The five human retinal specimens used in this study were obtained from voluntarily donated aborted fetuses, sourced from the Senior Department of Ophthalmology at the Third Medical Center of the Chinese PLA General Hospital. The Ethics Committee of the Third Medical Center of the Chinese PLA General Hospital approved this study (ID: KY 2021-021), and written informed consent was obtained from all participants. The procedures in this study adhered to the Helsinki Declaration of 1964 and its amendments, ensuring ethical integrity (World Medical, 2013).

The hESC line H9 were routinely cultured in Essential 8 medium (ThermoFisher, A1517001) on plates coated with Vitronectin (Gibco, A14700). For passaging, cells were treated with Accutase (Stemcell Tech, 07920). RO differentiation followed established protocols with minor modifications (Wang et al., 2021; Kuwahara et al., 2015). Aggregates were cultured under 40% O₂/5% CO₂ conditions (30 aggregates per 10-cm dish) from day 24 (D24), using an NR-differentiation medium comprising DMEM/F12 (Gibco, 10565018), KSR (Gibco, 10828028), N2 supplement (Gibco, A1370701), 0.1 mM taurine (Sigma, T0625), and 0.5 μM retinoic acid (Sigma, R2625). Under these conditions, RO continued to grow for several weeks.

Single guide RNAs (sgRNA) constructs targeting critical RAX2 were cloned into px459 plasmids (Addgene, 62988) for knockout cell generation. HESC were transfected with these sgRNA plasmids using the Lipofectamine Stem Transfection Reagent (Invitrogen, STEM00001) and exposed to 0.5 μg/mL puromycin for 48 h 2,000–3,000 surviving cells were plated on a 6 cm dish, and 96 single colonies were picked up to a 96-well plate. Genomic DNA was extracted for PCR using specific primers:

Fw: CTTAGGGCGTGAGAAGGGAT;

Rv: CCCCACGCCCAATTAACAGA.

The PCR products were validated by TA cloning and Sanger sequencing to confirm RAX2 gene deletions.

Our earlier investigation used an in vitro self-organization model of human RO derived from hESC, which mimicked human retinal development, to conduct an scRNA-seq analysis at five different time points during RO differentiation (D36, D66, D96, D126, and D186) (Wang et al., 2021). In this study, to delineate the role of the RAX2 in retinogenesis, we reanalyzed the scRNA-seq data. Canonical markers were used to distinguish 6 cell clusters: RPC, Proliferating-RPC, PC, RGC, MCs and RPE cells (Supplementary Figures S1A, S1B). RAX2 was primarily detected in the PC population (Figure 1A). A gradual increase in RAX2 expression correlating with PC emergence in RO was observed (Figures 1B–D), consistent with the immunofluorescence (IF) staining of human RO, which also revealed a progressive increase in RAX2-positive cells (Figure 1E; Supplementary Figure S2A). Moreover, RAX2 expression patterns aligned with canonical PC markers, including CRX, NR2E3 and NRL (Figures 1A,B). CRX-positive cells appeared at D36 in a human RO culture and gradually increased over time. The expression of NRL and NR2E3 significantly increased during the maturation of PC (Wang et al., 2021), and OTX2 was found to be involved in embryonic PC fate determination (Muranishi et al., 2011). To enhance our comprehension of RAX2 dynamics in retinal development, we obtained human retinal tissue from voluntarily donated aborted fetuses aged 12–24 weeks of gestation, and performed multi-immunofluorescence (multi-IF) staining to precisely track the temporal expression patterns of RAX2 (Figure 1F; Supplementary Figure S2B). A notable increase in RAX2-positive cells was observed from 20 to 24 weeks, coinciding with the reported initiation of PC development (Hu et al., 2019). These findings indicated that RAX2 may regulate PC maturation during retinal development.

To explore the influence of RAX2 on human retinal development, the CRISPR/Cas9 system was used to disrupt critical exons of RAX2 in hESC (H9 cell line). Seven sgRNAs were created to target different regions around the gene, and their effectiveness was evaluated using a surveyor assay (Supplementary Figure S3). Cas9/sgRNA-7 and Cas9/sgRNA-6, both of which exhibited notable cleavage efficiencies, were selected for subsequent gene editing (Figure 2A). Two homozygous mutants, RAX2 ^−/−-1 and RAX2 ^−/−-2, were successfully generated and validated through Sanger sequencing (Figure 2B). Evaluation of genomic copy number variation (CNV) (Supplementary Figure S4), ESC colony morphology (Figure 2C), pluripotency markers expression (Figure 2D), and cell proliferation (Figure 2E) showed no significant differences between RAX2 ^−/− and wild type (WT) hESC. Embryoid body (EB) formation assay (Figures 2F,G) revealed that RAX2 deficiency in hESC significantly reduced the expression of ectoderm markers in the derived EBs, including MAP2, PAX6, RAX, and SIX6 (Figure 2H). This finding underscored the essential function of RAX2 in the ectoderm-related differentiation process.

Using a previously established BMP4-induced RO self-organization protocol (Wang et al., 2021), WT and RAX2 ^−/− hESC were grown in a 3D culture for 66 days (Figure 3A), and twenty-four RO were harvested for scRNA-seq analysis from each of WT and RAX2^−/− group. Following a rigorous quality control evaluation and removal of doublets, a UMAP analysis revealed five primary cell clusters, with the cell types identified through enriched gene profiles and canonical markers (Figures 3B,C). A marked reduction in RAX2 expression was detected in all the RAX2 ^−/− hESC-derived RO cell clusters identified (Figure 3D). A significant decrease in the percentage of PC, RPC, and RPE cell populations was observed in RO derived from RAX2 ^−/− hESC compared to the those from WT hESC (Figure 3E). Considering the process of retinal development, we focused on RPC, RGC, and PC clusters (Figures 3F,G). A developmental pseudotime trajectory analysis was conducted, which helped reveal highly interconnected nodes potentially indicating the differentiation status (Figures 3H,I). Depletion of RAX2 significantly altered various cellular distributions. Differentiation into PC was notably affected by the absence of RAX2, leading to a bias towards RGC lineage commitment. Additionally, the PC population analysis revealed a decrease in pathways associated with PC differentiation (Figure 3J). RT-qPCR analysis demonstrated decreased expression of PC-specific markers (CRX, NRL, and NR2E3) in RO derived from RAX2 ^−/− hESC, alongside elevated levels of RGC markers (POU4F2 and THY1), consistent with the observed lineage bias (Supplementary Figure S5). These findings underscored the critical function of RAX2 in PC fate determination.

In our previous study, we observed that the proportion of each cell type, including PC, in RAX2 ^−/− hESC-derived RO differed from that of the WT RO (Figure 3E), suggesting that RAX2 influenced the differentiation state of the entire organoid. To elucidate the underlying mechanism, we analyzed differentially expressed genes and noticed that the expression patterns of PAX6 (Oron-Karni et al., 2008) and SOX2 (Diacou et al., 2022), both vital for eye development, were significantly altered by RAX2 deficiency (Figure 4A). RT-qPCR and Western blot analyses confirmed the reduced expression of PAX6 and SOX2 (Figures 4B,C). In addition, IF staining analysis of human RO at D66 revealed a marked decrease in the fluorescence intensity of PAX6 and SOX2 in RAX2 ^−/− hESC-derived RO (Figure 4D). Overall, our results suggested that RAX2 is critical for retinal development by modulating PAX6 and SOX2 expression.