An Eye Organoid Approach Identifies Six3 Suppression of R-spondin 2 as a Critical Step in Mouse Neuroretina Differentiation

Abstract

Recent advances in self-organizing, 3-dimensional tissue cultures of embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) provided an in vitro model that recapitulates many aspects of the in vivo developmental steps. Using Rax-GFP-expressing ESCs, newly generated Six3^-/- iPSCs, and conditional null Six3^delta/f;Rax-Cre ESCs, we identified Six3 repression of R-spondin 2 (Rspo2) as a required step during optic vesicle morphogenesis and neuroretina differentiation. We validated these results in vivo by showing that transient ectopic expression of Rspo2 in the anterior neural plate of transgenic mouse embryos was sufficient to inhibit neuroretina differentiation. Additionally, using a chimeric eye organoid assay, we determined that Six3 null cells exert a non-cell-autonomous repressive effect during optic vesicle formation and neuroretina differentiation. Our results further validate the organoid culture system as a reliable and fast alternative to identify and evaluate genes involved in eye morphogenesis and neuroretina differentiation in vivo.

Keywords: R-spondins; Six3; Six3 conditional knockout ESCs; Wnt; eye; mouse; neuroretina; optic vesicles; organoids; pluripotent stem cells.

Figure 1. Rspo2 expression is upregulated in D7 eye organoids

(A, B) Transillumination images of differentiating D4 (A) and D7 (B) WT Rax-GFP ESCs aggregates using the SFEBq organoid culture protocol. (A) A thicker neuroectodermal layer surrounds the developing organoids at D4. (B) Rax-GFP green fluorescence is detected over the evaginating optic vesicles at D7. (C) Microarray analysis of relative gene expression of D4 vs D7 Rax-GFP ESCs. List represents the top 50 genes with increased expression from D4 to D7. Gene symbols highlighted in red are R-spondins and members of the Wnt signaling pathway. Genes are ordered top to bottom by relative expression (D7 vs D4). Fold increase ratio of D7/D4 is shown in the right of the heat map. The typical eye field markers (Rax and Six3) are included in the top 50 candidate genes. However, analysis of the microarray raw data showed that Rax expression was increased at D7 (log₂=1.51) and no significant change was listed for Six3. (D, E) RT-qPCR analysis of Wnt8b and Rspo2 expression in D4 and D7 Rax-GFP ESCs aggregates confirms the increase in their expression during optic vesicle morphogenesis. Error bars represent S.E.M from 3 independent experiments. (F, H) Whole mount in situ hybridization against Wnt8b and Rspo2 in D7 Rax-GFP ESC aggregates. Expression is localized but absent from the optic vesicles (yellow arrows). (G, I). Upon cryosection of those aggregates, Rax immunostating was performed confirming that Wnt8b and Rspo2 expression was localized outside the Rax⁺ optic vesicles. Black arrows indicate optic vesicle-like structures. Scale bar; 100 μm (A, B, G, I).

Figure 2. Derivation of Six3 conditional null ESCs

(A) Schematic representation of the protocol used to generate the different Six3 conditional mutant ESCs indicating the genotypes of the mouse strains used to isolate E3.5 embryos. WT, Six3 conditional heterozygous and Six3 conditional null ESCs were derived from those embryos. (B, C) RT-qPCR analysis of Cre and Six3 expression in the generated D7 Six3^+/f and Six3^delta/f;Rax-Cre ESC aggregates. Cre expression is detected only in Six3^delta/f;Rax-Cre ESC aggregates. Six3 expression was strongly reduced in Six3^delta/f;Rax-Cre ESC aggregates. Error bars represent S.E.M from 3 independent experiments. (D–G″) Transillumination images of D4 (D–D″), D5 (E–E″), D6 (F–F″) and D7 (G–G″) Six3CKO ESC aggregates. Neuroepithelium formation was clearly seen in all three genotypes at D4 (D–D″). Starting at D6, optic-vesicle formation was severely disrupted in Six3CKO mutant aggregates (F″), and this arrest was more evident at D7, where few smaller and often still associated optic-like vesicles were detected (G″). Arrows indicates optic-vesicle like structures. (H) Quantification of the number of optic vesicle-like structures per aggregates. (I) Quantification of optic vesicles area by imageJ analysis. Six3^+/f (aggregates=85); Six3^delta/f;Rax-Cre (aggregates=67).

