Reelin is required for class-specific retinogeniculate targeting

Abstract

Development of visual system circuitry requires the formation of precise synaptic connections between neurons in the retina and brain. For example, axons from retinal ganglion cells (RGCs) form synapses onto neurons within subnuclei of the lateral geniculate nucleus (LGN) [i.e., the dorsal LGN (dLGN), ventral LGN (vLGN), and intergeniculate leaflet (IGL)]. Distinct classes of RGCs project to these subnuclei: the dLGN is innervated by image-forming RGCs, whereas the vLGN and IGL are innervated by non-image-forming RGCs. To explore potential mechanisms regulating class-specific LGN targeting, we sought to identify differentially expressed targeting molecules in these LGN subnuclei. One candidate targeting molecule enriched in the vLGN and IGL during retinogeniculate circuit formation was the extracellular matrix molecule reelin. Anterograde labeling of RGC axons in mutant mice lacking functional reelin (reln(rl/rl)) revealed reduced patterns of vLGN and IGL innervation and misrouted RGC axons in adjacent non-retino-recipient thalamic nuclei. Using genetic reporter mice, we further demonstrated that mistargeted axons were from non-image-forming, intrinsically photosensitive RGCs (ipRGCs). In contrast to mistargeted ipRGC axons, axons arising from image-forming RGCs and layer VI cortical neurons correctly targeted the dLGN in reln(rl/rl) mutants. Together, these data reveal that reelin is essential for the targeting of LGN subnuclei by functionally distinct classes of RGCs.

Figure 1.

Reelin is selectively expressed in vLGN and IGL. A, B, Schematic representation of non-image-forming RGC (A) and image-forming RGC (B) axons in LGN subnuclei. C, Differentially expressed guidance and targeting cues in either vLGN/IGL or dLGN, displayed in a heat plot and normalized to the mean intensity of each gene. Color scale depicts fold difference between vLGN/IGL and dLGN. RNA was purified from three sets of vLGN/IGL and dLGN samples, each of which was analyzed by Agilent microarray and is represented by a column in the heat plot. Each sample contained tissue pooled from at least five mice. D, Developmental regulation of differential gene expression in vLGN/IGL compared with dLGN (±SD) assessed by qPCR. RNA was isolated from five ages. In contrast to other genes, the highest levels of reln mRNA enrichment in vLGN/IGL coincided with the arrival of RGC axons. E–G, IHC for reelin on coronal sections of mouse thalamus. IHC confirmed the developmental regulation of reelin in vLGN and IGL during periods of retinogeniculate targeting. The dotted green and white lines outline vLGN and dLGN, respectively. Scale bar: (in G) E–G, 300 μm. d, dLGN; i, IGL; v, vLGN; OT, optic tract.

Figure 2.

Reelin-signaling components are expressed in the GCL of mouse retina. A, ISH for synaptotagmin 1 (syt1) mRNA in P13 wild-type retinal cross-sections demonstrated the distribution of neurons within the GCL. B, C, In contrast to the wide distribution of syt1 mRNA, ISH revealed that apoer2 (B) and vldlr (C) mRNAs were expressed by small subsets of cells within the GCL (see arrowheads in B, C). D, Dab1 IHC in P13 wild-type retinal cross-sections confirmed expression of Dab1 by amacrine cells in the INL (arrows) (Rice and Curran, 2000). Sparse labeling of Dab1 was also observed in a subset of cells in the GCL (arrowhead). E, Colabeling of Meln and Dab1 in P13 wild-type retinal cross-section. The arrowhead depicts Dab1-expressing ipRGC. The arrows indicate Dab1-expressing amacrine cells. The asterisk denotes nonspecific immunoreactivity. F, Reelin IHC in P13 wild-type retinal cross-sections. Many cells within the GCL express reelin; however, the arrowheads indicate cells with no significant reelin expression. G, H, Colabeling of P13 wild-type retinal cross-sections with Meln and reelin antibodies revealed that most ipRGCs do not express reelin (see arrowheads). Meln-expressing RGCs in G and H presumably represent different classes of ipRGCs since there dendrites stratify in different sublaminae of the IPL (see arrows). Both classes of ipRGCs appear to lack reelin expression. I, Reelin IHC was performed on P13 retinal cross-sections from opn4-tau-LacZ transgenic mice. M1 ipRGCs were labeled by LacZ IHC. The arrowhead highlights a M1 ipRGC that lacks reelin expression. For all images, nuclei were labeled by DAPI staining. Scale bar: A–C, E, 15 μm; D, F–I, 20 μm.

