Mutations in SPATA13/ASEF2 cause primary angle closure glaucoma.

Current estimates suggest 50% of glaucoma blindness worldwide is caused by primary angle-closure glaucoma (PACG) but the causative gene is not known. We used genetic linkage and whole genome sequencing to identify Spermatogenesis Associated Protein 13, SPATA13 (NM_001166271; NP_001159743, SPATA13 isoform I), also known as ASEF2 (Adenomatous polyposis coli-stimulated guanine nucleotide exchange factor 2), as the causal gene for PACG in a large seven-generation white British family showing variable expression and incomplete penetrance. The 9 bp deletion, c.1432_1440del; p.478_480del was present in all affected individuals with angle-closure disease. We show ubiquitous expression of this transcript in cell lines derived from human tissues and in iris, retina, retinal pigment and ciliary epithelia, cornea and lens. We also identified eight additional mutations in SPATA13 in a cohort of 189 unrelated PACS/PAC/PACG samples. This gene encodes a 1277 residue protein which localises to the nucleus with partial co-localisation with nuclear speckles. In cells undergoing mitosis SPATA13 isoform I becomes part of the kinetochore complex co-localising with two kinetochore markers, polo like kinase 1 (PLK-1) and centrosome-associated protein E (CENP-E). The 9 bp deletion reported in this study increases the RAC1-dependent guanine nucleotide exchange factors (GEF) activity. The increase in GEF activity was also observed in three other variants identified in this study. Taken together, our data suggest that SPATA13 is involved in the regulation of mitosis and the mutations dysregulate GEF activity affecting homeostasis in tissues where it is highly expressed, influencing PACG pathogenesis.


Introduction
Glaucoma is the most common cause of irreversible blindness worldwide, affecting nearly 80 million people [1]. It is characterised by an intermittently progressive optic neuropathy, often culminating in loss of vision if left untreated [2]. Two main subtypes are primary open angle glaucoma (POAG) [3] and primary angle closure glaucoma (PACG) [2]. PACG accounts for a quarter of all glaucoma, but causes half of all glaucoma blindness [4]. Physical obstruction of the aqueous humor outflow channels and consequent elevation of pressure inside the eye, are hallmarks of this disease. Typically PACG occurs in eyes that are smaller than average. [5].
Two PACG loci are recognized-GLC2A in a Singaporean family on 10q identified by genetic linkage analysis [6] and another locus mapped to the 3q27.1 region [7] using a population-based GWAS study in Singaporean Malay, Indian and Beijing Chinese populations. However, no causative gene has been identified for PACG at these loci. Vithana and co-workers found a genome-wide significant association between PACG and three loci: rs11024102 in pleckstrin homology domain containing family A member 7 (PLEKHA7) on chromosome 11p, rs3753841 in COL11A1on chromosome 1p, and rs1015213 located between PCMTD1 and ST18 on chromosome 8q in a mixture of Far East, Middle East, Indian and UK populations [8,9]. The two single nucleotide polymorphisms (SNPs) in PLEKHA7 and COL11A1 were replicated in a large PACG cohort in China [10], Australia and Nepal [11]. Variants in PLEKHA7, an adherens junction protein, increase the risk of sudden, symptomatic pressure rises in PACG [10]. This protein is expressed in iris, ciliary body, and choroid [12] and has GTPaseactivating protein (GAP) activity [13]. SNPs in other candidate genes, including HGF (hepatocyte growth factor) [14], HSP70 (heat-shock protein 70) [15], MFRP (membrane type frizzled related protein) [16], eNOS (endothelial nitric oxide synthase) [15] and MMP9 (matrix metalloproteinase-9) [17] also have significant association with PACG.
Next generation sequencing has renewed interest in using family data to identify causal gene(s) for otherwise complex disorders. It exploits information about co-segregation, thus helping to identify potentially causal variants. Rare variants of large effect influencing the susceptibility to the disease are present at a higher frequency in affected than in unaffected individuals in families, as well as the general population. Using this strategy, we have identified SPATA13 as the causative gene for PACG in a large multigenerational Caucasian family and shown that this protein is likely to influence tissue homeostasis.

Pedigree recruitment
A seven-generation Caucasian British family (Family 1) was recruited at Moorfields Eye Hospital (MEH). The proband V:8 (Fig 1; red arrow) attended with her sister V:15 and daughter The proband is V:8 (red arrow) and this pedigree shows the relationship of the 62 individuals examined (see S1 Table for detailed phenotype). There was no male-to-male transmission observed, and the only affected male was V:30, a first cousin of the proband. From branch B, two out of nine individuals examined were affected. There were no individuals examined from branches A and C affected with angle-closure disease. Twelve subjects, four of whom were affected first cousins, were genotyped for genetic linkage analysis, and some were analysed in greater detail by exome and whole genome sequencing. It is suspected that the phenotype in VII:2 and VI:7 is caused by another genetic defect as it is very likely that their disease is inherited from the married in spouse (V:11).
