The sonic hedgehog (SHH) gene (MIM# 600725) has a complex expression pattern, which is crucial to the developmental function of the secreted protein. This pattern of expression is dependent on a number of cis-regulatory elements that are scattered within the introns and throughout the 1 Mb of DNA upstream of the gene [Jeong et al., 2006; Jeong and Epstein, 2003; Lettice et al., 2003; Sagai et al., 2009]. Heterozygous mutations, which cause loss of Shh activity, are associated with holoprosencephaly spectrum (HPE; MIM# 142945) [for review see Solomon et al., 2010]. Both intragenic mutations and chromosomal rearrangements presumed to result in loss of cis-regulators cause a highly variable HPE phenotype [Belloni et al., 1996].
SHH has an important role during limb development. The zone of polarizing activity (ZPA) plays a key role in limb morphogenesis and is located along the posterior margin of developing fore and hind limbs [Hill, 2007]. The molecular function of the ZPA is to produce and secrete SHH, which acts as a morphogen to regulate the identity of each developing digit and to specify the number of digits that form. The cis-regulatory element that is responsible for this limb expression of SHH is termed the ZRS (for ZPA regulatory sequence) [Lettice et al., 2003]. The ZRS is both necessary and sufficient to regulate limb-specific Shh expression as shown by transgenic (TG) experiments in which the 800 bp, highly conserved ZRS recapitulates the Shh expression pattern [Lettice et al., 2007; Maas and Fallon, 2005]. In addition, deletion of the ZRS specifically eliminates Shh expression in the limb without affecting other Shh expression domains [Sagai et al., 2005]. Regulatory mutations involving the ZRS activate Shh expression at an ectopic site along the anterior margin of the limb bud. [Lettice et al., 2003] This misexpression is central to a group of limb defects, which are called the “ZRS associated syndromes” [Wieczorek et al., 2010]. These defects include preaxial polydactyly type 2 and type 3 (PPD 2/3, MIM#s 174500, 174600), triphalangial thumb polysyndactyly (TPTPS; MIM# 190605)), syndactyly type IV (SD4; MIM# 186200), and Werner's mesomelic syndrome (MIM# 188770). Well over 20 different point mutations in human, mouse, and the domestic cat are known to cause PPD [Albuisson et al., 2011; Dobbs et al., 2000; Farooq et al., 2010; Furniss et al., 2008; Gurnett et al., 2007; Lettice et al., 2003; Lettice et al., 2008; Sato et al., 2007; Wieczorek et al., 2010]. In addition, chromosomal rearrangements that result in genomic duplications of the ZRS are associated with TPTPS and SD4 [Klopocki et al., 2008; Sun et al., 2008]. Here, we identify a novel regulatory mechanism for generating ectopic SHH expression that occurs due to a large-scale intrachromosomal rearrangement that puts SHH under the influence of another limb enhancer. The resulting chromosomal configuration we refer to as “enhancer adoption” and postulate that this is a common cause of malformation in patients with chromosomal rearrangements.
Materials and Methods
Patient Samples and Breakpoint Mapping by FISH Analysis
Initial identification of the inversion was performed in the local Clinical Cytogenetics laboratories. Research samples were then obtained with full ethical approval and written consent was obtained for this study. Metaphase chromosome preparations were prepared from Epstein-Barr-virus-mediated lymphoblastoid cell lines established from peripheral blood of the proband. Mapping clones were chosen and were obtained from BACPAC Resources Center, CHORI, Oakland, CA, USA and the Wellcome Trust Sanger Institute, UK. Clone DNA was prepared by a standard mini-prep method and labeled with digoxigenin-11-dUTP or biotin-16-dUTP (Roche Diagnostics Ltd, Burgess Hill,UK) by nick translation. Probe labeling, DNA hybridization and antibody detection were carried out using methods described previously [Chong et al., 1997]. At least 10 metaphases were analyzed for each hybridization using a Zeiss Axioskop 2 microscope (Carl Zeiss Uk Ltd, Hertfordshire,UK) with the appropriate filters (#83000 for DAPI, FITC and rhodamine; Chroma Technology GmbH, Olching,Germany). Images were collected and merged using a Coolsnap HQ CCD camera (Photometrics, Tucson, AZ) and SmartCapture 2 (Digital Scientific UK, Cambridge,UK) software. For fine mapping of the 7q22.1 BP, oligos were designed (set1 GGGTCCAACATGCCACTCTCCTTGG and GTGCGCCTCCTGGAAGAGCTCAGAA; set2 TCAGGGCCTGGTTACGTCCGTTCTG and CAGCCTGGGATTGCTCAGGGAAGGT) to amplify two overlapping 10-kb probes using Expand Long Template polymerase chain reaction (PCR) kit (Roche Diagnostics Ltd) using the manufacturers conditions. The PCR products were then purified using a QiaQuick PCR purification kit (Qiagen Ltd, Crawley,UK). Probes were labeled with digoxigenin-11-dUTP (Roche Diagnostics Ltd) by nick translation. Fluorescent in situ hybridization (FISH) analysis of the probes was performed as described above.
