Molecular and cellular mechanisms of lamina-specific axon targeting

Abstract

The specificity of synaptic connections is directly related to the functional integrity of neural circuits. Long-range axon guidance and topographic mapping mechanisms bring axons into spatial proximity of target cells and thus limit the number of potential synaptic partners. Synaptic specificity is then achieved by extracellular short-range guidance cues and cell-surface recognition cues. Neural activity may enhance the precision and strength of specific circuit connections. Here, we focus on one of the final steps of synaptic matchmaking: the targeting of synaptic layers and the mutual recognition of axons and dendrites within these layers.

Figure 1.

Laminae are a fundamental organizing unit of neural circuits. Each column corresponds to a single topographic position (e.g., location on the retina). Within each column, different cell types (shown type A: blue, and type B: red) respond to different features in the visual world, such as motion or luminance. These pixels are repeated many times over and thus cover all of visual space. A simple rule of “Cell type A connects to Layer A, etc.” ensures that functional segregation is maintained in the connections from the retina to the target (parallel processing). Each pixel P1, P2, and P3 connects to a single column (C1, C2, and C3), establishing serial processing. Within each column, there are local circuits that, too, are layer-specific. Thus, laminae ensure functional specificity of both afferent-target connections and local circuit connections.

Figure 2.

Lamina-specific retina-to-target connections across organisms. (A) Drosophila. Photoreceptors 1–6 (R1–6; red) in the retina project to their target in the brain, the lamina. Retinal photoreceptors R7 (green) and R8 (blue) project to different sublaminae of the medulla. Lamina neurons L1–L5 also send layer-specific connections to the medulla. (B) Chick. Axons from different types of RGCs (blue, yellow, and red) enter the tectum through the stratum opticum (SO) and then dive into one of three laminae (“B, D, or F” of the SGFS), where they establish synaptic connections onto neurons whose somas and/or dendrites reside in that specific layer. (C) Zebrafish. Axons from different types of RGCs (blue, yellow, and red) enter the tectum from the rostral pole and directly target either the SO or one of the three sublaminae of the SFGS. Two deeper layers also receive sparse retinal input (not shown). The SO-projecting axons extend a collateral into the pretectal nucleus AF-7. (D) Mouse. Axons from different types of RGCs (blue, yellow, and red) all enter the tectum together through the SO. Some terminate in SO (yellow axons), whereas most turn dorsally to synapse either in the lower half (e.g., blue axon) or upper half (red axon) of the SGS (stratum griseum superficiale). The thin stratum zonale (SZ) receives sparse, if any RGC input.

Figure 3.

Developmental sequence leading to sublamination in the retinal inner plexiform layer. (A–C) Summary of time-lapse imaging experiments in the developing zebrafish retina. (A) At 48 hours postfertilization (hpf), all amacrine cells (dark blue and green cells) are confined to inner nuclear layer (INL) and extend rudimentary processes in all directions. RGCs (red, light blue, and yellow) reside in the ganglion cell layer (GCL) and project their axons into the nerve fiber layer (NFL) but are yet to extend dendrites into the inner plexiform layer (IPL). (B) At 60 hpf, amacrine cells project their dendrites into separate strata (blue and green zones) in the IPL and RGCs extend dendrites into the IPL. Some amacrine somas begin to push through the IPL, toward the GCL; these are future “displaced amacrine cells.” (C) By 72 hpf, different types of RGCs (red, an On type, blue, an Off type, and yellow, On-Off type) confine their dendrites to separate strata in the proximal (On, red) or distal (Off, blue) half of the IPL. Displaced amacrine somas are present in the GCL. The dendrites of conventional and displaced amacrines and RGCs costratify in the correct sublaminae where they form synapses. (D–F) Summary of static images of IPL stratification in mouse. Labeling conventions as in A–C. (D) Before vision, many RGCs target either the On or Off sublaminae of the IPL and cells that are destined to remain bistratified are bistratified at this stage (Huberman, unpubl. observations). (E) At eye opening (∼14 days of age in mouse), many more bistratified RGCs exist than will persist into adulthood, although some bias in purely On- or purely Off-sublamina targeting is already present. (F) With normal visual experience from eye opening, the IPL shows On-, Off- and On-Off targeted RGC dendrites.

Figure 4.

Molecular mechanisms of layer-specific axon targeting in the fly visual system. (A) Molecular matching of two different populations of photoreceptor axons (green or yellow) that each express different molecular recognition signals (red or blue). The target layers of these axons express either one recognition cue (blue) or a different molecular recognition cue (red), leading to a precise match between specific sets of axons and cells in specific laminae. (B) A temporal code could also underlie axon-target matching. One model is that a given set of axons (green) grow into the target and then turn on a surface recognition cue (blue), “locking” them into the most distal, bottom layer that expresses the matching cue. Later, a second group of axons (yellow) grow in to the target structure. They are already expressing the surface recognition cue (blue) and are directly targeting the first blue layer they encounter. (C) An alternative temporal code model is that axons from different populations of neurons arrive already expressing the relevant layer-specific recognition cue. However, the expression of the matching cue is dynamic in the target. Thus, slight differences in timing of ingrowth will cause early arriving axons (here green) to synapse in certain layers, and other later arriving axons (yellow) to synapse in other layers with delayed expression of the cue. (D) Layer-specific axon targeting based on matching of axons and target layers expressing different levels of ligands and receptors. Here, the high-expressing growth cones (green axons) target the darker, high expressing (dark) blue target layers. The low expressing growth cones (yellow) target the also low expressing (lighter blue) target layers.

Figure 5.

Established and putative molecular mechanisms for layer-specific axon targeting in the vertebrate tectum. (A) In zebrafish, RGC axons target the appropriate tectal laminae from the outset. This could be accomplished by guiding axons into their respective layers through repulsion and/or attraction. Guidance cues could be confined to narrow layers or expressed as gradients across the depth of the target (purple gradient from ventral to dorsal SFGC) and define domains of attraction and/or repulsion to subsets of RGC axons. (B) In mammals, some RGC axons initially target the wrong lamina and then refine to their correct target layer (light blue and red axons). Other RGC axons may target the correct lamina from the outset (yellow SO-targeting axon). The same targeting mechanisms presented in panel A could regulate these processes.