Distal limb deformities are congenital malformations with phenotypic variability, genetic heterogeneity and complex inheritance. Among these, split-hand/foot malformation is an ectrodactyly with missing central fingers, yielding a lobster claw-like hand, which when combined with long-bone deficiency is defined as split-hand/foot malformation and long-bone deficiency (SHFLD) that is genetically heterogeneous. Copy number variation (CNV) consisting of 17p13.3 duplication was identified in unrelated pedigrees, underlying SHFLD3 (OMIM 612576). Although the transcription factor Fingerin (bHLHA9) is the only complete gene in the critical region, its biological role is not yet known and there are no data supporting its involvement in mammalian limb development. We have generated knockout mice in which only the entire coding region of Fingerin was deleted, and indeed found that most null mice display some limb defects. These include various levels of simple asymmetrical syndactyly, characterized by webbed fingers, generated by incomplete separation of soft, but not skeletal, tissues between forelimb digits 2 and 3. As expected, hand pads of Fingerin null embryos exhibited reduced apoptosis between digital rays 2 and 3. This defect was shown to cause syndactyly when the same limbs were grown ex vivo following the apoptosis assay. Extrapolating from mouse data, we suggest that Fingerin loss-of-function in humans may underlie MSSD syndactyly (OMIM 609432), which was mapped to the same locus. Taken together, Fingerin gene dosage links two different congenital limb malformations, syndactyly and ectrodactyly, which were previously postulated to share a common etiology. These results add limb disorders to the growing list of diseases resulting from CNV.
Limb development during embryogenesis is a highly complex process that has to be precisely orchestrated. Since non-lethal genetic abnormalities and developmental malfunctions may take place, as many as 1 in 600 newborns present with some form of hand abnormality (1). Moreover, the developing limb has long been an excellent model for gaining insights into morphogenesis and organogenesis, and therefore it caught the attention of embryologists and developmental geneticists, since limbs display discrete visual patterns and are amenable to various experimental manipulations. However, many open questions remain to be answered to fully understand how multiple genes, transcriptional networks, and signals are dynamically integrated in order to regulate various molecular and cellular activities during limb development.
Syndactyly is the most common congenital malformation of the limbs, with an incidence of 3–40 in 10K births (2). Clinically, it is a heterogeneous developmental deformity, even in the same individual. Syndactyly may be isolated or syndromic, finger fusion unilateral or bilateral, affect hands and/or feet and be presented as symmetrical or asymmetrical. Syndactyly can be partial or complete due to the degree of fusion, or be cutaneous or skeletal and affects even distal bones in the forearm or foreleg (2). There are at least nine categories, six of which segregate as a Mendelian dominant trait, with only two autosomal recessive and one X-linked forms (2). The genes and mutations underlying all syndactyly types have not yet been fully determined (1,2).
Split-hand/split-foot malformation (SHFM), also known as ectrodactyly, is characterized by a deep V-shaped cleft situated in the center of the autopod. It is caused by the absence of one or more medial digits, with bordering fingers that may exhibit syndactyly; hence, the name lobster claw hand (3). This rare condition, affecting one in 8.5–25K newborns and accounting for 8–17% of all limb reduction defects, is extremely complex because of its variability in clinical presentation, irregularities in its inheritance pattern, and the heterogeneity of molecular and genetic alterations that are found in affected individuals (4). Currently, six loci (SHFM1–6) have been associated with human non-syndromic SHFM, and various environmental factors (e.g. teratogens) were shown to be involved (4,5).
SHFM, combined with long-bone (usually the tibia) deficiency (SHFLD), is a genetically heterogeneous group of disorders. One of its forms was recently defined as SHFLD3 (OMIM 612576) (5). A causative mutation was not identified, but a screen for copy number variation (CNV) revealed various genomic microduplications in 17p13.3 (5). An 11.8 kb minimal duplication region, arranged in direct tandem orientation, was defined, and was suggested to encode only one full putative gene, bHLHA9 (5) (referred to hereafter as Fingerin). Later, 44 more cases of ectrodactyly from 15 unrelated families were described, all containing 3 Fingerin alleles (6,7). Interestingly, a Chinese fetus diagnosed prenatally with a bilateral split-hand malformation was found to carry four copies of Fingerin, due to triplication in the mother (8).
