1. Introduction
CTNNB1 Syndrome is a severe autosomal dominant neurodevelopmental disorder usually caused by de novo loss-of-function mutations in the
CTNNB1 (Cadherin-associated protein, beta 1) gene [
1]. CTNNB1 Syndrome manifests itself in a variety of developmental disorders including Neurodevelopmental Disorder with Spastic Diplegia and Visual Defects (NEDSDV), and visual disorders including Familial Exudative Vitreoretinopathy (FEVR). NEDSDV is a neurodevelopmental disorder characterized by global developmental delay, impaired intellectual development with absent or very limited speech, craniofacial anomalies and microencephaly, axial hypotonia, and spasticity [
1,
2]. FEVR is an autosomal dominant disorder characterized by incomplete development of the retinal vasculature [
3]. De novo loss-of-function mutations in the
CTNNB1 gene were first discovered in 2012 after diagnostic exome sequencing of individuals with severe intellectual disability [
4], and since then the term CTNNB1 Syndrome has become the generic term for all disorders associated with
CTNNB1 haploinsufficiency. Currently, this disorder is diagnosed in approximately 300 patients worldwide, although this number is likely an underestimation due to misdiagnosis in cerebral palsy [
5,
6], leading to efforts to reevaluate the diagnoses of cerebral palsy patients to enable genomics-based changes in their clinical care.
The
CTNNB1 gene is located on chromosome 3 (locus 3p22.1, 41240942–41281939). It consists of 16 exons, with exons 2–15 (2346 bp) providing the coding sequence for β-catenin protein. β-catenin protein consists of 781 amino acids and belongs to the armadillo family of structural proteins involved in both embryonic development and adult homeostasis where it plays two essential roles: (1) as a transcriptional co-factor in the Wnt-signaling pathway, and (2) as an anchor in intracellular contacts and cell adhesion [
7]. When the Wnt pathway is not stimulated, most of the newly expressed β-catenin is depleted from the cytoplasm by the destruction complex, while the remaining undergraded β-catenin engages with E-cadherin and α-catenin in membrane complexes that serve as cellular anchors. Within the destruction complex, which consists of Axin, Adenomatous Polyposis Coli (APC), and CK140 proteins, β-catenin undergoes a series of consecutive phosphorylations by the glycogen synthase kinase 3β (GSK3β) and CK1 kinases, which ultimately leads to β-catenin ubiquitination by β-TrCP and its proteasomal degradation. In the course of the canonical Wnt signaling pathway, Wnt ligands bind the membrane Frizzled family receptor that stimulates Dishevelled protein to sequester destruction complex proteins [
8]. In this way, the degradation of β-catenin is inhibited, allowing the accumulation of free β-catenin which, transported to the nucleus, assists the T-cell factor/lymphoid enhancer factor (TCF-LEF) family of transcription factors in the transcription of various developmental genes, such as axin 1 and cyclin D. Structural and signaling roles of β-catenin are mutually exclusive, which is reflected in its protein structure. β-catenin consists of three regions with a distinct charge distribution: (1) an unstructured N-terminal region (130 amino acids), bearing amino-acid residues important for β-catenin degradation (S33, S37, Y41, S45); (2) a highly conserved central core region (550 amino acids) consisting of 12 armadillo repeats (each is a 42 amino-acid triple helix) [
9,
10] that form a positively charged groove [
9], where β-catenin interacts with more than 20 protein partners including E-cadherin, TCF and degradation complex proteins [
10,
11,
12]; and (3), the unstructured C-terminal region (100 amino acids), which is believed to enhance β-catenin stability by shielding the N-terminus from the destruction complex [
9,
10,
13]. The molecular mechanism of binding exclusivity for the various β-catenin partners remains elusive—it is thought that the occlusion of ligand binding may be achieved by back-folding of termini.
Given the low prevalence of CTNNB1 Syndrome and its relatively recent discovery, little is known about the effect of
CTNNB1 mutation type and exonic localization on the severity of clinical phenotypes. It is also not clear whether
CTNNB1 mutations are null (in which case the mutated transcripts undergo nonsense-mediated RNA decay—NMD [
14], or are translated into non-functional proteins) or, on the other hand, cause a partial loss of protein function due to the presence of an incompletely functioning protein. Another type of mutation can lead to the expression of proteins that interfere with the normal function of the protein from the wild type allele. These so-called antimorphic or dominant-negative mutations (mutated transcripts escape NMD and translate into truncated variants with potentially deleterious effects on the function of the healthy allele) are rare; however, given the variability of CTNNB1 Syndrome-associated mutations in terms of type and location, a production of auto-inhibitory truncated variants cannot be ruled out.
