Temporal MG cell morphology and gene expression.(a) Diagrammatic representation of the retina within the eye showing the positioning of MG cells. (b) Tg(TP1:Venus) transplanted MG cells showing the time course of MG cell differentiation that gives rise to the distinct MG compartments (OLM: outer limiting membrane; OPL: outer plexiform layer; INL: inner nuclear layer; IPL: inner plexiform layer, ILM: inner limiting membrane). (c) Heatmap (relative expression values by sample—CPM) of top 100 significantly expressed genes in MG (GFAP‐GFP cells) compared to control (GFP negative cells) retinal cells (known glial genes are green and * indicates previous reported expression in MG) (see Supporting Information Table S1 for normalized enrichments). (d) Hierarchical clustering of samples used for RNA‐seq demonstrating consistency between the three replicates used for each time point (MG—GFAP‐GFP sorted cells, C—GFP negative control tissue) (e) Representative gene ontology proportions of MG genes enriched at 48, 60, 72, 96,120, and 196hpf

Temporal gene expression dictates MG cell morphologies. (a) Heatmap to show the relative gene expression for genes tested. (b) Summary of phenotypes observed for genes enriched across windows of MG cell differentiation. Red—phenotype, blue—no‐phenotype. (c) slc24a5 CRISPR injected control animals have normal MG cell morphology that extends from the apical to the basal surfaces, forming the ILM (inner limiting membrane) and OLM (outer limiting membrane) on either side. MG cells are also regularly tilled across in the eye with their cell bodies mostly restricted to the middle of the INL (inner nuclear layer) and are highly branched within the IPL (inner plexiform layer) and OPL (outer plexiform layer). (d) nav1b CRISPR injected animals have defects in apico‐basal cell body position in the INL (inner nuclear layer), OLM (outer limiting membrane), OPL (outer plexiform layer), tiling, IPL (inner plexiform layer), and ILM (inner limiting membrane). (e) mapab1 CRISPR injected animals have defects in OLM, OPL, and IPL. (f) fat1b CRISPR injected animals have defects in OPL and IPL defects. (g) nphp1 CRISPR injected animals have defects in MG cell tiling. (h) Frequency (%) of phenotypes observed in each MG compartment in F0 CRISPR screen 1. Scale bars = 8μm

Discrete gene expression regulates MG cell compartment morphology. (a) Heatmap to show the relative gene expression for genes tested. These were all screen in F0 CRISPR injected mutants. (b) Summary of phenotypes observed for genes enriched across windows of MG differentiation. Red—phenotype, blue—no‐phenotype. (c) lamb4 mutants have defects in the apico‐basal distribution of MG cell bodies only. (d) dcaf8 mutants have defects in the OLM and OPL. (e) mmp28 CRISPR injected mutants have defects in the ILM only. (f) egr1 mutants have defects in the IPL and OPL. (g) slitkr2 mutants have defects in the OPL layer only. (h) icn2 mutants have tiling defects only. 1. Frequency (%) of phenotypes observed in each MG compartment in F0 CRISPR screen 2. Scale bars = 8 μm

A set of highly conserved genes that affect MG cell morphology. (a) Overlap of zebrafish MG enriched genes with previously reported MG transcriptomes from zebrafish, mouse, and fly (Macosko et al., 2015; Nelson et al., 2011; Qin, Barthel, & Raymond, 2009; Roesch et al., 2008; Sifuentes, Kim, Swaroop, & Raymond, 2016). MG—genes enriched in GFAP‐positive cells; C—Genes enriched in non‐GFP positive cells; * indicates significance (Bonferroni adjusted p‐value <0.001) by Fisher's exact test. (b) pax2a CRISPR injected animals have highly disorganized retinas with breaks in the OLM and ILM, abnormal tiling and apico‐basal distribution of the cell bodies, as well as much less branching in the IPL and OPL (see Supporting Information Table S4 for details). (c) Percentages of genes used in this study that either had or did not have a phenotype. * indicates significance by Fisher's exact test. (d) GO terms for the top 500 genes significantly (adjusted p < 0.05) up or down‐regulated pax2a mutants. (e) itga5 CRISPR injected animals have defects on the basal side of MG specifically in the ILM and IPL. (f) Itag6 CRISPR injected animals have defects on the apical side of the cell in the OLM and OPL. (g) F0 itb1a CRISPR injected animals have defects in cell body tiling and apico‐basal position, as well as in OLM and ILM. (h) Frequency (%) of phenotypes observed in each MG compartment in F0 CRISPR screen 3. Scale bars = 8μm

