Split-hand/foot malformation (SHFM), also known as ectrodactyly, is a rare congenital limb defect characterized by the median cleft of the autopod that results from hypoplasia/aplasia of the central phalanges, metacarpals, and/or metatarsals. This disorder occurs in approximately 1 in 18,000 live born infants and accounts for 8%–17% of all limb malformations (Czeizel et al., 1993). SHFM is clinically heterogeneous and presents inter- and intra-familial phenotypic variability that ranges from a mild manifestation, such as hypoplasia of a single phalanx, to the most severe form - monodactyly. Ectrodactyly may occur as an isolated malformation or in a syndromic form, which in general is less frequent but relatively common in certain types of the condition (Elliott and Evans, 2006; Gurrieri and Everman, 2013; Sowińska-Seidler et al., 2014b). There are eight known genetic loci associated with the corresponding types of the disorder, the majority of which undergo an autosomal dominant mode of inheritance with relatively frequent cases of reduced penetrance. Type 1 (SHFM1) is linked to 7q21.3 (OMIM 183600), SHFM2 is located at Xq26 (OMIM 313350), SHFM3 is associated with duplications at 10q24.3 (OMIM 246560), SHFM4 with heterozygous mutations in TP63 (3q27) (OMIM 605289), SHFM5 with deletions at 2q31 (OMIM 606708), SHFM6 with homozygous or compound heterozygous mutations in WNT10B (OMIM 225300), SHFM/SHFLD3 with duplications at 17p13.3 (BHLH9 gene) (OMIM 612576), and the recently identified EEC syndrome is associated with microdeletions at 19q13.11 or loss-of-function mutations in UBA2 (ubiquitin-like modifier-activating enzyme 2, OMIM 613295).

The SHFM1 type of disorder, mapped to locus 7q21.3, is characterized by high clinical heterogenicity, incomplete penetrance, and sex bias, pointing to a greater penetrance in male individuals than in female individuals (for isolated SHFM only) (Elliott and Evans, 2006; Sowińska-Seidler et al., 2014b; Rasmussen et al., 2016). Syndromic forms of the condition are frequently found. Accordingly, ectrodactyly is associated with intellectual disability (ID) in 33% of cases, with craniofacial malformations (CF) in 33% of cases, and with hearing loss (HL) (SHFM1D; OMIM 220600) in 35% of cases. Some patients have ectodermal dysplasia which usually enables the diagnosis of EEC syndrome (Elliott and Evans, 2006). The SHFM1 locus encompasses five protein-coding genes, namely, DYNC1I1, SLC25A13, SEM1, DLX5, and DLX6 (the distal-less homeobox 5/6), and a number of tissue-specific enhancers. Among the aforementioned genes, only DLX5/6 are known to be associated with the phenotypic manifestation of the disorder, which has been functionally demonstrated in mouse model studies. DLX5/6 form a digenic cluster, a part of a Dlx homeobox transcription factor family, and are essential in promoting osteogenesis. Both genes upregulate chondrocyte and osteoblast differentiation (Samee et al., 2008). The homozygous double knockout of both genes in mice (Dlx5/Dlx6 ^−/− ) results in the SHFM1 phenotype derived from the genes’ loss-of-function in the apical ectodermal ridge (AER) of the limb buds or in the otic vesicle (Acampora et al., 1999; Merlo et al., 2002a; 2002b; Robledo et al., 2002; Conte et al., 2016). The histological analysis revealed an impairment of the endochondral ossification of bones of the appendicular and axial skeleton in the mutants (Robledo et al., 2002). Additional findings that support these observations came with reports of SHFM1 patients harboring homozygous or heterozygous loss-of-function variants in the DLX5 gene (Shamseldin et al., 2012; Sowińska-Seidler et al., 2014a; Wang et al., 2014) and a single report of the heterozygous missense variant in DLX6 (Ullah et al., 2017). In humans, however, mutations that directly affect the gene sequence, point mutations or intragenic deletions, are not the most frequent pathogenic events. In fact, in the majority of SHFM1 cases, the disorder is related to a “position effect” which is associated with structural variants (SVs), i.e., deletions, translocations, inversions, and duplications, that disrupt the DLX5/6 regulatory environment of the genes’ topologically associated domain (TAD). There are at least 16 elements that play an established role in the regulation of tissue-specific Dlx5/6 expression, as shown in mouse and zebrafish enhancer assays, including the limb-, brain-, branchial arch-, otic vesicle-, and heart-specific enhancers (Zerucha et al., 2000; Pennacchio et al., 2006; Verzi et al., 2007; Kouwenhoven et al., 2010; Birnbaum et al., 2012). It has been shown that the clinical manifestation of the disorder in SHFM1 patients harboring pathogenic SVs depends on the genomic location of the variant’s breakpoints and, consequently, on the type and number of the tissue-specific DLX5/6 enhancers that have been dislocated from their target genes (reviewed in Birnbaum et al. (2012); Rasmussen et al. (2016)). Accordingly, the SHFM1 locus has been organized into three phenotypic subregions associated with: 1) isolated SHFM, 2) SHFM and HL, and 3) SHFM, HL, and CF (proposed by Rasmussen et al. (2016)). Nevertheless, the critical region for the SHFM1 phenotypic manifestation maps to a 103 kb segment, which encompasses two enhancers (eExon15 and eExon17) located within exons 15 and 17 of the DYNC1I1 gene, respectively, and therefore overlaps with subregion 1 for isolated SHFM (Tayebi et al., 2014).

Herein, we report on seven patients from five unrelated Polish families affected by SHFM1 and discuss two previously described families from our SHFM1 cohort (Tayebi et al., 2014). In all examined individuals, the disorder resulted from chromosomal rearrangements at the 7q21.3 locus, i.e., deletions and reciprocal translocations. The cohort was clinically heterogeneous since the patients presented with a variety of symptoms ranging from isolated ectrodactyly to EEC syndrome. Interestingly, in addition to the typical SHFM1 manifestation, one sporadic case, harboring a reciprocal translocation t(7;12)(q21.3;q21.2), presented features overlapping with Baker–Gordon syndrome (BAGOS) (OMIM 618218). This striking phenotype was associated with the translocation breakpoint mapped within the sequence of the SYT1 gene, which resulted in a gene loss-of-function mutation. Until now, SVs affecting SYT1 have never been reported to cause BAGOS, which highlights the novelty of our finding. In addition to conventional diagnostic assays that enabled the identification of SVs, we applied the whole-genome sequencing (WGS) approach in order to determine the breakpoints of the balanced translocations and search for a second pathogenic variant in one of the families with incomplete penetrance of the trait. In this paper, we provide a genotype–phenotype correlation of the patients from our cohort and discuss it in relation to the previously proposed phenotypic subregions within the SHFM1 locus.

In this paper, we present seven newly reported patients from five unrelated Polish families affected by SHFM1 and discuss two previously described families from our SHFM1 cohort (Tayebi et al., 2014). In all presented cases, the disorder resulted from chromosomal structural variants involving deletions or translocations with breakpoints located at the 7q21.2–q21.3 locus. In eight patients, the disease-causing mechanism was associated with a regulatory position effect consisting of the deletion or repositioning of tissue-specific enhancers. In one patient, the condition resulted from a large deletion encompassing both the regulatory elements and the target genes DLX5/6. The genotype–phenotype correlation presented herein is consistent with the previously published phenotypic subregions within the SHFM1 locus (Rasmussen et al., 2016) for all analyzed cases except one (patient P5). Rasmussen et al. (2016) presented a comprehensive review of the literature focusing on the genotype–phenotype correlation of 100 patients from 32 SHFM1 families, in which the disorder was caused by SVs at the 7q21.3 locus. The studied patients were grouped according to the clinical manifestation of the disorder, i.e., isolated SHFM, SHFM with HL, and SHFM with HL and CF anomalies. A thorough analysis of the variant’s breakpoints performed for the distinct phenotypic groups of patients, in relation to the location of the DLX5/6 