Figure 3.

Reelin is necessary for retinogeniculate targeting. A, B, Retinogeniculate projections in P14 control (A) or reln^rl/rl mutant mice (B) assessed by labeling of RGC axons with CTB. The left eyes were injected with Alexa Fluor 594-CTB and right eyes with Alexa Fluor 488-CTB. LGN from right hemispheres are shown. “Contra” denotes retinal arbors originating from cells in the contralateral eye, and “ipsi” denotes those originating from ipsilateral eye. In mutant LGN (B), note the near absence of retinal projections in IGL (arrows), the reduced pattern of vLGN innervation, and the presence of mistargeted axons (asterisk). C and D show high-magnification images of areas depicted by asterisks in A and B, respectively. Projections from both eyes have been merged into a grayscale image for improved visualization [CTB(contra+ipsi)]. E, F, Retinogeniculate projections in P1 control (E) or reln^rl/rl mutant mice (F). RGC axons from both eyes were labeled with the same fluorescently conjugated form of CTB [CTB(b)]. The arrowhead in F highlights mistargeted RGC axons. The arrows indicate IGL. d, dLGN; v, vLGN. Scale bars: (in A) A, B, 400 μm; (in C) C, D, 200 μm; (in E) E, F, 300 μm.

Figure 4.

Dab1 is necessary for retinogeniculate targeting. A, B, Retinogeniculate projections in P14 control (A) or dab1^scm/scm mutant mice (B, C) assessed by labeling of RGC axons with CTB. The left eyes were injected with Alexa Fluor 594-CTB and the right eyes with Alexa Fluor 488-CTB. LGNs from right hemispheres are shown. “Contra” denotes retinal arbors originating from cells in the contralateral eye, and “ipsi” denotes those originating from ipsilateral eye. In mutants, two different retinogeniculate phenotypes were observed. In one set of dab1^scm/scm mutants (B), a near absence of retinal projections was observed in IGL (arrows), whereas in another set, retinal projections were observed in the IGL (C). In both cases (B, C), mutants exhibited a reduced pattern of vLGN innervation and mistargeted axons were observed exiting the LGN (asterisks). A′, B′, and C′ show high-magnification images of areas depicted by asterisks in A–C, respectively. D–F, Retinogeniculate projections in P2 control (D) or dab1^scm/scm mutant mice (E, F). RGC axons from both eyes were labeled with the same fluorescently conjugated form of CTB (CTB[b]). Similar to the analysis at P14, two distinct phenotypes were observed in the pattern of IGL innervation by RGC axons in P2 dab1^scm/scm mutants (E, F, arrows). In both cases, misrouted axons were observed exiting mutant LGN and entering non-retino-recipient thalamic regions (E′, F′). D′, E′, and F′ show high-magnification images of areas depicted by the asterisks in D–F, respectively. The arrows indicate IGL. d, dLGN; v, vLGN. Scale bar, 150 μm.

Figure 5.

LGN subnuclei are present, remain distinct, and contain appropriate neural populations in the absence of functional reelin. IHC and ISH were performed on coronal sections of P14–P21 reln^rl/rl mutant or littermate control brains with subnuclei- or neuron-specific markers. A, B, NeuN and NPY IHC in P14 reln^rl/rl mutant or control LGN. NPY remained properly restricted to the IGL in the absence of reelin. C, D, GFAP IHC demonstrated the presence and confinement of astrocytes in P21 control and mutant IGL. E, F, IHC revealed a normal distribution of Gad65 in P21 reln^rl/rl mutant vLGN and IGL. G, H, ISH revealed a confinement of syt2-expressing neurons in vLGN of P21 mutants and controls. I, J, SMI32 immunoreactivity was appropriately enriched in vLGN and dLGN, but absent from IGL in P21 mutants and controls. K, L, Adamts15 mRNA, assessed by ISH, was correctly localized in P14 mutant and control dLGN. Sections were costained with NPY antibodies to demonstrate the location of the IGL. Note that NPY-positive cells did not invade dLGN in the absence of functional reelin. In C–J, retinal axons were anterogradely labeled with CTB (CTB[b]). In each image, the IGL is encircled by white dots. d, dLGN; v, vLGN; v_r, retino-recipient portion of vLGN; v_nr, non-retino-recipient portion of vLGN. The asterisks denote corticothalamic axon tracts. Single-color panels for each merged image can be found in supplemental Figure 6 (available at www.jneurosci.org as supplemental material). Scale bar, 400 μm.

