REPORT RAB23 Mutations in Carpenter Syndrome Imply an Unexpected Role for Hedgehog Signaling in Cranial-Suture Development and Obesity

Dagan Jenkins, Dominik Seelow, Fernanda S. Jehee, Chad A. Perlyn, Lus G. Alonso, Daniela F. Bueno, Dian Donnai, Dragana Josiova, Irene M. J. Mathijssen, Jenny E. V. Morton, Karen Helene rstavik, Elizabeth Sweeney, Steven A. Wall, Jeffrey L. Marsh, Peter Nurnberg, Maria Rita Passos-Bueno, and Andrew O. M. Wilkie
Carpenter syndrome is a pleiotropic disorder with autosomal recessive inheritance, the cardinal features of which include craniosynostosis, polysyndactyly, obesity, and cardiac defects. Using homozygosity mapping, we found linkage to chromosome 6p12.1-q12 and, in 15 independent families, identied ve different mutations (four truncating and one missense) in RAB23, which encodes a member of the RAB guanosine triphosphatase (GTPase) family of vesicle transport proteins and acts as a negative regulator of hedgehog (HH) signaling. In 10 patients, the disease was caused by homozygosity for the same nonsense mutation, L145X, that resides on a common haplotype, indicative of a founder effect in patients of northern European descent. Surprisingly, nonsense mutations of Rab23 in open brain mice cause recessive embryonic lethality with neural-tube defects, suggesting a species difference in the requirement for RAB23 during early development. The discovery of RAB23 mutations in patients with Carpenter syndrome implicates HH signaling in cranialsuture biogenesisan unexpected nding, given that craniosynostosis is not usually associated with mutations of other HH-pathway componentsand provides a new molecular target for studies of obesity.

Carpenter syndrome (MIM %201000), also known as acrocephalopolysyndactyly type II, is a rare autosomal recessive disorder characterized by craniosynostosis, obesity, polydactyly, and soft-tissue syndactyly (g. 1).13 Unlike other inherited craniosynostoses, which most commonly affect the coronal sutures, fusion of the midline (metopic and sagittal) sutures is typical in Carpenter syndrome; severe cases have cloverleaf skull. Other well-recognized features include brachydactyly with shortening or absence of the middle phalanges, molar agenesis, genu valgum, hypogenitalism, congenital cardiac defects, umbilical hernia, and learning disability.35 Although the molecular basis of many craniosynostosis syndromes is now well described, with mutations of broblast growth-factor receptors, ephrins, or the transcription factor TWIST most frequently identied,6 the etiology of Carpenter syndrome has been elusive. We mapped the disease locus, using a large family (family 1; subjects 3541, 3589, and 3593) from the United States.7 After approval from the Oxfordshire Research Ethics Committee and the local institutional review board,

DNA was collected with informed consent from six siblings (three affected and three unaffected) and both parents. The parents were not known to be related, but, of their eight grandparents, three of the fathers and two of the mothers had been born in the same city, raising the possibility of a distant consanguineous loop. We also analyzed an affected male (family 2; subject 3624) whose Danish parents were rst cousins; in addition to the classic features of Carpenter syndrome, subject 3624 had a lumbar myelomeningocele. Using the GeneChip Human Mapping 50K Array Hind240 (Affymetrix), we undertook genomewide SNP genotyping and performed parametric multipoint linkage analysis, allowing for heterogeneity, using GENEHUNTER-MODSCORE software.8 In the absence of consanguinity in family 1, no signicant linkage was obstained (the maximum possible heterogeneity LOD [HLOD] score of 2.7 was found for three different regions of the genome); however, when a consanguineous loop (conservatively assigned as second-cousin parents) was introduced, the HLOD score increased dramatically for just one of these regions, to a maximum of 4.8, for families 1 and

From the Weatherall Institute of Molecular Medicine, University of Oxford (D.J.; A.O.M.W.), and Craniofacial Unit, Oxford Radcliffe Hospitals National Health Service Trust (S.A.W.; A.O.M.W.), Oxford, United Kingdom; Cologne Center for Genomics and Institute for Genetics, University of Cologne, Cologne (D.S.; P.N.); Centro de Estudos do Genoma Humano, Departamento de Biologia, Universidade de Sao Paulo (F.S.J.; M.R.P.-B.), and Departmento de Morfologia, Universidade Federal de Sao Paulo (L.G.A.) and SOBRAPAR, Universidade de Campinas (D.F.B.), Sao Paulo; Division of Plastic Surgery, Washington University School of Medicine (C.A.P.), and Cleft Lip/Palate and Craniofacial Deformities Center, St. Johns Mercy Medical Center (J.L.M.), St. Louis; Academic Unit of Medical Genetics, University of Manchester, St Marys Hospital, Manchester, United Kingdom (D.D.); Clinical Genetics, Guys Hospital, London (D.J.); Department of Plastic and Reconstructive Surgery, Erasmus Medical Center, Rotterdam (I.M.J.M.); West Midlands Regional Genetics Service, Birmingham Womens Hospital, Birmingham, United Kingdom (J.E.V.M.); Department of Medical Genetics, Rikshospitalet-Radiumhospitalet Medical Centre, and Faculty Division Rikshospitalet, University of Oslo, Oslo (K.H..); and Merseyside & Cheshire Clinical Genetics Service, Liverpool Womens Hospital, Liverpool (E.S.) Received January 16, 2007; accepted for publication March 8, 2007; electronically published April 18, 2007. Address for correspondence and reprints: Dr. A. O. M. Wilkie, Weatherall Institute of Molecular Medicine, University of Oxford, John Radcliffe Hospital, Oxford OX3 9DS, United Kingdom. E-mail: awilkie@hammer.imm.ox.ac.uk Am. J. Hum. Genet. 2007;80:11621170. 2007 by The American Society of Human Genetics. All rights reserved. 0002-9297/2007/8006-0015$15.00 DOI: 10.1086/518047