tissue-specific enhancers, allowed for mapping each group to a common genomic region, thus suggesting the occurrence of three phenotypic subregions within the SHFM1 locus (Rasmussen et al., 2016). The region located most distant from DLX5/6 overlaps with the critical SHFM1 locus, encompassing eExon15 and eExon17 limb enhancers along with the eDlx#23 multi-tissue enhancer, and is associated with isolated SHFM. The intermediate subregion is related to SHFM with HL with no CF anomalies and contains five forebrain and/or branchial arch enhancers (hs1642, eDlx#19, eDlx#18, eDlx#16, and eDlx#14). The third subregion, associated with the full spectrum of clinical manifestation, i.e., SHFM with HL and CF anomalies, localizes closest to DLX5/6 and encompasses a single BS1 enhancer that displays activity in the limb, ear, and forebrain (Rasmussen et al., 2016) (Figure 2F). It has been shown that BS1 interacts with the Dlx6 promoter and with two Dlx5/6 enhancers, namely, I56i and I56ii, which regulate the gene expression in the forebrain and branchial arches (Zerucha et al., 2000; Ghanem et al., 2003; Kouwenhoven et al., 2010). According to the authors, the disruption of these interactions explains the hearing loss and craniofacial anomalies diagnosed in the third phenotypic group of SHFM1 patients (reviewed in Rasmussen et al. (2016)). The analysis of the results obtained for our patients allows us to reevaluate the position of the centromeric boundary of the third subregion that, accordingly, should also include the branchial arch enhancer eDlx#18 (Figure 2F). Our suggestion is substantiated by the results of the breakpoint sequencing analysis of the reciprocal t(7;12)(q21.3;q21.2) translocation found in the most severely affected patient in our cohort - P5. The patient displayed EEC syndrome with multiple additional abnormalities unspecific for SHFM1. The results of WGS, followed by Sanger sequencing evaluation, showed a complex chromosomal aberration involving not only a reciprocal translocation between long arms of chromosomes 7 and 12 but also additional reorganization and inversion of segments originating from chromosome 12 (see Figure 3C, Table 1, and Supplementary Figure S1G-I for the variant description). The translocation’s breakpoint on chromosome 7 was located within the SHFM1 7q21.3 locus and resulted in the dislocation of a number of tissue-specific enhancers, i.e., the limb, branchial arch (eDLX#19 and eDlx#18), brain, heart, and ear (otic vesicle) from the DLX5/6 genes (Figure 2F). Nevertheless, the interaction between BS1 and the DLX5/6 locus remained intact, yet the patient manifested CF anomalies, i.e., micrognathia and low-set posteriorly rotated and protruding ears (Figures 1O,P). Accordingly, we concluded that BS1 activity may not be sufficient to maintain proper DLX5/6 expression in the branchial arches and that possibly at least one other distant branchial arch-specific enhancer, i.e., eDLx#18 or 19, would probably need to be active in order to prevent the manifestation of CF anomalies. While the EEC syndrome observed in patient P5 resulted clearly from the 7q21.3 translocation breakpoint, the additional phenotypic features, which were not typical of SHFM1 clinical manifestation, resulted from the breakpoint within chromosome 12 in the second intron of the SYT1 gene (OMIM 185605; NM_005639.3). SYT1 encodes synaptotagmin 1, a protein that plays a critical role in mediating synchronous calcium-dependent neurotransmitter release via triggering the synaptic vesicle fusion. Heterozygous missense loss-of-function or dominant-negative mutations in SYT1 result in Baker-Gordon syndrome (BAGOS) (OMIM 618218), a neurodevelopmental disorder characterized by a global developmental delay, poor or absent speech, infantile hypotonia, ophthalmic abnormalities (i.e., epicanthal folds, almond-shaped eyes, strabismus, and nystagmus), facial dysmorphic features (e.g., prominent high forehead with V-shaped hairline, smooth philtrum, and thin upper lip), hyperkinetic movements, gastrointestinal abnormalities, sleep apnea, and EEG abnormalities in the absence of overt seizures (Baker et al., 2015; 2018). Interestingly, patient P5 