Transgenic Constructs and Embryos
The β-galactosidase (β-gal) expression TG constructs made use of a vector containing the β-globin minimal promoter (βGMP) and the bacterial lacZ reporter gene [Yee and Rigby, 1993] (vector p1230, a kind gift from Robb Krumlauf). The HNCE2 and HNCE1 were synthesized by PCR from human DNA. The primers used for amplifying the HCNE1 sequence (GATCAT AAGCTT GGCAGGGGCAGAAGCGCC and GATCAT AAGCTT TCCCACAGACGGAAGTGATC) and the HCNE2 sequence (GATCAT AAGCTT CCTGGCCCTGCCGTCAGCAG and GATCAT AAGCTT AAATAAAAGACATCCTGGAAC) contain HindIII sites (shown in bold) on the end of each for ease of cloning. The products were verified by sequencing. The Shh misexpression construct was assembled from the HNCE2 fragment, an oligo containing the βGMP and the genomic region of mouse Shh from the translational start to the end of the 3′ UTR of the cDNA. A control construct lacking the HNCE2 was also made. In all cases, the vector fragment was removed by NotI/SalI digestion and DNA for microinjection was prepared by electroelution (Elutrap, Schleicher and Schuell, Keene, NH). TG mice were made by pronuclear injection using standard protocols. G0 embryos were harvested at the appropriate stage or allowed to develop to term. TG males were subsequently used as studs with CD1 females. All embryos harvested had their yolk sacs retained to allow for PCR genotyping. β-gal staining and in situ hybridizations were carried out using standard techniques [Hecksher-Sorensen et al., 1998]. The probes used for in situ hybridization were Shh and Ptc (a kind gift from Andy McMahon and Chris Hayes, respectively).
Sequence Analysis of HCNE2
Genomic sequences were obtained from Ensembl release 60, November 2010 and manipulated using programs from the EMBOSS package [Rice et al., 2000]. For genomes where the annotation was not clear enough, we searched for orthologous regions using Blast [Altschul et al., 1997]. To judge conservation of the presumptive enhancer in different organisms, large genomic sequences were aligned and visualized using Pipmaker [Schwartz et al., 2000]. A multiple alignment of the conserved region in the different species was done using CLUSTALW [Thompson et al., 2002]. TRANSFAC professional 12.1 [Matys et al., 2006] matrices for E12, Sox9 and HOX5A with MATRIX-SCAN, a tool from the Regulatory Sequence Analysis Tool (RSAT) [van Helden, 2003] was used to look for potential binding sites in the enhancer sequence.
Rendering, Manipulation and Length Measurements of Optical Projection Tomograms
Optical projection tomography (OPT) was performed on whole mouse embryos mounted in 1% agarose, dehydrated in methanol, and then cleared overnight in BABB (1 part benzyl alcohol: 2 parts benzyl benzoate). The sample was then imaged using a Bioptonics OPT Scanner 3001 (Bioptonics, Edinburgh, UK) using bright field to detect the LacZ staining and for tissue autofluorescence (excitation 425 nm/emission 475 nm) to capture the anatomy [Sharpe et al., 2002]. The digital reconstructed images of embryos were acquired using Amira® v5.2.1 (Visage Imaging GmbH, Berlin, Germany), where surface rendering and volume rendering were done using the Isosurface and Volren modules, respectively. Embryo limbs were virtually dissected using the VolumeEdit module, while length measurements were taken using the three-dimensional Length tool, both in orthographic viewing mode. Original OPT scans were taken at different magnifications per embryo. At E11.5, 7 TGs and 12 wild-type (WT) embryos and at E12.5 11 TG and 13 WT embryos were examined.
