Abstract Mutations in the Connexin 26 (GJB2/Cx26)gene are responsible for more than half of all cases ofprelingual non-syndromic recessive deafness in many Cau-casian populations. To determine the importance of Cx26mutations as a cause of deafness in Turks we screened 11families with prelingual non-syndromic deafness, seven(64%) of which were found to carry the 35delG mutation.We subsequently screened 674 Turkish subjects with noknown hearing loss and found twelve 35delG heterozy-gotes (1.78%; 95% confidence interval: 0.9%–3%) but noexamples of the 167delT mutation. To search for possiblefounder effects, we typed chromosomes carrying the35delG mutation for closely linked polymorphic markersin samples from Turkey and United States and comparedthe allele frequencies with those of hearing subjects. Thedata showed a modest degree of disequilibrium in both pop-ulations. Analyses of two pedigrees from Turkey demon-strated both conserved and different haplotypes, suggest-ing possible founder effects and multiple origins of the35delG mutation.

Introduction

Approximately 50% of recessive non-syndromic deafnesscan be attributed to mutations in the Connexin 26 (Cx26)gene in many Caucasian populations (Cohn and Kelley

1999). Although more than 50 mutations have been de-scribed (http://www.iro.es/cx26deaf.html), one particularvariant, 35delG, accounts for up to 70% of the pathologicalleles in Northern and Southern European and AmericanCaucasian populations, with a carrier frequency rangingfrom 1.3% to 2.8% (Gasparini et al. 2000).

Because of structural similarities with other putative mu-tational hotspots and its haplotypic heterogeneity, the highfrequency of the 35delG mutation has been attributed torecurrent mutation (Carrasquillo et al. 1997; Denoyelle etal. 1997). However, the frequency of the 35delG mutationamong the deaf in Asian populations is low (Kudo et al.2000), raising the possibility that other factors may con-tribute to variation in the frequency of the 35delG muta-tion. It has recently been proposed that a combination ofrelaxed selection and assortative mating during the pasttwo centuries may also have made an important contribu-tion to the high frequency of Cx26 deafness in the UnitedStates and other countries with a long tradition of inter-marriage among the deaf (Nance et al. 2000). Finally, itshould be noted that the deletion of any one of six consec-utive guanine residues would give rise to indistinguish-able 35delG alleles.

Another Cx26 mutation, 167delT, has a high frequencyamong Ashkenazi Jews, where carrier frequencies of 3%–4% have been documented (Morell et al. 1998). A foundereffect has been demonstrated for this mutation, but not for35delG, among Ashkenazi Jews (Morell et al. 1998).

We have sought (1) to determine the prevalence of Cx26deafness in Turkey, (2) to estimate the carrier frequenciesfor the 35delG and 167delT mutations in control subjects,(3) to search for 35delG founder effects, and (4) to evalu-ate possible causes for the variation in the frequency ofCx26 deafness in different populations.

Materials and methods

DNA samples from 21 deaf individuals in 11 families (6 simplexand 5 multiplex) with severe to profound prelingual sensorineuralnon-syndromic deafness were included to determine the prevalenceof Cx26 deafness in Turkey. Signed informed consent was obtained

Mustafa Tekin · Nejat Akar · Șükrü Cin ·Susan H. Blanton · Xia Juan Xia · Xue Zhong Liu ·Walter E. Nance · Arti Pandya

Connexin 26 (GJB2) mutations in the Turkish population: implications for the origin and high frequency of the 35delG mutation in Caucasians

Hum Genet (2001) 108 :385–389DOI 10.1007/s004390100507

Received: 22 January 2001 / Accepted: 20 March 2001 / Published online: 27 April 2001

ORIGINAL INVESTIGATION

M. Tekin · S. H. Blanton · X. J. Xia · X. Z. Liu · W. E. Nance ·A. Pandya (✉ )Department of Human Genetics, Medical College of Virginia/Virginia Commonwealth University,1101 East Marshall Street, PO Box 980033, Richmond, VA 23298-0033, USA e-mail: apandya@hsc.vcu.edu, Tel.: +1-804-8289632, Fax: +1-804-8283760

M. Tekin · N. Akar · Ș. CinDepartment of Pediatrics, Ankara University School of Medicine,Ankara, Turkey

S. H. BlantonDepartment of Pediatrics, University of Virginia, Charlottesville,Va., USA

© Springer-Verlag 2001

386

Fig.1A–C Results of mutation and haplotype analysis in two pedi-grees representing six families from Turkey. Haplotype alleles arefor D13S250, D13S175, D13S141, and D13S1275 markers fromcentromere to telomere (T-A, G-G alleles for SNP markers CGAP-

C-50237 and CGAP-C-53447, respectively). A Two unrelated fam-ilies from Turkey (asterisk relationship is more distant than secondcousins). B Four unrelated families from Turkey. C Depiction ofdifferent haplotypes associated with the 35delG mutation

387

from each participant. For carrier screening, 674 anonymized DNAsamples were collected from subjects who had no known hearingloss and who had participated in unrelated research projects atAnkara University School of Medicine.

