Retinal Pigment Epithelium and Neural Retinal Progenitors Interact via Semaphorin 6D to Facilitate Optic Cup Morphogenesis

sema6d mutants do not display gross morphologic defects. A, Schematic of the two sema6d mutant alleles; sema6d^ca302 mutants contain a five-base pair deletion and two-base pair mismatch in exon6, and sema6d^ca303 mutants contain a 16-base pair insertion in exon10, both predicted to generate proteins that truncate prematurely within the SEMA domain and result in the loss of the transmembrane domain (TMD) and the intracellular domain (ICD). B–E, Lateral views of whole-mount embryos processed for bhlhe40 ISH, revealing normal gross embryonic morphology and somite labeling in both mutant alleles with respect to their wild-type siblings at 24 hpf. F, G, Eye area measured from lateral images of sema6d mutants and their respective wild-type siblings at 24 hpf [unpaired t test, df = 26 (F), df = 21 (G); error bars are SD]. Scale bar: 200 μm.

sema6d mutants display normal eye vesicle development and patterning before optic cup morphogenesis. Whole-mount RNA ISH for patterning genes indicates early development occurs normally in sema6d mutants (B, D, F, H, J) relative to wild-type siblings (A, C, E, G, I). In the bottom right of panels are the number of embryos of the total analyzed that exhibited the normal wild-type expression pattern. A, B, fgf8a expression in 10-ss embryos viewed laterally. C–H, Dorsal whole-mount images of the eye vesicles (dashed outlines) and neural keel of 12 ss processed by whole-mount ISH for the expression of early patterning genes. C, D, Early dorsal (future nasal) tissue with foxg1a. E, F, Early ventral (future temporal) tissue with foxd1. G, H, Early anterior (future ventral) tissue with vax2. I, J, RPE progenitors and neural crest cells express tfec at the appropriate time in wild-type siblings and sema6d mutants. K, The average anteroposterior length of the eye vesicles. L, Ratio (%) of RPE to eye vesicle anteroposterior lengths [bars a:b in J; p values are unpaired t test, df = 22 (K), df = 21 (L), error bars are SD]. Scale bars: 100 μm. A: anterior, C: caudal, ev: eye vesicle, h: hypothalamus, mhb: mid-hindbrain boundary, N: nasal, nc: neural crest, nk: neural keel, P: posterior, R: rostral, som: somites, T: temporal, tb: tailbud.

sema6d mutants display temporal neural retina defects post optic cup morphogenesis. A, B, D, E, Lateral images of 24-hpf optic cups processed for RNA ISH in wild-type (A, D) and sema6d mutant (B, E) embryos show similar expression of nasal (foxg1a) and dorsal (tbx5) markers. Of note, these domains derive from the dorsal and posterior eye vesicle, respectively. In the bottom right of panels are the number of embryos of the total analyzed that exhibited either a WT (A, B, D, E, G, J) or a disrupted (I,K,L) expression pattern. C, F, The angle formed by the lateral edges of the foxg1a (C; p values are unpaired t test, df = 22, error bars are SD) and tbx5 (F; p values are unpaired t test, df = 30 ex6+/+ vs ca302, df = 24 ex6+/+ vs ca303, error bars are SD) domains to the center of the lens (θ). G–L, foxd1 and vax2 RNA ISH viewed in whole mount. The shape and size of temporal eye (early ventral) markers, vax2 and foxd1, are disrupted in the sema6d³⁰² (H, K) and sema6d³⁰³ (I, L) mutants as compared with wild-type siblings (G, J). Black arrowheads point to aberrant vax2 and foxd1 label in the inner vesicle leaflet, seen through the depth of the eye in the transparent zebrafish embryo. M, N, P, Q, Transverse plastic sections (axis shown by red line in G) through the central retina of 24-hpf wild-type sibling and sema6d mutant eyes processed for foxd1 and vax2 RNA ISH. Note in mutants a smaller (compare t in J–L and length of bars in M, P vs N, Q) foxd1/vax2 domain in the temporal neural retina, and an open ventricle as compared to wild-type (white arrow in N). O, Blinded quantitation of the ratio of the width of vax2+ domain in temporal versus nasal optic cup (t and n in J–L), which captures the redistribution of vax2 expression in sema6d mutants (p values are one-way ANOVA, Dunnett’s multiple comparisons test, error bars are SD, df = 53). Scale bars: 50 μm (A, M). D: dorsal, il: inner leaflet, L: lens, N: nasal, ol: outer leaflet, T: temporal, V: ventral.

