Genetic Causes of Childhood Glaucoma Linked to Front Eye Development Problems

CYP1B1 encodes a 543-amino acid enzyme which is part of the cytochrome P450 superfamily and is involved in retinol [61,62], melatonin [61] and 17ß-oestradiol metabolism [62,63]. It regulates myocilin production by limiting estrogen-responsive activation of the MYOC promoter [63], and contributes to trabecular meshwork development through its modulation of periostin [64]. Pathogenic variants include missense, nonsense and frameshift mutations, and all have been identified throughout the gene [65]; the majority result in loss of function of metabolic activity [62,66,67], although some variants, such as c.1154C>A p.(Phe261Leu) and c.1776C>T p.(Arg469Trp) cause significantly increased retinol metabolism [62]. CYP1B1 is the most commonly identified genetic cause of primary congenital glaucoma (PCG) [65,68], causing 15–35% of PCG in European studies [12,66], and up to 100% of PCG in some ethnic groups [69]. To date, there have been 62 CYP1B1 variants associated with anterior segment dysgeneses, including the following: c.1064_1076del p.(Arg355Hisfs*69), c.1103G>A p.(Arg368His) and c.1169G>A p.(Arg390His) associated with Axenfeld–Rieger syndrome and Peters anomaly [25,28,29,31]; c.171G>A p.(Trp57*) and c.868dupC p.(Arg290Profs*37) associated with Peters anomaly and other corneal dystrophies [25,31,33,70]; and c.434_443del p.(Arg145Profs*4), c.756C>A p.(Asn252Lys) and c.1453T>C p.(Ser485Phe) associated with aniridia [30,71].

FOXC1, part of the forkhead box family of transcription factors, is expressed in the periocular mesenchyme and is necessary for separation of the cornea and lens in the developing eye [2,5,72]. It acts in conjunction with PITX2 to promote the development of the corneal stroma and endothelium, iridocorneal angle and choroid, with co-expression of both transcription factors necessary for normal differentiation of these cellular populations [73]. In the trabecular meshwork, FOXC1 activity is mediated by miR-204 and regulates MEIS2 [74], a homeobox transcription factor which in turn is a direct regulator of PAX6 [75]. FOXC1 variants have been identified in patients with childhood glaucoma associated with Axenfeld–Rieger anomaly [15,28,34], Peters anomaly [13,33] and aniridia [13]. In the context of Axenfeld–Rieger syndrome, FOXC1 variants cause significantly more congenital glaucoma than do PITX2 variants, which are more associated with juvenile- and adult-onset glaucoma [15]. Whole gene deletions account for approximately one third of anterior segment dysgenesis caused by FOXC1 [12,15]. There are 126 intragenic FOXC1 variants associated with Axenfeld–Rieger syndrome, of which c.482T>A p. (Met161Lys) was most commonly identified [15]. In a functional study, Medina-Trillo and colleagues showed that FOXC1 variants can result in either significantly increased or decreased transcriptional activity compared with the wild type; hypermorphic variants including c.141C>G p.(Tyr47*) and c.316C>T p.(Gln106*) were associated with isolated dominant glaucoma, whilst hypomorphic variants including c.377T>G p.(Ile126Ser) were identified in patients with Axenfeld–Rieger anomaly [76].