1162

The American Journal of Human Genetics

Volume 80

June 2007

www.ajhg.org

Figure 1. Clinical features of Carpenter syndrome. A, Affected sister of subject 4009, aged 6 years. Note metopic ridge and temporal bulging secondary to multisuture synostosis, arched eyebrows, epicanthic folds, anteverted nares, and broad thumbs and halluces with syndactyly, brachydactyly, clinodactyly, and polydactyly (postaxial in hands, central in feet). B, Severe bilateral clubfoot in subject 4009. C, Three-dimensional CT skull reconstruction of subject 3541, aged 4 wk, showing complete synostosis of the metopic, sagittal, and coronal sutures. D, Preoperative radiographs of the hands and feet of subject 3734, aged 11 years. Note characteristic longitudinally split epiphyses at bases of several proximal phalanges, central polydactyly of the feet, and biphalangeal digits II of the hands and II and III of the feet. 2 combined, at chromosome 6p12.1-q12 (g. 2A). The anking heterozygous SNPs in family 1 were rs7766181 and rs10498828, and the affected individual in family 2 was homozygous throughout this region. The interval of homozygosity shared by the two families contained 24 annotated genes (Ensembl Genome Browser). Initially, we considered BMP5 a candidate, because mutation of murine Bmp5 causes a range of skeletal defects resulting in the short ear phenotype9; however, no mutations were found. We next analyzed RAB23; recessive nonsense mutations in the orthologous murine Rab23 gene cause neural-tube defects, abnormal somites, polydactyly, and poorly developed eyes (opb [open brain] locus).1012 With use of the primers listed in table 1, direct sequencing

of the seven exons and surrounding intronic regions of RAB23 revealed an identical homozygous 434TrA transversion encoding an L145X nonsense mutation (g. 3A) in the four affected individuals from families 1 and 2, as well as in a further sporadic case (subject 3734). These ve subjects were all identically homozygous for 30 fully genotyped consecutive SNPs on the Affymetrix 50K array (from rs10484709 to rs1577630) (g. 2B), indicating that they were very likely to share a single ancestral mutation (see below). To expand the mutation spectrum in RAB23, we ascertained 12 additional unpublished Carpenter syndrome cases unrelated by family history. Including the three families described above, the 15 independent families included 17 affected individuals from whom DNA was available; 6 cases arose from known consanguineous unions. The families originated from the United Kingdom (ve), Brazil (ve), The Netherlands (three), and Denmark and the United States (one each). The clinical features of these 17 cases, which represent the largest series of subjects with Carpenter syndrome to date, are presented in table 2. Craniosynostosis was present in all individuals, with the sutures affected with relative frequency sagittal1metopic1coronal1lambdoid. Abnormalities of the hands included postaxial polydactyly (9 of 17 cases), broad or bid thumbs (6 of 17), cutaneous syndactyly (12 of 17), and absent middle phalanges (9 of 11). In the feet, preaxial or central polydactyly (16 of 17) and syndactyly (17 of 17) were nearly always present. High birth weight (9 of 9) and obesity (9 of 10) were prevalent. Other signicant complications included umbilical hernia (8 of 17), congenital heart disease (3 of 17), deformities of the knees (4 of 14) or ankles (4 of 17), and cryptorchidism or hypoplastic testes in males (6 of 8). Brain imaging showed abnormalities in 7 of 10 subjects; 3 of 16 had hydrocephalus requiring insertion of a shunt. Signicant learning disability was present in 6 of 13 individuals. The occurrence of an open neural-tube defect (family 2; subject 3624), although not previously described in Carpenter syndrome, is unlikely to be coincidental, because this is a cardinal feature of Rab23 mutation in the mouse; the mother of subject 3624 had taken periconceptional folic acid supplements. Pathogenic sequence variants were found in all individuals with the classic phenotype (table 3 and g. 3A), showing that mutations of RAB23 are the major cause of Carpenter syndrome. We identied ve different mutations, all of which predict a loss of function. Four (E48fsX7, Y78fsX30, E137X, and L145X) of the ve alleles are nonsense or frameshifting mutations that would generate truncated proteins. We identied a single missense mutation, C85R, encoding a nonconservative substitution from an uncharged to a charged amino acid; this residue is involved in b-sheet formation and is completely buried in the core of the protein (g. 4),13,14 suggesting that this substitution would disrupt normal folding of RAB23. This mutation was present in individual 3961, who was a compound heterozygote for the C85R and L145X alleles; all