https://doi.org/10.1371/journal.pgen.1008721.g001 VI:4. They gave a strong family history of glaucoma (S1 Table). Proband V:8 had been treated for ocular hypertension for 15 years from age 52. On examination all three, V:8, V:15 and VI:4, were found to have the same phenotype of plateau iris configuration (PIC , Fig 2A). The proband's mother IV:11 underwent surgery for PACG. Including her mother, the proband had nine other aunts and uncles, i.e. ten siblings in generation IV who lived to adulthood, consisting of seven sisters and three brothers. Establishing a detailed pedigree enabled us to expand the family tree to seven generations and contact 63 members. Twelve of these (including V:11, unrelated, spouse) had occluded drainage angles (primary angle-closure suspect-PACS), PAC (Primary angle-closure disease) or PACG (see S1 Text). The family had lived in the East London area since at least 1812.

Linkage analysis
All family members with occluded drainage angles or PACG were selected for linkage analysis (see S1 and S2 Text). An autosomal dominant disease transmission was observed in Family 1, with more affected females than males. This may be attributable due to the majority of the offspring of III:11 being female (seven daughters and three sons in adulthood; five out of seven daughters in generation IV were affected) but could also reflect the greater predisposition of women because of their smaller eyes. Individual VI:7 was not included in this analysis as her mother, V:11, not related to this family and married-in spouse of V:10, is affected. For similar reasons, her daughter (VII:2) was also not included as there could be another gene responsible for the disease, which could come through V:11. Eleven samples from this family (IV:32 Table). After removal of uninformative markers, highest positive LOD score was observed on chromosome 13 between rs9580111 and rs1160226 (Max LOD Score 2.90, NPL Spair 3.80, NPL ALL 18.7). LOD scores above 1 were also observed on chromosome 5 between rs1478449-rs3812042 (Max LOD score 1.25, NPL Spair 1.44, NPL ALL-0.75), chromosome 9 between rs10979427-rs419097 (Max LOD score 1.72, NPL Spair 1.47, NPL ALL 1.76). On chromosome 13 these values increased to Max LOD score 3.5, NPL Spair 4.7, NPL ALL 34.5 when the unaffected individual IV:32, who at 94 years of age did not show any signs of PACG, was included (allowing for 90% penetrance in the parametric model) ( Fig  2B). Including another unaffected subject V:28 also gave similar results (Max LOD score 3.760913, NPL-Spair 4.81, NPL ALL 37.12). A proximal crossover at rs10400673 and rs2166431 at the distal end in individual V:8 narrowed the genetic interval to 4.1 Mbp.

Whole exome and genome sequencing
Four "clearly affected" members of this family, V:8, V:30, V:32 and VI:12 were analysed further with exome sequencing using SureSelect Target Enrichment System, followed by sequencing on a HiSeq2000 (Illumina) with 100bp paired end reads. The average ratio of reads alignment to reference genome was 71%. When measured at a minimum depth of 20x, 93% of the intended targets were covered. Initial filtration of variants (total number of variants in 4 samples was 98,808) was performed to identify heterozygous variants (60,329) shared between the four affected individuals (reduced to the total number to 4,836 variants, i.e. 1,209 candidate variants per sample), followed by selecting all the shared variants with MAF of <0.01. Only one variant, SPATA13 (NM_001166271, chromosome 13q) c.1432_1440del; p.478_480del II.2, shown by an arrowhead carry the deletion. His mother, 5.I.2 was affected but no DNA was available for analysis. Proband has an affected daughter, 5.III.2, but her eye phenotype is different to his. Her mother, 5:II.3, has narrow but open angles and was hyperopic and it is therefore suspected that the daughter has inherited her disease from the mother. fulfilled the filtering criteria. On Sanger sequencing it was found that this variant segregated in all the nine clearly affected samples. The variant was absent in 11 of the 16 blood-related, unaffected family members and none of the spouses carried the variant. Results from five of the unaffected subjects were inconclusive due to poor quality of DNA. Two subjects with uncertain diagnosis and three subjects with other ocular conditions also carried this change. (S1 Table).
The 9 bp in-frame deletion in SPATA13 results in deletion of three amino acids, QSP. There are three repeats of QSP within this region; the mutation deleted one of them. This variant had a MAF of 0.0001158 in Kavier, 0.0006118 in gnomAD Exomes, and 0.0008671 in gno-mAD genomes. We also analysed V:15, V:30, IV:27 and V:28 by whole genome sequencing on HiSeqX with 30x coverage as some genes showed poor coverage on WES within the linked region on chromosome 13. The variants in the affected (V:15, V:30, IV:27) were compared with the unaffected V:28. The only variant with an allele frequency of less than 0.01 that was shared in all affected individuals within the linked regions was SPATA13 (NM_001166271) c.1432_1440del; p.478_480del.