Heterozygous pathogenic variants in PAX6 are typically associated with autosomal dominant aniridia, or are part of the Wilms tumor-aniridia-genital anomalies-range of developmental delay (WAGR) syndrome if the whole gene is deleted along with WT1 [11,18]. The PAX6 transcription factor is known to have a role in ocular development from gestational day 22 in the differentiation of the eye field and optic vesicle, through to the development of the optic cup and lens vesicle at day 37, and then in corneal, iris, ciliary body and trabecular meshwork development from day 47 onwards [2,77,78]. It has both paired and homeobox DNA-binding domains, and interacts with numerous other genes during this process, including SOX2, RAX, MEIS2 and ALDH1A3 [5,75,77]. Close regulation of PAX6 expression is required throughout embryogenesis to ensure normal development; underexpression results in the aniridia spectrum of iris, foveal and optic nerve hypoplasia with associated glaucoma, cataracts and keratopathy, whilst overexpression causes microphthalmia, microcornea and ciliary body maldevelopment [79,80,81]. Of the 467 pathogenic and likely pathogenic PAX6 variants reported to date [82], c.649C>T p.(Arg203*), c.760C>T p.(Arg240*), c.823C>T p.(Arg261*) and c.991C>T p.(Arg317*) are the most commonly identified [14]. Whilst there are reports implicating PAX6 in isolated juvenile open angle glaucoma and childhood glaucoma associated with Peters anomaly, childhood glaucoma caused by PAX6 variants is very rarely diagnosed in patients without aniridia [33,83].

Both LTBP2 and ADAMTSL4 contribute to cytoskeletal development through their interactions with fibrillin-1. LTBP2 encodes a 1821-amino acid extracellular matrix protein which interacts with fibrillin-1-containing microfibrils [84,85] and is expressed in the cornea, trabecular meshwork and lens [85]. A causative gene for Weill–Marchesani syndrome [28,36,84], which presents with short stature, brachydactyly, microspherophakia, ectopia lentis and myopia, LTBP2 has also been linked with congenital glaucoma associated with isolated microspherophakia [16], ectopia lentis (c.4825T>A p.(Cys1609Ser) and c.529T>C p.(Trp177Arg) variants) [37] and congenital cataracts (c.895C>T p.(Arg299*) variant) [12]. The glycoprotein ADAMTSL4 binds fibrillin-1 and facilitates microfibril assembly, and is expressed in many ocular tissues including the cornea, iris and ciliary body in the anterior segment, and the retinal ganglion cells, retinal pigment epithelium, choroid and sclera in the posterior segment [86]. The missense variant c.2236C>T p.(Arg746Cys) has been linked to congenital glaucoma with ectopia lentis [38].

CPAMD8 is expressed in the human eye from gestational week 9 onwards [87]. From weeks 9 through 22 of gestation, CPAMD8 expression increases in the lens and decreases in the retina, and strong expression is also observed in the iris and cornea at week 22 [87]. The pathways by which CPAMD8 acts to contribute to ocular development remain undefined, but recent work by Escribano and colleagues in zebrafish suggests an interaction between cpamd8 and adamtsl4 [88], implicating a potential contribution of CPAMD8 to the development of fibrillar structures in the anterior segment. CPAMD8 variants have been linked to anterior segment dysgenesis associated with both congenital and juvenile glaucoma [16,39,40,41], with the variants c.4298C>A p.(Thr1433Asn) and c.2352dupC p.(Arg785Glnfs*23) identified in more than one family [40].

Variants in COL4A1 and PXDN, which are both basement membrane components, have been identified in patients with anterior segment dysgeneses associated with childhood glaucoma, although only the c.2159G>A p.(Gly720Asp) has been reported in more than one family [13,33,42,43,55,56]. COL4A1 encodes the alpha-1 subunit of type IV collagen, which is the main component of endothelial and epithelial basement membranes [89]. Peroxidasin, the protein product of the PXDN gene, forms sulfilimine crosslinks between type IV collagen molecules, and is necessary for normal development of type IV collagen-, laminin- and fibronectin-based fibrillar networks in endothelial cells [90]. Type IV collagens are found throughout the developing eye, and by embryonic day 18.5 (E18.5), the alpha-1 isoform can be identified in all anterior segment tissues in mice [91].

One of two receptor tyrosine kinases in the angiopoietin/TEK signaling pathway, autosomal dominant mutations in TEK have been identified in patients with primary congenital glaucoma [16] and congenital glaucoma associated with sclerocornea (specific variant not reported) [36]. In mice, TEK is expressed in Schlemm's canal endothelial cells, and its ligand Angpt1 is expressed in the trabecular meshwork [92,93]. Supporting the role of TEK in iridocorneal angle and Schlemm's canal development, murine studies have shown that Tek haploinsufficiency and Angpt1 knockout result in severe hypoplasia of both Schlemm's canal and the trabecular meshwork, whilst Tek knockout results in complete failure of Schlemm's canal development [93,94].

