Array-CGH in patients with Kabuki-like phenotype: Identification of two patients with complex rearrangements including 2q37 deletions and no other recurrent aberration

Kabuki syndrome (KS), also known as Kabuki make-up syndrome (OMIM 147920), is a multiple congenital anomaly syndrome independently established by Niikawa [1] and Kuroki [2] with an estimated incidence of 1 in 32000 newborns [3]. KS is characterized by specific facial features including long palpebral fissures with eversion of the lateral one-third of the lower eyelid, arched eyebrows with sparseness of the lateral third, a short columella with a depressed nasal tip and prominent ears. This syndrome includes mild to moderate mental retardation, postnatal growth retardation with final short stature, skeletal abnormalities, and unusual dermatoglyphic patterns with prominent fingertip pads. The majority of reported patients have been sporadic with similar sex ratio. Several instances of vertical transmission of KS and KS-like features have been reported, making autosomal dominant inheritance a likely mode of transmission. No consensus diagnostic criteria have been established yet, but based on the clinical findings reported by Niikawa et al [1] five cardinal manifestations were defined to include patients in the KS group (peculiar face, skeletal anomalies, dermatoglyphic abnormalities, mild to moderate mental retardation and postnatal growth deficiency).

Although the cause of KS is currently unknown, numerous cytogenetic abnormalities have been reported in patients with a clinical diagnosis of KS [3]. Matsumoto and Nikawa [4] speculated on a possible microdeletion or microduplication involving contiguous genes as the cause of the disorder, given that the KS patients showed a wide spectrum of multisystemic manifestations and that the majority of patients were sporadic. Milunsky and Huang [5] using classic comparative genomic hybridization (CGH) and fluorescence in situ hybridization (FISH) on metaphase chromosomes, found a 3.5 Mb duplication at 8p23.1-p22 in 6 unrelated KS patients suggesting that this alteration may represent a common etiology for the disorder. However, this duplication has not been observed in other multiple patients studied by other groups [6–11] indicating that it is not a common finding in KS.

Maas et al. have also recently reported a de novo 250 kb deletion disrupting a single gene of unknown function in 20p12.1, C20orf133, detected by aCGH in a single KS patient out of 20 [12].

In order to search for putative microrearrangements responsible of KS, we have first screened our 16 patients referred to us with clinical suspicion of KS for 8p22-23 duplications by microsatellite analysis. Subsequently, in 10 patients with adequate DNA, we performed a genome wide survey of DNA aberrations with two different technologies: an in-house printed BAC array, with a global coverage of 23% of the genome and higher density in hot-spot candidate regions (located between segmental duplications and subtelomeric positions) and a commercial oligonucleotide array that spans coding and noncoding sequences with an average spatial resolution of ~100 kb (Agilent G4410B).

KS is a developmental syndrome with multisystemic involvement but variable expressivity, most cases being sporadic with a few exceptions of possible vertical transmission in an autosomal dominant pattern. This presentation highly suggests that KS could be caused by a recurrent genomic disorder affecting more than one gene [4]. To test this hypothesis, we have studied 16 Spanish patients with clinical features suggestive of a diagnosis of KS or KS-like, using molecular tools aimed at detecting cryptic genomic imbalances. We decided not to use strict clinical inclusion criteria with the idea that the molecular characterization of patients with overlapping phenotypes may provide clues to the final identification of the molecular basis of specific syndromes.

We first ruled out in all cases gene dosage alterations at 8p22-8p23.1, the region previously reported to be duplicated in six American KS patients (four Caucasian, one African-American, and one Haitian) [5]. Our results, as recently reported by others [6, 8–11, 31, 32], do not support the claim that genomic rearrangements of 8p22-8p23.1 are a common cause of KS.

To screen for putative submicroscopic genomic copy number alterations in our KS and KS-like population, we used two high throughput technologies based on aCGH with different genomic coverage, a BAC array with high density in putative hotspots for genomic disorders and a gene-based oligo array. We did not detect any copy number changes in the 20p12.1 region recently associated with KS, covered by two BACs in the HSBA and 15 probes in the G4410B array, further indicating that rearrangements at this locus are not a common cause of KS [12]. Surprisingly, we found rearrangements of terminal 2q in two patients (KS2 and KS14) with clinical manifestations overlapping with the KS phenotype. In patient KS2, we detected an inverted duplication with terminal 2q deletion (invdup-del2q, invdup 11 Mb-del 4.5 Mb). In patient KS14, who had a karyotype with an apparently balanced translocation (t(15,17)) inherited from his mother [13], we found two de novo unbalanced rearrangements: a 7.2 Mb terminal deletion of 2q37 in the paternal chromosome and a ~0.5 Mb interstitial deletion of 16p11.2 in the maternal chromosome. Patterns of complex rearrangements affecting several chromosomes and mostly generated in the paternal meiosis have been previously described by aCGH in some patients with apparently balanced de novo translocations [33, 34], suggesting a common mechanism for those rearrangements. However in KS14, no alterations were detected with any of the arrays in the proximity to the breakpoints of the cytogenetically balanced translocation inherited from the mother.