Prevalence of SPATA13 mutations in unrelated patient cohort
To identify additional variants in SPATA13, all 13 coding exons were screened by Sanger sequencing in a cohort of 189 unrelated patients: PACS (n = 14)/PAC (n = 106, 52 acute) and PACG (n = 69, 3 acute). One patient, (5:II:2, Fig 2C, S3 Table) with PACS carried the same 9 bp SPATA13 deletion (c.1432_1440del; p.478_480del) identified in Family 1 (S3 Table). Genotype analysis of SNPs within SPATA13 and its adjacent region show that this individual was not related to Family 1. The genotypes and segregation of this mutation in this family and the clinical details are given in S3 Text.
We also identified another eight variants in SPATA13 amongst eight unrelated patients ( Table 1, Fig 3A), seven of them were in exon 2, the largest in this gene. In our cohort the prevalence of individuals with a rare SPATA13 variant was 4.8%. The MAF of all variants were less than 0.002 in gnomAD, 1000Genomes and EXAC databases. There are two variants that were identified in two subjects of African ancestry, the rest were present in Caucasian subjects. One variant, p.R89P, is present in 1% of the African population, the other variant, p.S246T, was present in 1000 Genomes with a frequency of 0.00019. We did not screen a control African DNA panel as it was not available to us. The rest of the variants in SPATA13 were screened in a control cohort of 192 random European Caucasians. In this control cohort all the 13 exons of SPATA13 were screened by Sanger sequencing where we identified one subject carrying the c.C497T, p.P166L variant. This variant also had a CADD value lower than 10 so presumably this was a rare polymorphism. This shows that the rare variants identified in this study are significantly enriched in our PAC/PACG cohort (S4 Text).
The degree of conservation of all the variants in SPATA13 identified in this study is shown for all nine species (Fig 3B). The predicted secondary structure of SPATA13 shows that the first 800 residues at the N-terminus, where most of the variants are located, has higher percentage of random coil compared with the C-terminus (Fig 3C). The 9 base pair deletion makes the N terminus of SPATA13 more compact when compared with the wildtype protein (compare Fig 3D with 3E).

Expression of SPATA13 transcripts in human cell lines and eye tissues
Multiple SPATA13 transcripts differing in their 5' region have been reported and designated ASEF2a-d [18]. One of the short transcripts, ASEF2b (NM_153023), encoding a 652 amino acid protein (NP_694568), has been studied in detail [19][20][21]. This transcript also contains one of the nine mutations reported above. For simplicity, we will refer NP_001159743 as SP-1277 and NP_694568 as SP-652 based on the amino acid number they encode. We investigated the expression levels of these two SPATA13 transcripts by qPCR. Transcript specific primers were designed to detect NM_153023 (encoding SP-652) and NM_001166271 (encoding SP-1277) as well as total SPATA13 as shown in Fig 4A. Seventeen different human cell lines derived from the eye (retinal pigment epithelial, RPE-1), liver (HepG2), kidney (HEK293), cervix (HeLa), oesophagus (TR146), fibrosarcoma (HT1080), breast (SUM159, T47D), skin (N/Tert-1, HaCaT), head and neck cancers (HN4, HN8, SCC4, SCC9), oral dysplasia (SVpgC2a), and those produced from SVpgC2a by nicotine treatment (SVFN3 and SVFN10) [22] were analysed. As shown in Fig 4B, these transcripts were detected in all 17 cell lines. The transcripts for SP-1277 showed highest expression in RPE-1, SVFN3 and low in SCC9 whereas the transcript for SP-652 was highest in N/Tert-1, SCC4 and lowest in SCC9 (Fig 4B). These results suggest that the two transcripts are ubiquitously expressed at varying levels in multiple cell lines derived from different human tissues.
We also used qPCR to investigate the mRNA expression of SP-1277 and SP-652 in human iris, ciliary epithelium, retinal pigmented epithelium (RPE), retina, cornea and the lens ( Fig  4C). The transcripts for SP-652 showed highest expression in cornea and lens, whereas the transcripts for SP-1277 was highest in cornea and ciliary epithelium, suggesting that SP-1277 was the predominant transcript in tissues most affected in PAC/PACG.

SPATA13 antibodies characterisation and expression in the eye
There are several commercially available antibodies which recognise SP-1277 and SP-652. To confirm their specificities, we cloned SP-1277, SP-652 isoforms and the first N-terminal 625 residues of SP-1277 (SP-1277-N 625 ) (Fig 5A) as AcGFP fusion proteins and expressed them in HT1080 cells. Protein bands of around 166kDa (SP-1277), 100kDa (SP-652 and SP-1277-N 625 ) were detected with anti-AcGFP antibody (Fig 5B). In this study we have used four different antibodies, two against the N-and two against the C-terminal of SP-1277 (available from Abcam and ThermoFisher Scientific). The N-terminal antibodies recognised SP-1277 but showed no reactivity with SP-652 (representative blots of Abcam antibodies are shown in Fig  5B). As shown in Fig 5C, the Abcam C-terminal antibody recognised a band of 70kDa and several high molecular weight bands in the range of 80-180kDa (Fig 5C, lane 1). The N-terminal antibody from ThermoFisher reacted with several high molecular weight bands in the range of 80-180kDa (Fig 5C, lane 2) while the Abcam antibody recognised a single band of 180kDa (Fig 5C, lane 3) but did not bind to the 70kDa band showing the specificity of these reagents. To further test the specificity of the antibody on tissue culture cells, RPE-1 cells were transfected with AcGFP tagged SP-1277 (AcGFP-SP-1277) and immunostained with the Nterminal SPATA13 antibody. As shown in Fig 5D there was complete co-localisation of the AcGFP-SP-1277 with the N-terminal antibody. Similar results were obtained when the C-terminal SPATA13 antibody was used to identify the AcGFP-SP-1277. These results show that the N and C-terminal SPATA13 antibodies were able to recognise SP-1277 in western blots and immunostaining. In this study, the two N-terminal SPATA13 antibodies were used interchangeably, however, the data presented here are mostly with the Abcam antibody.