Pathogenic variants in SLC4A11 are the most commonly identified genetic cause of congenital hereditary endothelial dystrophy (CHED) [95], which presents with early onset corneal edema and, rarely, concomitant congenital glaucoma [16,32,46]. Homozygous missense variants cause approximately half of SLC4A11-related disease [95], and both families with congenital glaucoma whose variants have been reported had homozygous missense mutations; these were c.1343G>A p.(Gly448Asp) and c.2024A>C p.(Glu657Ala), respectively [32,46]. SLC4A11 localizes to the corneal endothelium, where it contributes to osmotic homeostasis of the cornea through its action as a transmembrane sodium-coupled hydroxyl transporter, and thus to maintenance of corneal transparency [95,96]. Slc4a11-knockout mice develop progressive corneal thickening with associated endothelial cell morphological changes and loss [97], but no groups have yet reported intraocular pressure in these murine models.

Part of the ADAMTS zinc-dependent metalloproteinase family, ADAMTS18 is expressed in adnexal, anterior segment and retinal cells during murine ocular development, with expression persisting postnatally in the lens and lacrimal gland [98,99]. It is a key factor in branching morphogenesis of the embryonic lacrimal gland [98], as well as in the development of the lung, breast and reproductive tract [24]. In relation to anterior segment development, it interacts with fibrillin-1 and is modulated by estrogen, raising the possibility of an interaction with CYP1B1 [24]. Pathogenic variants in ADAMTS18 cause microcornea, myopic chorioretinal atrophy, and telecanthus (MMCAT) [100] and Knobloch syndrome (high myopia, vitreoretinal degeneration, retinal detachment, cataract and occipital encephalocele) [101], but childhood glaucoma associated with microcornea has been reported in twins with compound heterozygous missense mutations of ADAMTS18 [c.3436C>T p.(Arg1146Trp) and c.1454T>G p.(Phe485Cys)] identified through panel testing [47]. Ocular defects seen in Adamts18 knockout mice include breaks in the lens capsule resulting in extrusion of material, but microphthalmia, microcornea and retinal anomalies were not seen [99]. However, morpholino oligonucleotide (MO) knockdown of adamts18 in zebrafish has been shown to cause macroscopic ocular defects [102].

The bicoid-type homeodomain transcription factors PITX2 and PITX3 have a role in embryonic anterior segment and lens development [103,104]. As discussed above, PITX2 interacts with FOXC1, and in mice, they are co-expressed in the periocular mesenchyme from E11.5 [73]. It is therefore unsurprising that the ocular phenotypes of PITX2-related disease are similar to those of FOXC1 and include childhood glaucoma associated with Axenfeld–Rieger anomaly [16,33] and Peters anomaly [13,16]. As with FOXC1, the majority of PITX2 variants have been identified in only one or two families, with the exception of the splicing variant c.253-11A>G and the intragenic variants c.191C>T p.(Pro64Leu), c.356del p.(Gln119Argfs*36) and c.475_476del p.(Leu159Valfs*39), which have all been reported in four or more families [15]. PITX3 is expressed in the lens placode at murine E10 and continues to be expressed in the lens throughout development, in addition to expression in the midbrain, tongue and other embryological craniofacial tissues [104]. The c.640_656dup p.(Gly220Profs*95) variant has been identified in two families with congenital glaucoma associated with microcornea and congenital cataracts [48,49].

