A homozygous SIX6 mutation is associated with optic disc anomalies and macular atrophy and reduces retinal ganglion cell differentiation

To the Editor: We report on a consanguineous family with three children who had optic disc anomalies and macular atrophy without microphthalmia or cataracts (Table 1 and Fig. 1a,b). Clinical examinations were otherwise unremarkable. Levels of thyroid hormones, gonadotropins, testosterone, estradiol, liver enzymes, kidney functions, glucose and electrolytes, CBC, and urinalysis were within normal limits. Physical and ophthalmological examinations were normal in all unaffected siblings and in parents. Genotyping eight family members via Affymetrix 5.0 SNP-chips showed an autozygous region at chr14: 59247009-75949378 (Hg19) (Fig. 1b). Two genes in this interval, CHX10 (MIM142993) and SIX6 (MIM606326), were previously reported to cause microphthalmia or anophthalmia in humans (1). Sanger sequencing of both genes showed that all three affected individuals were homozygous for a novel variant, c.110T>C (p.L37P) (NM_007374.2, NP_031400.2), in SIX6 (Fig. 1c). This variant co-segregated with the phenotype in the entire family as a recessive trait and was negative in 342 ethnicity-matched controls. It is also absent from the Exome Variant Server and dbSNP137. Leucine at position 37 is completely conserved in different species (Fig. S1, Supporting Information). None of the loci containing a gene known to cause recessive macular disease, including PRPH2 (MIM179605), ABCA4 (MIM601691), and CNGB3 (MIM605080) co-segregated with the phenotype. Two other autozygous regions >1 MB did not contain a gene known to cause an eye disease (Table S1, Supporting Information). We built and compared structural models of SIX6WT and SIX6L37P, using the crystal structure of SIX1 as a template (PDBID: 4EGC). SIX6 belongs to the SIX family of transcription factors with the TA-DB modular architecture, where TA is the N-terminal transactivation domain and DB is the C-terminal DNA-binding homeodomain (2). The transactivation of SIX family proteins depends upon their TA domain binding other nuclear proteins such as EYA2 and DACH2 (3). Our models suggest that the p.L37P mutation is located within the N-terminal α-hairpin, comprised the first two α-helices of the TA domain of SIX6 (Fig. 1d). In SIX6WT, this α-hairpin is maintained with a van der Waals contact between V14 and L37 at its open end. Introduction of the p.L37P mutation would lead to unfolding of the N-terminal α-helix within this α-hairpin by about one turn due to the inability of proline to participate in intramolecular backbone hydrogen bonding (Fig. 1d). This together with the limited conformational flexibility of proline would result in the loss of an intramolecular van der Waals contact with V14, thereby compromising the integrity of the α-hairpin that would potentially drive the assembly of SIX6 with its ligands. We further investigated for potential molecular and cellular mechanisms through which the p.L37P mutation could induce ocular phenotypes, using embryonic day 14 (E14) mouse retinal progenitor cells (RPCs) as a model. We first asked whether the p.L37P mutation affects nuclear localization by expressing SIX6WT and SIX6L37P fused to GFP tags. Both the normal and mutant SIX6 localized in the nucleus (Fig. 1e), suggesting that the mutation does not disrupt protein localization. The comparable levels of GFP-fusion protein expression and transduced cells across conditions also suggested that the SIX6L37P mutation did not greatly destabilize the protein or lead to significant cell toxicity. We next hypothesized that the mutation may affect retinal ganglion cell (RGC) differentiation during early development. E14 mouse RPCs were cultured and transduced with GFP, SIX6WT and SIX6L37P (this time with bicistronic fluorescent protein markers rather than fusion proteins), allowed to differentiate over 5 days in culture, and then immunostained for the RGC marker Brn3, GFP (to mark and quantify transduced cells), and EdU (to mark and quantify cells derived from progenitors that underwent at least one cell division in vitro) (Fig. S2). SIX6WT was not sufficient to further promote the differentiation of RGCs (Fig.1f), consistent with the expression of SIX6 in embryonic RPCs (4). However, the SIX6L37P mutant showed a reduction in the number of differentiated RGCs (Fig. 1f). While this effect could reflect differences between the human and mouse cells, relative levels of gene expression between expressed and endogenous alleles, or contextual milieus,