Fingerin was originally identified as a putative protein by virtual screening of sequence databases from various organisms for bHLH-containing predicted proteins and it is listed as BHLHF42 (9) and bHLHA9 (10). It encodes an evolutionally conserved protein of 235 and 231 amino acids in humans and mouse, respectively, which are localized to a single coding exon. Sequence similarity implies that within the super family of basic Helix-Loop-Helix (bHLH) transcription factor proteins, Fingerin belongs to a class/clade of tissue-specific regulators of transcription (9,10). Knockdown of bHLHA9 has been performed in zebrafish and resulted in shortening of the pectoral fins, which supports its involvement in limb development (5). However, direct evidence for the involvement of Fingerin in limb development in mammals is not yet available. Although it is the only full-length gene in the minimal critical region duplicated in SHFLD3 patients and some family member carriers, the involvement of one or more unrecognized regulatory elements of other genes that are present in this critical region cannot be ruled out. Hence, the involvement of Fingerin in limb and especially digit development remains to be determined.
Towards revealing the biological role of Fingerin in embryogenesis, we generated a Fingerin null mutant mouse (FingerinNull) by homologous recombination in ES cells (Fig. 1). The targeting vector was designed to precisely and completely eliminate the coding sequence of Fingerin and replace it by a promoterless nuclear lacZ reporter gene, without further affecting the endogenous Fingerin locus (Fingerin::LacZ). To confirm the validity of lacZ reporter expression, learn about the expression domains, and look for a phenotype, we characterized the spatiotemporal expression pattern of the Fingerin::LacZ reporter (Fig. 2).
Fingerin::LacZ expression in the limb bud was detectable as early as embryonic day (E)9.5 (not shown), and was very clear and tissue specific on days E10.5–E15.5 (Fig. 2). We could not stain at later developmental stages, because the skin acquires a barrier function and becomes impenetrable to X-gal. Closer inspection verified that staining in the upper and lower limb buds appeared in both the dorsal and ventral sides of the autopod, the future hand/foot plate. At E12.5–E13.5, when mesenchymal condensation, the first step in skeletogenesis, was manifested as darker structures, continuous X-gal staining was evident above the rays of the future fingers and above the interdigital regions. However, at E14.5 and E15.5, when morphogenesis had taken place and fingers were already sculptured, the expression domain had narrowed and reporter staining was displayed only above (distal to) the rays. Interestingly, at all developmental stages, no expression was detected at the dorsoventral border, appearing in the midline of the dorsoventral axis as a white band (Fig. 2 arrows).
To verify that reporter expression mimics the endogenous expression pattern, we performed in situ hybridization at E14.5 (Fig. 3) to complement previous reports (5). In both fore- and hindlimbs as well as section axes, X-gal and Fingerin expression were detected in the dorsal and ventral external-most layer, probably reflecting the non-ridge ectoderm. The pattern of a Fingerin-negative band flanked by two Fingerin-positive domains was present before the apical ectodermal ridge (AER) had been formed and after it was disintegrated, and at the start of digit formation (Figs. 2–3). Our data are in agreement with Figure 2C–H of Klopocki et al. (5), who also noted expression in the distal limb bud at E11.5 in the dorsal and ventral, but not the AER regions. However, after examining many sections, at various ages, by two methodologies, we did not detect Fingerin expression in the mesenchymal layer, but have always observed staining in the outermost layers.
Noting that Fingerin and its reporter are expressed during a time window in which limb morphogenesis is established, we examined the limbs of adult mice, looking for abnormalities caused by its genomic deletion. In contrast to wild type and FingerinHet, most FingerinNull mice exhibited various degrees of syndactyly, with an incomplete separation of digits 2–3, the index and middle fingers, in the forelimbs (Fig. 4A). Three exceptional individuals were identified out of dozens of mice examined, displaying webbing of hindlimb fingers 2–3, or forelimb fingers 3–4 or 2–3–4.