www.ajhg.org

The American Journal of Human Genetics

Volume 80

June 2007

1163

Figure 2. Linkage and haplotype mapping of Carpenter syndrome to chromosome 6p12.1. A, Genomewide HLOD scores from 50K microarray data for families 1 and 2 combined, with the assumption of a consanguineous loop (second-cousin parents) in family 1. HLOD scores are given along the Y-axis, relative to genomic position (cM) on the X-axis. Note the signicant peak (HLOD 4.8) in the centromeric region of chromosome 6. B, Representation of 50K SNP haplotypes (vertical bars) for chromosomes of affected individuals in families 1 and 2 and an additional sporadic case (subject 3734), all of whom are homozygous for the 434TrA (L145X) mutation in RAB23. Distinct haplotypes are represented by different shaded bars. On the basis of homozygosity, the critical region on chromosome 6 is dened by heterozygosity for SNPs rs7766181 (family 1) and rs1689237 (subject 3734). Within this, a smaller region (white bars) is identical in all affected individuals for 30 consecutive SNPs, suggesting a common ancestral origin of the L145X mutation. C, Genotyping of 13 selected SNPs spanning this identical segment in 10 individuals homozygous for the L145X mutation and 2 individuals homozygous for E137X. At left, the position of these SNPs is shown in relation to RAB23 and 8 additional genes within the 6p12.1q11 region. Note that all patients with the L145X mutation share a common haplotype for seven consecutive SNPs; this is interrupted proximally in two Dutch patients, probably because of a shared recombination. From top to bottom, the genotyped SNPs are rs1925179, rs2397214, rs9296842, rs1547625, rs6927258, rs6906792, rs3904827, rs6934928, rs1343391, rs1224703, rs1850417, rs2343013, and rs1689237. BrapBrazil; DenpDenmark; NethpThe Netherlands. other affected individuals appeared homozygous for their particular mutation. When samples were unavailable from both parents but we had sufcient proband DNA, we excluded the possibility that one allele harbored a deletion (table 3) by multiplex ligation-dependent probe-amplication (MLPA) analysis using synthetic oligonucleotide probes to RAB23 exons 1, 3, and 7 (MRC-Holland). All mutations were absent in 292 control chromosomes, as assessed by diagnostic restriction digests (g. 3A and table 4). The L145X mutation was apparently homozygous in 10 probands (3 each from the Netherlands and the United Kingdom, 2 from Brazil, and 1 each from the United States and Denmark), 3 of whom had been shown by the Affymetrix SNP analysis to share a common haplotype around the mutation (g. 2B). To check whether any of the other cases had arisen from an independent mutation, we genotyped them for a subset of 13 SNPs around RAB23 (table 5). All 10 patients shared a common haplotype comprising 7 SNPs and spanning 2.2 Mb. This haplotypewhich, according to HapMart (International HapMap Project), is present in only 11 of 120 Utah-CEPH chromosomes from HapMap16was delimited by ancestral recombinations, distally in the Danish family and proximally in two Dutch families, and contains only eight genes in addition to RAB23 (g. 2C). These data indicate that a founder effect,

1164

The American Journal of Human Genetics

Volume 80

June 2007

www.ajhg.org

Table 1.

Primers Used for PCR Amplication of RAB23

Primer Sequence (5 r3 ) Product Size (bp) 270 333 188 250 207 228 256 Wave Temperature(s) ( C) 59 56.4 54 56.5 and 58.5 54.5 55 57

Primer RAB23_1 RAB23_2 RAB23_3 RAB23_4 RAB23_5 RAB23_6 RAB23_7

Forward CTCCACCCTGGCATTTAGAC CCACAGATTTGAGAGGGAAGA TTACCAAAAACATTTTCCTTTACA TGTTAATGTAAATACCTTGATTGATTG AAACAAGCTATCAGAAGGCACC ATCATTGACCTGGTTCTGGG TAACTCAGGCGTGTCAGTGG

Reverse AACAGCCCTTTTCAGACCCT AGTTGCCACACCTCGAAATC GCCAAAATAATATGCCCAAA TATAGAATTACTGTCCCTCCTTCCC CAACACAATTTTAAAAGCGCA TCACTTTTAAATCACATTTCTGAAAGA ATGACAGCTGGATGGGTTTC

NOTE.DNA was obtained from whole-blood samples by phenol-chloroform extraction and was amplied in a total volume of 25 ml containing 15 mM TrisHCl (pH 8.0), 50 mM KCl, 2.5 mM MgCl2, 100 mM each deoxynucleoside triphosphate, 0.5 mM primers, and 0.75 units of Amplitaq Gold polymerase (Applied Biosystems). All PCRs were performed using an annealing temperature of 60 C. Cycling conditions consisted of an 8-min denaturation step at 94 C, followed by 35 cycles at 94 C for 30 s, at 60 C for 30 s, and at 72 C for 60 s, with a nal extension at 72 C for 10 min.