Phenotypes of gene mutants enriched over windows of MG cell differentiation. (a) slc45a5 controls have no observable MG phenotype. (b) f8 mutants have defects in cell body position, OLM, ILM, IPL, OPL, and tiling. (c) cdhr1 mutants have defects in cell body position, OLM, ILM, IPL, OPL, and tiling. (d) sptbn mutants have defects in OLM, IPL, OPL, and tiling. (e) mapa1b mutants have defects in OLM, IPL, and OPL. (f) xirp1 mutants have defects in OPL and tiling. (g) lamb2 mutants have defects in ILM, IPL, and tiling. (h) Cadm1b mutants have defects in ILM, IPL, OPL, and tiling. (i) sox10 mutants have defects in IPL and tiling. (j) nfat5 mutants have tiling defects. (k) snx19a mutants have tiling defects. (l) Percentages of individual phenotypes observed in all animals from this screen. Dashed lines represent levels of significance from Fisher's exact test after Boniforni multiple test correction (bottom = p < 0.05, top = p < 0.001). Scale bars = 8μm.

Phenotypes of gene mutants that are enriched at specific times of MG differentiation. (a) slc45a5 controls have no observable MG phenotype. (b) timp2bmutants have defects in cell body position. (c) prmt6 mutants have defects in cell body position and ILM. (d) vwde mutants have defects in cell body position, OLM, ILM, IPL, OPL, and tiling. (e) apcdd1l mutants have defects in IPL and OPL. (f) sypb mutants have defects in OLM, OPL, and tiling. G) Cux2b mutants have defects in ILM and IPL. (h) cx31.7 mutants have defects in tiling. (i) Mpp6b mutants have defects in tiling. (j) cacnb2a mutants have defects in tiling. (h) Percentages of individual phenotypes observed in all animals from this screen. Dashed lines represent levels of significance from Fisher's exact test after Boniforni multiple test correction (bottom = p < 0.05, top = p < 0.001). Scale bars = 8μm.

 Phenotypes of conserved highly conserved MG cell genes. (a) Schematic representation of how highly conserved genes we bioinformatically identified. (b) slc45a5 controls have no observable MG phenotype. (b) timp2b mutants have defects in cell body position. (c) nphs1 mutants have defects in ILM, IPL, and tiling. (d) kirrela mutants have defects in cell body position and tiling. (e) wt1 mutants have defects in cell body position, IPL, OPL, and tiling. (f) mmp2 mutants have defects in OLM, ILM, and tiling. (g) cadm4 mutants have defects in OPL, IPL, ILM, and tiling. (h) Cadm1a mutants have defects in IPL and tiling. (i) cadm2b mutants have defects in a cell body positing, IPL OPL and tiling. (j–l) Percentages of individual phenotypes observed in all animals from this screen. Dashed lines represent levels of significance from Fisher's exact test after Boniforni multiple test correction (bottom = p < 0.05, top = p < 0.001). Scale bars = 8μm.

CRISPR injection validation. (a) Cas9 only injected F0 fish have normal pigmentation at 120hpf while those injected with Cas9 and the slc45a5 guide RNAs are mostly devoid of pigment. (b) In control animals (GFAP:GFP) Pax2 is expressed in all MG by 120hpf. (c) F0 pax2a CRISPR injected animals lack Pax2 expression in most, but not all MG. (d) F1 pax2a CRISPR injected animals Pax2 is absent from all MG. (e) F1 CRISPR mutants with confirmed mutations have notably similar defects to those identified in F0 screen 3. (f) Percentages of animals with lethality and phenotypes after injections. Scale bars = 8μm.

ZFIN is incorporating published figure images and captions as part of an ongoing project. Figures from some publications have not yet been curated, or are not available for display because of copyright restrictions.