Wiesel, T. N. Hubel, D. H. (1980), respectively. B. Structural plasticity of individual geniculocortical axon arbors underlies experience-dependent circuit plasticity. Coronal view of geniculocortical arbors reconstructed in kittens in which one eye had been occluded for six to seven days. The borders of cortical layers 3 and 4 are indicated by arrowheads. Modified from Antonini and Stryker (1993).

Figure 5.

NMDA-R activity in postsynaptic tectal neurons is necessary to for eye-specific segregation. The effect of the NMDA receptor antagonist DL-APV on segregation of RGC afferents in doubly innervated Xenopus optic tecta. A. Image of a representative tectum from a sham-operated animal shows the stereotypical striped pattern representing the segregation of the supernumerary retinal afferents from the afferents of the normal eye. B. Image of a representative tectum from an animal treated with DL-APV shows desegregation of the striped pattern. Adapted from Cline et al 1987.

Figure 6.

Repeated directional stimulus shifts the receptive field. Receptive fields in the optic tectum from Xenopus tadpoles mapped before (left) and after (right) exposure of the animals to a bar moving repeatedly in the anterior to posterior direction for 20 minutes. The circle and the dotted lines show the receptive field (RF) center and boundary, respectively. The shift in the receptive field can be detected by both a shift in the center and boundary of the RF before (blue) and after (red) the visual experience conditioning. Modified from Lim, B.K., et al, 2010.

Figure 7.

Optic flow instructs retinotopic map formation through a spatial to temporal to spatial transformation of visual information A. Sequential order of RGC spikes encodes information about their soma’s locations in the retina. The temporal order of RGC activity evoked by anterior to posterior optic flow contains information directing the spatial order of axon projection sites. Visual stimuli moving from anterior to posterior (right) activate RGCs sequentially in temporal to nasal positions in the retina. RGCs in the temporal to nasal axis of the retina innervate the tectum along its rostral to caudal axis. B. Visual stimulation mapping protocols and to test whether anterior to posterior optic flow affects topography along the rostrocaudal tectal axis. After repeated exposure to with A->P or P->A moving bars, RGC somata and axon projections sites were determined. C, D. Schematics of results. A->P optic flow generated rostrocaudal topography (C) whereas P->A stimuli blocked formation of rostrocaudal topography, determined by analysis of RGC axon arbor positions. Modified from Hiramoto, M. and Cline, H.T (2014).

Figure 8

Neurons that fire in sequence wire in sequence: A spatiotemporospatial (STS) transformation of visual information instructs map formation. A. Comparison of the STS transformation with the classical Hebb model. Schematics of Hebbian plasticity (left) and the spatial to temporal to spatial transformation of sensory information (right). While Hebbian model considered that temporal proximity guides topological organization of connections, experiments show that the sequential order of afferent activity is also used to organize the spatial distribution of projections. In both models, the axonal inputs of afferent cells, A, B,…E, use activity-dependent cues to establish connections in the target, such that the temporal proximity of afferent activity is transformed into the spatial proximity of their projection sites. In the Hebbian model, afferents with correlated activity terminate close to each other: neurons that fire together wire together. In TS, the temporal sequence of afferent activity is a critical parameter in organizing connectivity, such that the temporal sequence of input activity is transformed into the spatial order of axon projections: neurons that fire in sequence wire in sequence.

Figure 9

STS maintains topography challenged by STDP. Schematic of modifications of the retinotectal projection in response to repeated exposure to a single direction motion stimulus. The map in the middle shows the starting point. Left: Modifications based on STDP. When afferent inputs are activated repeatedly in a particular sequence, synaptic connections that receive input from neurons activated earlier are strengthened and the connections from the later activated inputs are weakened due to STDP. Eventually, the synaptic connections receiving earlier input dominate the connectivity. In the schematic, following repeated anterior to posterior activity, inputs from the anterior most RGCs dominate the retinotectal connections, distorting retinal topography. Right: Modification based on STS. STS maintains calibrated map connectivity across the target area. The relative spatial positions of projection sites are established as a transformation of the temporal sequence of active inputs.

Figure 10.

Spontaneous retinal waves travel from temporal to nasal retina. Recordings of spontaneous retinal waves from eyes of P8–10 mice demonstrate a predominant temporal to nasal direction. The chart shows the frequency distributions of propagation directions. Modified from Ge et al 2021.