Lamination Speeds the Functional Development of Visual Circuits

Abstract

A common feature of the brain is the arrangement of synapses in layers. To examine the significance of this organizational feature, we studied the functional development of direction-selective (DS) circuits in the tectum of astray mutant zebrafish in which lamination of retinal ganglion cell (RGC) axons is lost. We show that although never laminar, the tuning of DS-RGC axons targeting the mutant tectum is normal. Analysis of mutant tectal neurons at late developmental stages reveals that directional tuning is indistinguishable from wild-type larvae. Furthermore, we show that structural plasticity of tectal dendrites and RGC axons compensates for the loss of lamination, establishing connectivity between DS-RGCs and their normal tectal targets. However, tectal direction selectivity is severely perturbed at earlier developmental stages. Thus, the formation of synaptic laminae is ultimately dispensable for the correct wiring of direction-selective tectal circuits, but it is crucial for the rapid assembly of these networks.

Figure 1

Tuning of DS-RGCs Innervating the Tectum Is Normal in astray Mutants Functional imaging in WT (n = 8, total of 24 optical sections) and ast^ti272z (n = 9, total of 27 optical sections) larvae expressing the presynaptic reporter of neural activity, SyGCaMP3, specifically in RGC axons (Tg(Isl2b:Gal4;UAS:SyGCaMP3)). (A–D) Cumulative histograms summarizing the incidence of preferred angles of all DS voxels in WT larvae at 4 dpf (A) and 7 dpf (B), and ast^ti272z larvae at 4 dpf (C) and 7 dpf (D). Overlaid colored curves show fitted von-Mises distributions used to identify three functional subtypes of DS-RGC. Color-coded arrows represent peak preferred angles of each subtype. R² values relate to the summed von-Mises distributions. Insets show polar plots of the mean (solid line) ± 1 SD (dashed line) normalized response profile for voxels of each DS-RGC subtype. (E and F) Quantification of DS voxel numbers shows no significant difference between WT and ast^ti272z RGCs at 4 dpf (E) or 7 dpf (F). All graphs show mean values ± SEM. ns, not significant, unpaired t test. See also Figure S1, Table S1, Movies S1 and S2.

Figure 2

Lamination of DS-RGC Axons in the Tectum Is Lost in astray Mutants (A–D) Composite parametric maps generated from all Tg(Isl2b:Gal4;UAS:SyGCaMP3) fish imaged showing the spatial distribution of DS-RGC subtypes in the tectal neuropil of WT larvae (n = 8, total of 24 optical sections) at 4 dpf (A) and 7 dpf (B), and ast^ti272z larvae (n = 9, total of 27 optical sections) at 4 dpf (C) and 7 dpf (D). Voxel brightness is proportional to the summed incidence of each functional subtype across all larvae imaged. The standard space template image derived from the mean fluorescence image of SyGCaMP3-expressing axons (grayscale) provides an anatomical reference. Dashed lines indicate the position of the skin overlaying the tectum. Scale bar represents 20 μm. A, anterior; L, lateral; SFGS, stratum fibrosum et griseum superficiale. (E–H) Line plots generated from the composite parametric maps in (A)–(D) illustrating the distribution of DS-RGC subtypes along the laminar axis of the tectum in WT larvae at 4 dpf (E) and 7 dpf (F), and ast^ti272z larvae at 4 dpf (G) and 7 dpf (H). Line plots show normalized intensity of each DS-RGC subtype as a function of its distance from the skin. (I–N) Pairwise comparisons showing the degree of spatial overlap between upward and downward (I and L), upward and forward (J and M), and downward and forward (K and N) DS-RGC subtypes within WT (I–K) and ast^ti272z (L–N) tecta at 4 dpf. Dotted area represents the area of intersection between the two subtypes. Values shown represent the fraction of downward (for I and L) or forward (for J, K, M, and N) DS voxels that spatially overlap with upward or downward DS voxels. See also Figures S2 and S3.