FOXE3 variants are associated with a wide spectrum of anterior segment phenotypes and have been identified in patients with childhood glaucoma associated with Peters anomaly [50,51,52], congenital cataracts/aphakia [51,53] and microphthalmia [51,54]. The most common variants are c.21_24del p.(Met7Ilefs*216), c.244A>G p.(Met82Val) and c.720C>A p.(Cys240*) which have all been reported in four or more families [51]. The main role of the FOXE3 transcription factor during embryogenesis is in the development of the lens, where it is expressed in mice from E12.5 onwards and directly regulates the autophagy-associated protein DNAJB1 [50]. Functional studies of FOXE3 missense variants and small deletions isolated from patients with recessive disease showed significantly reduced transcriptional activity with homozygous or compound heterozygous variants, but no disease phenotype in their heterozygous parents, indicating that the presence of a single functional allele is sufficient for normal anterior segment development [53].

The SOX11 SRY-Box transcription factor is a close homolog of SOX2 [105], which is known to play a key role in ocular development from induction of the eye field at gestational day 21 onwards [77]. During Xenopus embryogenesis, sox11 is expressed in the optic vesicle, retina and anterior segment, and MO knockdown of sox11 results in decreased pax6 expression in the eye field, loss of retinal lamination and a microphthalmic phenotype [106]. In Sox11 knockout mice, Pax6 expression is unaffected, but BMP7 expression is reduced, and the phenotype includes lenticular defects, Peters anomaly and anterior ocular coloboma [107]. SOX11 variants, including c.251G>T p.(Gly84Val), have been reported in patients with congenital glaucoma associated with Peters anomaly and aniridia [36,57], although the gene is more commonly associated with intellectual developmental disorder with microcephaly with or without ocular malformations or hypogonadotropic hypogonadism (IDDMOH; Coffin Siris syndrome), of which congenital glaucoma can be a feature [28,57].

GJA8 encodes connexin 50 (Cx50), a gap junction protein of lens fiber cells which is necessary for maintaining lens transparency and cellular homeostasis [28,108]. Studies of the missense mutation c.217T>C p.(Ser73Pro) in transfected human lens epithelial cells have shown aggregation of the abnormal protein in the cytosol with loss of localization to the plasma membrane, resulting in reduced function of gap junction channels and increased apoptosis [108]. Although GJA8 variants are more commonly associated with congenital cataracts [11], the missense variants c.280G>C p.(Gly94Arg) and c.151G>A p.(Asp51Asn) have been identified in patients with congenital glaucoma associated with sclerocornea [28,33,36].

KERA encodes keratocan, a keratan sulfate proteoglycan of the corneal stroma with a central role in maintaining collagen spacing and thus corneal transparency through its interaction with keratan sulfate glycosaminoglycans and collagen fibrils [109,110]. In adults, it is localized to stromal keratocytes [109]; during embryological development, studies of chicks by Conrad [111] and Gealy [112] have shown expression of Kera in the periocular mesenchyme from E4.5, with subsequent localization to the anterior corneal stroma. Keratocan-deficient mice have a thinner cornea and narrower iridocorneal angle than wild-type mice, with larger collagen fibrils in a less regular arrangement [110]. Mutations in KERA are predominantly linked to cornea plana, in which there is flattening of the normal convex curvature of the cornea resulting in hyperopia [113], but one patient with anterior segment dysgenesis and congenital glaucoma associated with a homozygous nonsense KERA mutation c.937C>T p.(Arg313*) has been reported [39].

CDH2, KDM5C and TFAP2A are involved in neural development, with roles including axonal pathfinding [114], neuronal differentiation [115] and neural crest migration [116], respectively. All are reported to be causative of congenital glaucoma associated with Peters anomaly [58,59,60]. CDH2 encodes N-cadherin, part of the cadherin family of Ca²⁺-dependent transmembrane glycoproteins which regulate cell–cell adhesion in multiple tissues [58,114]. CDH2 is expressed in murine lens cells and the lens stalk at E10.75-11.0, and in lens and corneal endothelial cells at E17.5 [58]. In conjunction with CDH1 and CDH3, it controls separation of the lens vesicle from the surface ectoderm during anterior segment development; thus, its association with Peters anomaly is unsurprising [117]. Knockout of N-cadherin from the lens vesicle in mice results in a spectrum of anterior segment anomalies, including severe iris hyperplasia, microphthalmia and small, underdeveloped lenses with abnormal lens fiber development [117]. The missense variants c.485T>A p.(Val162Asp), c.1574A>G p.(Asp525Gly) and c.1807C>T p.(Pro603Ser), and splicing variant c.702+1G>A, have each been identified in one patient with Peters anomaly and congenital glaucoma [58].