Figure 3. RAB23 mutations in Carpenter syndrome. A, Sequence chromatograms and conrmatory restriction digests for the ve pathogenic mutations identied. Note that patient 3961 is a compound heterozygote for the C85R and L145X mutations. B, top, The exon/intron organization of RAB23, with the coding part of the cDNA (GenBank [accession number NM_183227.1]; Ensembl Genome Browser [reference OTTHUMG00000014918]) in black and the UTRs in white (alternatively spliced 5 noncoding exons omitted). Plain numbering refers to the rst nucleotide of each exon, starting from the initiation codon, and italic numbering indicates the length of introns. Bottom, Functional domains in the 237-aa protein,13,14,17 color coded, with the location of human mutations causing Carpenter syndrome and those found in open brain mice12 indicated by red and green dots, respectively. GDPpguanosine diphosphate.

Table 2. Clinical Features of Patients with Mutations in RAB23

The table is available in its entirety in the online edition of The American Journal of Human Genetics.

rather than a recurrent mutation, underlies Carpenter syndrome in patients of northern European descent and does not support the possibility that the L145X mutation has particular functional consequences (see below). Two patients from eastern Brazil both appeared homozygous for the E137X mutation, which resides on a shared haplotype spanning at least 5.8 Mb (g. 2C). The E48fsX7 and Y78fsX30 mutations were each found in one patient only. To explore whether RAB23 mutations play a more general pathological role either in craniosynostosis or in limb malformations, we screened respective patient panels by a combination of Wave denaturing high-performance liquid chromatography (Transgenomic) and diagnostic restriction digests for mutations identied in Carpenter syndrome. DNA from 256 patients with craniosynostosis (negative for the common mutations in the FGFR1, FGFR2, FGFR3, and TWIST1 genes),6 202 patients with limb malformations requiring plastic surgery, and 163 control individuals was analyzed using the assays detailed in tables 1, 4, and 6. None of the Carpenter syndrome mutations (table 3) were identied in any of these cohorts. Although six novel allelesincluding an amino acid substitution, an amino acid deletion, and a nonsense mutation (all in the heterozygous state)were found in the patient groups (table 6), none appeared related to the clinical phenotype. An additional common nonsynonymous SNP c.619GrA (p.G207S; rs1040461) was present, for which all three genotypes were identied in unaffected individuals in the HapMap panel.16 These results show that RAB23 mutations Table 3.

do not frequently contribute either to craniosynostosis or to limb malformations. Rab23, rst isolated from the mouse in 1994,17 belongs to the RAB family of 160 small guanosine triphosphatases (GTPases) that regulate intracellular trafcking of membrane-associated proteins1315; other family members for which germline mutations cause human disorders are RAB7 (Charcot-Marie-Tooth disease type 2B, dominant inheritance [MIM #600882])18 and RAB27A (Griscelli syndrome type 2, recessive inheritance [MIM #607624]).19 Our nding of RAB23 mutations in Carpenter syndrome is unexpected, because similar nonsense mutations of the orthologous murine Rab23 gene (encoding K39X and R80X) in opb mice cause recessive embryonic lethality with exencephaly.1012 It is unlikely that the human RAB23 mutations represent partial loss-of-function alleles of lesser severity than do the murine ones, for two reasons. First, prenylation at a consensus site in the C-terminus of RAB proteins by RAB geranylgeranyl transferase is essential for their correct membrane targeting,15 predicting that all truncating mutations should result in complete loss of function. Second, two of the human truncating mutations (E48fsX7 and Y78fsX30) occur upstream of the opb mutation R80X (g. 3B),12 yet neural-tube defects were absent in the affected individuals. Consistent with this, we did not nd any clear genotype-phenotype correlation for the human mutations, although the fetus 3965, with the most N-terminal truncation (E48fsX7), was the only subject terminated antenatally and might represent a more severe phenotype (table 2). The original identication of Rab23 mutations in opb mice was driven by genetic studies to identify modiers of hedgehog (HH) signaling in the neural tube. In mammals, there are three paralogous HH genesShh, Ihh, and

RAB23 Mutations in Patients with Carpenter Syndrome

Mutation at Allele Country of Origin United Kingdom Brazil United Kingdom Brazil Brazil United States United States United States Denmark United Kingdom Netherlands Netherlands Brazil United Kingdom Netherlands Brazil United Kingdom Sample(s) Analyzeda C and M C and M C, M, and F C and M C, M, and F C, M, and F C, M, and F C, M, and F C, M, and F C and M C, M, and F C, M, and F C, M, and F C, M, and F C C, M, and F C Maternal DNA 140_141insA 232delT 434TrA 408_409insT 408_409insT 434TrA 434TrA 434TrA 434TrA 434TrA 434TrA 434TrA 434TrA 434TrA 434TrA 434TrA 434TrA Protein E48fsX7 Y78fsX30 L145X E137X E137X L145X L145X L145X L145X L145X L145X L145X L145X L145X L145X L145X L145X Paternal DNA 140_141insA Sameb 253TrC 408_409insT Same 434TrA 434TrA 434TrA Same 434TrAb 434TrA Same Same 434TrA 434TrA Same 434TrA Protein E48fsX7 C85R E137X L145X L145X L145X L145X L145X

Subject 3965 3983 3961 3978 3980 3541 3589 3593 3624 3734 3947 3971 3975 3985 4008 4009 4059

Sex M F F F F M F M M F M M M F F M M

Parental Consanguinity 1st cousin

Ethnicity White African/white White African/white White White White White White White White White White White White African White

1st cousin once removed

(Family (Family (Family (Family

1) 1) 1) 2)

1st cousin

1st cousin 1st cousin

L145X L145X L145X

1st cousin

NOTE.Nucleotide numbering of RAB23 cDNA is based on GenBank sequence NM_183227.1 but starts from the rst base of the initiation codon. For cases in which the two parental alleles are very unlikely to be independent (due to known consanguinity), the paternal allele is denoted Same. a Cpchild; Mpmother; Fpfather. b Complete deletion of one allele was not formally excluded by MLPA analysis.

