Mutations in GJB6 cause hidrotic ectodermal dysplasia

brief communications

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


Fig. 2 Prevalence of E. coli enzymes in gene-fusion events. Distribution of detected component proteins involved in fusion events for the set of 636 metabolic enzymes of E. coli (red) and 30 control sets of other proteins (blue). Each of the control sets contains 636 randomly selected proteins, extracted from the remaining 3,654 ORFs of the published E. coli genome sequence8 (total number of ORFs, 4,290). Each of the 31 sets was subject to fusion analysis3 by comparing all 636 sequences with 22 genomes (Fig. 1). Comparing this set with the non-redundant database is prohibitively expensive in terms of computation time. The analysis for the set of metabolic enzymes detected 106 component proteins, whereas the sets of randomly selected proteins detected on average 28±7.4 component proteins. Counts of detected component proteins are shown on the x axis (bins of 5) and the frequency of the sets corresponding to these counts is shown on the y axis.

ite proteins are Saccharomyces cerevisiae, Bacillus subtilis, Caenorhabditis elegans and M. tuberculosis (Fig. 1). It seems that the detection of composite proteins is a complex function of genome size, paralogy and phylogenetic distance. Almost one-half of the composite proteins are exclusively obtained from eukaryotes S. cerevisiae and C. elegans (with the largest and most paralogous genomes). S. cerevisiae provides more composite proteins than C. elegans, possibly due to the presence of the homologous enzymes (and by implication the corresponding biochemical pathways). Similar species exhibit similar distributions of the fusion cases. For example, composite proteins from the two Mycoplasma or Chlamydia species are identical (Fig. 1). Species closely related to E. coli yield more composite proteins: C. elegans, having a much larger genome than B. subtilis, does not yield as many composite proteins homologous to E. coli enzymes. To test whether any randomly chosen protein is also as likely to participate in a fusion with another protein, we obtained 30 sets of 636 randomly selected proteins, excluding the 636 enzymes, from the E. coli

complete genome sequence8. By performing an identical analysis against the same number of genomes3, we found that an average of 28±7.4 components (out of 636, or an estimated average of 4.4%) seem to be involved in this process (Fig. 2). Due to the fact that the control sets may contain unknown metabolic enzymes, this frequency can be seen as an overestimate for the true (but unavailable) number of nonenzymes involved in gene fusion. This observation suggests that the set of metabolic enzymes in E. coli exhibits a strong tendency to participate in genefusion events, with a more than threefold difference over a set of control proteins. A statistical test for comparing the proportions with equal sample size produces a Z-score of 7.2, which is highly significant (Fig. 2). The possible implication of this fact is predictive power: for any two E. coli genes of unknown function identified by gene-fusion analysis, there is a higher chance that they are metabolic enzymes. This may also be true for other classes of proteins, but selecting those from databases on the basis of their functional roles is not currently possible.

Mutations in GJB6 cause hidrotic ectodermal dysplasia ere we present genetic evidence that different missense mutations of the gene GJB6, encoding connexin-30 (CX30), cause hidrotic ectodermal dysplasia (HED, MIM 129500; also known as Clouston syndrome). This disease is an autosomal dominant skin disorder characterized

H

142

by palmoplantar hyperkeratosis, hair defects (from partial to total alopecia), nail hypoplasia and nail deformities1. HED occurs worldwide with a very low frequency, but is more common in FrenchCanadians2, wherein a strong founder effect has been demonstrated3.

Although our analysis is restricted to E. coli, its applicability may extend to similar species with high precision. Unfortunately, no other database allows the extraction of enzyme components involved in small-molecule metabolism. We conclude that metabolic enzymes exhibit a much higher tendency to participate in multiple gene-fusion events. We hypothesized previously3 that this effect may be due to metabolic channelling of substrates9 or other physical constraints. Our results may have implications for more precise metabolic reconstruction from the genome sequence10 and the construction of artificial multifunctional enzymes for metabolic engineering11. Acknowledgements

We thank A.J. Enright for technical advice and P.D. Karp for help and comments. This work was supported by the European Molecular Biology Laboratory and the TMR Programme of the European Commission DGXII (Science, Research and Development). C.A.O. acknowledges further support by the Medical Research Council (UK) and IBM Research. Sophia Tsoka & Christos A. Ouzounis Computational Genomics Group, Research Programme, The European Bioinformatics Institute, EMBL Cambridge Outstation, Cambridge, UK. Correspondence should be addressed to C.A.O. (e-mail: [email protected]). 1. 2.