KDM5C is a cause of X-linked syndromic intellectual developmental disorder, but the missense variant c.1204G>A p.(Asp402Asn) has been identified in one patient with Peters anomaly and congenital glaucoma [59]. Involved in transcriptional regulation through demethylation of H3K4 residues, studies of kdm5c in Xenopus have revealed that it is strongly expressed throughout neural and ocular development, particularly in the lens and retina [118]. MO knockdown studies in Xenopus have confirmed the essential role of kdm5c in ocular development, with loss of kdm5c resulting in microphthalmia, coloboma and abnormal retinal lamination [118].

In human eyes, expression of the transcription factor TFAP2A is seen in the anterior lens epithelium from gestational day 35 and in the equatorial lens epithelium, secondary lens fibers and neural retinal by day 54 [119]. Whilst the detailed mechanism by which TFAP2A acts has not yet been elucidated, conditional knockout of Tcfap2a in the lens placode of mice has been shown to result in failure of lens separation from the surface ectoderm with associated corneal anomalies, accompanied by loss of Foxe3 and Pitx3 expression from the lens stalk, suggesting that TFAP2A may be an upstream regulator of these genes as they pertain to lens separation [120]. This is consistent with the identification of a TFAP2A splicing variant (c.1025+2T>A) in a patient with Peters anomaly associated with congenital glaucoma [60].

Molecular-based therapies under development for glaucoma are predominantly targeted towards modification of the aqueous production/outflow pathways and minimizing damage to retinal ganglion cells (RGCs) [135], as these are key factors in the development of multifactorial adult-onset glaucoma. Sulak and colleagues [136] identified 153 studies into gene-based therapies for glaucoma and noted that there were twice as many therapies targeting RGC neuroprotection compared with those targeting aqueous production and outflow. Strategies to modulate the molecular pathways involved in aqueous regulation are in development, with targets including MYOC, AQP1 and matrix metalloproteinases (MMPs).

Mutations in the MYOC gene, encoding the extracellular matrix protein myocilin, are the most commonly identified single gene cause of primary open angle glaucoma (POAG) and juvenile open angle glaucoma (JOAG), identified in 2–4% and 10–30% of cases, respectively [135]. Dominant gain-of-function mutations result in intracellular accumulation of the abnormal protein, resulting in increasing intraocular pressure (IOP) and resultant glaucomatous damage to the retinal ganglion cells and optic nerve [127]. Mice with the MYOC p.(Tyr437His) variant (Tg-MYOC^Y437H) can be used as a model for POAG, developing these same pathogenic features, and may be employed for in vivo confirmation of the therapeutic benefit [127,129].

Gene editing approaches targeted at specific myocilin mutations are in development. Jain [127] used a clustered regularly interspaced palindromic repeats (CRISPR)-CRISPR-associated protein 9 (Cas9) approach with an adenoviral Ad5 vector to introduce a truncating frameshift mutation into Tg-MYOC^Y437H mice. They demonstrated significant reduction in IOP into the normal range in both younger (age ≤ 1 month) and older (age ≥ 9 months) mice, with results sustained for at least 2 months post-injection. The group have subsequently shown positive results using the same CRISPR-Cas9 system with a lentiviral vector [129], with IOP reduction into the normal range in Tg-MYOC^Y437H mice maintained for at least 6 weeks post-injection. They additionally tested an AAV2 vector in vitro on human TM3 cells with p.(Tyr437His) or p.(Gly364Val) MYOC mutations, but the reduction in accumulated myocilin was only 34% lower than untreated levels with the AAV2 vector compared with 62% reduction for the lentivirus. A non-randomized interventional trial of a BD113 virus-like particle, a lentiviral-based CRISPR-Cas9 gene therapy targeted to knock down MYOC, has commenced recruitment of adults with myocilin-related POAG refractory to initial treatment and will primarily assess IOP reduction following a single intracameral injection [137]. Another mechanism that has been proposed for reducing the pathogenicity of MYOC mutations is to use small interfering RNAs (siRNAs) to disrupt the expression of the gene. The preliminary results in TM5 cells demonstrated significant suppression of both wild-type and mutant myocilin expression to 60% of control levels [138]. Whilst further work is needed, these approaches hold promise for potential gene-targeted therapies for MYOC-related glaucoma to become available in the future.

