letter - Nature

letter

© 2000 Nature America Inc. • http://genetics.nature.com

Mutations in the human Delta homologue, DLL3, cause axial skeletal defects in spondylocostal dysostosis Michael P. Bulman1*, Kenro Kusumi2*, Timothy M. Frayling1, Carole McKeown3, Christine Garrett4, Eric S. Lander5, Robb Krumlauf2, Andrew T. Hattersley1, Sian Ellard1 & Peter D. Turnpenny6


*These authors contributed equally to this work.

Spondylocostal dysostosis (SD, MIM 277300) is a group of vertebral malsegmentation syndromes with reduced stature resulting from axial skeletal defects. SD is characterized by multiple hemivertebrae, rib fusions and deletions with a non-progressive kyphoscoliosis. Cases may be sporadic or familial, with both autosomal dominant and autosomal recessive modes of inheritance reported1. Autosomal recessive SD maps to a 7.8cM interval on chromosome 19q13.1–q13.3 (ref. 2) that is homologous with a mouse region containing a gene encoding the Notch ligand delta-like 3 (Dll3). Dll3 is mutated3 in the Xray–induced mouse mutant pudgy (pu), causing a variety of vertebrocostal defects similar to SD phenotypes. Here we have cloned and sequenced human DLL3 to evaluate it as a candidate gene for SD and identified mutations in three autosomal recessive SD families. Two of the mutations predict truncations

b

a

c

within conserved extracellular domains. The third is a missense mutation in a highly conserved glycine residue of the fifth epidermal growth factor (EGF) repeat, which has revealed an important functional role for this domain. These represent the first mutations in a human Delta homologue, thus highlighting the critical role of the Notch signalling pathway and its components in patterning the mammalian axial skeleton.

We previously used homozygosity linkage mapping in two consanguineous Arab-Israeli and Pakistani SD pedigrees to identify a critical genetic interval of 7.8 cM at 19q13.1–q13.3 between D19S570 and D19S908 (ref. 2). We have now confirmed this region in an additional consanguineous Pakistani kindred (maximum lod score 3.1 at θ=0). On the basis of the similarity of the phenotypes in human and mouse (Fig. 1a–f) and the homologous region (Fig. 1g), we hypothesized that a human Dll3 orthologue would be a candidate gene for the SD locus. Therefore, we cloned the previously unidentified human gene DLL3 and confirmed its localization to 19q13. Using a mouse Dll3 probe, we identified several cDNA clones representing exons 2–10 of human DLL3 (Fig. 2a). Comparison of the predicted amino acid sequence of DLL3 shows 79% identity to mouse Dll3, with EGF repeat 5 varying by only one residue (Fig. 2b). For genomic analysis, we isolated a human PAC clone and determined exon-

g d

e

f

Fig. 1 Clinical features of spondylocostal dysostosis. a, An affected child from pedigree 2 showing short trunk and short stature, with abdominal protrusion. b, An affected newborn from pedigree 1 showing truncal and neck shortening with secondary abdominal distension. c, Radiograph of an affected infant from pedigree 2. The vertebral dysgenesis is most marked in the thoracic region with multiple misaligned ribs. d, Affected homozygous SD adult from pedigree 2. The trunk is shortened and asymmetric; normally formed limbs look excessively long. e, Radiograph of an affected adult from pedigree 1. Individual vertebrae cannot be readily distinguished within a spinal column demonstrating fixed curvatures and restricted movement. f, Vertebral and rib malformations as revealed by Alizarin red/Alcian blue skeletal preparation of a neonatal Dll3pu/pu mouse. There is no loss of segments, but compressions of vertebral bodies are evident in the lumbo-sacral region. Skeletal defects also encompass other sclerotomal derivatives, including bifurcations and fusions of ribs and delayed ossification of the occipital plate. g, Comparison of the regions of syntenic conservation on human chromosome 19q13.1 in the SD genetic interval2 and mouse chromosome 7 surrounding the Dll3pu locus3.c, e, reproduced with permission of The University of Chicago Press. © 1999 by The American Society of Human Genetics.