Statistical analysis of the syndactyly phenotype in FingerinNull mice revealed that although the severity of the phenotype was highly variable (Fig. 4A), the penetrance was as high as 93% (Fig. 4B). About 40% of the mice displayed syndactyly in both forelimbs, 29% only in the left, 25% only in the right and 7% were normal. However, a breakdown of the results by sex revealed some gender-related differences. Whereas in males, the majority (46%) of the adult mice displayed syndactyly only in the left forelimb; in females, the majority (48%) displayed syndactyly in both forelimbs. Interestingly, in human patients with duplication, the severity of the disease was gender-biased, but it increased in males (5,7).
Next, we investigated the nature of the syndactyly, and whether it involves soft and/or skeletal tissues. Consequently, we performed Alizarin red and Alcian blue staining of bone and cartilage, respectively (Fig. 4C). No apparent skeletal abnormalities were found in the hand plates and arms, and therefore mechanisms involving skeletal patterning defects, identity of tissues and other anomalies were ruled out. Conversely, the soft tissue (cutaneous) syndactyly pointed to interdigital apoptosis as the affected developmental process in FingerinNull mice. To support our suggestion, we first examined digit separation in FingerinNull embryos. Indeed, incomplete separation of forelimb digits 2–3 was evident as early as E14.5 (Fig. 5A), which is in agreement with the timing of a massive wave of interdigital apoptosis, which normally takes place in mouse forelimbs at E13.5. The fact that syndactyly was also observed at E15.5 (Fig. 5A) and in adult mice (Fig. 4A) ruled out a developmental delay and suggested the existence of an uncorrectable developmental malformation.
To more directly test the hypothesis that interdigital programmed cell death by apoptosis was affected, we performed whole-mount staining of the limbs using the apoptotic reporter LysoTracker. Briefly, limbs were excised at E13.5 prior to digit separation, and forelimbs and hindlimbs were separately pooled and stained, and thereafter placed in a multi-well plate for identification purposes, and photographed (Fig. 5B). Then, limbs were separately grown for 4 days as an ex vivo organ culture and re-examined. Indeed, stained limbs from wild-type and FingerinHet embryos displayed massive interdigital apoptosis, with stained cells flanking all digital rays. In contrast, the hand pads from FingerinNull embryos were different, and displayed various levels of reduced apoptosis between digits 2 and 3 (Fig. 5B). To verify that lack of apoptosis in the FingerinNull limbs was a characteristic that could be used to predict syndactyly, we examined digit separation in the same limbs after culturing. Indeed, after 4 days of incubation, we observed that apoptosis-defective limbs did not completely separate fingers 2–3 and displayed various levels of syndactyly, thus supporting our hypothesis. The molecular mechanism linking transcriptional regulation by Fingerin to apoptosis is yet to be discovered.
In sum, here we show for the first time a biological role for the transcription factor Fingerin in distal limb development. Complete deletion of the murine ortholog of Fingerin causes reduced apoptosis and subsequent syndactyly, both in vivo and ex vivo. Clinically speaking, this Fingerin-dependent syndactyly in mice is non-syndromic, simple, cutaneous, incomplete and asymmetrical, whereas genetically, it is autosomal recessive with sex bias and incomplete penetrance.
Both in situ hybridization and Fingerin-driven reporter expression in mouse embryos have indicated the dynamic expression of Fingerin, starting as a crescent in the distal hand/foot plates, and narrowing down to the distal-most regions covering the developing digits, but not the interdigital mesenchyme. In contrast to proteins with an unequal dorsoventral distribution, such as Wnt7a (1), Fingerin is expressed in both the dorsal and ventral surfaces. Moreover, at all stages examined, Fingerin was not identified in a strip of cells at the dorsoventral borderline. This Fingerin-negative region is identified both before and after the formation and disintegration of the AER. To evaluate our notion that this expression profile is distinctive, we examined the original images depicting the expression of various key limb markers such as Fgf4/8/9, Hoxd13, Hoxa13, Tbx2, MSx1/2, Shh, Grem1, Jag1 and others (e.g. (11–14)). Comparison criteria included limb specificity, the polarity of the expression domain (anterior–posterior and dorsal–ventral), the timing of the appearance and/or disappearance, the expression in AER or the underlying mesenchyme, etc. Overall, to the best of our knowledge, the unique spatiotemporal expression profile of Fingerin differs from that of other genes underlying limb morphogenesis.