To investigate whether SP-1277 protein is expressed in the eye we used the N-terminus antibody on murine (Fig 5E & 5F) and human (Fig 5G) eye sections using

PLOS GENETICS
Primary angle closure glaucoma and SPATA13 immunohistochemistry. As shown in Fig 5E-5G, we observed specific reactivity with both pigmented and non-pigmented ciliary epithelia, iris sphincter and dilator muscles, corneal epithelium and retinal outer nuclear, inner nuclear and ganglion cell layers.

SPATA13 shows nuclear and cytoplasmic localisation in RPE-1 cells
Having established expression of SPATA13 in the eye, we investigated the cellular localisation of the endogenous SP-1277 isoform using the antibodies characterised in Fig 5. With the Nterminus antibody, a predominantly grainy nuclear staining was observed in RPE-1 cells (Fig  6A). The C-terminus antibody, which should recognise all the SPATA13 isoforms, also gave almost identical staining (Fig 6B). Similar observations have been reported on HeLa cells using SPATA13 antibodies [23]. As SP-652 has been implicated in actin function [23], we investigated the co-localisation of SPATA13 with F-actin in RPE-1 cells. Fig 6A and 6B show that both the N-and C-terminal antibodies gave reactivity in the nucleus as well as in the cytoplasm with little co-localisation with F-actin. We also observed no co-localisation of SPATA13 staining with nuclear actin (Fig 6C).
To investigate the functions of each isoform individually, we transfected AcGFP-SP-1277 and its N-terminus 625 residues (AcGFP-SP-1277-N 625 ) in RPE-1 cells. AcGFP-SP-1277 was primarily localised as globules in the nucleus with some cytoplasmic staining (Fig 6D), which could be due to a single 14 residues bipartite nuclear localisation signal (NLS) 574-RTPKRRWGSGRRPR-587 identified by NLStradamus [24] (Fig 5A). When AcGFP-SP-1277-N 625 was ectopically expressed in RPE-1 cells, it gave homogenous nuclear AcGFP expression (Fig 6E). Four of the nine variants (S292I, S473N, 478-480d and P964L) reported here were introduced by site-directed mutagenesis and expressed in RPE-1 cells. The nuclear localisation of these variants did not change when compared with the wildtype suggesting the mutations do not affect localisation. (S1 Fig).

Role of SP-1277 in mitosis
While investigating the cellular expression of SPATA13 isoforms, we observed that some cells undergoing cell division showed very speckled localisation of SP-1277 in the equatorial region of the dividing cells. To investigate this further, we examined the localisation of SP-1277 in RPE-1 cells at different stages of cell division (Fig 7). SP-1277 showed granular nuclear signal at interphase with a faint background cytoplasmic signal which remained visible throughout mitosis (Fig 7A). Cells were co-stained with acetylated α-tubulin, to mark the microtubules, centrosomes and spindles. At the beginning of prophase, SP-1277 coalesced into intense speckles within the nucleus, which condensed further and moved towards the equatorial region during metaphase. At anaphase, SP-1277 speckles localised specifically along the kinetochore. The SP-1277 signal dispersed as the daughter cells separated, with only a faint background signal remaining.