Disrupted RPE morphogenesis in sema6d mutants. A–C, Morphogenesis of the RPE is disrupted in sema6d^ca302 and sema6d^ca303 eyes as compared with wild-type, with mutants displaying ectopic bhlhe40 staining apparent through the transparent lens (white arrowhead in B, C). In the bottom right of panels are the number of embryos of the total analyzed that exhibited either a WT (A) or a disrupted (B, C) expression pattern. A’–C’, F, Transverse sections reveal full expansion of the bhlhe40+ RPE to abut the nasal lens (black arrows), while the RPE of the temporal eye fails to reach the lens in mutants (compare red arrows). RPE cells are expanded in the inner leaflet of wild-type embryos, but cuboidal in mutants (compare red asterisks). In most mutants the bhlhe40+ RPE not only failed to abut the lens, but bhlhe40 was ectopically expressed in the lateral portion of the temporal neural retina (white arrowheads in B’, C’), while in some sema6d mutants, bhlhe40 expression was restricted entirely to the inner eye vesicle leaflet (red arrow in F). D, E, Wild-type (D) and sema6d^ca302 mutant (E) embryos bred on a Tg(tfec:EGFP) background to label RPE progenitors. Mutants display ectopic EGFP expression in the temporal neural retina (arrowheads in E) and EGFP+ RPE cells in the temporal (early ventral) inner leaflet are cuboidal and not extended as in wild-type (compare arrows in D, E). G, H, Transverse sections of 22-hpf eyes of wild-type (G–G’’) and sema6d^ca302 (H–H’’) mutant embryos on a Tg(tfec:EGFP) background that were processed for GFP immunohistochemistry (G, H) and RNA ISH for the temporal neural retina marker foxd1 (G’, H’). Blended images (G’’, H’’). EGFP+ cells that co-express foxd1 are present in the distal temporal neural retina of both wild-type and mutant eyes (arrowheads in G, G’’, H, H’’). A few foxd1+ cells in the wild-type inner leaflet (asterisk in G’) have not yet moved around the distal rim, with many more present in the mutant inner leaflet (asterisk in H’). Scale bars: 50 μm (A, A’). D: dorsal, il: inner leaflet, L: lens, N: nasal, ol: outer leaflet, T: temporal, V: ventral.

Temporal neural retina is disorganized in sema6d mutants. A–F, Immunohistochemistry for the apical marker aPKC (A, B) of wild-type and sema6d mutant (C, D) Tg(rx3:GFP) embryos. Magnified view of merge of labels in temporal eye (boxed area in C; E, F). E’–F’, Schematic of E, F showing ventricle as marked by aPKC immunoreactivity (red), labeled GFP (blue) cells, and their orientation (black lines). In sema6d mutants, there is a failure of the temporal ventricle to seal (arrowheads in B) and disorganization of temporal neural retina cells (D, F, F’). G, H, Distributions of the angles made by the long axis of DAPI-labeled nuclei with the basal temporal retinal neuroepithelium at 24 hpf. Average of the distributions in mutants (I, n = 12 embryos) and wild-type sibling (H, n = 13 embryos) are significantly different (p < 0.0001, χ² contingency test). Circles indicate the average numbers of nuclei found in each 15° bin. Scale bars: 50 μm (A) and 20 μm (E). N: nasal, T: temporal.

