Background
Primary congenital glaucoma (PCG, OMIM 231300) is the most common type of childhood glaucoma due to the abnormal development of structures in the anterior segment of the eye including the trabecular meshwork, Schlemm’s canal and the anterior chamber angle. It manifests during neonatal or early infantile period (before 3 years of age) and is characterized by elevated intraocular pressure (IOP), increased corneal diameter, enlarged globe, Haab’s striae, corneal edema, and optic nerve head cupping. Symptoms are photophobia, ephiphora and blepharospasm [
1]. PCG represents a diagnostic and therapeutic challenge as it can lead to irreversible blindness in the first years of life if untreated.
Inheritance is primarily autosomal recessive, although pedigrees with dominant inheritance or sporadic cases have been described [
2]. Several chromosomal loci have been so far mapped for the recessive form of PCG, but up to now only two genes have been identified: cytochrome P4501 subfamily 1B1 (CYP1B1, OMIM 601771) on the GLC3A locus and latent transforming growth factor beta binding protein 2 (
LTBP2, OMIM 602091) on the GLC3D locus.
CYP1B1 is the most common identifiable cause of PCG worldwide. In the European population the prevalence of
CYP1B1 mutations ranges from 20 to 30 % of all PCG cases [
3]. In contrast,
LTBP2 mutations in classical cases of PCG are much rarer, being reported only for a few cases from Pakistan and in patients of Gypsy ethnicity [
4].
Mutations in
CYP1B1 can infrequently underlie the autosomal dominant, juvenile open-angle glaucoma (JOAG, OMIM 137750) and even adult-onset forms of primary open-angle glaucoma (POAG, OMIM 137760) [
5]. However, across most populations, the most common identifiable cause of JOAG remains heterozygous mutations in myocilin gene (
MYOC, OMIM 601652), underlying up to 7–15 % of cases. JOAG manifests in the first decades of life and is characterized by elevated IOP, progressive glaucomatous optic neuropathy, severe visual field defects (VFD) and is frequently associated with severe myopia [
6].
A further autosomal dominant form of infantile/developmental glaucoma is a phenotypic aspect of ocular anterior segment dysgenesis (ASD), a genetically heterogeneous group of complex developmental disorders including Axenfeld-Rieger’s anomaly, Peters’ anomaly, aniridia, iris hypoplasia and iridogoniodysgenesis [
7]. Approximately 50 % of individuals with ASD generally develops glaucoma due to malformations of tissues responsible for the IOP regulation and aqueous humor drainage including the iris, cornea, lens, Schlemm’s canal and trabecular meshwork (TM) [
7,
8]. Due to the malformations in Schlemm’s canal and TM, PCG is sometimes grouped together with the ASD disorders. In ASD glaucoma may develop during childhood (developmental glaucoma), but it is more common during adolescence or at the beginning of adulthood (early-onset glaucoma). Glaucoma associated with ASD progresses rapidly, is difficult to manage and may result in severe damage of the optic disc and visual field [
9]. To date mutations in Forkhead Box C1 (
FOXC1, OMIM 601090), Paired-Like Homeodomain transcription factor 2 (
PITX2, OMIM 601542), and Paired Box Gene 6 (
PAX6, OMIM 607108) genes are the most common cause of glaucoma in ASD [
8]. Interestingly, mutations in
CYP1B1 are also associated with rare cases of Peters’ anomaly.
Currently, mutations in six genes,
CYP1B1,
LTBP2,
MYOC,
PITX2,
FOXC1 and
PAX6, can explain only a fraction of all congenital/infantile/early-onset glaucoma cases worldwide suggesting the involvement of other candidates [
10].
The identification of disease-causing variants in known or novel genes in children with glaucoma can have a significant impact on establishing proper diagnosis, disease risk assessment and clinical care. In fact, variable expressivity, phenotypic overlap and limited follow-up among these different early-onset glaucoma forms have often led to an incorrect or delayed definitive diagnosis. Moreover subsequent treatment may be inadequate because of advanced disease. As a consequence childhood glaucoma still causes a disproportionately high percentage of childhood blindness worldwide. Hence, an early and reliable diagnosis is essential to prevent unwanted vision loss and also to reduce the burden of childhood blindness [
11].
Whole exome sequencing has been demonstrated to be highly successful in identifying disease-causing variants in rare ocular diseases [
12,
13]. Thus, we decided to apply this approach to selected cases of congenital/infantile glaucoma presenting no disease-causing variants in the known associated genes.
Here, we report on the novel association of compound heterozygous variants in collagen type I alpha 1 gene (COL1A1, OMIM 120150) in one patient diagnosed with PCG and retinal detachment. Furthermore, three COL1A1 heterozygous variants were identified in three patients: two with an early onset glaucoma form and one with congenital glaucoma, two of them presenting also an early-onset cataract and mild form of osteogenesis imperfecta (OI). In addition, we provide protein modeling based evidence for the pathogenicity of the variants identified.
Altogether, these findings expand the role of
COL1A1 in different forms of developmental/early onset glaucoma and show that disease-causing variants in
COL1A1 may act also as recessive alleles confirming data of previous mouse models [
14]. In addition, our results support and double the number of patients in the literature with glaucoma observed in osteogenesis imperfecta (OI) [
15] and serve to advise that glaucoma is also an important complication of OI.