The sharp boundary between Fingerin-positive and -negative expression domains raises the possibility that Notch signaling plays a role in this patterning event. The Notch signaling pathway is a conserved intercellular signaling mechanism that is crucial for proper embryonic development in numerous organisms and plentiful tissues (15–17). bHLH factors (e.g. proneural genes) are downstream targets of Notch signaling, and the expression of all components of the transcriptional network is regulated by lateral inhibition with feedback (18). Indeed, independent mutations in two ligands of the Notch family of transmembrane receptors suffer from AER defects and syndactyly, thus supporting the involvement of Notch signaling in boundary formation and interdigital apoptosis. Mice homozygous for the syndactylism (sm) mutation develop a hyperplastic AER, resulting in abnormal dorsoventral thickening of the limb bud, subsequent merging of the mesenchymal condensations that give rise to cartilage and bone in the digits, and eventually fusion of fingers (19). The gene mutated in sm mice encodes the Notch ligand Serrate2. In another study, knockout of another Notch ligand, Jagged2, resulted in null embryos that exhibit syndactyly in the fore- and hindlimbs (20). Their AER is hyperplastic, the Fgf8 expression domain is expanded, and Bmp2/7 expression and apoptotic interdigital cell death are reduced in the foot plates of the mutants. Taken together, we suggest the existence of crosstalk between AER-expressed Notch ligands and the adjacent cells expressing Fingerin and the Notch receptor. Future work will determine the effect of Fingerin deletion on the level of Notch signaling components, as we have previously identified in the cerebellum of mice lacking Math1, a sequence-related bHLH factor (21).
We hypothesize that a Notch-independent mechanism may also affect the AER-induced interdigital apoptosis and finger separation. The primary function of Fingerin is cell autonomous, since as a bHLH transcription factor it probably modifies the expression of downstream targets, some of which may serve as morphogens. Upon extracellular secretion, these diffusible proteins may interact with other proteins, which play a role in limb morphogenesis and apoptosis. This secondary non-cell autonomous function links the non-ridge ectoderm, the AER, and interdigital apoptosis. A second mechanism may utilize the fact that besides the bHLH domain, Fingerin also contains a Proline-rich carboxy-terminus, not typically found in bHLH factors. Whereas in the mouse Fingerin the frequency of Proline in the C-terminal is 15.7% (18/115), in the human ortholog it is as high as 21.7% (25/115). Moreover, 13 of the Prolines are positionally conserved between the two species. Proline-rich domains participate in protein–protein interactions, for example, in signaling proteins (22–24). Therefore, Proline-based protein–protein interactions may also sequester the function of proteins that provide the apoptotic signal.
Here, we show for the first time that Fingerin plays a role in distal limb development, and therefore it strengthens the suggestion that a genomic microduplication in 17p13.3 gives rise to SHFLD3 in humans owing to extra copies of Fingerin per se. Notably, the generation of the Fingerin knockout mouse involves only a microdeletion of a single exon, which encodes the entire coding region of Fingerin. This targeted deletion affected digit separation, and therefore ruled out the involvement of unidentified regulatory sequences in the critical region, such as enhancer elements of distal genes, non-coding RNA and others. Transgenic mice with extra copies of Fingerin, which we will generate, will not only provide the final proof for the role of Fingerin in ectrodactyly but will also allow to decipher the underlying molecular mechanisms.