Figure 3. Optic vesicle formation and NR differentiation are disrupted in Six3 conditional null ESC aggregates

(A–C) RT-qPCR analysis of Rax, Pax6 and Mitf expression in D7 Six3^+/f and Six3^delta/f;Rax-Cre ESC aggregates. Rax and Mitf expression were strongly decreased in Six3^delta/f;Rax-Cre ESC aggregates, while that of Pax6 was partially reduced. Error bars represent S.E.M from 3 independent experiments. (D–F″) Immunostaining of D7 Six3CKO ESC aggregates. Six3^+/f and Six3^+/f;Rax-Cre ESC aggregates show typical Rax, Pax6 and Mitf expression in the optic vesicles. Instead, expression of these three markers was strongly reduced in Six3^delta/f;Rax-Cre aggregates. Arrows indicate optic-vesicle like structures. Insets in D″ and F″ show single channel images of Rax and Mitf. (G–G″) Transillumination images of D10 ESC aggregates. Neuroretina formation (arrows) was evident in Six3^+/f and Six3^+/f;Rax-Cre aggregates, but not in Six3^delta/f;Rax-Cre ESC aggregates. Pigmentation (RPE) was also evident in Six3^+/f and Six3^+/f;Rax-Cre aggregates, but the pigmented region was smaller and mis-localized in Six3^+/f;Rax-Cre aggregates (arrowhead). (H) Quantification of the number of aggregates showing pigmentation (see also Figure S3). (I–J″) Immunostaining of D10 aggregates using Mitf and Vsx2 (Chx10) antibodies. Mitf expression is normally detected in the region behind the developing neuroretina (arrows) in control aggregates (I, I′); instead, in Six3^delta/f;Rax-Cre ESC aggregates the RPE-like structure is seen in the reduced size and morphologically abnormal optic vesicle-like structures (I′, arrowheads). Vsx2-expressing neuroretina-like structures are seen in Six3^+/f and Six3^+/f;Rax-Cre ESC aggregates (J, J′, arrows). In contrast, Vsx2 expression was absent in Six3^delta/f;Rax-Cre aggregates (J″, arrowhead). (K–L″) D14 aggregates were immunostained against the retina marker Vsx2 and the ganglion cell marker Brn-3b. Even at this late stage, Vsx2 expression was not detected.(K″, arrowheads) in null aggregates, and no obvious expression of Brn-3b was observed (L″, arrowhead) in Six3^delta/f;Rax-Cre ESC aggregates. DAPI signal in the vesicle structures (arrowheads) are shown in white insets (K″ and L″).

Figure 4. R-spondin 2 negatively regulates neuroretina differentiation in vitro

(A) Rax-GFP fluorescence is seen in in the putative eye field region of D5 Rax-GFP ESC aggregates. (B) Diagram of the protocol used to evaluate the role of Rspo2 during neuroretina differentiation in cultured organoids. 250 ng/mL of commercially available Rspo2 recombinant protein was added to the culture at D5. (C–H′) Expression of NR markers (Rax-GFP, Six3, Vsx2 and Sox2) and RPE markers (Pax6 and Mitf) in D10 Rax-GFP control aggregates. Six3, Vsx2 and Sox2 expression levels were reduced by the addition of recombinant Rspo2 protein. Instead, no obvious changes were seen in the expression of Pax6 and Mitf. (I, J) RT-qPCR analysis of D10 tissues with or without Rspo2 recombinant protein. Axin2 was increased in Rspo2 treated conditions; in contrast, addition of Rspo2 decreased expression of Vsx2. Arrows indicate the neuroretina-like territory.

Figure 5. Ectopic expression of R-spondin 2 inhibits NR differentiation in transgenic embryos

(A–B′) Anterior expansion of Rspo2 expression in the dorsal diencephalic region was observed in Six3 conditional null embryos at E9.0 and E9.5 (arrowhead). (C, C′) Fezf2-YFP–driven ectopic expression of Rspo2 was analyzed in E10.5 transgenic mouse embryos. Different levels of ectopic expression of YFP (Rspo2) were detected in some of the analyzed transgenic embryos in the forebrain territory. (D–I′) Expression of the neuroretina markers (Rax, Six3, Vsx2 and Sox2) and RPE markers (Pax6 and Mitf) was analyzed in E10.5 control embryos. In Fezf2-Rspo2-YFP transgenic embryos, neuroretina differentiation was inhibited, as indicated by the absence (or decreased expression) of those markers. However, expression of the RPE markers Pax6 and Mitf appeared normal.

