Visual information is delivered to the brain by >40 types of retinal ganglion cells (RGCs). Diversity in this representation arises within the inner plexiform layer (IPL), where dendrites of each RGC type are restricted to specific sublaminae, limiting the interneuronal types that can innervate them. How such dendritic restriction arises is unclear. We show that the transcription factor Tbr1 is expressed by four mouse RGC types with dendrites in the outer IPL and is required for their laminar specification. Loss of Tbr1 results in elaboration of dendrites within the inner IPL, while misexpression in other cells retargets their neurites to the outer IPL. Two transmembrane molecules, Sorcs3 and Cdh8, act as effectors of the Tbr1-controlled lamination program. However, they are expressed in just one Tbr1+ RGC type, supporting a model in which a single transcription factor implements similar laminar choices in distinct cell types by recruiting partially non-overlapping effectors.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Z. He and C. Wang at Boston Children’s Hospital Viral Core for their generous support (5P30EY012196). This work was supported by NIH grants NS29169 and EY022073 to J.R.S., NS34661 to J.L.R., and MH094589 to B.C. J.L. was funded by an Agency for Science, Technology and Research (A*STAR) fellowship from Singapore.
Integrated supplementary information
(a) Whole mounts of retina stained for Tbr1 plus Tbr2, FoxP2, FoxP1 or Satb1. (b) Effective radius and cell density of each Tbr1-RGC type. n=4 fields per retina, 3 retinas per type, each retina from a different animal. Each point indicates mean + standard error for both x and y axes. Scale = 25µm. (c-f) Distribution of somata of each Tbr1-RGC type. (g) Proportions of each Tbr1-RGC type as a fraction of all RGCs (RBPMS-positive cells) in a retina. (h) Whole mount of a retina stained with Tbr1, Brn3c, calretinin, Brn3b and Opn. All Tbr1+ cells express one of these four markers. Open triangles, closed triangles and arrows indicate Tbr1+ cells expressing Brn3c/calretinin, Brn3b or Opn respectively. Scale=25µm. (i-l) Whole mounts of each Tbr1-RGC type and the markers they express. Scale=25µm. (m, n) Whole mount showing that YFP-positive RGCs in the TYW3 and Hb9-GFP transgenic lines are Tbr1-negative. Scale=25µm. (o) Coverage factor, defined as product of mean soma density (from b) and mean dendritic field area (from p). Bars and brackets indicate mean + standard error. (p) Dendritic field area of each Tbr1-RGC type. Top and bottom whiskers represent 10th and 90th percentiles. Upper and lower box limits represent 25th and 75th percentiles. Center line marks median. n=19, 87, 19, 36 cells for J-RGC, α-OFF-s-RGC, Tbr1-S1-RGC and Tbr1-S2-RGC respectively. Each experiment in a, c-f and h-n was repeated independently in three animals with similar results.
(a) Retinal cross-sections at indicated embryonic ages stained for Tbr1, Brn3b/c. Tbr1 (yellow arrowheads) appears in RGCs by E17.5. (b-c) Retinal cross-sections at E17.5, showing Tbr1, Tbr2 and FoxP2 protein in non-overlapping populations. Scale=20µm. Each experiment in a-c was repeated independently in three animals with similar results.
(a) Tbr1 gene and targeting vector used to generate the Tbr1loxP allele. The wildtype Tbr1 allele has six known exons (numbered black boxes, 1–6); the initiation codon is in exon 1, and the termination codon is in exon 6. White boxes indicate the 5’ and 3’ UTR. Red arrowheads correspond to the location of LoxP sites; black box with ‘F’ inside are Frt sites. Germline mutants were mated to a Flp-expressing transgenic to remove the Neomycin expression cassette (grey box with Neo inside). Upon Cre recombination, exons 2 and 3 are deleted to generate Tbr1 mutant allele. (b) Cross-section showing normal retinal architecture in the absence of Tbr1. RBPMS marks all RGCs; Syt2 marks primarily type II bipolar cells, VGlut3 marks a S3-laminating amacrine cell type, Calbindin (Calb) marks horizontal cells and subsets of amacrine and ganglion cells. Scale=25μm. Experiment was repeated independently in three animals with similar results.
