Molecular Vision 2022; 28:57-69
Received 18 December 2021 | Accepted 15 May 2022 | Published 17 May 2022
Evangelia S. Panagiotou,1 Narcis Fernandez-Fuentes,2 Layal Abi Farraj,1 Martin McKibbin,1,3 Nursel H. Elçioglu,4,5 Hussain Jafri,6 Eren Cerman,7 David A. Parry,1 Clare V. Logan,1 Colin A. Johnson,1 Chris F. Inglehearn,1 Carmel Toomes,1 Manir Ali1
The last two authors contributed equally to the conduct of this research.
1Division of Molecular Medicine, Leeds Institute of Medical Research, University of Leeds, United Kingdom; 2Institute of Biological, Environmental & Rural Sciences, Aberystwyth University, United Kingdom; 3Eye Clinic, St. James's University Hospital, Leeds, United Kingdom; 4Department of Pediatric Genetics, Marmara University Medical School, Istanbul, Turkey; 5Eastern Mediterrean University of Medical School, Cyprus, Turkey; 6Fatima Jinnah Medical University, Lahore, Pakistan; 7Department of Ophthalmology, Marmara University Medical School, Istanbul, Turkey
Correspondence to: Manir Ali, Division of Molecular Medicine, Leeds Institute of Medical Research, University of Leeds, Wellcome Trust Brenner Building, St. James’s University Hospital, Leeds LS9 7TF, UK; Phone: 44 113 343 8420; FAX: 44 113 343 8603; email: firstname.lastname@example.org
Purpose: To investigate the molecular basis of recessively inherited congenital cataract, microcornea, and corneal opacification with or without coloboma and microphthalmia in two consanguineous families.
Methods: Conventional autozygosity mapping was performed using single nucleotide polymorphism (SNP) microarrays. Whole-exome sequencing was completed on genomic DNA from one affected member of each family. Exome sequence data were also used for homozygosity mapping and copy number variation analysis. PCR and Sanger sequencing were used to confirm the identification of mutations and to screen further patients. Evolutionary conservation of protein sequences was assessed using CLUSTALW, and protein structures were modeled using PyMol.
Results: In family MEP68, a novel homozygous nucleotide substitution in SIX6 was found, c.547G>C, that converts the evolutionarily conserved aspartic acid residue at the 183rd amino acid in the protein to a histidine, p.(Asp183His). This residue mapped to the third helix of the DNA-binding homeobox domain in SIX6, which interacts with the major groove of double-stranded DNA. This interaction is likely to be disrupted by the mutation. In family F1332, a novel homozygous 1034 bp deletion that encompasses the first exon of SIX6 was identified, chr14:g.60975890_60976923del. Both mutations segregated with the disease phenotype as expected for a recessive condition and were absent from publicly available variant databases.
Conclusions: Our findings expand the mutation spectrum in this form of inherited eye disease and confirm that homozygous human SIX6 mutations cause a developmental spectrum of ocular phenotypes that includes not only the previously described features of microphthalmia, coloboma, and congenital cataract but also corneal abnormalities.
Developmental eye defects such as microphthalmia, anophthalmia, and coloboma (MAC) are often associated with a spectrum of ocular phenotypes that involve the cornea, lens, iris, optic nerve, or retina . Although often difficult to determine, these additional features could represent the primary defect or they may be secondary to the underlying MAC phenotype. These conditions are genetically heterogeneous and can affect common developmental pathways including transcription factor activity, TGFβ/BMP, and retinoic acid signaling .