De novo mutations of the
CTNNB1 gene have been associated with neurodevelopmental disorders, with cases of intellectual disability and speech delay [
4]. Addressing these open questions through phenotype–genotype correlation studies is essential in order to develop targeted interventions and focused clinical care, specific to the mutational context in the affected individuals [
15]. The availability of data for such studies has been aided by genomic microarray technology, which has tremendously changed diagnostic approaches in children with neurodevelopmental disorders. Genetic testing can identify the genetic etiology in approximately 40% of cases of cerebral palsy (CP) cases, particularly those diagnosed with autism spectrum disorders (ASD) and intellectual disability (ID) with no apparent causative factor related to CP [
5,
16]. Access to a large number of patients who have been reliably and systematically assessed is fundamental for understanding CTNNB1 Syndrome. For the first time, this study provides a systematic review of previously reported cases in which we a) analyze the prevalence of clinical manifestations, and b) classify mutations according to their type (missense/nonsense/frameshift/splicing), exonic location, associated clinical features, and disease severity. Based on the analysis of the collected data, the genotype–phenotype correlations for CTNNB1 Syndrome are explored in detail. These may serve as a classification standard for new case studies and as a reference for researchers working to develop personalized therapeutic approaches.
4. Discussion
The present paper provides a comprehensive and up-to-date review of published cases of CTNNB1 Syndrome, an analysis of the prevalence of the most common symptoms, and a classification of
CTNNB1-associated mutations according to the severity of their respective phenotypes (
Table 5,
Table 6 and
Table S1, Supplement A). Based on the availability of sufficient data, 35 patients were included in the analysis. While we acknowledge that the size of the analyzed cohort was too small to perform a statistically significant genotype–phenotype correlation, we believe that the classification performed in our study may nevertheless be informative for future correlation studies and provide the classification basis for further data collection and analysis.
The main finding of this paper is that there is substantial variability within genotypes and phenotypes of patients with CTNNB1 Syndrome. Regarding genotype, we found that mutations associated with CTNNB1 Syndrome are scattered throughout the coding sequence of the gene, with the exception of the first coding exon (exon 2), although we cannot exclude the possibility that pathological mutations also occur in this exon. In terms of phenotype, we were able to classify patients into a spectrum of disease severity (severe, severe-moderate, moderate, mild, and normal). Our analysis of the mutations available in each phenotype category suggested certain relationship between phenotype severity and mutation location and type. The majority of the mutations analyzed were associated with moderate or severe disease phenotypes, manifested by facial dysmorphisms, microcephaly, various motor disabilities, speech and cognitive impairments, and behavioral difficulties. From a biochemical perspective, this was expected because the critical interaction surface of β-catenin is large, extending from armadillo repeat 3 to 9 encoded in exons 5 to 10, respectively (
Figure 3,
Figure 4,
Figure 5 and
Figure 6).
Consistent with these expectations, the analyzed nonsense and missense mutations occurring in exons 14 and 15, which are part of the C-terminal domain of β-catenin, were classified in the normal clinical outcome as they presented with eye abnormalities only, and otherwise had a normal phenotype. Both mutations were found in all family members, which suggests that they were inherited rather than occurred de novo [
35]. Interestingly, a classic ophthalmological feature of CTNNB1 Syndrome—FEVR—was found in many of these patients (
Table S1, Supplement A), suggesting that it may be caused by alterations at the uncharacterized C-terminal domain. Biochemically, nonsense mutations impose a premature stop codon that results in a truncated protein, while missense mutations lead to the substitution of a single amino-acid in the protein sequence, in which case the severity of the consequences depend on the structural integrity of the mutant protein and possible disruption of binding sites for interacting proteins. Lack of disease severity for mutations in exons 14–15 is most likely attributed to the fact that a large part of the protein, should the transcript circumvent nonsense-mediated decay, is transcribed and is likely to be, at least partially, functional. Such hypomorphic mutations have also been observed in other genes (e.g., DMD, APC) [
45,
46]. Furthermore, helix C, a critical structural motif of the C-terminal domain with a role in the co-transcriptional activity of β-catenin [
47], is encoded by exon 13; therefore, it lies upstream of mutations in exons 14–15. Accordingly, patients with mutations in exon 13 exhibit an array of additional cognitive and motor impairments (
Table 5). One of the reported nonsense mutations in exon 13 is a mutation of a tyrosine residue at position 654 (Y654;
Table S1, Supplement A), which, in its phosphorylated state, is directly linked to stabilizing helix C, and thus allows accessibility to co-activators of β-catenin-mediated transcription leading to nuclear localization of β-catenin [
48]. However, the mutations in exon 13 (c.2038_2041dup, p.Ser681* and c.1981C > T, p.Arg661*) were at worst classified into the mild-moderate phenotype category, which can be attributed to the fact that, while important for signaling, helix C appears to be completely dispensable for the structural role of β-catenin in cell-cell adhesion [
9].