Marcotte, E.M. et al. Science 285, 751–753 (1999). Marcotte, E.M., Pellegrini, M., Thompson, M.J., Yeates, T.O. & Eisenberg, D. Nature 402, 83–86 (1999). 3. Enright, A.J., Iliopoulos, I., Kyrpides, N.C. & Ouzounis, C.A. Nature 402, 86–90 (1999). 4. Doolittle, R.F. Nature Genet. 23, 6–8 (1999). 5. Overbeek, R., Fonstein, M., D’Souza, M., Pusch, G.D. & Maltsev, N. Proc. Natl Acad. Sci. USA 96, 2896–2901 (1999). 6. Sali, A. Nature 402, 23–26 (1999). 7. Karp, P.D., Riley, M., Paley, S.M., Pellegrini-Toole, A. & Krummenacker, M. Nucleic Acids Res. 27, 55–58 (1999). 8. Blattner, F.R. et al. Science 277, 1453–1474 (1997). 9. Welch, G.R. & Easterby, J.S. Trends Biochem. Sci. 19, 193–197 (1994). 10. Karp, P.D., Krummenacker, M., Paley, S. & Wagg, J. Trends Biotechnol. 17, 275–281 (1999). 11. Bülow, L. Biochem. Soc. Symp. 57, 123–133 (1990).

A locus for HED was mapped by linkage analysis to the pericentromeric region of chromosome 13q (ref. 4). Recombination mapping of HED families allowed the refinement of the candidate region to an interval flanked by D13S1828 proximally and D13S1830 distally3. The construction of a physical map of this chromosomal region allowed us to map 15 genes and expressed sequence tags5 (ESTs), among which we found GJB2 and GJB6, encoding CX-26 and CX-30, respectively. GJB2 and nature genetics • volume 26 • october 2000



Fig. 1 HED families used in this study. Pedigrees of the two French HED families initially screened are shown. a, Family HED1 (ref. 12). b, Family HED2. c, Pedigree of the Welsh HED10 family. All numbered individuals were examined and their DNA samples collected.

a


b

GJB6 are mutated in different types of deafness6–9. Hearing impairment has been reported in few cases of HED (ref. 10); consequently, GJB6 seemed to be a good candidate for the disease, as the involvement of GJB2 in HED has been excluded6. In addition, GJB6 has been shown to be expressed in mouse epidermis11 and we detected a GJB6 transcript of 1.9 kb in different human epidermal cells (data not shown). Fig. 2 Identification of GJB6 mutations in HED patients. a, Sequence analysis in HED patients of both French families demonstrate a transition of one allele (arrow) corresponding to an alteration of the normal glycine codon to an arginine codon. Note that the sequence (top) is presented in the 3´ to 5´ direction. The same mutation was found in one Spanish, one African, one French and three French-Canadian patients. b, BclI digestion after PCR obtained with a mutagenesis primer designed to create a restriction site in the mutant, but not in wild type (G11R mutant allele, 80 bp; normal allele, 119 bp; size marker, 100-bp ladder). The lanes are numbered to correspond to the pedigrees shown in Fig. 1a,b. c, Sequence analysis in HED patients from this HED10 family demonstrates a C→T transition of one allele (arrow) corresponding to alteration of the normal arginine codon to a valine codon. The same mutation was found in Indian and Malaysian patients. d, HaeII digestion of a PCR product containing the whole GJB6 coding sequence. The C→T transition in position 263 removes an HaeII site. A88V mutant allele, 850 bp; normal allele, 310 and 540 bp. The lanes are numbered to correspond to the pedigrees shown in Fig. 1c. Ind, Indian HED patient; Mal, Malaysian HED patient; size marker, 100-bp ladder. e, Comparison of the amino acid sequence of human CX-30 (HuCx30, PID g4127986), mouse Cx-30 (MmCx30, PID g2493805), human CX-26 (HuCx26, PID g585020), human CX-31 (HuCx31, PID g3445287), human CX-31-1 (HuCx31-1, PID g4336903) and human CX-32 (HuCx32, PID g117688) in the cytoplasmic N-terminal domain and second transmembrane domain (M2). A high degree of cross-species homology is seen, including the glycine 11 and the arginine 88 residues, which are also conserved in CX-26. Amino acids in blue are strongly conserved (at least in four of six sequences).