Figure 6. Chimera aggregates using Six3^−/− iPSCs exhibit neuroretina differentiation defects

(A) E13.5 fibroblasts from Six3^LacZ/+ and Six3^LacZ/LacZ embryos were transfected with Dox-induced lentiviral particles expressing the 4 Yamanaka factors to generate iPSCs. FB, forebrain. (B, C) X-gal stained D7 Six3^LacZ/+ iPSCs exhibit ß-gal expression in the evaginating optic vesicles. Weak ß-gal staining was also seen around the optic vesicles. ß-gal expression was stronger in D10 aggregates in the neuroretina-like structures surrounded by pigmented tissues. (D–E′) Immunostaining of D7 and D10 Six3^LacZ/+ and Six3^LacZ/LacZ aggregates against Six3 shows expression in the optic vesicles or neuroretina-ike structures (arrows) in Six3^LacZ/+ but not in Six3^LacZ/LacZ aggregates. (F) Scheme of the chimera assay using Rax-GFP and Six3^LacZ/+ or Six3^LacZ/LacZ cells in a 2:1 ratio. (G–I) RT-qPCR analysis of chimera aggregates. Six3^LacZ/LacZ had the highest expression for Wnt8b, Rspo2 and Axin2. (J–J″) Rax-GFP signal in D10 chimera aggregates is used as control. Arrowheads indicate Rax-GFP⁺ vesicle structures. Control and Six3^LacZ/+ chimera aggregates showed strong Rax-GFP⁺ signals and neuroretina-like morphology. Instead, Six3^LacZ/LacZ chimera aggregates had weak GFP signal and small vesicle-like structures. (K–M″) Immunostaining of D10 control and chimera aggregates using antibodies against GFP, Mitf and Six3. Expression of Rax-GFP was clearly seen in control aggregates. Its expression is also detected in Six3^LacZ/+ chimera aggregates but in a mosaic pattern over the neuroretina-like structures, suggesting that Six3^LacZ/+ cells were incorporated into these neuroretina structures. Conversely, Six3^LacZ/LacZ chimera aggregates showed weak Rax-GFP and Six3 signals (K″, L″, M″). Mitf expression level was similar to that of control and Six3^LacZ/+ chimera aggregates (M″). Arrows and open arrowheads indicate neuroretina-like structures (K, K′, L, L′, M, M′) and vesicle like structures (K″, L″, M″), respectively.

Figure 7. Schematic representation of Six3 functional role during eye development

(A) Diagram showing the regional expression pattern of Six3 and Rspo2 in eye organoids and E9.5 mouse heads. In the head, Rspo2 expression is normally outside the eye territory. The 3D SFEBq eye organoid culture system faithfully recapitulates the in vivo process of eye morphogenesis and its molecular expression profile. (B–C) Six3 is a key player in the process regulating neuroretina differentiation. This process is driven by Six3-mediated repression of Wnt8b and Rspo2 such that their expression does not expand anteriorly into the prospective eye forming territory. In Six3 conditional mutants, lack of Six3 activity allows the abnormal anterior ectopic expansion of Wnt8b and Rspo2 expression. As a consequence, optic vesicles fail to acquire a Vsx2⁺ neuroretina fate but they maintain Mitf expression. Using a chimeric aggregate assay that combines WT ESCs with Six3 null cells, we determined that Six3 repressive effects act in a non-cell autonomous manner. In its absence, it is likely that the activity and/or range of different Wnt signaling family members (i.e Wnt8b, Rspo2) get enhanced such that they repress normal optic vesicle morphogenesis and neuroretina differentiation in WT Rax-GFP ESCs. (D) Six3 null cells potentially exert inhibiting effects on WT neuroretina differentiation by secreting Wnt8b and Rspo2. This effect is likely non-cell autonomous, resulting in the arrest of neuroretina differentiation. WT, wild-type; KO, knockout; OV, optic vesicle; Tel, telencephalon; Di, diencephalon; MB, midbrain; HB, hindbrain.