Sine oculis homeobox homolog 6 (SIX6; OMIM 606326) is a 2-exon gene that translates into a 246 amino acid protein and belongs to a family of evolutionarily conserved transcription factors (SIX1-SIX6) that all originated by gene duplication and divergence from a common ancestor . It has a protein-interacting SIX1_SD domain that forms complexes with cofactors to mediate transcriptional regulation, and a helix-turn-helix structural homeobox domain that binds DNA. To date, only four studies have reported SIX6 mutations associated with recessively inherited eye disease in humans [4-7]. Two studies [4,5] reported a 5 bp homozygous deletion in SIX6, c.532_536del, that created a frameshift mutation, p.(Asn178Profs), which would be expected to lead to a null phenotype since part of the homeobox domain and the C-terminal end of the protein would be lost. Aldahmesh et al.  identified this mutation in two siblings affected with microphthalmia associated with anterior and posterior segment abnormalities, although there was no evidence of coloboma. The younger child at 16 months of age had buphthalmos, corneal scarring, optic nerve cupping, and retinal detachment in the right eye and an extremely small globe without retinal detachment in the left eye. His 5-year-old sibling had bilateral retinal dystrophy and optic nerve dysplasia. More recently, Deepthi et al.  described this mutation in two affected brothers, suggesting a common founder effect. The younger child at 10 years old had anophthalmia of the right eye and microphthalmia, sclerocornea, and cataract but no coloboma in the left eye. The older sibling examined at 11.5 years old had bilateral anophthalmia, severe psychomotor delay, and intellectual disability. In another report, Yariz et al.  described a homozygous nucleotide substitution in SIX6, c.110T>C, that changed a leucine at the 37th amino acid in the protein to a proline, p.(Leu37Pro). This leucine residue is located in the SIX1_SD transactivation domain, which binds other nuclear proteins and is required for the DNA-binding activity of SIX6. The mutation to a proline residue disrupts this interaction, preventing optimal SIX6 function in the patient. This mutation was found in three affected children, aged 14, 5, and 2, of a consanguineous family that had iris and chorioretinal colobomas with macular atrophy and optic disc anomalies but without microphthalmia and cataracts. Additionally, Rudilla et al.  identified a 38-year-old male with congenital cataracts, strabismus, color blindness, and severe coloboma who had a homozygous SIX6 variant, c.86G>C, that changed the evolutionary-conserved 29th amino acid located in the transactivation domain from an arginine to a proline, p.(Arg29Pro). This mutation is likely to affect normal function by disrupting the interaction of SIX6 with other proteins and reducing its capacity to bind DNA. Here, we report previously undescribed mutations in SIX6 in two consanguineous families, allowing us to expand the mutation spectrum and clinical phenotype associated with this gene.
This study was approved by the Leeds East Research Ethics Committee (project numbers 03/362 and 17/YH/003) and adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all adult participants. In total, four affected (one male and three females) and three unaffected (two males and one female) subjects were recruited. Pedigree structures are depicted in Figure 1. The families were assessed by an experienced ophthalmologist and found to have no other obvious abnormalities apart from problems with their vision. Blood samples were obtained from both affected and unaffected family members where available, and genomic DNA was extracted using standard procedures. Genomic DNA from a panel of 196 patients with various ocular developmental phenotypes spanning the MAC spectrum were gathered from local, national, and international collaborations.
Whole-genome homozygosity mapping was performed using Affymetrix 6.0 SNP arrays (AROS Applied Biotechnologies, Aarhus, Denmark). The data were analyzed and presented using AgileMultiIdeogram software.
Genomic DNA was processed according to Agilent’s SureSelectQXT Library Prep protocol (Agilent Technologies Limited, Wokingham, UK) to enrich exonic regions using SureSelect All Exon V5 capture reagent. The processed libraries were subjected to 100 bp paired-end sequencing on a HiSeq2500 (Illumina, Little Chesterford, UK). The resulting reads were aligned to the human reference genome (GRCh37, hg19) using Novoalign short-read alignment software and processed in the SAM/BAM format using Picard and the Genome Analysis Toolkit (GATK) [8-10]. SNPs and indels were called in the variant call format using the HaplotypeCaller function of GATK. Variants were excluded if they were located outside the exons and flanking splice site regions, had a read depth of ≤10, were synonymous, or had a minor allele frequency >1% in EVS, 1000 Genomes, or gnomAD databases. The remaining variants were annotated using ANNOVAR software (version July 2014) . To predict the pathogenicity of missense variants, PolyPhen2, Mutation Taster, and scaled CADD scores (Combined Annotation Dependent Depletion, version 1.3) were investigated.
ExomeDepth  and Integrative Genomics Viewer (IGV) were used to detect copy number variations in exome data. To check for homozygous regions in the exome data, the sequence variants were analyzed using AgileVariantMapper and presented as a MultiIdeogram.
Specific primer pairs used for PCR amplification and direct sequencing of the coding regions and flanking splice sites of SIX6, as well as the identification of the deletion breakpoint, were designed using Primer3 and are listed in Appendix 1. An aliquot of the PCR was checked for a product using agarose gel electrophoresis. The remaining product was digested with ExoSAP-IT (GE Healthcare, Freiburg, Germany) and sequenced using a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems). The sequencing reactions were resolved using an ABI3130xl Genetic Analyzer according to the manufacturer’s instructions (Applied Biosystems).