The proband was born at full term by normal delivery with a birth weight of 3.86 kg (75th centile) and a head circumference of 34 cm (ninth centile) following an unremarkable pregnancy. She is the eldest of three siblings, her parents are unrelated, and there is no relevant family history. Antenatal ultrasound scan at 20 weeks was normal. At birth complete syndactyly of the hands and feet was noted with all nails being fused to form a single band in each limb (Fig. 1A–D). X-rays of the hands showed a normal number of digits in the hands but with fused terminal phalanges and an abnormal terminal phalanx of both thumbs while both feet showed PPD (Fig. 1E and F). Contractures were present at both elbows and knees. Echocardiography revealed a small atrial septal defect that later resolved. She is hypoteloric (Fig. 1A) and has a solitary maxillary central incisor (SMCI), which are facial features associated with HPE spectrum disorder but a magnetic resonance imaging scan showed no detectable structural brain anomalies. Her height and weight followed the second to ninth centiles but her head circumference at the age of 4 years was well below the 0.4th centile. There was some mild speech delay but otherwise intellectual development was normal.
Cytogenetic Analysis Reveals a Breakpoint that Disrupts SHH Gene Regulation
Cytogenetic analysis revealed a de novo, apparently balanced inversion on chromosome 7 with breakpoints (BPs) at 7q22.1 and 7q36.3. BPs mapping by metaphase FISH showed that the 7q36.3 BP lies within BAC RP11-161A19 (hg19 chr7:155,556,692-155,723,755) (Fig. 1G, Supp. Table S1) and FOSMIDS G248P86063E9 (hg19 chr7:155,633,516-155,671,683) and G248P85537F4 (hg19 chr7:155,638,129-155,677,858) (Fig. 1I, indicated by pink bar) thus localizing the BP to a 33.5-kb region between hg19 ch7:155,638,129-155,671,683 (Supp. Table S1). The transcription start site of the gene (SHH maps ∼50 kb from the 7q36.3 BP. The 7q22.1 BP maps within RP5-1059M17 (Fig. 1H) (chr7:101,190,619-101,362,597 hg19). Fine mapping using 10-kb PCR products (indicated in Fig. 1I) amplified from RP5-1059M17 suggested that the BP lies in the telomeric half of hg19 chr7:101,265,853-101,275,861 either disrupting or immediately 5′ of the gene MYL10 (myosin light chain 10, regulatory) (Fig. 1I, Supp. Table S2).
7q36.3 BP in apparently balanced chromosomal rearrangements (ABCRs) has been associated with HPE spectrum disorders (HPE, SMCI, and hypotelorism) [Roessler et al., 1997] most likely due to the loss of cis-acting early forebrain enhancers of SHH [Jeong et al., 2006] (Fig. 1I and J). In addition, there may be associations with enhancers further upstream such as the dental placode enhancer (MRCS1 in Fig. 1I) [Sagai et al., 2009] in the observed SMCI. However in human, limb malformations are not a reported feature of BPs in this region. This is in agreement with the targeted deletion of the ZRS in mouse [Sagai et al., 2005] that show that loss of a single copy of this regulator does not generate a developmental phenotype.
Analysis of Conserved Noncoding Elements as Candidates for Enhancer Adoption
The expression of SHH in the limb is central to defining digit number. Misexpression of SHH at ectopic sites in the limb is due to dominant gain of function mutations, which produces polydactyly [Lettice et al., 2007] but is also associated with syndactyly in TPTPS and SD4 [Klopocki et al., 2008; Sun et al., 2008]. Therefore, the proband's malformations are consistent with novel, ectopic expression of SHH at crucial stages in limb development. Since the relocated SHH gene was no longer under the regulation of the ZRS (Fig. 1J), we investigated the possibility that the chromosomal inversion placed the gene under the influence of an “adopted cis-regulator” at 7q22.1. We used a comparative genomic approach to identify candidate enhancers in the 500-kb centromeric to the 7q22.1 BP (Fig. 1J). By comparing the orthologous regions in the human and chick genome, we identified two highly conserved noncoding elements (HCNEs) that were >200 bp in length and >80% sequence identity; HCNE1 (hg19 chr7:101,268,896-101,269,123) within an intron of MYL10 and HCNE2 (hg19 chr7:101,129,817-101,130,058) within an intron of EMID2 (MIM# 608927). The human HCNE2 is highly conserved in vertebrates (Supp. Fig. S1A and B) being found in evolutionary distant species such as amphibians. A number of potential, conserved binding sites for factors found in limb mesenchyme such as Sox9 were predicted by TRANSFAC (Supp. Fig. S1B).