exhibited symptoms that were overlapping with the BAGOS phenotypic manifestation, i.e., global developmental delay, absent speech, hypotonia, defect in the atrial septum, high forehead with V-shaped hairline, strabismus, almond-shaped eyes, smooth philtrum, and thin upper lip (Figure 1O-Q), that most likely resulted from SYT1 loss-of-function. Up to date, only five variants have been reported to cause BAGOS, all of which are missense mutations clustered in one of the two highly conserved calcium-binding domains of the synaptotagmin-1 protein, the C2B domain (Baker et al., 2015; 2018; Cafiero et al., 2015). These pathogenic variants were associated with variable phenotypic severity of the disorder relating mostly to the spectrum of neurological abnormalities. This mutation-specific impact on protein expression, function, and localization at nerve terminals resulted in a variable degree of disturbance to synaptic vesicle kinetics (Baker et al., 2018). At least one of the reported mutations (I368T) has been shown to act through a dominant negative mechanism (Baker et al., 2015); however, the mechanism of action of the other variants remains elusive and requires further studies. Nevertheless, for mutations associated with the mildest neurodevelopmental impairment (e.g., M303K), a loss-of-function mechanism is considered. The results of our findings support this hypothesis since the translocation breakpoint identified in the patient P5 was mapped to the second intron of SYT1, which is a part of gene’s 5′-UTR. This, in turn, leads to the dislocation of the entire coding part of SYT1 from its promoter. This regulatory mutation resulted in a 70% reduction in gene expression in the proband’s PBMCs compared to controls (Supplementary Figure S1D), pointing to its loss-of-function effect. Apart from the typical BAGOS symptoms, our patient was diagnosed with the atrial septal defect that, surprisingly, was also observed in a single patient from the cohort described by Baker et al., 2018 (patient 4). Our findings, therefore, confirm that this defect may be a characteristic feature of BAGOS.

Although developmental delay (DD) was reported in 33% of SHFM1 patients (Elliott and Evans, 2006), the abnormality was restricted to large deletions or translocations of which at least one breakpoint was mapped outside the SHFM1 locus (Haberlandt et al., 2001; Bernardini et al., 2008; Saitsu et al., 2009; van Silfhout et al., 2009; Brown et al., 2010; Vera-Carbonell et al., 2012; Delgado and Velinov, 2015; Sivasankaran et al., 2016; Rai et al., 2019). Since the feature was absent in patients carrying a deletion limited to the DLX5/6 TAD, neither was it identified in patients harboring SNVs in DLX5 (reviewed in Rasmussen et al. 2016)), it seems unlikely that DLX5/6 contribute to DD. Consequently, the global DD observed in patient P5 was most likely a manifestation of BAGOS rather than SHFM1. In addition to this finding, DD was also not present in the sporadic patient P1 from our cohort, who carried a large 4.6 Mb deletion encompassing the entire SHFM1 locus, including the DLX5/6 genes and their regulatory units. As suspected, the patient was diagnosed with the full spectrum of SHFM1 symptoms, including the EEC syndrome (i.e., bilateral ectrodactyly of hands and feet, enamel and nail hypoplasia, and tendency for tooth decay), hearing loss, and craniofacial anomalies with the characteristics of the 7q21.3 locus, namely, micrognathia and malformation of the ears (i.e., overfolded helix) (Figures 1A–D). Among the deleted genes, 10 were disease-associated; however, only one, namely, SGCE (OMIM: 604149), was likely dosage-sensitive and associated with an autosomal dominant phenotype. Heterozygous truncating mutations in the SGCE gene lead to myoclonus-dystonia syndrome (OMIM: 159900), characterized by myoclonic jerks affecting mainly proximal muscles. Nevertheless, the gene is maternally imprinted and, consequently, expressed only from the paternal allele; thus, reduced penetrance is observed upon maternal transmission (Schüle et al., 2004). This finding presumably explains the lack of phenotypic manifestation of myoclonus-dystonia syndrome in patient P1.