(a-b) Dendritic area and length in control and Tbr1J/J J-RGCs. n=10 cells from 3 retinas per genotype, each retina from a different animal. Bars represent mean ± standard error. Circles represent individual cells. (c) Proportions of ventrally asymmetric J-RGCs in control versus Tbr1J/J retinas. n=132–710 control cells and 77–299 Tbr1J/J cells, 4 retinas per genotype, each retina from a different animal. Bars represent mean+ standard error, circles represent individual retina. p=0.62, 0.40 and 0.39 for area, length and proportion of ventrally asymmetric J-RGCs, two-tailed Student’s t-test, ns=not significant. (d) Axonal projections of Tbr1+/J and Tbr1J/J J-RGCs into the superior colliculus. Scale=100µm. (e) Cross-sections of Tbr1J/J J-RGCs (arrowheads) showing no loss of markers normally expressed by J-RGCs (in yellow box) and no gain in markers for other RGC types. Each box outlined in black encloses corresponding images. Pcsk2 was identified from the J-RGC RNAseq data. All J-RGCs shown here were labeled by tamoxifen at P0, except cells in the first row, which were labeled by tamoxifen at E14.5. Scale = 25µm. (f-g) Dendritic area and length of α-OFF-s-RGCs in control and Tbr1ret/ret retinas. n=15 and 13 cells from 6 control and 9 Tbr1ret/ret retinas respectively. Two-tailed Student’s t-test, p=0.75 and 0.18 for area and length respectively. Bars represent mean ± standard error. Circles represent individual cells. Each experiment in d-e was repeated independently in three animals with similar results.
(a) Experimental timeline for early (magenta) and late (red) ablation of Tbr1. Green arrowhead marks when J-RGC dendrites become restricted to S128. (b) P4 retinal cross-sections from controls and Tbr1J/J retinas following deletion of Tbr1 at P0. Scale=25µm. (c) Distribution of fluorescence intensities of J-RGC dendrites from images such as those in b. Gray arrowheads indicate S2 and S4 as labeled by VAchT immunostaining. p=0.00018, Cochran-Armitage test. n = 11 and 18 sections from control and Tbr1J/J animals respectively, 3 animals per genotype. (d) P20 retinal cross-sections from control and Tbr1J/J retinas where Tbr1 is ablated after laminar restriction. Scale=25µm. (e) Distribution of fluorescence intensities of J-RGC dendrites from images such as those in d. p=0.096, Cochran-Armitage test. n = 8 and 14 sections from 3 control and 4 Tbr1J/J animals respectively.
(a-d) Averaged responses (in black) to 2-second full-field flashes (in red) for 22, 18, 13 and 15 J-RGCs, α-OFF-s-RGCs, Tbr1-S1-RGCs and Tbr1-S2-RGCs respectively, as identified by Tbr1 immunostaining. x and y axes are time in seconds and Δf/f respectively. J-RGCs give ON-OFF responses as shown previously in Figure 2d from Kim et al. 200813; it appears to be due to a strong surround. Tbr1-S1-RGCs also showed ON-OFF responses to full field flashes. Tbr1-S2-RGCs were poorly responsive with the stimulus parameters used. α-OFF-s-RGCs gave strong OFF responses (data re-plotted from Figure 3c for comparison). (e-g) Polar plots of maximum calcium responses evoked by bars moving in 8 directions. Outermost gray circle is Δf/f=1. e, Tbr1+ Opn+ control α-OFF-s-RGCs. f, Tbr2- Brn3c- Opn+ control α-OFF-s-RGCs. g, Tbr2- Brn3c- Opn+ Tbr1ret/ret α-OFF-s-RGCs. (h-i) Distribution of z-score (h) and quality index (i) for all RGCs recorded from the retinas shown in Figures 3b-m (n= 1083 RGCs from 3 control and 3030 RGCs from 6 mutant retinas). These statistics were calculated as described in Methods. Responsiveness of control and mutant RGCs were similar by both measures.