Protein sequences were downloaded from the National Center for Biotechnology Information (NCBI) databases. The conservation of amino acids was assessed by aligning protein sequences from different species using CLUSTALW.
The structural model of human SIX6 protein (Uniprot accession number O95475) was derived by homology modeling using M4T [13,14]. The crystal structure of the human SIX1-EYA2 protein complex  and the crystal structure of human Brn-5 transcription factor in complex with DNA  were used as templates. The mutation of the residue was modeled using Dunbrack’s rotamer-dependent library. The quality and stereochemistry of the model were assessed using Prosa-II and PROCHECK and represented using PyMol.
Two families, one from Pakistan (MEP68) and the other from Turkey (F1332), were investigated in this study. The Pakistani family had previously been partly described by Khan et al. . Three members of this family affected with visual impairment since early childhood were all clinically examined in their fifth decade. Two of the three individuals (Figure 1A, IV:1 & IV:3 and Table 1) had nystagmus and low vision (could count fingers only). Both had microcornea with corneal vascularization, peripheral corneal opacity, and cataract (Figure 2A). One of them (IV:1) also had iris coloboma. Indirect ophthalmoscopy showed no fundus abnormalities. The other affected member in this family (IV:2) had sclerocornea totalis, which prevented fundus viewing. Ocular ultrasound was not available, so the axial length of each eye could not be ascertained. The proband in the Turkish family (Figure 1B, IV:3, and Table 1) was diagnosed at the age of 1 with bilateral microphthalmia, corneal vascularization, sclerocornea, congenital cataract, and reduced/absent fundus reflexes. Follow-up at three years old showed microphthalmia, sclerocornea totalis with cornea plana, and flat anterior chambers (Figure 2B). B-scan ultrasonography revealed no posterior segment anomalies. Clinical examination of the affected cases from both families found no neurologic or other systemic abnormalities. The family history and evidence of consanguinity suggested autosomal recessive inheritance (Figure 1).
To identify the causative mutation in this family, genomic DNA from two affected cases, IV:1 and IV:3, were analyzed by conventional autozygosity mapping using SNP microarrays. Several homozygous regions on chromosomes 1, 6, 7, 8, 11, 13, 14, and 20 were shared between both individuals (Figure 3A). To identify the pathogenic mutation, DNA from patient IV:1 was whole-exome sequenced. The resulting dataset of 63,009 variants that differed from the human genome reference assembly GRCh37, hg19, was filtered to select variants present at less than 1% in the public databases and that altered the protein-coding sequence or consensus splice signal of genes. Of these, only five homozygous variants, as listed in Table 2, mapped within the autozygous regions. Pathogenicity prediction software prioritized three of these variants (GPR137, SIX6, and ARHGAP40) for further study based on their potential to be disease-causing. These variants were investigated in the genomic DNA of family members by Sanger sequencing to check for segregation of the alleles. Only the missense variant in SIX6 (NM_007374: c.547G>C: p.(Asp183His)); ClinVar accession SCV000928400.1) segregated in the family as expected for a recessively-inherited condition (Figure 1A). This nucleotide substitution, which is not present in the EVS, 1000 Genomes, or gnomAD databases, is predicted to convert the evolutionarily conserved 183rd amino acid from an aspartic acid to a histidine (Figure 3B,C). The aspartic acid residue maps to the homeobox domain in SIX6. Modeling the DNA binding of SIX6 shows that the aspartic acid residue, which is negatively charged, is located in the third helix, also called the recognition helix, and is in close contact with residues that directly interact with the major groove of double-stranded DNA. It seems likely that the histidine substitution, with its positively charged and bulkier imidazole side chain, disrupts the stabilization of the recognition helix, preventing normal protein function (Figure 3D).