Discussion
The collagen alpha-1(I) chain protein belongs to the collagen complex superfamily in which each component has a specific function, or set of functions, and there are extensive interactions with other connective tissue components [
46]. Thus the clinical phenotypes resulting from collagen mutations are wide-ranging in their manifestations and severity [
47]. Of the 28 known vertebrate collagen types, type I collagen is the most abundant and widely expressed collagen in humans. It is a heterotrimer comprising two alpha 1 (I) chains and one alpha 2 (I) chain. The alpha 1 (I) and alpha 2 (I) chains of type I collagen are encoded at the unlinked loci
COL1A1 and
COL1A2, respectively [
48]. The most striking feature of these alpha chains is that they consist of repeating Gly-Xaa-Yaa tri-peptides motifs. The presence of glycine every third amino acid is essential to allow the alpha chain to adopt the characteristic collagen triple helix. The mutation profiles of these genes are not restricted to any specific region but are scattered throughout the entire structural domains and show enormous diversity. So far, more than 800 mutations have been reported only for
COL1A1 [
45], mainly associated with skeletal and dermatological conditions such as OI, Ehlers-Danlos syndrome (EDS), bone mineral density variation, osteoporosis and Caffey disease. However, in the literature there are different descriptions in which both OI and EDS patients present also a variety of ocular abnormalities. Characteristic ocular findings described are usually blue sclera, thin cornea, microcornea, myopia, keratoconus, congenital absence of Bowman’s layer, retinal detachment, glaucoma [
49‐
51]. This is not surprising since up to 80 % of eye tissues are composed of different collagen proteins, and the most abundant is the type I collagen, in particular in cornea, sclera, iris, ciliary body, trabecular meshwork and optic nerve [
52]. Additionally, genome wide association studies (GWAS) have shown that variants at collagen-related genes, including
COL1A1, influence one of the main glaucoma risk factor such as central corneal thickness (CCT) [
53,
54].
Mutations in certain collagen genes have also been associated with glaucoma manifesting as part of a systemic disorder. Stickler syndrome (STL, OMIM 108300), for example, is a group of diseases primarily caused by mutations in the fibril-forming collagen type II gene. Mutations causing premature stop codons in exon 2 of
COL2A1 lead to ocular-only phenotypes including retinal detachment and high myopia, but with few or no other systemic manifestations [
55]. A recent GWAS study found significant association of a single nucleotide polymorphism in
COL11A1 with primary angle-closure glaucoma [
56]. Further, genetic variants in collagen XV, alpha 1 (
COL15A1) and collagen XVIII, alpha 1 (
COL18A1) have been shown to modify the age of onset of both JOAG and POAG [
57].
Collagen proteins are also a pivotal component of the extracellular matrix (ECM) of the TM, Schlemm’s canal (SC) and lamina cribrosa (LC), which represent the ocular tissues involved in glaucoma development. Morphological and ultrastructural ECM changes involving elastic-fibers and microfibrils have been already implicated in glaucoma pathogenesis [
58]. In addition, mutations in multiple genes encoding elastic microfibril components have been linked to glaucoma [
59]. For example, fibrillin-1 (
FBN1) mutations cause Marfan syndrome (OMIM 154700) and ocular abnormalities including ectopia lentis, myopia and glaucoma [
49,
60]. Mutations in
LTBP2, which binds to fibrillin-containing microfibrils, cause not only PCG, but also POAG and pseudoexfoliation glaucoma [
61,
62]. Altogether, these observations underline the importance of collagen genes, including
COL1A1, in the eye and their putative role in glaucoma pathogenesis. Nevertheless, except for a recent report by Wallace et al., [
15] there is no detailed documentation about the association of
COL1A1 mutations and glaucoma with and/ or without sign of OI and EDS.
In the present study we describe
COL1A1 variants in four patients associated with different glaucoma forms: two presenting with PCG, two with JOAG and/or an early form of POAG. Both patients with early form of open angle glaucoma present also a diagnosis of OI: in particular a mild OI type I in the MZ-2 patient, while the third patient CA-3 has a more severe but not classified OI form (Table
2).
Of particular interest is the novel finding of
COL1A1 compound heterozygous variants (p.(Met264Leu) and p.(Ala1083Thr)) in the patient MI-1 presenting a novel phenotype of PCG, retinal detachment and light ligaments laxity (Table
2). Molecular modelling suggests that these two amino acid changes impair the collagen protein complex interaction with two different binding proteins: Hsp47 and fibronectin. Thus, the mechanism of these two variants is different from the one of the most frequently reported missense variants causing mainly OI and EDS that is usually dominant and affects the formation and stabilization of the collagen alpha-1(I) chain triple-helix itself similarly to the case of our three other patients described (Table
2). Instead, in the recessive mode the collagen alpha-1(I) chain protein interactions with binding partners seem to be affected. Hsp47 is important for the proper assembly of the triple-helical procollagen molecules and has an important role in eye morphogenesis [
38]. Fibronectin has been found to be expressed in TM, ciliary body, choroid, basement membrane of the corneal epithelium, corneal stroma and Descemet’s membrane. It has been shown to be essential in cornea morphogenesis and in stabilizing the vitreoretinal attachment [
63]. Impaired binding of collagen protein to Hsp47 and fibronectin could thus affect the correct formation of eye tissues important for outflow, IOP regulation and retinal attachment. Thus our results suggest that concomitant impairment of both copies of COL1A1 with respective binding proteins could influence at a very early stage the development of the eye more than other tissues. This might explain the more severe ocular phenotype observed in patient MI-1 compared to the other patients carrying variants with dominant effect.