The HCNE2 appeared to be a good candidate to examine as a limb enhancer. The EMID2 gene, which accommodates this conserved element, is expressed in the embryonic limb [Diez-Roux et al., 2011] and HCNE2 was reported to be bound by p300, a factor indicative of enhancer activity, in embryonic limb tissue at E11.5 [Visel et al., 2009]. Therefore, HCNE2 may normally function as an enhancer responsible for EMID2 expression in embryonic limbs. Both the HCNE1 and two enhancers were tested for limb expression activity using TG constructs containing the bacterial-derived reporter gene encoding β-gal. Stable reporter TG lines were made. Consistent with the p300 data, four independent stable lines carrying HCNE2 (out of nine established lines) showed a pattern of expression in the developing limb bud. Three of the lines (Lines A, B, and C) displayed similar expression patterns (Fig. 2A–I) while Line D showed expression in the limb but in a broader pattern (data not shown). At E11.5 (Fig. 2A, D, and G) expression is in the condensing mesenchymal core of the limb and appears in all three lines. Expression continues in embryogenesis and is similar amongst all three lines at E12.5 (Fig. 2B, E, and H). At E14.5, Lines A and B expression reflect the pattern reported for EMID2 [Diez-Roux et al., 2011] (Fig. 2C and F); however, Line C (Fig. 2I) expression is difficult to detect by this stage. Strong expression of the HNCE2 reporter is also seen in the craniofacial region. In contrast, HNCE1 showed no developmental expression at E11.5, E12.5, or E14.5 in any of the eight stables lines analyzed.
HCNE2 Activity Produces Digit Abnormalities
To determine if HNCE2 has a role in the proband's limb malformation, we used transient transgenesis in mouse embryos to model the cis-regulatory effect of the inversion. The entire genomic transcription unit of mouse Shh (9.41 kb; hg19 chr7:155,595,558-155,604,967) was amplified and inserted 3′ of the β-globin minimal promotor (βGMP) either alone (control vector) or with HCNE2 5′ of βGMP (test vector) (Fig. 3A). The heterologous βGMP is known to have very low background activity and in accord the control vector [Lettice et al., 2007; Yee and Rigby, 1993] gave no observable phenotype. We examined 10 control TG embryos at E14.5 to look for phenotypic differences and found no significant difference to WT. In contrast, embryos that were TG for the HNCE2-βGMP-Shh construct had PPD when examined at E14.5 (2/12) (Fig. 3C and D) or had enlarged hand plates at E12.5 (6/14) and E11.5 (9/24) (Supp. Table S3). Ectopic expression of Shh (Fig. 3F and G) and its receptor, Ptc (Fig. 3H and I), which is induced by Shh expression and is a sensitive read-out of functional SHH signaling, was found at E11.5. Ectopic, anterior patches of Shh were detected and corresponding expression of Ptc in the ectopic domain was seen (arrows in Fig. 3H and I). The Ptc expression also showed further ectopic expression in the distal mesenchyme (Fig. 3I) of the hind limbs. At E12.5, Shh expression persists in the distal mesenchyme and is found in the digit and interdigital tissue (Fig. 3J and K). Some embryos also showed facial clefting (Fig. 3E). Each TG embryo was further analyzed by OPT and each limb was digitally reconstructed for further quantitative analysis (Fig. 4A). The width of each limb plate (LP) and the crown to rump length was measured (Fig. 4B, C at E11.5 and Fig. 4D, E at E12.5) and the TG embryos revealed a significant increase in the LP:crown rump (LP:CR) ratio (Fig. 4F) in the TG group with no alteration in the overall body size of the TG embryos (Fig. 4G). Hence, the activity of the HCNE2 regulator is sufficient to drive ectopic Shh in the limb to cause footplate expansion and eventual polydactyly. In addition, at E14.5 the embryos with the affected limbs showed the persistence of soft tissue between the digits (arrows in Fig. 3D), which may manifest as syndactyly similar to that seen in the hands and feet of the child. These analyses in mouse embryos show that the HCNE2/SHH TG construct is capable of generating a suitable phenocopy of the child's limb abnormalities.