1166

The American Journal of Human Genetics

Volume 80

June 2007

www.ajhg.org

Figure 4. Sequence conservation and structural context of C85R substitution. A, Amino acid sequence comparison of the Switch 2 region of human RAB23 (top) with 13 other species. The consensus sequence is shown at the bottom, and the position of the mutated C85 residue is indicated with an arrow. B, Structure of human RAB2314 (Protein Data Bank [number 1Z22]), showing the C85 residue located in a b-strand (blue) and completely buried in the core of the protein. The bound Mg-GDP is shown in yellow. The structure was modeled using the Protein Workshop tool (Protein Data Bank).

Table 4. Primers and Restriction Enzymes Used for Conrmation of RAB23 Mutations
Product Size (bp) 157 AAAGACTACAAGAAAACCATTGCCATTG AGTTGCCACACCTCGAAATC 188 TTACCAAAAACATTTTCCTTTACA GCCAAAATAATATGCCCAAA 156 TTTGAATGGATAAAAGTTGCCC TTCCCTATCTGTGGTAGAGAACTCGAGC 207 AAACAAGCTATCAGAAGGCACC CAACACAATTTTAAAAGCGCA 120 AAACAAGCTATCAGAAGGCACC TTCTTTCACTGATGTTCTGTAGAATGTT HpaI ( ) HindIII ( ) BssSI ( ) StuI ( ) Restriction Enzyme XcmI ( )

Mutation and Primer E48fsX7: E48fsXdigF RAB23_2R Y78fsX30: RAB23_3F RAB23_3R C85R: C85RdigF C85RdigR E137X: RAB23_5F RAB23_5R L145X: RAB23_5F RAB23_L145XdigR

Primer Sequence (5 r3 )a

NOTE.PCRs were performed using the same conditions as described in table 1. a Nucleotides shown in bold represent mismatches incorporated into primers to engineer diagnostic restriction sites.

Table 5. Primers and Restriction Enzymes Used for Haplotype Analysis

Primer Sequence (5 r3 ) GGCAGCTTTCTTCCTGACTG TTACATTTCAAAGGGGTGGC GGCCTGTGATTTGAGTGGTT TCTCTTGTGACCAGATGCCA CTGAAGTTGCATTCTTGCCA AAATTAAAAATCAGTGTCCTGCAA CCTGGCCTCATCCTACCATA TGCCTGAGAAAAATGAGGCT TTCTGATCATGATGTAGTGCCA GGGCTTGGCATCTCTGAGTA CAGAAACTTGGCAACAAAATG TTCCTGGAATTTAAAAGGTAGCA ACTGCATACCGCTTACCAAA TCCAGACAAACAAAGGCTGA AAGTGGCTTATTTCCTCCAAGA GCCCAAATCCATGTAACTTCT AAGGAGAGGGAGAGACCGAG AGCACATGATATGCCCACTT AATAGGGCAGAAGGGTGCTC ACCCACCAAAGAGACGTGAG CTTTAACTCCATTTTAAGGGACAG AACAAAGCTTGGAGAAGCAAA TGTGTTCCCAAACTGCTGAA CATCTCCCCCGGTTAAACTT TGAGGGATCTGGGATGCTAC TTTAGCTCTCACTGCATGGC Product Size (bp) 188 165 165 154 218 182 275 180 187 195 190 221 157

Primer rs1925179F rs1925179R rs2397214F rs2397214R rs9296842F rs9296842R rs1547625Fseq rs1547625Rseq rs6927258F rs6927258R rs6906792Fseq rs6906792Rseq rs3904827F rs3904827R rs6934928F rs6934928R rs1343391Fseq rs1343391Rseq rs1224703F rs1224703R rs1850417F rs1850417R rs2343013F rs2343013R rs1689237Fseq rs1689237Rseq

Assaya Hpy188I MnlI BtsCI Seq MboI Seq Tsp509I RsaI Seq Tsp509I EarI MwoI Seq

NOTE.PCRs were performed using the same conditions as described in table 1. a Seqpanalysis by DNA sequencing.