To find the molecular basis of anterior segment anomalies and microphthalmia in the proband of family F1332, whole-exome sequencing was undertaken but failed to identify any homozygous variants in any of the known genes. A list of the known genes in which pathogenic variants had previously been found to cause nonsyndromic and syndromic MAC are shown in Appendix 2. As the presence of consanguinity does not preclude other patterns of inheritance, heterozygous variants were also analyzed. After filtering the exome variant list, 332 heterozygous variants remained with a read depth greater than 5, of which 113 were prioritized as candidates because they had a CADD score greater than 15 (Appendix 3), where a score of 20 or above was indicative of the variant being among the top 1% of deleterious changes in the human genome. Of the 113 heterozygous variants, none mapped to any of the known genes in which mutations had previously been implicated in MAC disease. Another possibility to consider was that there might be a mutation in a known gene not adequately covered during exome analysis. Mutations in forkhead box E3 (FOXE3; OMIM 601094) [18,19] cause a phenotype similar to that in the patient described here, yet it is often not well covered during exome sequencing, and the mutation may not be detected. However, the coding sequence of FOXE3, which maps to chr1:47,881,988–47,882,947 (GRCh37, hg19), was shown to be suitably covered in the exome sequence of the proband. The region 47,881,903 to 47,883,084 on chromosome 1, which contains FOXE3, was shown to have an average read count of 28.48 for each nucleotide, with 93.1% of the region above read count 10 and 99.9% above read count 5.
To detect any large copy number changes that would have been missed using the variant analysis filtering pipeline, ExomeDepth software was used on the exome dataset. This approach compared the read depth of each exon of the proband’s exome to seven unrelated exomes that were prepared using the same targeted enrichment reagent and ran in the same lane on the sequencer. The analysis revealed a homozygous deletion at the SIX6 locus (Figure 4A) that was shown using the Integrative Genomics Viewer to remove all of exon 1, as the proband showed no sequence read coverage at this position (Appendix 4). Homozygosity mapping with the exome data confirmed a 33 Mb homozygous region at chromosome 14 (Chr14: 56,142,480–89,338,619, GRCh37), which contains the SIX6 gene (Figure 4B). To find the deletion breakpoints in the patient’s genomic DNA, oligonucleotide primers were designed to amplify the flanking sequences spanning the mutation until a suitable primer pair was identified (Appendix 1). PCR and Sanger sequencing identified a homozygous 1034 bp deletion in the proband (chr14:g.60975890_60976923del, c.-227_572+235del1034; ClinVar accession SCV000929754.1) that removed all 572 coding base pairs of exon 1 as well as 227 bp of the 5′ UTR and 235 bp of the first intron of SIX6 (Figure 4C). This mutation is likely to lead to the complete absence of SIX6 protein in the patient, as there is no downstream methionine in exon 2. PCR combined with agarose gel electrophoresis was used to confirm that both parents, who were asymptomatic, were carriers of the deletion mutation (Figure 4D).
Both exons of SIX6 and their flanking sequences were sequenced in a panel of 196 patients with various developmental ocular phenotypes similar to the affected cases in families MEP68 and F1332, but no further mutations were found. The F1332 exon 1 deletion breakpoint PCR primers were also used to screen this panel, but no further deletions were identified.
Here, we present two families, each with a novel homozygous SIX6 mutation that causes non-syndromic eye disease in humans. Previously, only three recessively inherited SIX6 mutations had been described [4-7]. The mutations, which included null and missense alleles, were mapped schematically (Figure 5). A clinical comparison of the phenotypic presentation of patients with biallelic SIX6 mutations showed that they present with a spectrum of ocular developmental defects that may or may not include microphthalmia, corneal defects, cataracts, iris coloboma, optic nerve anomalies, and retinal abnormalities (see Table 1). Due to the small number of affected cases that have been studied so far, it is difficult to determine whether there are any obvious genotype–phenotype correlations. Nevertheless, it appears that patients with null mutations (family F1332) [4,5] represent the more severe end of the developmental condition with microphthalmia and little or no light perception, while patients with missense changes in the protein (family MEP68)  only have reduced vision in the absence of microphthalmia, although more studies are required to support these initial observations. The lack of MAC spectrum features in MEP68 is worth noting, as it is a novel observation.
Animal studies of mice, fish, frogs, and flies have also pointed to a role for SIX6 in eye specification, eye size, and retinal development. Six6 is expressed in the developing and adult retina, optic stalk, hypothalamus, and pituitary gland in mice . Mice lacking Six6 have increased expression of cyclin-dependent kinase inhibitors causing hypoplasia of the retina and pituitary, as well as absence of the optic nerves and chiasm . Such mice have also been shown to have reduced fertility due to decreased gonadotrophin-releasing hormone  and irregular circadian rhythms associated with abnormal development of the suprachiasmatic nucleus . Knockdown of the SIX6 ortholog, six6b, in zebrafish leads to a small eye phenotype due to an underdeveloped lens and reduced optic nerve diameter , whereas overexpression of the SIX6 ortholog, XOptx2, in Xenopus embryos causes the development of large eyes . Intriguingly, the Drosophila SIX6 ortholog, optix, can induce ectopic eye development in nonretinal tissues . These studies highlight overlapping features observed in human patients with SIX6 mutations who also show anterior segment defects.