To further establish a role for Fingerin in human limb abnormalities, and especially syndactyly, we have searched in various databases for CNVs involving 17p13.3. Ten cases with a copy number gain spanning 70–590 kb in this locus were identified in DECIPHER (Database of Chromosomal Imbalance and Phenotype in Humans Using Ensembl Resources). These patients display limb defects: ectrodactyly of the hands/hands and feet, which may be accompanied by aplasia/hypoplasia of the tibia/fibula, in agreement with published reports on SHFDL3. In contrast, a single patient with syndactyly and a copy number loss in the relevant region was identified in ISCA (International Standards for Cytogenomic Arrays Consortium). This case was diagnosed with a loss of 490 kb (nssv577586, nsv532083 and chr17:758 644–1 248 811) in a region spanning Fingerin and a few additional genes, and 2–3 toe syndactyly was the only reported phenotype. Thus, additional cases with a complete loss of a smaller region were of need to strengthen the involvement of Fingerin in syndactyly.
A consanguineous Pakistani family and an unrelated large inbred Turkish family, with individuals exhibiting a novel form of non-syndromic, autosomal-recessive syndactyly were reported (25,26). Patients presented with synostosis, in which the 3rd and 4th metacarpal palm bones were fused, along with reduction of phalanges, and proximal webbing of toes, but it was not characterized by bizarre arrangement of skeletal elements of hand/foot. This phenotype was termed mesoaxial synostotic syndactyly with phalangeal reduction (MSSD, type IX syndactyly, OMIM 609432). MSSD was linked in both families to 17p13.3 (27), where Fingerin is mapped. Moreover, within this locus, Fingerin was localized to a region flanked by D17S1533 and D17S596 markers that show high-LOD scores in both families (27). Thus, we suggest that Fingerin may be the causative gene for MSSD Syndactyly type IX.
Deletion of Fingerin causes syndactyly in mouse (and maybe also in humans), whereas three (and even four) Fingerin alleles were detected in a small genomic interval flanking Fingerin in patients with ectrodactyly. Thus, Fingerin is an interesting example of a CNV involving a single gene that leads to two different diseases. This notion is in agreement with the suggestion that central polydactyly, syndactyly and ectrodactyly (presented with excess, conjoined or missing fingers) share a common etiology, which result from abnormal maintenance of the central AER (3). Two lines of data support a common etiology hypothesis. First, the same patient or pedegree may display the various malformations (28). Second, the same teratogen, when injected into pregnant rats at different time points, can cause different disorders (29). For example, busulfan was shown to disturb the central autopod and affect apoptosis, causing different outcomes in different regions (1). Increased apoptosis in the AER resulted in partial digit duplication, whereas reduced apoptosis in the mesodermal interdigital region caused finger webbing, and cell death in both regions, which gave rise to a cleft, lobster claw-like hand. The human and mouse limb phenotypes, when considered together, support this notion.
An interesting case of ‘one gene—two disorders’ was recently solved (30). Growth and differentiation factor 5 (GDF5), a BMP family member, plays a pivotal role during limb development. Mutations in GDF5 lead to a wide variety of skeletal limb malformations, ranging from complex syndromes to isolated forms of brachydactylies (short digits) or multiple synostoses (fused bones) syndrome 2 (SYNS2). Recently, a family with an autosomal dominant inherited combination of SYNS2 and additional brachydactyly type A1 (BDA1) was identified. Functional studies revealed a dual mechanism, involving simultaneously gain and loss of function. One mutation in the overlapping interface of the antagonist and receptor-binding site in GDF5 caused two phenotypes: BDA1 and SYNS2 (30). In Fingerin-related disorders, the different diseases are not caused by distinct mutations, but rather by increased or decreased allele numbers. Overall, Fingerin-related limb malformations are added to a growing list of diseases caused by gene dosage variation (31,32). Moreover, Fingerin joins the small list of disorders with a bidirectional gene dosage effect that results in different diseases.