Another CRISPR-based approach employing a lentivirus vector carrying single guide RNA (sgRNA) to target the TGFβ2 promoter has been developed by Mao's group [139]. Administration of the sgRNA lentivirus 2.5 weeks prior to induction of ocular hypertension prevented IOP increases, maintaining an IOP level below 20 mmHg for at least 8 weeks post-injection, compared with an increase to over 30 mmHg in controls. AQP1, which encodes aquaporin 1, has also been targeted with sgRNA. This transmembrane water channel is present in many tissues including the ciliary body, and deletion of Aqp1 has been shown to significantly reduce IOP in mice [140]. Introduction of two AQP1-indel-inducing sgRNAs within a single AAV vector given by intravitreal injection has been shown to transduce the murine ciliary body and reduce IOP by 22% after 3 weeks, with neuroprotective effects sustained to at least 7 weeks post-injection [141]. There was also effective transfection of ex vivo human ciliary body epithelial cells. AQP1 has therefore been proposed as a potential therapeutic target for glaucoma.

MMP activity is known to be reduced in glaucoma, and one of the key actions of the prostaglandin analog drops used to treat open angle glaucoma is to increase MMP expression in the ciliary body [142]. Using an AAV9 vector to transduce corneal endothelial cells, O'Callaghan [143] introduced tetracycline-inducible murine MMP3 into Tg-MYOC^Y437H and dexamethasone-induced ocular hypertensive mice. In both groups, IOP 2 weeks after doxycycline induction was significantly reduced in the treated mice compared with controls. They also demonstrated significant increases in MMP3 expression and aqueous outflow in non-human primates in vivo, and increased aqueous outflow in human donor anterior segments in vitro using this approach [143]. Similar effects have been achieved using intracameral IκBα-silencing siRNA to transfect the anterior segment of nonhuman primates, with upregulation of MMP2 and MMP9 and resultant 49% reduction in IOP [144]. A lesser IOP reduction has been achieved through the use of the microRNA miR-21-5p, which is predicted to have downstream effects including the activation of MMP9. Whilst preliminary results show potential, the maximal IOP reduction following administration was 17.8% [145], lower than the 20–35% achieved by applying current topical treatments [146].

Novel gene-targeted approaches acting on established therapeutic pathways are also under development. Rho kinase inhibitors act directly on the trabecular meshwork to inhibit phosphorylation of proteins including myosin light chains, which reduces trabecular cell adhesion and resistance and promotes phagocytosis of extracellular debris by trabecular reticulum cells, thus increasing aqueous outflow and reducing IOP [147]. Current commercial formulations are administered topically, but a Phase 1 study of GVB-2001 gene therapy in adults with POAG is open to recruitment [148]. GVB-2001 consists of a human RhoA gene with a dominant negative mutation carried in an AAV vector, an approach that has been successful in transducing the trabecular meshwork and reducing nocturnal IOP rises in rats through inhibition of the Rho pathway [149].