Ventral inner leaflet cells fail to move appropriately around the distal rim during optic cup morphogenesis. A, B, Confocal optical sections, and their corresponding schematics, acquired from the nasal/temporal plane between the 14 ss and 23 hpf of wild-type and sema6d mutant Tg(rx3:GFP) eyes (labels all eye progenitors). Dotted yellow line(s) indicates the separation between the inner and outer leaflets. C, D, Higher magnification of inset boxes (early ventral/future temporal retina) in A, B, respectively. In wild type (A), ventral progenitors move around the distal rim of the optic cup and come to reside in the temporal neural retina at 23 hpf. In contrast, sema6d mutants display temporal defects (follow arrow in B). While a ventral inner leaflet cells in wild type (C) moves around the distal rim (pink asterisk), a corresponding cell in the sema6d mutant (D) begins to move toward the distal rim, but stalls and, together with other progenitors, protrudes from the inner leaflet (blue asterisk). E, F, Mean of the speed of individual cells in each wild-type and sema6d mutant embryo tracked over the period of optic cup morphogenesis. Tracked were ventral neural progenitors located at the distal eye vesicle tip in the Tg(rx3:GFP) background (E) and brightly EGFP labeled presumptive RPE cells in the Tg(tfec:EGFP) background (F). In both reporter lines, sema6d mutant cells move more slowly than their wild-type counterparts, and stall at ∼18–20 hpf. ns are number of embryos assayed (2–7 cells/embryo), and error bars are SEM. Mean speeds at individual time points were compared statistically between wild-type and heterozygous embryos by a Mann–Whitney U test (p < 0.05; E, F). A: anterior, il: inner leaflet, L: lens, N: nasal, nr: neural retina, ol: outer leaflet, P: posterior, pRPE: presumptive RPE, RPE: retinal pigment epithelium, T: temporal, ve: ventricle.

Plxna1a loss-of-function recapitulates the temporal eye defects observed in sema6d mutants. A, B, RT-PCR confirming morpholino mis-splicing (outlined in schematic) of plxna1a transcript (A), and knock down of plxna1a transcript in CRISPRi-injected embryos (B). β-Actin mRNA as loading control. C, D, Normal expansion of the bhlhe40-expressing RPE progenitor domain at the 14 ss. Mean anteroposterior eye vesicle length (C) and percent RPE expansion (D; RPE bhlhe40+ domain length/anteroposterior eye vesicle length) are not significantly different in plxna1a morphants as compared with controls [N = 2, control n = 17, plxna1 MO n = 17; p values are unpaired t tests, df = 33 (C), df = 18 (D), error bars are SD]. E, F, Tg(vsx1:GFP) expression in the nasal retina. G, H, vax2 mRNA in lateral views of a control (G) and a plxna1a morphant (H) 24-hpf eye. I, J, Transverse sections of whole-mount foxd1+ RNA ISH of 24-hpf eyes. Early ventral (future temporal) foxd1+ tissue undergoes rim movement into the outer leaflet in control (I), but in a plxna1a morphant remains partially in the inner leaflet (J; red asterisk and compare bars). Also evident is an open ventricle (arrow in J) in the morphant. K, The average angle formed by the lateral edges of the vsx1 (J) domain to the center of the lens (θ) is similar between controls and plxna1a morphants (N = 2; control n = 16, plxna1a MO n = 19, p value is unpaired t test, df = 33). L, Ratio of the width of the temporal to nasal (t and n in G) vax2 whole-mount RNA ISH domain measured in images of lateral eyes (unpaired t test, p < 0.0001, control n = 13, plxna1a MO n = 23, error bars are SD, df = 34). M–P, Transverse eye sections of whole-mount RNA ISH for bhlhe40+ performed on 24-hpf control embryos (M, O) or embryos injected at the one-cell stage with either an antisense plxna1a morpholino (N) or a sgRNA against exon5 of plxna1a along with dead-cas9 mRNA (CRISPRi; P). RPE bhlhe40+ progenitors elongate in control embryos (arrows in M, O) to line the back of the eye, and abut the lens, while the bhlhe40 signal is expressed ectopically in the morphant and CRISPRi-injected embryos (arrowheads N, P), and RPE cells retain a cuboidal shape (arrows N, P). Embryos in I, J, M, N were treated with 1-phenyl 2-thiourea to inhibit pigmentation of the RPE. Scale bars: 75 μm (E–H) and 50 μm (I, J, M–P). A: anterior, D: dorsal, L: lens, N: nasal, P: posterior, T: temporal, V: ventral, ve: ventricle.