Figure 3

Functional Development of DS Tectal Cells Is Delayed in astray Mutants Functional imaging of tectal cell responses in WT (n = 12, total of 36 optical sections) and ast^ti272z (n = 12, total of 36 optical sections) larvae with pan-neuronal expression of GCaMP5G (Tg(elavl3:GCaMP5G)). (A–D) Cumulative histograms summarizing the incidence of preferred angles for DS tectal cells in WT larvae at 4 dpf (A) and 7 dpf (B), and ast^ti272z larvae at 4 dpf (C) and 7 dpf (D). Overlaid colored curves in (A), (B), and (D) show fitted von-Mises distributions used to identify the four DS subtypes and the color-coded arrows represent peak preferred angles. R² values relate to the summed von-Mises distributions. Insets in (A), (B), and (D) show polar plots of the mean (solid line) ± 1 SD (dashed line) normalized responses of cells within each DS subtype. Note that the same functional subtypes are present at 4 and 7 dpf in WT tecta. (E and F) Cumulative plots generated from the data in (A)–(D) comparing the incidence of preferred angles for DS tectal cells between WT and ast^ti272z larvae at 4 dpf (E) and 7 dpf (F). (G and H) Quantification of DS tectal cell number at 4 dpf (G) and 7 dpf (H). Note the significant reduction in DS tectal cell numbers at 4 dpf, but recovery at 7 dpf. All graphs show mean values ± SEM. ^∗p < 0.05; ns, not significant, unpaired t test. See also Figures S4 and S5, Table S2, Movies S3 and S4.

Figure 4

Morphology of DS Tectal Cell Dendrites Is Disrupted in the astray Mutant (A and B) Two alternative models of how direction selectivity recovers in the tectal cell population in ast^ti272z mutants, functional plasticity (A) or structural plasticity (B). SO, stratum opticum; SFGS, stratum fibrosum et griseum superficiale; SGC, stratum griseum centrale; SAC, stratum album centrale; SPV, stratum periventriculare. (C and D) Polar plots illustrating functional subtypes of WT (n = 24 in 22 larvae) (C) and ast^ti272z (n = 19 in 16 larvae) (D) tectal cells singly labeled with GCaMP6F in the 7 dpf tectum following co-electroporations of FoxP2.A:Gal4FF + 5UAS:GCaMP6F DNA constructs. Dashed lines indicate polar plots from individual cells, and filled color-coded polar plots represent the mean for each functional subtype. Pie charts to the right indicate the relative proportions of functional subtypes and number of neurons analyzed. (E and F) Morphologies of tectal cells selective for forward motion. Two examples from WT (E) and ast^ti272z tecta (F) are shown. Note that the laminar profile of DS tectal cells tuned to forward motion is lost in ast^ti272z animals (arrowheads). Dashed lines indicate the position of the skin overlaying the tectum. Scale bar represents 30 μm. A, anterior; D, dorsal; L, lateral. (G and H) Line plots comparing the distribution of forward tuned DS-RGC axons and forward tuned DS tectal cell dendrites along the laminar axis of the tectum at 7 dpf in WT (G) and ast^ti272z (H) animals. Line plots show normalized intensity of each cell type as a function of its distance from the skin. Note the alignment of DS-RGC axons and DS tectal dendrites in both conditions. (I–L) Comparison of four morphological parameters extracted from all tectal cells (Figure S4A). For total arbor length (I), 682.6 ± 62.4 for WT and 724.6 ± 50.5 for ast^ti272z; for distance from skin (J), 15.2 ± 1.3 for WT and 12.6 ± 1.2 for ast^ti272z; for distal arbor anterior-posterior span (K), 42 ± 3.6 for WT and 42.3 ± 2.2 for ast^ti272z; and for distal arbor laminar extent (L), 14.9 ± 1.2 for WT and 28.2 ± 2.4 for ast^ti272z. Data points are color coded according to functional subtype as in (C) and (D). All graphs show mean values ± SEM. ^∗∗∗p < 0.001; ns, not significant, Mann Whitney test. (M) Proportions of cells with laminar and diffuse (distal arbor laminar extent < 16 μm and > 16 μm, respectively) arbors in WT and ast^ti272z tecta. WT: 67% laminar, 33% diffuse arbor; and ast^ti272z: 16% laminar, 84% diffuse arbor. See also Figures S6 and S7 and Movies S5 and S6.