1Molecular Genetics, School of Postgraduate Medicine and Health Sciences, Barrack Road, Exeter, UK. 2Division of Developmental Neurobiology, National Institute for Medical Research, The Ridgeway, London, UK. 3Clinical Genetics Unit, Birmingham Women’s Hospital, Edgbaston, Birmingham, UK. 4Kennedy Galton Centre, Northwick Park and St. Mark’s Trust, Harrow, Middlesex, UK. 5Whitehead Institute for Biomedical Research, Cambridge, Massachusetts, USA. 6Clinical Genetics, Royal Devon & Exeter Hospital, Barrack Road, Exeter, UK. Correspondence should be addressed to P.D.T. (e-mail: [email protected]).

438

nature genetics • volume 24 • april 2000


letter

Fig. 2 Organization of human DLL3. a, Genomic organization of exons 2–10 of DLL3, using the exon designations of the mouse orthologue Dll3 (ref. 3). The locations of the exon-intron junctions within the predicted amino acid sequence are identical, although there is variability in the size of introns. In addition, the intron following exon 9 observed in mouse Dll3 is absent from human DLL3, resulting in a large terminal exon which is designated 9/10. b, Alignment of the predicted amino acid sequence of human and mouse DLL3. The location of exon-intron boundaries and the main protein motifs within the transcript are indicated on the left and include the following: the signal sequence (SS), the DSL domain, the EGF repeats and the transmembrane (TM) region. Conserved residues are shown in grey. Conserved glycine and cysteine residues of the EGF repeats are indicated in yellow and red, whereas red highlights the highly conserved glycine residue mutated in SD pedigree 3.

a

b exon

protein


Dll3 Dll3

intron boundaries using inter-exon PCR based on the mouse structure. The exon-intron junctions within the predicted amino acid sequence are identical to those of mouse Dll3, with the exception of the terminal exon, which corresponds to a fusion of mouse exons 9 and 10 (Fig. 2a). This difference would result in the human protein having 32 additional amino acids (Fig. 2b). To search for SD mutations, we sequenced the coding region and splice sites of DLL3 in affected individuals and identified a unique mutation in DLL3 in each of the three pedigrees (Fig. 3). In pedigree 1, there is a 5-bp insertion in exon 5 predicted to truncate the protein in the Delta-Serrate-Lag2 (DSL) domain before EGF repeat 1 (Figs 3a and 4). In pedigree 2, there is a 2-bp deletion in the fourth EGF repeat domain that predicts a truncation immediately after EGF repeat 3 (Figs 3b and 4). In pedigree 3, we discovered a missense mutation in which aspartic acid replaces glycine in EGF repeat 5 (Fig. 3c and 4). This residue (Fig. 2b) is highly conserved in Delta proteins from Drosophila melanogaster to humans and the substitution of a charged polar for a nonpolar residue may disrupt the conformation of the DLL3 protein. Testing of all available family members in each pedigree confirmed that affected SD individuals were homozygous and obligate carriers heterozygous for the mutation. These mutations were not observed in 72 random population controls. nature genetics • volume 24 • april 2000

Mutations in DLL3 result in a core SD phenotype involving malformations of the axial skeleton, with affected individuals displaying similar skeletal defects (Fig. 1). In the mouse, mutation analysis has shown that genes in the notch pathway have a role in the formation of boundaries in somitic mesoderm (the precursor to adult vertebrae), ribs and axial musculature3–7. Therefore, the skeletal defects in SD pedigrees reflect primary disruptions of early embryonic mesodermal patterning dependent upon DLL3-mediated notch signalling. Dll3 expression is not limited to somites in the mouse8, suggesting that other nonskeletal defects in the human SD pedigrees may represent additional primary defects in notch signalling. Dll3 is expressed in the brain and spinal cord of the developing nervous system8, and defects in neural patterning have been observed in the Dll3pu mouse3. Clinical examination did not identify any neurological abnormalities in affected SD individuals, however, and none had mental retardation. Hence, further neurological examination and detailed neuroimaging analysis may uncover previously undetected defects in SD pedigrees. We have observed additional defects associated with the core SD skeletal phenotype. In pedigree 1, an individual succumbed in infancy, and at necropsy was found to have a large patent ductus arteriosus and a membranous diaphragm9. Hence DLL3 might have a direct role in the development of cardiovascular mesodermal tissues. With respect to other abnormalities, some affected males in pedigree 1 display unilateral or bilateral inguinal herniae, but this may be a secondary effect of truncal shortening. In general, it appears that non-skeletal organ system involvement is unusual in familial SD, but there have been reports of anal and urogenital anomalies10, congenital heart disease11,12 and inguinal herniae in males9,13. Further detailed comparisons between the human phenotypes and the Dll3pu mouse model may help distinguish between direct and indirect effects. In mammals, four paralogues of the notch receptor have been identified14–18, which bind to two distinct families of ligands––delta and jagged/serrate. Two delta (refs 8,19), two jagged (refs 20–23) and three Fringe (a notch signalling modulator; refs 24,25) genes have been reported. Given the competitive and overlapping interactions between Delta and Serrate in Drosophila development, variability in the human SD phenotypes and in comparison with Dll3pu may arise due to interactions with other members of the Notch signalling pathway. The importance of the Notch pathway in human patterning is demonstrated by the autosomal dominant NOTCH3 CADASIL syndrome26, associated with strokes and dementia, and the JAG1 Alagille syndrome20,21, associated with developmental abnormalities including hemivertebral defects. The dominant Alagille mutations lead to truncations before the transmembrane domain and similar mutations have been shown to be dominant-negative Notch pathway components in Drosophila27–30. In contrast, similar SD truncations are recessive, although in pedigree 1 (Fig. 3a) a heterozygous carrier exhibits mild defects, including short stature, left hemifacial microsomia, a mild mid-thoracic scoliosis 439

