We report the identification of
MKS3 mutations in eight of 14 (57%) JSRD families with congenital liver fibrosis, expanding the allelic spectrum of
MKS3 to include COACH syndrome. This mutation frequency is notably higher than the 7% figure observed in MS (
Consugar, 2007;
Kaddour, 2007), indicating a major role for
MKS3 within this specific JSRD subtype. These findings add a relevant contribution to the emerging gene-phenotype correlates in JSRDs, that are leading to a novel clinical-molecular classification based on the degree of multiorgan involvement and the outcome of large mutation screens of known genes (
Valente et al., 2008). Besides pure JS and JS plus retinopathy (for which the major gene is
AHI1), JS plus renal involvement (mostly caused by
NPHP1 or
RPGRIP1L mutations) and cerebello-oculo-renal phenotypes (strongly associated to
CEP290 mutations), we now suggest to include a fifth subgroup termed “JS plus CHF”, that encompasses the COACH acronym. In this subgroup, which major gene is
MKS3, CHF is the only mandatory criterion while other COACH-related features such as colobomas and renal involvement are possible additional manifestations. Interestingly, none of the 12 mutated patients had Leber congenital amaurosis or other forms of retinal dystrophy, that are frequently detected in other JSRD subgroups. This is unlikely to reflect a selection bias since patients were ascertained on the basis of CHF associated with JS signs, regardless of ocular abnormalities. Indeed, two of the six
MKS3-negative patients presented with retinopathy in the absence of chorioretinal coloboma.
In our cohort, CHF could be histologically confirmed in most cases by liver biopsy, and only in two families it was diagnosed based on elevated liver enzymes, hepatomegaly and intrahepatic bile duct dilatation at liver MRI. The clinical presentation of CHF appears to be extremely variable and often subtle in young children, with liver function and ultrasound that may remain normal or just show minor abnormalities for several years before becoming symptomatic, even acutely. In light of these findings, young JSRD patients with hepatomegaly and/or persistent elevation of liver enzymes should always undergo a detailed assessment of hepatic function, since an early diagnosis of CHF is crucial for a timely management of complications.
The clinical variability observed in our
MKS3-mutated families, related not only to the occurrence of ocular and renal involvement but also to the extent and severity of neurological and liver disease, still remains unexplained. A possible explanation comes from Bardet-Biedl syndrome (BBS; MIM# 209900), a ciliopathy consisting of retinopathy, polydactyly, obesity, hypogenitalism and posterior fossa defects due to mutations in at least 12 distinct genes. Recent studies have unmasked an oligogenic way of inheritance, in which mutations at different BBS loci can epistatically interact to cause and/or modify the phenotype (
Badano et al., 2006), and such mechanism has recently been demonstrated also for NPH genes (
Hoefele et al. 2007). Thus, epistatic effects of mutated alleles in other JSRD/MS genes are likely to explain at least in part the observed variability, as it has been already suggested for
NPHP1,
AHI1 and
CEP290 genes (Tory, 2007). Of note, Wolf et al. (2007) reported two patients with JS plus CHF and renal involvement who carried a single mutated allele in the
RPGRIP1L gene. It is tempting to speculate that these patients carry distinct mutations in
MKS3 or in another, still unidentified gene, and that mutations in
RPGRIP1L could represent modifier factors for NPH development. A similar speculation could apply to our family COR191, in which the living proband had a typical COACH phenotype while the aborted fetus met the diagnostic criteria for MS. In this family only one
MKS3 mutated allele could be identified in the proband, and DNA was not available from the fetus for molecular analysis. Although a second
MKS3 mutation unidentified by conventional sequencing cannot be excluded, the possible co-occurrence of mutations in other JSRD/MS genes is currently being tested.
Out of seven
MKS3 compound heterozygous families reported here, five showed an association of splicing or truncating mutations with missense changes, while two were compound heterozygous for missense variants. Interestingly, one splice site mutation resulted in the simultaneous skipping of two consecutive exons (19 and 20). A possible explanation for this unusual phenomenon is that the mutation-induced skipping of one exon could result in a loss of exonic splice enhancers (ESE) required to stimulate splicing efficiency of flanking adjacent exons. This is true especially in case of small exons/introns and weak splice sites, as in
MKS3 exon/intron 19 (
van Wijk et al., 2004).
None of the patients carried two mutations leading to premature truncation of meckelin, in line with previously reported
MKS3-mutated JS and Meckel-like patients (
Baala et al., 2007b). Conversely, abolition of meckelin activity is frequently reported in MS patients (
Smith et al., 2005;
Consugar et al., 2007;
Khaddour et al., 2007), supporting the hypothesis that complete loss of function could lead to a more severe, early lethal phenotype while patients retaining some protein activity would develop a milder JSRD phenotype. Notably, hypomorphic mutations in the
NPHP3 gene are responsible for juvenile NPH with retinal dystrophy and liver fibrosis (
Olbrich et al., 2003), while loss of function mutations in the same gene have been recently found to cause an early lethal Meckel-like syndrome with CHF, cystic dysplastic kidneys, variable laterality defects, and CNS malformations (
Bergmann et al., 2008).
In our cohort, missense mutations were found throughout the protein, in contrast with MS-associated missense mutations that mostly cluster in the extracellular domain of meckelin. A possible explanation is that the extracellular domain, containing a cleavable peptide and a cystein-rich repeat region superficially similar to EGF, EGF-CA and laminin EGF repeats, is more critical to meckelin function than other protein domains. This would be in line with the proposed role of meckelin as a receptor, based on structural evidences and on similarities to the G-protein coupled and Frizzled receptor families (Smith et al., 2005).
Meckelin has been shown to locate to proximal renal tubules and biliary epithelial cells where it plays an essential role in formation of the primary cilium, a sophisticated organelle found in most epithelial tissues and also in developing neurons (
Dawe et al., 2007). Increasing evidence points to a fundamental role for primary cilia in bile duct morphogenesis and renal tubulo-epithelial differentiation during embryogenesis, as well as in regulating key pathways of embryonic development, such as those involving Sonic Hedgehog and Wnt signaling (
Davenport and Yoder, 2005;
Singla and Reiter, 2006). These intriguing findings support a unifying hypothesis for the pathogenetic mechanisms related to primary cilia dysfunctions, that explain the multiorgan involvement and phenotypic variability observed in most ciliopathies.