Unlike the aforementioned patient P5 with EEC, harboring the t(7;12)(q21.3;q21.2), all carriers of the t(7;10)(q21.3;q22.2) translocation from family 4 (P4.1, P4.2, and P4.3) were diagnosed with an isolated SHFM with reduced penetrance. The index patient P4.1—the mother and her son P4.2 manifested bilateral ectrodactyly of hands and feet, whereas the daughter (P4.3) was affected with mild brachydactyly of feet. When comparing the location of the rearrangement breakpoints at the 7q21.3 locus in both families, we found that the disrupted region was diversified by only two branchial arch enhancers, eDlx#19 and eDlx#18, yet associated with significantly distinct phenotypic expression. Both enhancers remained intact within DLX5/6 TAD in individuals from family 4 and dislocated from the target genes in patient P5. This finding further supports the crucial role of eDlx#19 and/or eDlx#18 in the pathogenesis of craniofacial anomalies and EEC syndrome. Thorough analysis of WGS results, aiming to search for genetic modifiers associated with reduced penetrance in the members of family 4, did not reveal any variants that could potentially explain this phenomenon. Nevertheless, brachydactyly - especially phalangeal aplasia or hypoplasia - is, same as SHFM, a form of limb reduction defect and, in this case, can be regarded as a manifestation of the SHFM spectrum, most likely resulting from 7q21.3 locus disruption. Incomplete penetrance is relatively common in SHFM1 familial cases; however, it is limited only to isolated forms and associated with a disruption of eExon15, eExon17, and eDlx#23 enhancers. There is a clear sex bias among the mutation carriers since it has been demonstrated that the penetrance is higher in males vs. females and that the disorder is overtransmitted from affected fathers to sons (reviewed in Rasmussen et al., (2016)). This observation can possibly result from the functionally supported hypothesis of DLX5/6 maternal imprinting in osteoblasts (Rattanasopha et al., 2014).

Two unreported sporadic patients from our cohort (P2 and P3), along with individuals from two previously published Polish families (family 6 and 7 - this study; published as family 1 and 3, respectively, in Tayebi et al. (2014)), who harbored the smallest microdeletions, were diagnosed with isolated SHFM. In all cases, the deletion encompassed a common cluster of regulatory elements, i.e., two limb enhancers, eExon15 and eExon17, along with two multiple-tissue enhancers, eDLX#24 and eDLX#23, which was consistent with the “isolated SHFM” phenotypic subdomain proposed by Tayebi et al. (2014) and Rasmussen et al. (2016). Interestingly, despite the highly overlapping location of the CNVs (especially for patient P2 and members of family 6 - see Figure 2F), the cohort presented variable expressivity as ectrodactyly was limited to feet in P2 and P3 individuals (hands were unaffected), whereas all affected members of families 6 and 7 manifested SHFM in all four extremities. Additionally, the disorder was highly variable among individuals from family 7, ranging from a severe cleft to mild hypoplasia of the central digit and mild defect of the feet. Variable phenotypic expression of identical or highly similar genetic mutations is relatively frequent in many limb malformation phenotypes. In a number of cases, this phenomenon could be explained by differences in the gene’s regulatory environment, while in other studies, it remains unexplained and requires further studies involving epigenetic modifications and contributions from environmental factors.

In conclusion, in this paper, we present seven novel cases from five Polish families diagnosed with variable features of the SHFM1 spectrum, all resulting from chromosomal rearrangements at the 7q21.3/7q21.2–q21.3 locus. We performed a thorough genotype–phenotype correlation of the cyto-molecular results described herein against previously suggested subregions of the SHFM1 locus, which enabled us to refine the regulatory units, critical for specific clinical manifestation of the disorder. Consequently, we suggest that the dislocation of the branchial arch-specific enhancers, eDlx#18 and/or eDlx#19 from the DLX5/6 target genes, is critical for the pathogenesis of craniofacial anomalies in EEC syndrome. Additionally, we present the first evidence that a loss-of-function regulatory mutation within the SYT1 gene results in the BAGOS-like phenotype since these features, along with EEC syndrome, were present in a sporadic patient harboring a reciprocal translocation that disrupted both SHFM1 and SYT1 loci. In our work, we highlight the need for sequence-based approaches in defining the exact breakpoints of pathogenic SVs in the molecular diagnostics of patients whose features point to the disruption of certain tissue-specific enhancers, such as in SHFM1.