letter



a

c

b

Fig. 3 Mutations in DLL3 identified in SD pedigrees. Partial pedigrees are shown with mutation status indicated as follows: NN, homozygous normal; NM, heterozygous mutation; MM, homozygous mutation. Affected and unaffected individuals are represented by filled and open symbols, respectively. The uncertainty regarding the affection status of one individual in pedigree 1 is indicated by a question mark. Double lines indicate consanguineous marriages. Electropherograms documenting each mutation are shown in an unaffected (top) and an affected (bottom) individual. a, Mutation 953ins GCGGT in the ArabIsraeli pedigree. b, Mutation 945delAT in the Kashmiri pedigree. c, Mutation G385D in the Rawalpindi pedigree.

and minor vertebral dysostosis of the lower lumbar vertebrae. Although the other nine heterozygotes in this pedigree were phenotypically normal and no heterozygous defects have been detected in the adult Dll3pu mouse, this finding raises the possibility of haploinsufficiency and modifier loci. Based on our work, genes encoding human notch signalling pathway components are candidates for examining SD pedigrees that do not show linkage to DLL3/19q13. LFNG is a candidate given that mutations in the mouse gene yield axial skeletal defects

SD1

SD2

SD3

very similar to those of Dll3pu mice6,7. Finally, our report of the first mutations in a human Delta homologue underscores the key roles of the notch signalling pathway in normal patterning and diseases of the human axial skeleton.

Methods Pedigrees. We studied three large consanguineous pedigrees with autosomal recessive SD. The large Arab-Israeli kindred2,9 and a second family from Rawalpindi, Pakistan2 have been described. The third pedigree originates from Kashmir, Pakistan and has a phenotype similar to that of the two previously reported families. Controls. We recruited 72 controls from two populations: 33 Arab-Israelis from Nazareth and 39 ethnic Pakistanis from the West Midlands region of the United Kingdom. Linkage analysis. We selected microsatellite markers for the linked region of 19q from the Généthon linkage set2. PCR products were electrophoresed on an ABI 377 DNA sequencer and analysed with Genescan and Genotyper software. We constructed haplotypes and calculated two-point linkage analysis between affected status and the haplotype apparently segregating with the disease using the MLINK program (version 5.2) of the LINKAGE package. An AR mode of inheritance and a disease-allele frequency of 0.0001 were used.

Fig. 4 A diagram representing the mutations in human and mouse DLL3. Mutant alleles in SD pedigree 1 and 2 and Dll3pu would not be expected to produce functional DLL3 product. A missense mutation in a highly conserved glycine residue within SD pedigree 3 is predicted to disrupt EGF repeat 5, highlighting the functional importance of this domain.