(a) En face view of Tbr1-misexpressing RGCs showing a variety of dendritic arborization patterns, contrasting with the similarity in laminar restriction shown in Figure 4. Arrowheads mark Tbr1-overexpressing RGCs. Note that XFP-positive cells include Brn3bhigh, Brn3blow and Brn3b-negative types. Scale=50μm. Experiment was repeated independently in two animals with similar results. (b-d) Tbr1 mis-expressing cells in the inner nuclear layer showing robust overexpression of Tbr1. These cells are amacrine cells (Rbpms-negative, AP2-positive). Scale=25μm. Experiment was repeated independently in three animals with similar results.
(a) Integrative Genomics Viewer plots showing alignment of Tbr1 ChIP-seq peaks32 to the Cdh8 (top panel) and Sorcs3 (bottom panel) genomic loci. For each gene, the top row represents viewing window in kilobases and the gene locus. Bottom two blue traces are peaks from n = 2 Tbr1 ChIP-seq replicates. Peaks associated with Cdh8 and Sorcs3, but not Jam2, Alcam or Smo, were within 50kb upstream of the annotated transcriptional start site, and may therefore be associated with regulatory regions. (b-c) Retinal cross-sections from control and Tbr1ret/ret stained for Alcam and Neo1, showing no difference. Scale=25µm. (d) Expression of lacZ within the ganglion cell layer (GCL) of Cdh8lacZ retinas from P3 to P7. Scale=50µm. (e) Retinal cross-section stained for Sorcs3 and Protein kinase C (PKC), a rod bipolar cell marker. Yellow arrows mark co-localization of Sorcs3 with dendrites of rod bipolar cells. Scale=25µm. (f) Expression of Sorcs3 in retinal cross-sections from P4 to P16. DR, animals dark-reared from birth. Upregulation of Sorcs3 in RGCs after eye-opening appears to be light-dependent. Scale=25µm. (g) Whole mounts from a Cdh8lacZ; TYW7 retina. RGCs labeled in this line, including α-OFF-s-RGCs, do not express lacZ. Scale =25µm. Each experiment in b-g was repeated independently in three animals with similar results.
(a) Average CPM of genes encoding cell-surface molecules that are significantly different between wildtype and mutant J-RGCs (two-tailed Fisher’s Exact Test, p < 0.001). The larger the -log10(p-value), the greater the statistical significance. n= 5 samples from 4 Tbr1+/+ animals and 6 samples from 4 Tbr1J/J animals. (b) Average CPM of markers normally expressed by J-RGCs (highlighted green) and of markers expressed by other specific RGC types (highlighted magenta). Tbr1 mutant J-RGCs retained expression of markers normally expressed by wildtype J-RGCs and did not gain expression of markers for other RGC types. Bars and brackets in a-b indicate average CPM and standard error. (c) Heatmap showing relative microarray values across different groups of RGCs for the differentially expressed genes in a. Most of these genes are present in other non-Tbr1 RGC types, except for Jam2, Sorcs3 and Cdh8 (yellow box).
(a) Cross-sections of OFF alpha RGCs labeled by TYW7; Cdh8+/lacZ and TYW7; Cdh8lacZ/lacZ retinas. (b) Distribution of GFP intensities from OFF alpha RGC dendrites along IPL depth. n=10 and 8 sections from Cdh8+/lacZ and Cdh8lacZ/lacZ animals respectively, 3 animals per genotype. p=0.12, Cochran-Armitage test. Scale=25µm. Lines and brackets indicate mean and standard error. (c) DNA construct for AAV encoding short-hairpin RNA against Sorcs3 (shSorcs3). (d-e) Whole mount view of GCL of a retina infected with control AAV expressing Td-Tomato (TdT), or AAV encoding shSORCS3. Sorcs3 protein is greatly reduced in shSORCS3-treated retinas. Scale=25µm. Each experiment in d-e was repeated independently in 3 and 5 animals respectively with similar results.