Dhhencoding the sonic, Indian, and desert HH proteins, respectively.20,21 The opb locus was initially described in a sporadically occurring mouse mutant with exencephaly10 (a variably expressed phenotype observed on both C57Bl/ 6 and C3H backgrounds).10,11 Subsequently, a second allele was isolated in an ethylnitrosourea mutagenesis screen for recessive embryonic lethal mutations22; nonsense mutations of Rab23 were demonstrated in both strains.12 Homozygosity for these opb mutations rescues many of the morphological defects in Shh / mice, with Shh/opb double-mutant mice largely resembling opb single mutants, Table 6.

indicating that the Rab23 mutations bypass the requirement for Shh.12 In the neural tube, mutation of Shh has opposite consequences to mutation of Rab23, causing loss (Shh) and expansion (Rab23) of ventral markers; Patched1, a transcriptional target of Shh signaling, is activated in opb mice, showing that Rab23 is a negative regulator of HH signaling.11,12 Genetic analysis of epistatic relationships shows that Rab23 acts downstream of the key HH signaling intermediate Smoothened but upstream of both the effector transcription factors Gli2 and Gli323 and the intraagellar transport proteins (such as those encoded by Ift88/polaris and Ift172/wim), required for their capacitation.24 As such, Rab23 is one of a number of genes (including iguana, talpid3, Fkbp8, and Ift family members) implicated in the regulation of Gli transcriptionfactor processing specically in vertebrates21,25 and is the rst of these implicated in a human disorder. Rab23 localizes to membranes26,27 and is expressed at multiple sites in the mouse, including embryonic neural tube, limb bud, branchial arches, tooth and palate,12,28,29 and adult brain27; however, its precise membrane-transport activity has not been dened. Given the evidence that Rab23 regulates the HH pathway, it is not surprising that some aspects of the phenotype of Carpenter syndrome resemble other human disorders associated with disturbed HH signaling. Most notably, the combination of postaxial polysyndactyly of the hands and preaxial polysyndactyly of the feet is very similar to the pattern that occurs in Greig syndrome (MIM #175700), which is due to haploinsufciency of GLI3,30 and is consistent with the observed reduction in the proportion of Gli3 repressor in Rab23-mutant embryos.23 The brachydactyly present in Carpenter syndrome, characterized by hypoplasia or absence of the middle phalanges, resembles brachydactyly type A1, which is caused by heterozygous missense mutations in IHH that appear to cluster on one surface of the protein and disrupt phalangeal patterning by an unknown mechanism.31 However, the viability of the human RAB23 homozygous mutations in Carpenter syndrome uncovers several phenotypes (table 2), not previously observed in the lethal opb mouse mu-

RAB23 Variants Identied in Patients with Craniosynostosis and Limb Malformations

No. of Heterozygotes

dbSNP Accession Number ss69357975 ss69357972 ss69357968 ss69357974 ss69357973 ss69357971 ss69357970 ss69357969

Nucleotide c.149CrG c.39_41del c.119ArG c.155 44CrA c.24215_-12del c.247CrT c.301TrG c.574 28GrA

Amino Acid

Exon (Intron) (1) 2 2 (2) (3) 4 4 (6)

With Craniosynostosis (N p 256) 2 1 0 0 0 1 5 0

With Limb Malformation (N p 202) 0 0 0 0 1 0 2 1

Controls (N p 163) 0 0 1 1 0 0 2 1

Present in Unaffected Parenta NT NA

Restriction Digest Hpy188III XcmI ( NT NT NT BfaI ( MwoI ( NT ( ) )

p.V13del p.K40R

p.Q83X p.S101A

NA NA 2 (5 NA) NA

) )

NOTE.NTpnot tested. a NApone or both parents unavailable for testing.

1168

The American Journal of Human Genetics

Volume 80

June 2007

www.ajhg.org

tants,1012,22 that are not well recognized features of perturbed HH signaling.20 Particularly interesting are the craniosynostosis and tendency to postnatal obesity, which may provide new clues for dissecting the pathophysiology of these phenotypes. Relatively little is known about the role of HH signaling in the cranial sutures. In mice, Shh is expressed in the osteogenic fronts of the parietal bones and sagittal sutures only at a relatively late stage of suture development (embryonic day 17),32 and Shh / mice die too early to assess the developmental contribution to the cranial sutures. Although endochondral ossication is characteristically decient in Ihh / mice, membranous ossication of the skull vault is maintained33; however, there are no published data on the expression pattern of Ihh in the sutures. Our work should stimulate efforts to identify the active HH ligand(s) and to explore the extent to which the welldocumented developmental relationship among HH signaling, twist, and FGF receptors in the limbs34 is recapitulated in the cranial suture. There is a potential pathophysiological link with Antley Bixler syndrome (MIM #207410), another recessively inherited craniosynostosis caused by mutations in P450 oxidoreductase (POR).35 POR is the single avoprotein involved in electron transfer to all cytochrome P450 enzymes, including several involved in cholesterol biosynthesis, defects of which disrupt HH signaling.36 Although postnatal obesity (which has a central distribution) is practically universal in subjects with Carpenter syndrome, its neuroendocrinological and biochemical bases are not known. Obesity has not been described elsewhere in mammalian disorders of HH signaling20 and was not observed in opb mice, probably because of embryonic lethality. Pharmacological approaches have previously suggested that HH signaling may regulate adipogenesis, but the results have been conicting as to whether this effect was inhibitory or stimulatory.37 The association of RAB23 mutations with obesity may provide new insight into the role of HH signaling in the control of fat metabolism. Alternatively, this phenotype may reect either the regulation of RAB23 itself (possibly by bone morphogenetic proteins)12 or downstream HH-independent processes. The possible interaction of RAB23 with cilia21 suggests an overlap with the Bardet-Biedl syndromes, which are ciliopathies that also feature obesity and polydactyly.38

through the National Genome Research Network grant 01GR0416 (to P.N.), and the Wellcome Trust (support to A.O.M.W.).