MATERIALS AND METHODS
Generation of Fingerin knockout mice
A targeting vector was constructed by inserting 6.5 and 4.8 kb of mouse genomic DNA from 129SvEv, located 5′ and 3′ to the open reading frame (ORF) of Fingerin, respectively, into a pNZTK2 plasmid. The construct was electroporated to a hybrid ES cell line (C57BL/6N and 129SvEv) at InGenious Targeting Laboratory (iTL, Ronkonkoma, NY, USA), which served as a contractor. Five recombinant clones that survived positive and negative selection were expanded, and homologous recombination was further verified using Southern analysis. Four chimeras that gave rise to fertile heterozygous mice were used to establish the colony. F0 heterozygotes were bred to C57BL/6 mice, and the strain was maintained on a mixed C57BL/6-129SvEv background. In parallel, mice were repeatedly backcrossed to C57BL/6 mice to obtain a pure C57BL/6 background. Subsequent genotyping of Fingerin mice was carried out by PCR. The primers used are available upon request. Embryos were staged by timed mating of mice with various genotypes of Fingerin and the morning of the vaginal plug was designated as E0.5. All experiments were approved by Institutional Animal Care of the Hebrew University, which is an AAALAC internationally accredited institute.
X-gal staining and histological analysis
Whole embryos were stained as previously described (21,33,34). After postfixation, whole embryos were visualized and photographed using an Olympus SZ-40 stereomicroscope, and subsequently stored in 70% ethanol. For histological analysis, fore- and hindlimbs of staged embryos were dissected, paraffin-embedded and 12 µm sections were cut, counterstained with eosin, visualized under a Zeiss Axioskop2 microscope, and photographed using an Olympus DP71 camera. Skeletal staining was performed as in (34).
In situ hybridization
In situ hybridization was carried out as previously described (35). Probes were synthesized from PCR-amplified products of the Fingerin gene (NM_177182), which were ligated into a pBluescriptII-KS+ vector. Primers were designed according to Gray et al. (36) and are available upon request.
Lysotracker staining was carried out according to the standard protocol. Briefly, mice were time mated and embryos were staged. Limbs were dissected on ice in Hank's balanced salt solution (HBSS), and forelimbs and hindlimbs were placed in separate vials to facilitate distinguishing between them later. Limbs were washed twice for 5 min with HBSS, and then placed in a 500 nm LysoTracker® Red DND-99 solution (Molecular Probes, Life technologies) and gently rocked in the dark at 37°C for 1 h. Limbs were then washed twice with HBSS at room temperature and immediately visualized under an Olympus MVX10 fluorescent stereomicroscope and photographed using a QIClick™ digital CCD camera (QImaging).
Ex vivo limb cultures
The limbs were cultured using a standard tissue culture procedure. The protocol is based on Paradis et al. (37) with modifications. Briefly, each limb was placed in a separate well of a 6-well plate, washed twice with HBSS and replaced with 3 ml/well of minimum essential medium Eagle, Earle's Salt Base without Phenol Red (Biological Industries, Beit-Haemek, Israel), supplemented with 10% fetal calf serum (Biological Industries, Beit-Haemek, Israel). Plates were placed on a gentle shaker in a 5% CO2 humidified incubator at 37°C, and limbs were grown for 4 days, with fresh 1 ml of medium added 2 days after beginning the culture. After culturing, limbs were visualized under an Olympus SZ-40 stereomicroscope and photographed with a Nikon E5000 digital camera.
This work was supported by grants from the Israel Science Foundation (431/07 to N.B.A.) and the Legacy Heritage Biomedical Science Partnership Program of the Israel Science Foundation (1914/08 to N.B.A.).
We thank Noah Shroyer for referring us to the putative gene bHLHA9, Sonia Canterini, who was involved in the early stages of the work, Israel Vlodavsky for reagents, Alon Zaslaver for sharing equipment, and Sophie Khazanov, Tamar Golan-Lev and Vitaly Kliminski for assistance in the experimental work. Special thanks go to Abed Mansour for advice, guidance and support.
Conflict of Interest statement: None declared.
- congenital abnormality
- gene dosage
- genetic heterogeneity
- mice, knockout
- open reading frames
- genetic pedigree
- transcription factor
- mendelian inheritance in man, online
- limb deformity
- genetic inheritance
- cleft hand, congenital
- long bone
- limb development
- copy number polymorphism