Targeted disruption of β2-adrenoceptors has also shown promise. Loma [146] topically administered siRNAs designed to silence β2-adrenoceptors to rabbits and compared the effects to commercially available topical dorzolamide, latanoprost and timolol. The siRNAs significantly reduced IOP by 30 ± 5% compared to controls, with effects detectable 24 h after instillation and sustained for 5 days. The level of IOP reduction was comparable to that achieved by all commercial agents, but with a longer onset and offset time. A commercial compound, bamosiran, has recently completed its phase II trial. This RNA interference approach designed to prevent β2-adrenoceptor production showed non-inferiority to timolol in patients with a baseline IOP ≥ 25 mmHg, but not across all patient groups [150]. However, the positive results achieved overall by using these diverse approaches to molecular modulation of aqueous regulation lay a strong foundation for molecular therapies to become available for patients within the near future.

A novel approach to the treatment of aniridia is nonsense suppression therapy, using drugs that impair recognition of premature termination codons (PTCs) during translation to allow the production of a full-length protein [151]. In aniridia patients in whom a genetic cause is identified, dominant pathogenic PAX6 mutations are responsible for approximately 90% of cases [14]; of these, nonsense and frameshift mutations resulting in a PTC comprise 70% of pathogenic and likely pathogenic PAX6 variants [82], so a treatment that targets these mutations has the potential to benefit a large proportion of these patients. Promising results using nonsense suppression therapies, which work on in-frame nonsense mutations, have been reported in vitro [152] and in the Sey mouse model [153,154]. Sey mice with a heterozygous PAX6 p.(Gly194*) nonsense mutation treated with topical nonsense suppression drugs from postnatal day 14–60 developed partial phenotypic rescue, including reversal of corneal, lens and retinal abnormalities [153,154]. However, these benefits have not yet been translated into positive results in clinical trials. At completion, the Study of Ataluren in Participants With Nonsense Mutation Aniridia (STAR) unfortunately failed to meet its primary endpoint, with no significant difference in change in reading speed at 48 weeks between the treatment and placebo groups [155]. STAR used an oral formulation of ataluren; a topical alternative has been developed with proven physical stability and microbiological sterility over 60 days [156] which may have better ocular bioavailability and improved efficacy, and the outcomes from these assessments are awaited.

Gene editing is under investigation as another potential avenue of treatment for aniridia. Work in Sey mice has shown successful germline correction of the PAX6 p.(Gly194*) mutation in vivo via microinjection of the targeted CRISPR-Cas9 complex into murine zygotes at day 0.5 post-coitus, with inheritance of the corrected gene by the offspring of these transgenic mice [157]. The progeny had normal eye development and normal PAX6 levels in neural tissue at embryonic day 18.5, when microphthalmic changes are observed in Sey mice [157]. Further work has demonstrated the potential use of AAV vectors for delivery of gene therapy in aniridia, with results suggesting that the AAV9 capsid is able to safely transduce corneal and retinal cells of Sey mice [126].

Work to expand the therapeutic options for management of aniridia-associated keratopathy has shown that duloxetine and ritanserin, both anti-serotonin drugs which inhibit the MEK/ERK (mitogen-activated protein kinase-ERK kinase/extracellular signal related kinase) signaling pathway, can enhance endogenous PAX6 expression in haploinsufficient corneal limbal epithelial stem cells [158,159,160]. A regulatory microRNA, miR-204-5p, has also been reported that upregulates PAX6 in PAX6-knockdown corneal limbal stem cells and mature differentiated corneal epithelial cells in vitro; there was no significant change in PAX6 expression in primary human limbal epithelial cells, human limbal stem cells or in the corneas of Pax6^(Sey/+) mice transfected with miR-204-5p [161]. Further work is needed to determine the effects of these modulating factors in other ocular cell types, but they provide another potential avenue for therapeutic development. However, any strategies to increase PAX6 expression during embryogenesis need to ensure that induced levels remain carefully modulated. Both under- and overexpression of PAX6 can lead to ocular developmental anomalies as discussed earlier [80,81]; therefore, ensuring that corrected PAX6 expression levels are as close as possible to the normal embryological state is essential when correcting pathogenic variants. To address this, Simpson's team [162] have designed PAX6 MiniPromoters and reported their robust and specific expression within the PAX6-positive RGC, amacrine, horizontal and Müller glial cells, with significantly better regulation of PAX6 expression than achieved by the ubiquitous smCBA promoter.