Figure 5

Robo2 Directs Laminar Growth of Tectal Cell Dendrites (A) Experimental procedure used to assess the role of RGC axons and/or Robo2 in regulating the morphology of tectal cell dendrites. Optic vesicles were removed at 12–14 hpf. Single cells were labeled by co-electroporating FoxP2.A:Gal4FF + 5UAS:tdTomato DNA constructs into the deprived tectum at 4 dpf. Single tectal cells were imaged at 7 dpf. (B–E) Comparison of morphological parameters of tectal cells in four conditions: WT (n = 24 cells in 22 larvae), WT enucleated (n = 7 cells in 6 larvae), ast^ti272z enucleated (n = 10 cell in 9 larvae), and ast^ti272z (n = 19 cells in 16 larvae). For total arbor length (B), 682.6 ± 62.4 for WT, 567.4 ± 66.7 for WT enucleated, 681.7 ± 110.4 for ast^ti272z enucleated, and 724.6 ± 50.5 for ast^ti272z; for distance from skin (C), 15.2 ± 1.3 for WT, 5.4 ± 0.5 for WT enucleated, 8.3 ± 3.4 for ast^ti272z enucleated, and 12.6 ± 1.2 for ast^ti272z; for distal arbor anterior-posterior span (D), 42 ± 3.6 for WT, 41.4 ± 4.1 for WT enucleated, 44.6 ± 5.6 for ast^ti272z enucleated, and 42.3 ± 2.2 for ast^ti272z; and for distal arbor laminar extent (E), 14.9 ± 1.2 for WT, 11.8 ± 0.9 for WT enucleated, 22.4 ± 2.9 for ast^ti272z enucleated, and 28.2 ± 2.4 for ast^ti272z. All graphs show mean values ± SEM. ^∗∗p < 0.01; ^∗p < 0.05; ns, not significant, Kruskal-Wallis and Dunn’s multiple comparison tests. (F) Proportions of cells with laminar and diffuse (distal arbor laminar extent < 16 μm and > 16 μm, respectively) arbors within each experimental group. WT: 67% laminar, 33% diffuse arbor; WT enucleated: 86% laminar, 14% diffuse arbor; ast^ti272z enucleated: 20% laminar, 80% diffuse arbor; and ast^ti272z: 16% laminar, 84% diffuse arbor.

Figure 6

Robo2 Is Required in RGC Axons and Tectum for Rapid Development of DS Responses in Tectal Cells (A) Experimental procedure used to assess the cellular requirements for Robo2 in the functional development of DS tectal cells. Optic vesicles from donor embryos were transplanted into Tg(elavl3:GCaMP5G) hosts at 12-14 hpf and functional imaging of tectal cell responses was performed at 4 and 7 dpf. (B and C) Quantification of average number of DS tectal cells per group. Total number of DS cells at 4 dpf (B), 53.4 ± 11.5 for WT-to-WT (n = 12 larvae, total of 36 optical sections), 21.9 ± 7.9 for WT-to-ast^ti272z (n = 12 larvae, total of 36 optical sections), and 15.8 ± 6.1 for ast^ti272z-to-WT (n = 8 larvae, total of 24 optical sections). Total number of DS cells at 7 dpf (C), 14.4 ± 6.6 for WT-to-WT (n = 9 larvae, total of 27 optical sections), 8.1 ± 3.8 for WT-to-ast^ti272z (n = 7 larvae, total of 21 optical sections), and 9.3 ± 2 for ast^ti272z-to-WT (n = 6 larvae, total of 18 optical sections). All graphs show mean values ± SEM. ^∗p < 0.05; ns, not significant, Kruskal-Wallis and Dunn’s multiple comparison tests. See also Figure S8.