Analysis of human ABCR has been crucial to the identification of long-range cis-regulatory control of developmental gene expression [Kleinjan et al., 2001; Kleinjan and van Heyningen, 1998]. These were first recognized as distant genomic rearrangements, usually translocations, which generated the same phenotype as mutations within the coding region of the gene. Rearrangements of the PAX6 locus associated with aniridia have become the paradigm for this mechanism. Mutations in the SHH enhancer ZRS causing PPD remains the best example of gain of function cis-regulatory mutations [Lettice et al., 2008]. Here, we present a case in which both gain- and loss-of-function cis-regulatory mechanisms are likely to be acting. The HPE spectrum disorder in this child is almost certainly related to the loss of SHH enhancers as a result of the inversion. We also present experimental evidence for a gain of function effect as a cause of the preaxial synpolydactyly. We have used the term “enhancer adoption” to describe this mechanism of human developmental disease.
The synpolydactyly phenotype observed in the child is reproduced in the mouse model by the ectopic expression of the SHH gene. The ectopic Shh expression in the anterior limb bud margin produced in the animal model strongly relates to the polydactyly seen in the feet of the child but the hands and feet also show an unusual, extensive syndactyly. The mouse model provides further insights into this aspect of the phenotype. We found that ectopic Shh expression continues up to at least E12.5 detectable in the digital and interdigital mesenchyme. Cell death within the interdigital mesenchyme is an important step in generating individual, separated digits. Studies in chick show that SHH activity has the capacity to rescue this tissue from cell death [Sanz-Ezquerro and Tickle, 2000]. Application of SHH-soaked beads to interdigital mesenchyme of chick limb buds leads to persistence of the soft tissue between digits resulting in syndactyly. We suggest that, similar to the chick, the continued expression of Shh in the interdigital tissue of the mouse model causes persistence of the interdigital soft tissue and in the patient, is the basis for the extensive syndactyly.
We cannot however, discount the possibility that other genes near the inversion BPs contribute to the phenotype. For example, at one end of the BP the Cux1 gene is relocated to a position such that its regulation would potentially be under the influence of the ZRS (Fig. 1J). Based on evidence from mouse genetic analysis, we do not predict that Cux1 plays a contributing role. Firstly, targeted deletions of the Cux1 gene in mice do not produce an overt limb phenotype [Ellis et al., 2001; Luong et al., 2002] showing that Cux1 does not normally play a role in limb development. Secondly, and perhaps more relevant, studies in which Cux1 is deregulated in TG mice result in a number of hypoplastic affects on organ systems but does not detectably cause limb abnormalities [Ledford et al., 2002]. At the opposite end of the inversion, the 5′ end of MYL10 is close to or overlaps the BP and the expression is likely to be affected. It is less clear what role the haploinsufficiency of MYL10 would play in the phenotype; however, MYL10 is described as highly tissue specific being found predominantly in precursor B and T lymphocytes [Oltz et al., 1992] and therefore may be irrelevant to the limb. Finally, the gene (or genes) normally under the influence of the HCNE2 enhancer may be altered when its activity is usurped by the relocation of the SHH gene. It is difficult to predict how such an event may affect surrounding genes but to understand neighboring regulatory affects and how long-range regulators are affected by the influence of a new regulatory environment will take the examination of further mouse models.
The acquisition of long-range cis-regulatory elements by genomic rearrangement may prove to be a relatively common disease mechanism in ABCR-related phenotypic anomalies. In fact, another example of misregulation of the Shh gene due to a chromosomal inversion was reported in the mouse Dsh (short digit) mutant also involving a BP between the ZRS and the Shh gene [Niedermaier et al., 2005]. The dominant heterozygous limb phenotype is due to ectopic Shh expression, which occurs at late stages in limb development disrupting chondrogenesis in the phalanges. The limb defects phenocopy those of human brachydactyly type A1. Shh misregulation was postulated to be due to either loss of a repressor of limb expression or perhaps more likely, especially in light of the present data, the gain of new regulatory information. By whatever means, it is clear that there are a number of mechanisms for generating regulatory mutations of the Shh gene in the limb. Point mutations, ZRS duplications, and enhancer adoptions can produce a range of limb abnormalities, the nature of the phenotype depending on the spatial or temporal misregulation of SHH. Comparative genomics and analysis of chromatin modifications in embryonic tissues has led to the identification of many more bona fide developmental enhancers in recent years [Pennacchio et al., 2006; Visel et al., 2009]. Using the transient TG approach that we describe here it becomes practical to develop animal models of putative cis-regulatory effects of individual chromosomal rearrangements associated with developmental disorders.
We would like to thank the family for their participation. We would also like to thank the staff at the Evans Building for all their expert technical assistance. These studies were supported by an MRC Core Grant.