To investigate this novel observation, we examined whether SP-1277 was colocalising with the kinetochore. We co-stained RPE-1 cells for SP-1277 and two kinetochore markers, polo-  Endogenous SP-1277 isoform during mitosis. Endogenous SP-1277 protein was localised in dividing RPE-1 cells, during mitosis using the N-terminus antibody and counterstained using anti-acetylated α-tubulin antibody (A), or counter stained with two mitotic markers PLK-1 (B) and CENP-E (C). (A) SP-1277 was detected in the nucleus as speckles at G0, which intensified at the onset of prophase, aligning along the kinetochore at metaphase, and anaphase, before returning to its G0 localisation pattern. (B) SP-1277 globules show co-localisation with the kinetochore marker PLK-1 at prophase, metaphase and anaphase. At cytokinesis the PLK-1 staining is like kinase 1 (PLK-1) and centrosome associated protein E (CENP-E). As shown in Fig 7B, SP-1277 starts to co-localise with PLK-1 at the beginning of prophase and remains co-localised along the kinetochore complex in pro-metaphase and metaphase. At cytokinesis SP-1277 distributes in the cytoplasm with strong staining at the midbody. Similar colocalisation of SP-1277 was observed with CENP-E (Fig 7C). To quantify the degree of colocalisation we selected RPE-1 at two different stages of mitosis, prophase (Fig 8A) and anaphase (Fig 8F). Visual methods (described in S5 Text) indicated colocalisation of SP-1277 (green) and CENP-E (red) at prophase as shown in Fig 8A and in Fig 8B and 8C. We also observed that the colocalisation showed strong linear distribution in the scatterplots (Fig 8D and 8E). Automatic thresholding algorithm of Costes [25] showed localisation of CENP-E was 84% with SP-1277 and 75% of SP1277 localised with CENP-E. Similar results were obtained with ImageJ colocalisation plugins and the image analysis software package Imaris (Fig 8D & 8E). At anaphase SP-1277 and CENP-E showed 84% colocalisation (Fig 8F), with more colocalisation at the kinetochores of the separating chromosome and very little at the metaphase plate.

Influence of mutations on the GEF activity of SP-1277
As SP-652 has been shown to have GEF activity, we investigated if additional 625 residues at the N-terminus of SP-652 would affect this activity. GEF activity was measured by co-transfecting AcGFP-SP-1277 (wildtype or mutants) and RAC-1 in RPE-1 cells (Fig 9). In these experiments, SP-652 lacking the first 204 residues (SP-652del204) was used as positive control as it has been shown that removing 204 residues from the N-terminus increases its GEF activity [26]. There was a 1.5 fold increase in GEF activity of SP-652 compared with RAC-1 only, which increased to 2.5 fold when the first 204 residues were deleted.
SP-1277 had lower GEF activity than SP-652 (Fig 9) suggesting the random coil region in SP-1277 had a negative impact on its GEF activity. We also investigated whether the 9 bp deletion affected the GEF activity of SP-1277. There was more than 2 fold increase in GEF activity due to the 9 bp deletion. We further investigated the GEF activity in another three of the nine variants (S292I, S473N, P964L). The highest change was observed in S473N and P964L. The activity of SP-652 containing P964L mutation was also increased by about 2.5 fold suggesting that this mutation clearly affected the GEF activity.

Discussion
Pathogenesis of angle closure glaucoma is complex with multiple heritable factors affecting the disease progression and severity. Singling out individual factors contributing to the disease is crucial for unravelling the pathway leading to the development of PACG. The lack of suitably large families with multiple affected individuals, together with disease heterogeneity, has hampered classical genetic linkage analyses and positional mapping studies of PACG. However, recent advances in medical imaging techniques, such as anterior segment optical coherence tomography (AS-OCT) have improved phenotypic classification of the angle-closure disease. We used this technique to supplement clinical examination of the relationship between the angle and iris and lens.
In this study, we have used detailed phenotyping in combination with linkage analysis of the largest reported PACG family and next generation sequencing to identify SPATA13/ ASEF2 as a gene that harbors a disease-causing mutation leading to PACG. In this family primarily localized at midbody between the dividing cell which also co-stains for SP-1277. (C) SP-1277 show strong co-localisation with another kinetochore marker PACG was associated with a 9 bp deletion in SPATA13 showing variable expression and reduced penetrance (S6 Text). There were 39 blood related subjects that were clinically examined and genotyped, of these 5 were unaffected, giving us a reduced penetrance of 87.2%. However, of the 5 unaffected, 2 males, VI:25 (42y) and VI:26 (48y), may develop PAC/PACG later, so excluding them from the analysis would increase the penetrance to 92.3%. There is a possibility, although very small, that there might be another variant deep in the intron or in another gene in linkage disequilibrium with SPATA13 as the filtration of variants were prioritised for coding regions. However, we identified additional 8 individuals with rare variants in To evaluate whether colocalization occurs, a line was drawn connecting different globules (B). The green and red intensity was measured using ImageJ in globules along the line and plotted (C). The data was analysed using Image J and Imaris softwares and showed colocalization of 84% (tM0.84) of CENP-E with SP-1277 and 75% (tM = 0.75) of SP-1277 with CENP-E. Fluorescence intensity data of two images were distributed linearly as shown in the scatterplots (Fig 8D & 8E). The colocalization of SP-1277 with CENP-E in anaphase is shown in (F). The regions of interest showing highest colocalization are labelled as 1 and 2. The degree of colocalization of the two molecules was analysed using ImageJ (1a, and 1b for region 1) and (2a and 2b for region 2) and gave a colocalization of more than 80%.
https://doi.org/10.1371/journal.pgen.1008721.g008 SPATA13, one of whom carried the same 9 bp deletion, strengthening the genetic evidence ( Table 1). Analysis of SNPs within SPATA13 showed that the second proband with the 9 bp deletion is not related to Family 1. In this family the penetrance appears to be low, but we believe that the phenotype of the daughter is inherited from her mother, giving a penetrance of 100%. This is explained in detail in S3 Text.