Figure 7

robo2^+/+ RGC Axons Provide Positional Cues for robo2^−/− Tectal Cell Dendrites (A) Schematic showing the experimental procedure followed. Optic vesicles from donor embryos were transplanted into hosts at 12–14 hpf. Single cells were labeled by co-electroporating FoxP2.A:Gal4FF + 5UAS:tdTomato DNA constructs into the tectum receiving retinal input from transplanted eyes at 4 dpf and imaged at 7 dpf. (B–E) Comparisons of the main morphological parameters of tectal cells in four conditions: WT (n = 24 in 22 larvae), WT-to-WT (n = 7 in 6 larvae), WT-to-ast^ti272z (n = 6 in 5 larvae), and ast^ti272z-to-WT (n = 5 in 4 larvae) tectal cells. For total arbor length (B), 682.6 ± 62.4 for WT, 423.6 ± 60 for WT-to-WT, 399.5 ± 71.9 for WT-to-ast^ti272z, and 421.8 ± 64.5 for ast^ti272z-to-WT; for distance from skin (C), 15.2 ± 1.3 for WT, 9.4 ± 3.4 for WT-to-WT, 9.3 ± 1.4 for WT-to-ast^ti272z, and 9 ± 1.9 for ast^ti272z-to-WT; for distal arbor anterior-posterior span (D), 42 ± 3.6 for WT, 29.2 ± 2.9 for WT-to-WT, 28.2 ± 5.5 for WT-to-ast^ti272z, and 29.9 ± 5.8 for ast^ti272z-to-WT; and for distal arbor laminar extent (E), 14.9 ± 1.2 for WT, 15.3 ± 2.6 for WT-to-WT, 13.2 ± 1.8 for WT-to-ast^ti272z, and 12.8 ± 1.3 for ast^ti272z-to-WT. All graphs show mean values ± SEM ns, not significant, Kruskal-Wallis and Dunn’s multiple comparison tests. (F) Proportions of cells with laminar and diffuse (distal arbor laminar extent < 16 μm and > 16 μm, respectively) arbors within each experimental group. WT: 67% laminar, 33% diffuse arbor; WT-to-WT: 71% laminar, 29% diffuse arbor; WT-to-ast^ti272z: 67% laminar, 33% diffuse arbor; and ast^ti272z-to-WT: 60% laminar, 40% diffuse arbor.

Figure 8

robo2^+/+ Tectal Cell Dendrites Provide Positional Cues for robo2^−/− RGC Axons (A) Schematic showing the experimental procedure followed. Optic vesicles from Tg(Isl2b:Gal4;UAS:SyGCaMP3) donor embryos were transplanted into hosts at 12–14 hpf and functional imaging of donor RGC axons within the tectum was performed at 7 dpf. (B–D) Composite parametric maps across all fish imaged showing the spatial distribution of forward tuned DS voxels in the tectal neuropil of WT-to-WT (n = 8 larvae, total of 24 optical sections) (B), WT-to-ast^ti272z (n = 4 larvae, total of 12 optical sections) (C), and ast^ti272z-to-WT (n = 4 larvae, total of 12 optical sections) (D). Within individual parametric maps, voxel brightness is proportional to the summed incidence of the functional subtype across all larvae imaged. The standard space template image derived from the mean fluorescence image of SyGCaMP3-expressing axons (grayscale) provides an anatomical reference. Dashed lines indicate the position of the skin overlaying the tectum. Scale bar represents 20 μm. A, anterior; L, lateral. (E) Line plots generated from the composite parametric maps in (B)–(D) illustrating the laminar organization of forward tuned DS voxels within each group. Line plots show normalized intensity of DS voxels as a function of distance from the skin. (F) Pairwise comparisons showing the extent of spatial overlap in the distribution of forward tuned DS voxels in the different experimental groups. Note the increased fraction of correctly positioned DS voxels in transplanted ast^ti272z axons (ast^ti272z-to-WT group) compared with the ast^ti272z condition.