The anterior segment dysgeneses are congenital conditions with phenotypic features resulting from abnormal oculogenesis. Consideration therefore needs to be given to the challenges presented when administering any molecular therapies to a pediatric population and determination of trial endpoints suited to this group.

The first challenge is confirming the genetic diagnosis. Whilst molecular diagnostic rates for aniridia now reach 90–98% [163,164], rates for other ASDs remain much lower at less than 25% [12,23]. For gene-based therapies, accurate molecular diagnosis is essential, particularly given the overlapping phenotypes caused by many of the genes listed above. Secondly, these are rare diseases; therefore, identification of a sufficient number of patients for clinical trials may require collaboration across multiple centers and potentially multiple countries. Local and national rare disease registries and collaboration with charitable networks can aid in this.

Trial endpoints need to be carefully considered. Appropriate functional tests of acuity and fields vary with age, and it may be difficult to acquire quantitative results [165]. For IOP testing, Goldmann applanation tonometry remains the gold standard [166] but requires the patient to sit at the slit lamp. IOP testing with handheld devices such as the iCare tonometer can allow for monitoring in children of all ages, but their accuracy at pressures above 23 mmHg is reduced [166]; therefore, their use in trials of patients with childhood glaucoma needs to be carefully considered. For ASDs, monitoring of phenotypic changes may be necessary, and in some children, this may require examination under anesthesia, which should be incorporated into routine clinical care wherever possible.

Finally, there are the challenges inherent in the gene therapies themselves. Intracameral administration of therapies targeting the corneal endothelium and trabecular meshwork negates the need for corneal penetrance required by topical formulations and reduces the potential for systemic off-target effects, but this method of delivery to pediatric patients requires general anesthesia and introduces the potential for intraocular infection. Adenoviral vectors can provoke immune responses and inflammation whereas AAVs and lentiviruses rarely elicit these reactions [125]. Although less likely than with retroviruses, concerns have been raised regarding the potential for lentiviral vectors to cause insertional mutagenesis, or for additional mutagenic effects to occur with CRISPR-Cas9 delivery systems [125], but Jain [127] identified no alternative myocilin isoforms following treatment with a CRISPR-Cas9 approach. Studies of transduction of target cells in vitro have been promising, with up to 70% transduction efficacy of human trabecular meshwork cells by an Ad5 vector using CRISPR-Cas9 to target mutant MYOC, and reduction in IOP using this therapy in Tg-MYOC^Y437H mice [127]. However, exposure to AAVs in childhood is common and therefore pre-existing circulating antibodies may reduce transduction efficacy in human subjects [165].

Future research directions in this field should focus on both targeting the underlying biological pathways affected and on improving the delivery methods in the pediatric population. To date, aniridia and adult-onset glaucoma have been the primary foci for therapeutic development, but the approaches developed in these fields have the potential to be applicable to other anterior segment dysgeneses with a similar genetic basis. For example, 13% of known pathogenic variants of CYP1B1 are nonsense mutations, but these account for 21% of CYP1B1-related anterior segment dysgenesis. Studies of nonsense suppression therapies in aniridia have shown promising results, with partial phenotypic rescue in mice treated post-partum; similar trials in animal models of other nonsense-mediated anterior segment dysgeneses are an avenue that warrants further exploration. Likewise, whilst primary congenital glaucoma is thought to be predominantly driven by iridotrabeculodysgenesis and changes in Schlemm canal development, the juvenile-onset glaucoma seen in aniridia and Axenfeld–Rieger syndrome may be amenable to modulation of TGFβ2, AQP1 or MMPs. If these treatments progress smoothly through Phase 1 and 2 trials in adult-onset glaucoma, trials in glaucoma secondary to anterior segment dysgeneses should be considered.