Web Resources
Accession numbers and URLs for data presented herein are as follows: dbSNP, http://www.ncbi.nlm.nih.gov/SNP/ (for SNPs, including rs1040461, rs1925179, rs2397214, rs9296842, rs1547625, rs6927258, rs6906792, rs3904827, rs6934928, rs1343391, rs1224703, rs1850417, rs2343013, and rs1689237) Ensembl Genome Browser, http://www.ensembl.org/ (for RAB23 [reference OTTHUMG00000014918]) GenBank, http://www.ncbi.nlm.nih.gov/Genbank/ (for human RAB23 cDNA reference sequence [accession number NM_183227.1]) International HapMap Project, http://hapmart.hapmap.org/ BioMart/martview (for HapMart) MRC-Holland, http://www.mrc-holland.com/pages/indexpag .html (for information on MLPA reagents and methods) Online Mendelian Inheritance in Man (OMIM), http://www.ncbi .nlm.nih.gov/Omim/ (for Carpenter syndrome, Charcot-MarieTooth disease type 2B, Griscelli syndrome type 2, Greig syndrome, and Antley Bixler syndrome) Protein Data Bank, http://www.rcsb.org/pdb/home/home.do (for RAB23 structure [number 1Z22] and Protein Workshop)

References
1. Carpenter G (1901) Two sisters showing malformations of the skull and other congenital abnormalities. Rep Soc Study Dis Child Lond 1:110118 2. Carpenter G (1909) Case of acrocephaly, with other congenital malformations. Proc Roy Soc Med 2:4553 3. Temtamy SA (1966) Carpenters syndrome: acrocephalopolysyndactyly. J Pediatr 69:111120 4. Robinson LK, James HE, Mubarak SJ, Allen EJ, Jones KL (1985) Carpenter syndrome: natural history and clinical spectrum. Am J Med Genet 20:461469 5. Cohen DM, Green JG, Miller J, Gorlin RJ, Reed JA (1987) Acrocephalopolysyndactyly type IICarpenter syndrome: clinical spectrum and an attempt at unication with Goodman and Summit syndromes. Am J Med Genet 28:311324 6. Morriss-Kay GM, Wilkie AOM (2005) Growth of the normal skull vault and its alteration in craniosynostosis: insights from human genetics and experimental studies. J Anat 207: 637653 7. Perlyn CA, Marsh JL (2007) Craniofacial dysmorphology of Carpenter syndrome: lessons from three affected siblings. Plast Reconstr Surg (in press) 8. Dietter J, Mattheisen M, Furst R, Ru schendorf F, Wienker TF, Strauch K (2007) Linkage analysis using sex-specic recombination fractions with GENEHUNTER-MODSCORE. Bioinformatics 23:6470 9. King JA, Marker PC, Seung KJ, Kingsley DM (1994) BMP5 and the molecular, skeletal, and soft-tissue alterations in short ear mice. Dev Biol 166:112122 10. Gunther T, Struwe M, Aguzzi A, Schughart K (1994) open brain, A new mouse mutant with severe neural tube defects, shows altered gene expression patterns in the developing spinal cord. Development 120:31193130 11. Eggenschwiler JT, Anderson KV (2000) Dorsal and lateral fates

Acknowledgments
We thank all the families for their participation in this study; S. Balci and A. Richieri-Costa for referring patient samples; C. Becker (Nurnberg laboratory) for expert technical assistance in providing the SNP genotype data from Affymetrix microarrays; other members of the Wilkie laboratory for their support, especially D. Furniss for access to samples from patients with limb malformations and I. Taylor for DNA extraction; K. Clark for DNA sequencing; and J. Eggenschwiler and B. St-Jacques for discussions. This work was funded by the E. P. Abraham Cephalosporin Fund (support to D.J.), German Federal Ministry of Sciences and Education

www.ajhg.org

The American Journal of Human Genetics

Volume 80

June 2007

1169

12.

13.

14.

15. 16. 17.

18.

19.

20. 21.

22.

23.

24.

25.

26.