Further evidence for the involvement of SPATA13 in PACG comes from the protein expression data shown in Fig 5. This is the first study where SP-1277 expression is shown in iris, cornea, ciliary epithelium and retina of the human and murine eyes, the tissues most affected in PACG.
A number of GWAS studies have implicated SP-1277 in several neurological disorders including intellectual disability [27], comorbid depression syndrome and alcohol dependence [28] and anorexia nervosa in human [29] and social hierarchy and nocturnal activity in mice [30]. It has also been implicated in differentiated thyroid cancer risk [31] and BRCA1-like early onset breast cancer [32]. The SPATA13 gene product has never been implicated in any eye disease, this is the first study implicating this protein in PACG. The gene produces several transcripts, here we have shown that two SPATA13 transcripts are ubiquitously present in cell lines derived from different human tissues as well as in the eye tissues including cornea, iris, retina, RPE and the lens. This raises an interesting question; if SPATA13 is a ubiquitously expressed protein, why do mutations in this gene only affect the eye? The obvious possibility is that the mutations do cause pathogenesis in other tissues, but they have not been diagnosed or investigated. It is also conceivable that in other organs the effect of mutations is overcome by tissue-specific expression of other proteins. However, there are reports where mutations in a ubiquitously expressed protein only affects the eye, e.g., mutations in PRPF31 only cause retinitis pigmentosa and does not affect any other part of the body [33].
The SP-652 isoform has been reported to regulate cellular events such as reorganisation of cytoskeletal structures, cell migration [23] and angiogenesis [34,35]. The other isoform SP-1277 which has an additional 625 residues at the N-terminus due to alternative splicing of SP-652, has never been investigated and its function is not known. SP-652 contains an adenomatous polyposis coli (APC)-binding region (ABR) followed by a Src homology 3 (SH3) domain, a Dbl homology (DH) domain and a Pleckstrin (PH) domain (Fig 5A), which is also present in SP-1277. The ABR-SH3 domain binds to APC, which activates its GEF activity to mediate downstream cellular events [18]. At the protein level we show that endogenous SP-1277 localises to both nucleus and cytoplasm in interphase, which is consistent with previous reports for endogenous SP-652 expression [36]. In silico 3-D structural prediction analysis showed that SP-1277 contains 16 intrinsically disordered regions (IDR) (Fig 3C), the longest of which is 157 residues. It has been reported that SP-652 is normally in an inactive conformation until the Armadillo (ARM) domain of APC binds to the ABR and SH3 domain and facilitate the Cdc42 exchange [26]. It is likely that post translational modification or binding of APC to the ABR-SH3 domain of SP-1277 releases the DH domain activating its GEF activity and that the IDR acquires a tertiary structure creating binding sites for other proteins and/or DNA/RNA for its additional downstream functions. SP-652 lacks this IDR and therefore binding of APC will result in downstream functions different to SP-1277. The 9 bp deletion (p.478_480del) is predicted to make the IDRs more compact (compare Fig 3D with 3E), which would likely affect their interactions with binding proteins including APC.
In this study we have shown dramatic redistribution of SP-1277 during different stages of cell division suggesting a role in mitosis. Although there are fundamental anatomical and structural differences between the mouse and human eyes, such as lack of a lamina cribrosa or a macula, which may lead to different neurodegenerative pathways, two independent studies have shown that complete knockout of all SPATA13 isoforms in mouse was not embryonic lethal and had no ocular phenotype [30,34]. This suggests that SPATA13 was not essential for mitosis, but instead may have a regulatory role. Similar results have been reported with myocilin (MYOC), a gene associated with POAG pathogenesis [37]. Heterozygous and homozygous mice lacking myocilin show no ocular phenotype, however, missense mutations in this protein are not tolerated and lead to POAG [38].
Rho GTPases are involved in major cellular functions including cell growth, adhesion, motility, polarity and differentiation by controlling different cellular processes such as cytoskeletal remodeling, microtubule dynamics, gene transcription and phospholipid metabolism. Rho signaling responses in cells are highly regulated spatiotemporally to ensure homeostasis within the cell. They switch between active GTP bound and inactive GDP bound states with the help of RHO-GEFs, RHO-GAPs and RHO-guanine nucleotide dissociation inhibitors (RHO-GDIs). The localised modulation of Rho GTPases by GEFs, GAPs and GDIs depends on cell type and presents a complex network of highly integrated regulatory mechanisms. Rho GTPases are also key components of neuronal cell degeneration pathways [39]. The Rho GTPase signaling pathway has been shown to modulate aqueous humour outflow by regulating the contractile properties of trabecular meshwork and Schlemm's canal cells in humans [40,41]. Inactivation of Rho GTPases significantly stimulated/increased adult ciliary epithelial cell proliferation in situ [42]. One of the key features of RHO-GEFs and RHO-GAPs is that their activities are autoinhibited and are sensitive to local activation. SP-652 has been shown to be inactive for RHO-GEF and is activated by binding to APC. In this report we only investigated 4 of the 8 SPATA13 variants for their effect on GEF activity. We have shown that SP-1277 exhibits less GEF activity than SP-652 and the 9bp deletion increases the RAC1-dependent GEF activity. This has also been observed for three additional variants identified in SPATA13 (Fig 9), which suggested that conformational changes introduced by mutations in these variants affect the function through altering the GEF activity. It is likely that the rest of the variants that have not been tested might behave differently and may not show any difference in their GEF activity. It should be noted that PLEKHA7, another protein implicated in PACG (19), has GAP activity and therefore could interact with RAC1 similar to SP-1277. It is, therefore, also likely that both SP-1277 and PLEKHA7 are involved in Rho-GTPase pathway, indicating this pathway could play an important role in PACG disease pathogenesis.