440

Human DLL3 assembly. We used the mouse Dll3 sequence to identify the following human EST sequences by BLAST analysis: AI1880178 (IMAGE 1957080), AI214075 (IMAGE 1956799), AI356220 (IMAGE 2016894) and THC283856. Exon-intron boundaries for human DLL3 were extrapolated



by comparison with mouse Dll3 and later confirmed by sequencing of genomic DNA. We screened the human RPCI-1 PAC library by PCR, identifying DLL3-positive clone 265F22. Introns were PCR amplified from this clone and human genomic DNA, and sequenced using dye terminator chemistry (PE Biosystems). Genomic draft sequence, deposited after our initial characterization of DLL3 and identification of SD mutations, was used to confirm genomic organization.


DNA sequencing. We amplified genomic DNA by PCR and purified products using QIAquick purification columns (Qiagen). Both strands were sequenced using a BigDye Terminator Cycle Sequencing kit (PE Biosystems) according to the manufacturer’s recommendations. Reactions were analysed on an ABI Prism 377 DNA Sequencer (PE Biosystems). Mutation assays. The G385D mutation results in the loss of a restriction site for Eco0109I. Control samples were tested by amplification with primers 5´–GCAGCGATGACAGAGCTG–3´ (forward) and 5´–TTGAGT GCATCCGGGAGT–3´ (reverse) and digested with Eco0109 I to yield fragments of 480, 106 and 39 bp for the normal allele, and products of 586 and 39 bp from the mutant allele. The 2-bp deletion (945delAT) mutation was tested by using a length polymorphism assay with PCR primer sequences 5´–TGAATGCACCTGCCCGCGTG–3´ (forward) and 5´–GCAGATGTAG GCAGAGTCAG–3´ (reverse). The forward primers were labelled with the fluorescent dye FAM. PCR products were electrophoresed on an ABI 377 DNA sequencer and analysed with Genescan and Genotyper software. The

1.

2. 3. 4. 5. 6.

7. 8.

9.

10.

11. 12.

13. 14.

15.

Mortier, G.R., Lachman, R.S., Bocian, M. & Rimoin, D.L. Multiple vertebral segmentation defects: analysis of 26 new patients and review of the literature. Am. J. Med. Genet. 61, 310–319 (1996). Turnpenny, P.D. et al. A gene for autosomal recessive spondylocostal dysostosis maps to 19q13.1–q13.3. Am. J. Hum. Genet. 65, 175–182 (1999). Kusumi, K. et al. The mouse pudgy mutation disrupts Delta homologue Dll3 and initiation of early somite boundaries. Nature Genet. 19, 274–278 (1998). Conlon, R.A., Reaume, A.G. & Rossant, J. Notch1 is required for the coordinate segmentation of somites. Development 121, 1533–1545 (1995). de Angelis, M.H., McIntyre, J.I. & Gossler, A. Maintenance of somite borders in mice requires the Delta homologue Dll1. Nature 386, 717–721 (1997). Evrard, Y.A., Lun, Y., Aulhehla, A., Gan, L. & Johnson, R.L. lunatic fringe is an essential mediator of somite segmentation and patterning. Nature 394, 377–381 (394). Zhang, N. & Gridley, T. Defects in somite formation in lunatic fringe-deficient mice. Nature 394, 374–377 (1998). Dunwoodie, S.L., Henrique, D., Harrison, S.M. & Beddington, R.S.P. Mouse Dll3: a novel divergent Delta gene which may complement the function of other Delta homologues during early pattern formation in the mouse embryo. Development 124, 3065–3076 (1997). Turnpenny, P.D., Thwaites, R.J. & Boulos, F.N. Evidence for variable gene expression in a large inbred kindred with autosomal recessive spondylocostal dysostosis. J. Med. Genet. 28, 27–33 (1991). Casamassima, A.C. et al. Spondylocostal dysostosis associated with anal and urogenital anomalies in a Mennonite sibship. Am. J. Med. Genet. 8, 117–127 (1981). Delgoffe, C. et al. Dysostoses spondylocostales et cardiopathies congénitales. Ann. Pédiat. 29, 135–139 (1982). Simpson, J.M., Cook, A., Fagg, N.L.K., MacLachlan, N.A. & Sharland, G.K. Congenital heart disease in spondylothoracic dysostosis: two familial cases. J. Med. Genet. 32, 633–663 (1995). Bonaime, J.L. et al. Le syndrome de dysostose spondylothoracique ou spondylocostale. Pédiatrie 33, 173–188 (1978). Robbins, J., Blondel, B.J., Gallahan, D. & Callahan, R. Mouse mammary tumor gene int-3: a member of the notch gene family transforms mammary epithelial cells. J. Virol. 66, 2594–2599 (1992). Lardelli, M., Dahlstrand, J. & Lendahl, U. The novel Notch homologue mouse Notch 3 lacks specific epidermal growth factor-repeats and is expressed in


letter

normal and mutant alleles gave products of 129 bp and 127 bp, respectively. The 5-bp insertion (953insGCGGT) mutation was tested in controls by sequencing of exon 5. GenBank accession numbers. DLL3 genomic sequences, AF241367– AF241373; Dll3, AF068865; DLL3 genomic draft sequence, AC011500. Acknowledgements