in the mouse neural tube require the cell-autonomous activity of the open brain gene. Dev Biol 227:648660 Eggenschwiler JT, Espinoza E, Anderson KV (2001) Rab23 is an essential negative regulator of the mouse sonic hedgehog signalling pathway. Nature 412:194198 Pereira-Leal JB, Seabra MC (2000) The mammalian Rab family of small GTPases: denition of family and subfamily sequence motifs suggests a mechanism for functional specicity in the Ras superfamily J Mol Biol 301:10771087 Eathiraj S, Pan X, Ritacco C, Lambright DG (2005) Structural basis of family-wide Rab GTPase recognition by rabenosyn5. Nature 436:415419 Pereira-Leal JB, Seabra MC (2001) Evolution of the Rab family of small GTP-binding proteins. J Mol Biol 313:889901 The International HapMap Consortium (2005) A haplotype map of the human genome. Nature 437:12991320 Olkkonen VM, Peterson JR, Dupree P, Lutcke A, Zerial M, Simons K (1994) Isolation of a mouse cDNA encoding Rab23, a small novel GTPase expressed predominantly in brain. Gene 138:207211 Verhoeven K, De Jonghe P, Coen K, Verpoorten N, AuerGrumbach M, Kwon JM, FitzPatrick D, Schmedding E, De Vriendt E, Jacobs A, et al (2003) Mutations in the small GTPase late endosomal protein RAB7 cause Charcot-Marie-Tooth type 2B neuropathy. Am J Hum Genet 72:722727 Menasche G, Pastural E, Feldmann J, Certain S, Ersoy F, Du puis S, Wulffraat N, Bianchi D, Fischer A, Le Deist F, et al (2000) Mutations in RAB27A cause Griscelli syndrome associated with haemophagocytic syndrome. Nat Genet 25:173 176 Nieuwenhuis E, Hui C-c (2005) Hedgehog signaling and congenital malformations. Clin Genet 67:193208 Huangfu D, Anderson KV (2006) Signaling from Smo to Ci/ Gli: conservation and divergence of Hedgehog pathways from Drosophila to vertebrates. Development 133:314 Kasarskis A, Manova K, Anderson KV (1998) A phenotypebased screen for embryonic lethal mutations in the mouse. Proc Natl Acad Sci USA 95:74857490 Eggenschwiler JT, Bulgakov OV, Qin J, Li T, Anderson KV (2006) Mouse Rab23 regulates Hedgehog signaling from smoothened to Gli proteins. Dev Biol 290:112 Huangfu D, Liu A, Rakeman AS, Murcia NS, Niswander L, Anderson KV (2003) Hedgehog signalling in the mouse requires intraagellar transport proteins. Nature 426:8387 Davey MG, Paton IR, Yin Y, Schmidt M, Bangs FK, Morrice DR, Gordon Smith T, Buxton P, Stamataki D, Tanaka M, et al (2006) The chicken talpid3 gene encodes a novel protein essential for Hedgehog signaling. Genes Dev 20:13651377 Evans TM, Ferguson C, Wainwright BJ, Parton RG, Wicking

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37. 38.

C (2003) Rab23, a negative regulator of hedgehog signaling, localizes to plasma membrane and the endocytic pathway. Trafc 4:869884 Guo AC, Wang T, Ng EL, Aulia S, Chong KH, Teng FYH, Wang Y, Tang BL (2006) Open brain gene product Rab23: expression pattern in the adult mouse brain and functional characterization. J Neurosci Res 83:11181127 Miletich I, Cobourne MT, Abdeen M, Sharpe PT (2005) Expression of the Hedgehog antagonists Rab23 and Slimb/bTrCP during mouse tooth development. Arch Oral Biol 50:147151 Rice R, Connor E, Rice DPC (2006) Expression patterns of Hedgehog signalling pathway members during mouse palate development. Gene Expr Patterns 6:206212 Biesecker LG (2004) GLI3 and the Pallister-Hall and Greig cephalopolysyndactyly syndromes. In: Epstein CJ, Erickson RP, Wynshaw-Boris A (eds) Inborn errors of development. Oxford University Press, Oxford, United Kingdom, pp 257 264 Hellemans J, Coucke PJ, Giedion A, De Paepe A, Kramer P, Beemer F, Mortier GR (2003) Homozygous mutations in IHH cause acrocapitofemoral dysplasia, an autosomal recessive disorder with cone-shaped epiphyses in hands and hips. Am J Hum Genet 72:10401046 Kim H-J, Rice DPC, Kettunen PJ, Thesleff I (1998) FGF-, BMPand Shh-mediated signalling pathways in the regulation of cranial suture morphogenesis and calvarial bone development. Development 125:12411251 St-Jacques B, Hammerschmidt M, McMahon AP (1999) Indian hedgehog signaling regulates proliferation and differentiation of chondrocytes and is essential for bone formation. Genes Dev 13:20722086 ORourke MP, Soo K, Behringer RR, Hui C-c, Tam PP (2002) Twist plays an essential role in FGF and SHH signal transduction during mouse limb development. Dev Biol 248:143 156 Huang N, Pandey AV, Agrawal V, Reardon W, Lapunzina PD, Mowat D, Jabs EW, Van Vliet G, Sack J, Fluck CE, et al (2005) Diversity and function of mutations in P450 oxidoreductase in patients with Antley-Bixler syndrome and disordered steroidogenesis. Am J Hum Genet 76:729749 Cooper MK, Wassif CA, Krakowiak PA, Taipale J, Gong R, Kelley RI, Porter FD, Beachy PA (2003) A defective response to Hedgehog signaling in disorders of cholesterol biosynthesis. Nat Genet 33:508513 Rosen ED (2006) New drugs from fat bugs? Cell Metab 3:1 2 Badano JL, Mitsuma N, Beales PL, Katsanis K (2006) The ciliopathies: an emerging class of human genetic disorders. Annu Rev Genomics Hum Genet 7:125148

1170

The American Journal of Human Genetics

Volume 80

June 2007

www.ajhg.org