It is important to note that a proportion of SP-1277 is always in the cytoplasm even in dividing cells suggesting that it can also play a role in cell adhesion and migration. The mutations reported here enhance GEF activity of SP-1277, which could dysregulate mitosis with impact on symmetric/asymmetric cell division, cell adhesion and migration as has been reported for SP-652 [19,21,23,43,44]. Dysfunction in cell adhesion has been identified as a genetically modulated risk factor in PACG (14)(15)(16)(17)(18)(19) and is likely to cause its effect through degradation of the blood aqueous barrier. PLEKHA7 encodes an apical junctional protein that is expressed in the non-pigmented ciliary epithelium, a key component of the blood aqueous barrier (BAB). Lee et al found that PLEKHA7 is down regulated in the iris of PACG patients that carry the C risk allele at SNP 11024102 [13]. Silencing of PLEKHA7 in non-pigmented ciliary epithelium (NPCE) affected actin cytoskeleton organization in vitro. Consistent with the regulatory role of RAC-1 and Cdc42 in maintaining tight junction permeability, silencing of PLEKHA7 compromises the paracellular barrier between NPCE cells in vitro. The authors concluded that downregulation of PLEKHA7 in PACG may affect BAB integrity [13].
Other evidence supporting the involvement of SPATA13 in PACG comes from our expression studies where it is highly expressed in iris, cornea, ciliary body and retina (Fig 5C-5F), the tissue most affected by the disease. The malformation of the ciliary epithelium would also impact aqueous humour secretion as well as attachment of lens zonules which would affect the positioning of the lens as observed in PACG [45]. Disturbances in tissue homeostasis of the uvea may increase the propensity for shedding of pigment from the pigmented ciliary or iris epithelia. We observed increased trabecular meshwork pigmentation in all patients carrying the 9 bp deletion, except the youngest two males aged 24y and 26y, in whom disease signs might develop later.
APC and the kinetochore associated proteins have been investigated as targets for developing therapeutics for cancer treatment [46]. APC, which interacts with the two close homologues, ASEF1 and ASEF2 (41), is a peptidomimetic target for colon cancer treatment [47]. Cell cycle inhibitors are being investigated in targeting preferentially aneuploid cancer cells while sparing normal diploid cells. Similar strategies may offer the prospect of treatment for PACG involving APC-SPATA13/ASEF2 interactions.
In conclusion, we have identified SPATA13 as a gene in which mutations cause primary angle-closure disease, which in turn puts those affected at risk of developing glaucomatous loss of vision. There is variable expression and incomplete penetrance of the phenotype associated with this gene. Detailed functional characterisation of SP-1277 and mutations identified in this study will provide insights into the mechanism of disease pathogenesis, strengthening the hypothesis that abnormalities in cell division and/or cell adhesion in the anterior segment of the eye might be a common mechanism in the development of PACG. Identification of SPATA13 as one of the disease-causing genes for PACG opens new avenues for identifying additional genes in this pathway causing PACG.

Methods and materials
Informed, written consent was obtained from all participants. This study followed the principles of the Declaration of Helsinki. The study was approved by the Moorfields NHS Research Ethics Committee (06/Q0504/8).
Family history and pedigree recruitment. The index patient for each family (proband), underwent pedigree charting, enquiring about a family history of glaucoma and/or angle-closure. The accuracy of family history information was confirmed by corroboration of relatives attending the clinics. All family members of the proband, aged 20 years and older were invited to attend ophthalmic examinations. The Society of Genealogists (http://www.sog.org.uk/) helped identify distant relatives using publicly available genealogy and census data. A family member from of the five largest families sent letters of invitation explaining the research to their distant relatives, inviting them to participate.
Clinical examination. Detail clinical examination procedure and diagnostic classification is given in S2 Text. Each participant examined in research clinics underwent comprehensive ophthalmic examination including dark room gonioscopy, visual field testing as well as a clinical examination of the optic disc, and supplementary imaging of the optic disc and peripapillary retina. The primary defining feature of cases was contact between the iris and trabecular meshwork identified on dark room gonioscopy and/or anterior segment imaging. Primary angle-closure glaucoma was diagnosed when there was evidence of iridotrabecular contact and glaucomatous damage to the optic nerve, as detailed under the ISGEO diagnostic standards [2].