Therefore, we have detected overlapping rearrangements of the long arm of chromosome 2 in two KS-like patients. Although the rearrangements appear complex involving dosage imbalance for additional chromosomal regions in both cases, the region of overlap consists of a terminal 2q37 deletion of ~4,5 Mb. The long arm of chromosome 2 has a specific structure related to the ancestral fusion site of two primate subtelomeric regions at 2q13 and a subtelomeric duplicated region of DNA enriched with CpG islands [35]. To date, more than 60 patients with 2q37 deletions and a few rare cases with inverted duplications of 2q have been described. In fact, inverted duplications are thought to be recurrent rearrangements always associated with distal deletions [36]. Almost all patients with 2q37 deletions have some degree of mental retardation, some with autistic features, and facial dysmorphism. Brachymetaphalangism has been reported in approximately of 50% of patients while congenital heart defects, predominantly atrial or ventricular septal defects or more complex defects, are found in 20% of patients with 2q37 monosomy [37]. Phenotype-genotype correlations have defined candidate regions to harbor the genes responsible for congenital heart defects, brachymetaphalangysm and neurobehavioural problems [37]. However, considerably clinical variability was apparent even for patients with similar breakpoints pointing to the existence of modifiers, epigenetic or environmental phenomena also involved in phenotypic variability.

Even though their specific common facial dysmorphism (a round face with flat nasal bridge, dysmorphic ears often prominent, deep-set eyes, anteverted nares and thin upper lip) do not include long palpebral fissures with lateral eversion of the lower third, bulbous depressed nasal tip, or prominent ears with ear pits, characteristic of KS, the private features of these two patients (Figure 1) made the diagnosis of KS a possibility for our clinicians, considering brachymetaphalangy with prominent fingertip pads, a depressed nasal tip and prominent ears, mild to moderate mental retardation and heart defects were present in both. The phenotype in both patients was likely caused by additive effects of gene dosage imbalance of the 2q37 deletion that included the critical interval for most phenotypic features, and the additional rearrangement (2q36 duplication in KS2 and 16p11.2 deletion in KS14). However, given this phenotypic overlap, we believe that the diagnosis of a 2q37 deletion should be considered in patients with KS-like phenotypes where the characteristic eyes are less evident, a feature that is often obvious only after 3 years of age.

The 16p11.2 interstitial deletion additionally found in KS14 corresponds to a region flanked by segmental duplications that has been reported recently as one of the most common recurrent genomic imbalances associated with autism and developmental delay [38–40]. Interestingly, both de novo and inherited deletions and duplications of the region have been reported in ~1% of the ASD cases, indicating that the 16p11 region carries a substantial susceptibility risk for autism. The same region was found included in a new microdeletion syndrome in 16p11.2-p12.2 reported in patients with mental retardation, dysmorphic face and eye anomalies, even though the 0.5 Mb deletion of our patient is not part of the common interval but is included in the largest deletion reported in just one patient [41]. However, KS14 did not show autistic features nor eye anomalies, even though also several 2q37.3 deletion syndrome patients have been reported to have autistic features.

This region harbors multiple genes that could be dosage sensitive and involved in the regulation of developmental processes, one example being TBX6. Mouse embryos with reduced levels of Tbx6 show defective somite patterning of the paraxial mesoderm [42]. Therefore, it is tempting to speculate that some of the phenotypic features of KS14, such as the abnormal ribs, could be related with additive effects involving haploinsufficiency for TBX6. The finding of different parental origins for the chromosome 2q27 and the chromosome 16p11.2 deletions suggests that the rearrangements are not mechanistically related and their association must have occurred by chance.

The array technologies allow a rapid and automated evaluation of CNVs with genome-wide resolution.

We have detected three additional genomic alterations in KS patients not found in other samples and not reported as polymorphic CNVs in the published studies [21–27]. However, all three were inherited from a phenotypically normal progenitor in each case, suggesting that they are most likely polymorphic and non pathogenic rare CNVs. Given that most CNVs are quite rare in the population, an alternative hypothesis to the classic genomic disorder that could be the cause of disease, KS in this case, would be the additive effects of two or more rare CNVs. In our patient population, we detected the presence of a total of 85 CNVs, 66 of which were previously reported as polymorphic [30] and the other 19 were identified as new variants in our control populations. No patient specific combination of CNVs was found in the patients and the frequencies of the 85 detected CNVs between KS and controls showed no significant (p > 0.05) associations, suggesting those regions do not contribute to the KS phenotype.

A major conclusion of our study is that we were unable to find any common rearrangement causally related to KS. The combination of a high resolution BAC array specifically enriched for putative hotspot regions and a commercial 44 k oligo array (average resolution of 100 kb) failed to identify any common deletion or duplication in typical KS and KS-like patients, as previous reports using 1 Mb resolution BAC arrays. The inclusion of two patients in which a terminal 2q37 deletion syndrome could have been clinically suspected, suggests there is some phenotypic overlap among the two conditions, and this diagnosis should be kept in mind when evaluating patients with KS-like phenotypes. This study also supports that the 3.5 Mb duplication at 8p23.1-p22 is not the principal cause nor a common cause of KS [6, 8–11, 31, 32].

Although limited by the resolution of our arrays, our results indicate that a recurrent copy number genomic rearrangement affecting contiguous genes is unlikely to be the common cause of KS. Other type of structural changes, such as cryptic inversions, can not be ruled out with the applied technology and remain a possible explanation. Alternatively, KS could be caused by de novo genetic or epigenetic dominant alterations affecting a gene with pleiotropic effects, given that both concordant and discordant monozygotic twins have been reported [43, 44]. Further studies with higher resolution array platforms are indicated for typical and atypical patients suspected of having KS, since no strict diagnostic criteria have yet been established and no common etiology has yet been found.