nature genetics • volume 26 • october 2000

c

We studied two unrelated French families segregating a typical form of HED (Fig. 1a,b; ref. 12). We carried out mutation scanning by direct sequencing of DNA from the patients. A transition, 31(G→A), was found in the coding sequence, leading to a missense mutation (G11R) in each affected individual (Fig. 2a). To confirm

the involvement of GJB6 in HED, we analysed ten additional unrelated affected individuals of various geographic origins4 (1 Scottish-Irish, 1 Indian, 1 Malaysian, 1 African, 1 Spanish, 3 French-Canadian, 1 additional French and 1 Welsh). The same point mutation was found in all but three of the affected persons. We detected another transition, 263(C→T), leading to a missense mutation (A88V) in patients originating from India, Malaysia and Wales (Fig. 2c). Patients carrying A88V and patients carrying G11R present the same typical HED phenotype. By rapid detection of restriction sites created or removed by the mutations (Fig. 2b,d), we showed that the 2 heterozygous missense mutations segregated with the disease and were not observed in 118 unrelated normal subjects, excluding the possibility that these mutations represent a polymorphism. These results strongly suggest that mutation of GJB6 is responsible for HED.

a

b

c

d

e

143




HED joins the growing list of inherited skin disorders caused by mutant members of the connexin family of proteins. A mutation (D66H) in GJB2 was detected in Vohwinkel syndrome13 and a missense mutation (G12R) in GJB3 was observed in erythrokeratodermia variabilis14 (EKV). The glycine residues at positions 11 and 12 of CX-30 and in CX-31, respectively, are conserved and lie in a stretch of highly conserved residues located in the cytoplasmic amino-terminal domain (Fig. 2e). The introduction of a positively charged arginine in the N-terminal domain or the mutation A88V, introducing a highly hydrophobic residue in the transmembrane M2 domain (Fig. 2e), may change the polarity of connexin channels and affect communication between cells. The GJB6 mutations might also cause HED through haploinsufficiency of gap-junction channels, or by dominant-negative effect on normal CX-30 activity. The central question raised by our data is the involvement of CX-30 in two different pathologies: non-syndromic autosomal dominant deafness, caused by a T5M mutation9, and HED, here shown to be caused by different mutations of the same gene. In this respect, the different mutations in GJB6 are comparable to those occurring in other genes encoding members of the connexin family. Mutations in GJB3 are responsible for both EKV (ref. 14) and an autosomal dominant hearing impairment15. It is possible that the mutations detected in HED

patients affect an epidermal-specific connexin function, and it may be that the multiple connexins expressed in keratinocytes do not compensate for the effect of the mutant CX-30. On the other hand, other connexins in auditory cells may be able to compensate for this effect. Additional studies are needed to understand how mutant connexins affect hearing cells in deafness or epidermis, nails and hair in HED. Acknowledgements

We thank the families for participation; E. Denise and S. Dubus for technical assistance; and Foundation Jean Dausset–CEPH, particularly M. Legrand and C. Billon of the CEPH DNA laboratory, for DNA samples used in polymorphism exclusion. This work was supported by grants from Genethon, Association Française contre les Myopathies, French MENRT and Swiss National Science Foundation.

1.

Jérôme Lamartine1, Guilherme Munhoz Essenfelder1, Zoha Kibar2, Isabelle Lanneluc1, Edwige Callouet3, Dalila Laoudj1, Gilles Lemaître1, Colette Hand2, Susan J. Hayflick4, Jonathan Zonana4, Stylianos Antonarakis5, Uppala Radhakrishna5, David P. Kelsell6, Arnold L. Christianson7, Amandine Pitaval1, Vazken Der Kaloustian8, Clarke Fraser8, Claudine Blanchet-Bardon9, Guy A. Rouleau2 & Gilles Waksman1 1Laboratoire

de Génomique et Radiobiologie du Kératinocyte (EA 2541: Université d’Evry/CEA), Service de Génomique