DNA isolation and Sanger sequencing. DNA was extracted from peripheral blood using Gentra PureGene kit (Qiagen, UK). Sanger sequencing was performed with BigDyev3.1 (Life Technologies) on ABI3730 according to manufacturer's protocol. 192 human random control DNAs were purchased from Merk.
SNP chip array and linkage analysis. Single nucleotide polymorphism (SNP) genotyping was undertaken on 11 individuals from Family I. Processing of SNP microarrays (Illumina, Human CytoSNP-12, 298,199 markers) was performed according to the manufacturer's protocol. The relationship status of participants was checked against their genotype using the GRR software (http://csg.sph.umich.edu/abecasis/grr/) to identify errors in pedigree charting and possible non-paternity. Overall call rates were >98% for all samples and no gender mismatch was found. The marker sets were reduced to the 3,547 most informative markers distributed uniformly throughout the chromosomes to avoid artificially inflating LOD scores due to linkage disequilibrium between closely spaced SNPs. All genotype calls were checked for Mendelian inconsistencies and two-point and multipoint linkage analyses of genome wide SNP data were performed using Superlink-Online SNP-1.1 [48] using an autosomal dominant model, 100% penetrance in the affected-only analysis, 90% penetrance in other analyses.
Exome and whole genome sequencing. Genomic DNA from the proband of family I was fragmented enriched for exomic sequences using the Agilent SureSelect Whole Exome hybrid capture (Agilent, UK). The resulting enriched sequence library was sequenced with 76bp paired end reads across two lanes of Illumina GAIIx flowcell (v2 chemistry). The sequence reads were aligned to the reference genome (hg18) with the Novoalign aligner. Duplicate reads, resulting from PCR clonality or optical duplicates, and reads mapping to multiple locations were excluded from downstream analysis. Single nucleotide substitutions and small insertion/deletions were identified, and quality filtered within the SamTools software package.
Whole genome sequencing was performed on HiSeqX with 30x coverage at Edinburgh Genomic Centre, UK. The sequence reads were aligned to hg19 and the data was obtained as BAM and VCF files. VCFTools were used to filter the variants within the genetically linked region. Pathogenicity of the variants were evaluated by VEF (Ensemble) and CADD.
Molecular cloning and mutagenesis. Total mRNA was isolated from cultured cells using the Dynabeads mRNA DIRECT Purification Kit (Ambion, Paisley, UK) using manufacturer's instructions and transcribed into cDNA using the Transcriptor Reverse Transcriptase System (Roche, Burgess Hill, UK). The coding regions of SP-652 and SP-1277 were amplified using high fidelity Q5 DNA polymerase (New England Biolabs, UK) and ligated into pLPC_NAcGFP using standard cloning techniques [49]. The retroviral expression vector pLPC_NMyc (gift from Professor Titia de Lange (Addgene plasmid #12540) was re-engineered to replace N-Myc tag with AcGFP. For site-directed mutagenesis phosphorylated primers harbouring the desired mutations were used to amplify the SPATA13 constructs with Q5 DNA polymerase and ligated with Taq DNA ligase. Sanger sequencing was used to verify all cloned inserts and mutations.
SDS electrophoresis and western blotting. SDS electrophoresis and western blotting were performed according to protocol described previously [50]. mRNA extraction, cDNA synthesis and quantitative PCR. Total RNAs from pooled human eye tissues, ciliary body, iris, retinal epithelium and retina were commercially obtained from 3H Biomedical AB (Sweden) and isolated from cornea and lens tissues using RNeasy Mini Kit (Qiagen) as per manufacturer's instructions. The RNA samples were reverse transcribed into cDNAs as described above. Specific primers for three different forms of SPATA13 (NM_001286792; NM_001166271 and NM_153023; were designed spanning exon-exon junctions with Tm of forward and reverse primers within ±1˚C. NM_001286792; F: AAGGACCC TCTGAGGACGTAG; R: ACAGGTCACCCCTAGCTGG, NM_001166271; F: ACGCTGAC TTTGTAGGCTCC; R: TCATCTCCAGGAATGCCGTC, NM_153023; F: GACACGCTGAC TTTGTAGGC; R: ATCCTGTGATGAGCTTCGCC, SPATA13_Total; F: GCTCAAAAGG CAGGACATGG; R: CATAGTGATGTGGCGCTGGT. Quantitative PCR (qPCR) was performed on the LightCycler 480 qPCR System (Roche, UK) according to protocol described by Bose et al. [51]. The qPCR analyses for mRNA expression in cell lines were performed in three independent replicates, with three technical repeats for every sample. However, for human eye tissues the expression was performed only in three technical replicates.
Prior to immunofluorescence (IF) studies, cells were seeded in 24-well plates on glass coverslips at 5.0 x 10 4 cells per well and incubated for 24 h. The cells were processed for immunocytochemistry as described previously [50].