Figure 6.

Reelin is not required for the generation or development of non-image-forming RGCs, nor their projections to non-LGN retino-recipient nuclei. A, B, Non-image-forming ipRGCs were labeled by Meln IHC in P12 control and reln^rl/rl mutant retinal whole mounts. The distribution and morphology of ipRGCs appeared similar in controls and mutants. A′, B′, Dendritic stratification of melanopsin-immunolabeled ipRGCs (green) in OFF- and ON-sublaminae of the IPL (arrows) appeared similar in cross-sections of P12 reln^rl/rl mutant and littermate control retinas. Moreover, mutant and control ipRGCs (green) demonstrated equal capacities to endocytose CTB (red). C, D, Retinal projection to other retino-recipient nuclei innervated by ipRGC appear normal in the absence of functional reelin. RGC projections to the OPN (C, D) and SCN (E, F) were assessed by injections of different fluorescently conjugated CTB into each eye of P12 controls and mutants. The patterns of innervation of these nuclei were indistinguishable in mutants or controls. G, H, PLRs remained present in reln^rl/rl mutants. P14 mutant and control mice (n = 3 each) were dark adapted for >1 h and then exposed to a 30 s 1.7 mW/cm² light. Pupil constriction was captured on video, and images were used to measure pupil size (G, white circles) before the onset of light and after 30 s of light. In H, the percentage pupil constriction after light exposure was calculated by comparing to pupil size during dark adaptation. Scale bars: (in A) A, B, 100 μm; (in A′) A′, B′, 25 μm; (in C) C–F, 200 μm.

Figure 7.

Non-image-forming ipRGCs require reelin and Dab1 for retinogeniculate targeting. A–E, IpRGC projections in P11 controls (A), reln^rl/rl;opn4-tau-LacZ^+/+ mutants (B), and dab1^scm/scm;opn4-tau-LacZ^+/− mutants (D). M1 ipRGCs were labeled by LacZ IHC (green). All RGC axons were labeled by binocular injection of CTB (CTB[b]; red). The arrowheads in B–E indicate mistargeted RGC axons containing LacZ. The arrow in B′ and D′ highlight misrouted LacZ-containing M1 ipRGC axons invading dLGN in mutants (compare with arrow in A′). C–C″ and E–E″ are high-magnification images of mistargeted axons depicted by asterisks in B and D. Note that LacZ immunoreactivity is reduced in dab1^scm/scm;opn4-tau-LacZ^+/− mutants (D) compared with images in A and B, since opn4-tau-LacZ heterozygotes were analyzed. F–H, By P21 mistargeted ipRGC axons have extensive arbors within dLGN in reln^rl/rl;opn4-tau-LacZ and dab1^scm/scm;opn4-tau-LacZ mutants, but not controls (compare arrows in F–H). Scale bars: (in A) A, B, D, F–H, 300 μm; (in C) C, E, 25 μm.

Figure 8.

Reelin is dispensable for dLGN targeting by axons derived from image-forming RGCs. A, B, VGLUT2-immunolabeled dLGN-projecting retinal terminals appear similar in P12 controls (A) and reln^rl/rl mutants (B). C, D, Calretinin (Calr)-immunolabeled RGC axons appeared similarly confined to dLGN in P12 control (C) and reln^rl/rl mutant (D) LGN. The arrows indicate abrupt border of Calr-containing arbors at the dLGN–IGL border. E, F, Misrouted CTB-labeled RGC axons in adjacent non-retino-recipient thalamic nuclei do not contain Calr in P14 reln^rl/rl mutants (see arrowheads). Scale bars: (in A) A–D, 300 μm; (in E) E, F, 15 μm.

Figure 9.

Reelin is dispensable for dLGN targeting by corticothalamic axons. A, Schematic demonstrating layer VI cortical axons pass through the internal capsule (IC), bypass vLGN/IGL, and selectively innervate dLGN. B, C, Layer VI cortical neurons were labeled by GFP IHC in P12 reln^+/+;golli-tau-GFP controls (B) or reln^rl/rl;golli-tau-GFP mutants (C). Normal cortical layering and the location of white matter tracts (WM) are depicted in B. Layer VI cortical neurons and their projections in transgenic mice were labeled by GFP IHC. D, E, Although corticothalamic axons have an altered course in reln^rl/rl;golli-tau-GFP mutants (see asterisks in E), they selectively target dLGN as in reln^+/+;golli-tau-GFP controls. d, dLGN; v, vLGN; i, IGL. Scale bar, 300 μm.