An additional explanation for the milder phenotype in mutations in exons 14–15 could be that helix C, which is the most important feature of the C-terminal domain, critical for the co-transcriptional role of β-catenin [
47], is encoded upstream, in exon 13. Accordingly, patients with mutations in this region exhibit a range of additional cognitive and motor impairments (
Table 5). One of the reported nonsense mutations in exon 13 occurs at the Y654 residue (
Table S1, Supplement A), which, in its phosphorylated state, is directly linked to keeping helix C accessible to co-activators of β-catenin-mediated transcription, and is consequently responsible for the nuclear localization of β-catenin [
48]. Still, the mutations in exon 13 (c.2038_2041dup, p.Ser681* and c.1981C > T, p.Arg661*) were at worst classified into the mild-moderate phenotype category, which can be attributed to the fact that, while important for signaling, helix C appears to be completely dispensable for the structural role of β-catenin in cell-cell adhesion [
9].
In contrast to nonsense and missense mutations, we found that frameshift mutations in the C-terminal region can cause a severe disease phenotype. A patient with a frameshift mutation (c.2273delA, p.His758Leufs*30) in exon 15 (
Table S1, Supplement A) presented with severe intellectual disability and symptoms of autism spectrum disorder [
38]. Frameshift mutations can lead to a variety of changes, from the introduction of a premature stop codon to a protein extension beyond its normal stop codon. The consequences of these changes are difficult to predict without experimental analysis; however, in general, such mutations are considered deleterious [
49] because they can undermine the structural integrity of the entire protein. Furthermore, splicing mutations have also been found to cause severe or moderate disease. In our analysis, splice mutations were located at the 3′ acceptor splice sites of intron 5 and 7, with patients presenting mainly with severe cognitive disability and lack of eye contact, while the most robust characteristics of the cases categorized as moderate-severe included their inability to walk (exons from 8–12). Splicing mutations likely behave like frameshift mutations in that they can cause retention of the intron during the process of splicing, which can be detrimental for the reading frame. Interestingly, a pair of patients with identical mutations (c.2092_2096dup, p.Ile700Leufs*37) exhibited varied symptoms [
24]. Thus, the influence of a so-called modifier gene and/or additional environmental factors may play an important role in disease manifestation.
An additional interesting observation was that gross gene deletions resulted in a moderately pathological phenotype. Considering that these types of mutations likely result in null activity of β-catenin, we would expect them to cause the most severe phenotype possible. However, because many point mutations from our review appear to lead to a more severe phenotype, this is a possible indicator of dominant-negative effects. Dominant-negative effects are a consequence of mutations that lead to the expression of truncated proteins that obstruct the normal function of the protein from the wild type allele. It has been reported that β-catenin truncations without a C-terminal domain, which lack transcriptional activity, nevertheless bind co-transcription activators, and thus interfere with their binding to the wild type variants [
50].
The present study showed that mutations associated with CTNNB1 Syndrome are also found in exon 3, although mutations in this region are usually associated with cancer, because exon 3 bears important amino-acid residues for β-catenin degradation, such as the CK1-α phosphorylation site (S45), GSK3-β phosphorylation sites (S33, S37, and T41), and Fbw1 (D32 and G34). These are essential components of the inhibitory destruction complex that controls the levels of the free cytosolic β-catenin [
51,
52]. Missense mutations in these residues have been associated with colorectal cancers and other types of malignancies (e.g., melanoma, brain tumors) as they lead to the accumulation of nuclear β-catenin and tumorigenesis [
53]. In our data, four CTNNB1 Syndrome-causing nonsense and frameshift mutations in exon 3 were reported (G34, E54, Q72, and Q78), which were associated with a moderate phenotype. There is no evidence that these mutations cause tumor growth and progression.
Our review could not confirm the results reported by Rossetti et al. (2021) that missense mutations are associated with vitreoretinopathy [
22]. This condition was found in 13 cases (see
Table S1, Supplement A), with no correlation with mutation type; however, we observed a higher prevalence of this condition in the Chinese population (found in 69.2% of Chinese patients) [
24,
29,
33,
35,
39,
40]. The FEVR condition affects the retina by preventing blood vessels from forming at the edges of the retina. It is predicted that reduced β-catenin levels increase the probability of causing FEVR, although the exact pathophysiology is yet unclear [
35]. This condition is progressive in nature. Thus, regular screening of the retina can prevent unnecessary vision loss in patients with CTNNB1 Syndrome.
Some methodological considerations should be taken into account. The proportion of incomplete clinical data is perhaps the most significant and troubling issue in the current study. Most of the included studies were case reports or small case series, so referral, selection, and publication biases could occur. The results of our systematic review of the genotype–phenotype correlation are based on the data reported in the previous studies. It is, therefore, necessary to validate these results in prospectively gathered data and cell models. Importantly, this systematic review makes a clear demarcation of the exon borders of β-catenin, based on the article of Nollet et al. 1996 [
17], and can guide future studies. The organization of the clinical criteria based on the prevalence of the phenotypical findings may help neurologists to determine whether they will screen their patients for CTNNB1 Syndrome. Based on the phenotype, this gene can be added to the appropriate panels.