It is important to note that previous studies in humans have suggested that the haploinsufficiency of SIX6 was the likely disease mechanism for mutations in this gene [27-32]. For example, large heterozygous interstitial deletions at chromosome 14q22-q23 that encompass SIX6 have been found in cases of anophthalmia with other developmental abnormalities [27-32]. However, since those earlier reports, heterozygous mutations in orthodenticle homeobox 2 (OTX2; OMIM 600037), which is 3.7 Mb proximal to SIX6 and within the deleted regions, have been found in cases with anophthalmia and pituitary defects [33,34]. Furthermore, heterozygous mutations in bone morphogenetic protein 4 (BMP4; OMIM 112262), which is 6.6 Mb proximal to SIX6, have been identified in patients with ocular abnormalities and polydactyly , similar to those patients described with 14q22 deletion and polydactyly . Another study described a heterozygous SIX6 missense variant in a case of bilateral anophthalmia and pituitary anomalies, although the patient inherited the variant from her unaffected father, suggesting that the variant may be of unknown significance or that her mother also carried a SIX6 mutation that was not identified .
Studies have also reported associations of single nucleotide polymorphisms (SNPs) around the SIX6 locus in cohorts of studies on late-onset, multigenic condition, primary open-angle glaucoma in Caucasian [37-40] and East Asian [41-43] subjects. The associated SNPs at the SIX6 locus have included an intergenic variant, rs10483727 (T/C; T-allele associated), that is located approximately 94 kb downstream of SIX6 [37-39,41,42], a variant 4562 bp downstream of SIX6, rs12436579 (C/A; C-allele associated),  and one that changes an amino acid residue, rs33912345, (c.421C/A, p.(His141Asn)); C-allele associated) in SIX6 [40,42,43]. Patients homozygous for the functional p.His141 risk allele were found to have a thinner retinal nerve fiber layer compared to patients homozygous for the non-risk p.Asn141 allele . Transient transfection experiments of the p.His141 risk variant, when compared to the p.Asn141 variant, in isolated rat retinal ganglion cells showed increased transcription of the cell cycle regulator, cyclin dependent kinase inhibitor, CDKN2A isoform, INK4a (p16INK4a), and the senescence-associated secretory marker interleukin 6 . Similar experiments in cultured human fetal progenitor cells induced cell senescence that coincided with increased p16INK4a expression . The authors concluded that increased p16INK4a expression, caused by promoter binding of the SIX6 p.His141 variant, causes retinal ganglion cells to senescence and consequently leads to cell death in glaucoma. The p.His141 risk allele has even been associated with thinner retinal nerve fiber layer in general population-based controls [45,46]. Based on these observations, it is worth highlighting that although the parents of the affected cases described here are heterozygous carriers for a SIX6 mutation and do not show any obvious ocular symptoms following clinical examination, they might be at increased risk of developing late-onset glaucoma.
To summarize, we report previously undescribed mutations in the gene SIX6 as the cause of the ocular phenotype in two consanguineous families. The phenotype includes congenital cataract and various corneal defects (opacity, microcornea, sclerocornea, and cornea plana) in the presence or absence of iris coloboma and microphthalmia. This report not only increases the spectrum of SIX6 mutations in this form of eye disease but also reports corneal abnormalities as part of the phenotype.
Appendix 1. Primers for the PCR and Sanger sequencing of SIX6.
Appendix 2. List of 102 known genes in which pathogenic mutations cause all forms of MAC disease.
Appendix 3. The 113 heterozygous variants identified with a read count >5 and CADD score >15 after filtering the exome sequencing data of patient IV:3 from family F1332.
Appendix 4. Integrative Genomics Viewer of whole exome sequence from the proband of family F1332 at the SIX6 locus.
Funding for the work described in this manuscript was provided by the European Union’s Seventh Framework Programme for research, technological development, and demonstration under agreement no. 317472 (EyeTN). ESP was an EU Early Stage Researcher (grant number 317472) and LAF was a Fight for Sight PhD student (grant number 1835) during this work. NFF acknowledges support from the Biotechnology & Biological Sciences Research Council (BBSRC) (grant number BBS/E/W/0012843D), and CAJ and CFI were supported by a Sir Jules Thorn Charitable Trust Biomedical Award (grant number 09/JTA).