Although, so far
COL1A1 gene mutations in humans are overwhelmingly dominant in their action, there is a single example of a recessively-inherited case of OI caused by a homozygous missense mutation in
COL1A2 gene [
64]. Remarkably, three studies on transgenic mice with targeted missense mutations in
COL1A1 (not-affecting positions of amino acid Gly in the triple-helices) revealed a glaucoma-like phenotype supporting the causality of the variants identified in our patient. The mice developed sustained elevation of IOP, progressive optic nerve damage and showed a reduced outflow facility. In addition, they do not present typical clinical signs of OI and EDS like in our first patient. Further, the heterozygous mice carriers have been reported to be without any signs of OI and glaucoma similar phenotype as the parents of our MI-1 patient analyzed [
14,
65,
66].
The p.(Arg253*) identified in patient MZ-2 introduces a premature stop codon in the
COL1A1 gene coding sequence. This mutation has been previously reported in a 13 year old patient diagnosed with a mild form of OI and blue sclera [
67]. Glaucoma was not described for that patient maybe due to the earlier age (13 years) compared to the onset age of 29 years in our MZ-2 patient with JOAG and mild OI form. A large number of frameshifts/nonsense mutations in this gene have been reported to be associated with the mild OI type I, following classical dominant inheritance. Frameshift and nonsense sequence variants usually result in nonsense mediated decay (NMD) of affected transcripts leading to a reduced levels of mRNA. In practice, this means that little or no truncated protein product is formed resulting in the production of about half the amount of normal type I collagen [
68]. Decreased amount of collagen alpha-1(I) chain protein in the eye could thus explain the thin and abnormal sclera and the glaucoma detected in our patient MZ-2.
The mutation identified in patient CA-3 and TU-4 removes a glycine (Gly) at position 154 and 767 introducing a valine (Val) and serine (Ser) respectively. Mutations affecting glycine position which produce an abnormal collagen protein are usually reported in more severe phenotypes than nonsense do. This could explain the presence of a severe OI form in the patient CA-3 in which bone deformities started already in adolescence and the presence of a congenital glaucoma in the fourth patient identified. Clinical signs of OI or EDS for this last patient, TU-4, were not yet reported maybe due to the early age of the patient’s glaucoma diagnosis (4 years).
We suggest that the range of
COL1A1 mutations presenting as either dominant or recessive differentially affect the collagen alpha-1(I) chain protein function resulting in a spectrum of connective tissue related phenotypes of varying severity including glaucoma features as recently proposed [
15]. Nevertheless, we cannot exclude the possible presence of unidentified genetic variants located outside the regions covered by WES, acting in concert with
COL1A1 to modulate the phenotype.
One of the possible limitations of this study due to the rareness of these combined phenotypes could be the small number of affected people screened. A further limitation might be also the absence of unaffected relatives in the three patients, carriers of single heterozygous variants, to prove the dominant inheritance and penetrance of these variants. However, taking into account that these variants are extremely rare (Table
2) and that they show a putative impairing effect on the protein function (Fig.
4), we can suggest that these are probably pathogenic. Undoubtedly, it is necessary widening the screening of
COL1A1 in different, larger cohort of patients with glaucoma and additionally also OI to replicate and confirm these findings.
Abbreviations
ASD, anterior segment dysgenesis; BCVA, best correct visual acuity; CYP1B1, cytochrome P4501 subfamily 1B1; COL1A1, collagen type I alpha 1 gene; EDS, Ehlers-Danlos syndrome; FOXC1, forkhead Box C1 gene; Hsp47, heat-shock protein 47; IOP, intraocular pressure; GLC3A, GLC3D etc. primary congenital glaucoma locus A, D; JOAG, juvenile open angle glaucoma; LTBP2, latent transforming growth factor beta binding protein 2; MYOC, myocillin gene; MLPA, multiplex ligation-dependent probe amplification; OI, osteogenesis imperfecta; PAX6, Paired Box Gene 6; PCG, primary congenital glaucoma; PCR, polymerase chain reaction; PDB, protein database; POAG, primary open angle glaucoma; PITX2, paired-like homeodomain transcription factor 2; WES, whole exome sequencing
Acknowledgments
The authors thank all the patients and family members for participating in this study and for providing all the available clinical data. We thank Olga Zwenger for special assistance with sequencing. We would like to thank Professor emeritus Eugen Gramer from the University Eye Hospital in Wuerzburg for the clinical characterization of the 24 German PCG patients included in this study.