We thank the families for cooperation; K. Maruthainar, K. Dewar and B. Birren for undertaking sequencing efforts; and the management and laboratory staff of The Nazareth Hospital, Israel for their help. This work was supported by the British Scoliosis Research Foundation, the Medical Research Council (UK), Action Research, the Skeletal Dysplasia Group (UK), the Children’s Research Fund, the Darlington Charitable Trust, the Royal Devon & Exeter NHS Healthcare Trust and the University of Exeter. The assistance of the DNA Laboratories of the West Midlands Regional Genetics Service, Birmingham, the Yorkshire Regional Genetics Service, Leeds and the Kennedy-Galton Centre, London is appreciated. K.K. is supported by a Hitchings-Elion Fellowship of the Burroughs Wellcome Fund.

Received 28 December 1999; accepted 16 February 2000.

proliferating neuroepithelium. Mech. Dev. 46, 123–136 (1994). 16. Lardelli, M. & Lendahl, U. Motch A and motch Btwo mouse Notch homologues coexpressed in a wide variety of tissues. Exp. Cell. Res. 2, 364–372 (1993). 17. Ellisen, L.W. et al. TAN-1, the human homolog of the Drosophila notch gene, is broken by chromosomal translocations in T lymphoblastic neoplasms. Cell 66, 649–661 (1991). 18. Weinmaster, G., Roberts, V. & Lemke, G. A homolog of Drosophila Notch expressed during mammalian development. Development 113, 931–941 (1991). 19. Bettenhausen, B., Hrabe de Angelis, M., Simon, D., Guenet, J.L. & Gossler, A. Transient and restricted expression during mouse embryogenesis of Dll1, a murine gene closely related to Drosophila Delta. Development 121, 2407–2418 (1995). 20. Li, L. et al. Alagille syndrome is caused by mutations in human Jagged1, which encodes a ligand for Notch1. Nature Genet. 16, 243–251 (1997). 21. Oda, T. et al. Mutations in the human Jagged1 gene are responsible for Alagille syndrome. Nature Genet. 16, 235–242 (1997). 22. Sidow, A. et al. Serrate2 is disrupted in the mouse limb-development mutant syndactylism. Nature 389, 722–725 (1997). 23. Lindsell, C., Shawber, C., Boulter, J. & Weinmaster, G. Jagged: a mammalian ligand that activates Notch1. Cell 80, 909–917 (1995). 24. Johnston, S.H. et al. A family of mammalian Fringe genes implicated in boundary determination and the Notch pathway. Development 124, 2245–2254 (1997). 25. Cohen, B. et al. Fringe boundaries coincide with Notch-dependent patterning centres in mammals and alter Notch-dependent development in Drosophila. Nature Genet. 16, 283–288 (1997). 26. Joutel, A. et al. Notch3 mutations in CADASIL, a hereditary adult-onset condition causing stroke and dementia. Nature 383, 707–710 (1996). 27. Greenwald, I. Structure/function studies of lin-12/Notch proteins. Curr. Biol. 4, 556–562 (1994). 28. Thomas, U., Speicher, S.A. & Knust, E. The Drosophila gene Serrate encodes an EGF-like transmembrane protein with a complex expression pattern in embryos and wing discs. Development 111, 749–761 (1991). 29. Kopczynski, C.C., Alton, A.K., Fechtel, K., Kooh, P.J. & Muskavitch, M.A.T. Delta, a Drosophila neurogenic gene, is transcriptionally complex and encodes a protein related to blood coagulation factors and epidermal growth factor of vertebrates. Genes Dev. 2, 1723–1735 (1988). 30. Artavanis-Taskonas, S. Alagille syndromea notch up for the Notch receptor. Nature Genet. 16, 212–213 (1997).

441