A view of Neandertal genetic diversity he retrieval of mitochondrial DNA (mtDNA) sequences from the Neandertal type specimen from Feldhofer Cave in western Germany1,2 made possible a comparison of DNA sequences from an extinct hominid with those from modern humans. Recently, a second mtDNA sequence from a Neandertal child found in Mezmaiskaya Cave in the northern Caucasus was determined and found to be similar to the type specimen3. To further study the Neandertal mtDNA gene pool, we analysed the amino acid composition and extent of amino acid racemization in 15 Neandertal bones found in the G3 layer4,5 in Vindija Cave, Croatia. Seven samples proved to have a high content of amino acids, an amino acid composition similar to that of contemporary bone, and a low level of racemization of aspartic acid, alanine and leucine6, all features com-

T

144

Fonctionnelle, Département de Radiobiologie et Radiopathologie, Evry, France. 2Centre for Research in Neurosciences, McGill University and the Montreal General Hospital Research Institute, Montreal, Quebec, Canada. 3Genethon, Evry, France. 4Department of Molecular and Medical Genetics, Oregon Health Sciences University, Portland, Oregon, USA. 5Division of Medical Genetics, University of Geneva Medical School and University Hospital, Geneva, Switzerland. 6Centre for Cutaneous Research, St. Bartholomew’s and the Royal London Hospital, London, UK. 7Department of Human Genetics and Developmental Biology, Faculty of Medicine, University of Pretoria, Pretoria, South Africa. 8The F. Clarke Fraser Clinical Genetics Unit, Division of Medical Genetics, Montreal Children’s Hospital, Montreal, Quebec, Canada. 9Institut de Recherche sur la peau, Hôpital Saint-Louis, Paris, France. Correspondence should be addressed to G.W. (e-mail: [email protected]).

patible with DNA preservation7. We dated one of the samples (Vi-75-G3/h-203; ref. 5) by accelerator mass spectroscopy to over 42,000 years before present (Ua-13873) and used it for five DNA extractions. In three of the extractions, we included N-phenacylthiazolium bromide (PTB), a compound that

2. 3. 4. 5. 6. 7. 8. 9. 10.

11. 12. 13. 14. 15.

Ando, Y., Tanaka, T., Horiguchi, Y., Ikai, K. & Tomono, H. Dermatologica 176, 205–211 (1988). Clouston, H.R. Can. Med. Assoc. J. 21, 18–31 (1929). Kibar, Z. et al. Eur. J. Hum. Genet. 8, 372–380 (2000). Kibar, Z. et al. Hum. Mol. Genet. 5, 543–547 (1996). Lamartine, J. et al. Genomics 67, 232–236 (2000). Kelsell, D.P. et al. Nature 387, 80–83 (1997). Zelante, L. et al. Hum. Mol. Genet. 9, 1605–1609 (1997). Denoyelle, F. et al. Nature 393, 319–320 (1998). Grifa, A. et al. Nature Genet. 23, 16–18 (1999). Worobec-Victor, S.M., Bene-Bain, M.A., Shanker, D.B. & Solomon, L.M. in Pediatric Dermatology (eds Schachner, L.A. & Hansen, R.C.) 328 (Churchill Livingstone, New York, 1988). Dahl, E. et al. J. Biol. Chem. 271, 17903–17910 (1996). Lamartine, J. et al. Br. J. Dermatol. 142, 248–252 (2000). Maestrini, E. et al. Hum. Mol. Genet. 8, 1237–1243 (1999). Richard, G. et al. Nature Genet. 20, 366–369 (1998). Xia, J.H. et al. Nature Genet. 20, 370–373 (1998).

has been shown to improve DNA retrieval from late Pleistocene coprolites, probably due to its ability to cleave sugar-derived condensation products in which DNA may be entrapped8. We amplified multiple overlapping mtDNA fragments by PCR from the extracts, cloned the products and sequenced the insertions of multiple clones. When using primers that allowed both modern

Table 1 • Mitochondrial DNA sequence variation among three Neandertals, humans, chimpanzees and gorillas Population Neandertals Humans Europeans Africans Asians Native Americans Oceanians Chimpanzees Gorillas

Individuals

Mean

3 5,530 1,433 919 1,633 1,388 157 359 28

3.73 3.43 2.21 3.91 3.03 3.06 3.80 14.81 18.57

Minimum – 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.40

Maximum

s.d.

– 10.16 7.61 8.52 9.63 9.66 9.62 29.06 28.79

– 1.21 0.92 1.16 0.98 1.05 1.14 5.70 5.26

The variation is expressed as the percentage of sequence positions that has changed in trees relating three mtDNA sequences.

nature genetics • volume 26 • october 2000