The coding exon (exon 2) of the Cx26 gene was sequenced inthe 21 deaf subjects with an ABI 377 sequencer, whereas the 5’ un-translated region (exon 1) was screened by single-strand conforma-tional polymorphism analysis with silver staining. Polymerase chainreaction (PCR)-amplified DNA samples from hearing control sub-jects were tested for only the 35delG and 167delT mutations fol-lowing digestion with the BslI (Storm et al. 1999) and PstI restric-tion enzymes, respectively. Homozygous and heterozygous controlswere run in each gel as positive controls. All positive results ob-tained by the PCR/restriction fragment length polymorphism (PCR-RFLP) method were confirmed by direct sequencing.

To evaluate for founder effects, we performed genotyping onseven unrelated Turkish 35delG homozygotes and on 113 35delGnegative control subjects by using three microsatellite markers,D13S250, D13S175, and D13S141 (centromere to telomere), whichmap within a 2-cM region centromeric to the Cx26 gene (Car-rasquillo et al. 1997; Zelante et al. 1997; http://marshmed.org/ge-netics). In the presence of parental consanguinity (noted in three sub-jects), an allele was counted only once if it was identical by descent.In the two large pedigrees (Fig.1A, B), an additional microsatellitemarker (D13S1275), located 1 cM distal to D13S141 and probablyflanking the Cx26 gene (Zelante et al. 1997; http://marshmed.org/ge-netics) was typed and used in constructing the haplotypes. In addi-tion, the unrelated homozygotes for the 35delG mutation and 21normal control subjects were typed for the CGAP-C-50237 andCGP-C-53447 single nucleotide polymorphism (SNP) markers(http://lpg.nci.nih.gov), which are within the 3’ untranslated regionof the Cx26 gene, by direct sequencing with an ABI 377 auto-mated sequencer.

Thirty-five unrelated deaf US Caucasian probands, homozy-gous for the 35delG mutation, and 44 hearing controls were alsotyped for the three microsatellite markers (D13S250, D13S175,and D13S141). Sixteen homozygotes and 15 controls from the USwere also screened for the two above-mentioned SNPs. These sam-ples were obtained from the National DNA Repository for the Studyof Hereditary Deafness, at the Department of Human Genetics atthe Medical College of Virginia, VCU.

Primer sequences used in all these experiments are available onrequest.

Results and discussion

The 35delG mutation was detected in seven of the eleven(64%) Turkish families with prelingual non-syndromichearing loss. Sixteen of the 17 deaf subjects in these sevenfamilies were homozygous for the mutation. One deaf sub-ject from a hearing-by-hearing marriage in a multiplexfamily was heterozygous for 35delG and did not carry the167delT mutation as shown by PCR-RFLP; repeated at-tempts at sequencing in order to detect a second mutationwere unsuccessful. A heterozygous V27I change was notedin a deaf subject from a family that did not have any otherCx26 mutation.

The 35delG mutation in the Cx26 gene is a commoncause of prelingual deafness in Turkey. Our sample in-cluded four deaf-by-deaf marriages, all of which had onlydeaf offspring and may therefore represent noncomple-mentary matings between individuals with the same formof recessive deafness (Fig.1A, B). All of these four (100%)marriages were found to be Cx26 (+)/Cx26 (+). If Cx26mutations accounted for half of all recessive deafness, thislocus should explain 1/2×1/2=1/4 of all non-complementarymatings. Our findings therefore suggest that there can beno other yet-to-be discovered form of recessive deafnessthat has a higher frequency than Cx26 deafness in Turkey.

Twelve of the 674 Turkish control subjects were het-erozygous for the 35delG mutation (1.78%; 95% confi-dence interval: 0.9%–3%), but no examples of the 167delTmutation were encountered. Based on the frequency of the35delG heterozygotes, the gene frequency would be 0.009,and the estimated frequency of homozygotes would be 8.1in 100,000, assuming random mating. In the presence ofassortative mating, however, this figure could represent asubstantial underestimate of true phenotype frequency.

Fig.1C

The genotyping results are shown in Table 1, whereasthe haplotypes associated with the 35delG mutation intwo relatively large Turkish kindreds are shown in Fig. 1A,B and C. The six nuclear families in these two pedigreesare not known to be related and had not known each otherbefore the marriage of deaf members of the families. In-terestingly, only one of the four consanguineous couplescarried identical 35delG haplotypes (family 2), and a totalof eight different 35delG haplotypes were noted (Fig.1C).Clearly, consanguinity and even identity in state do notnecessarily imply identity by descent.

The observed haplotypic heterogeneity of the 35delGmutations supports the view that recurrent mutation maycontribute to the high frequency of Cx26 deafness in theTurkish population. On the other hand, our samples from35delG-positive and -negative subjects from Turkey andthe US suggest a modest degree of linkage disequilibriumfor the 35delG allele in both countries (Table 1). The pres-ence of identical or potentially derived haplotypes in sev-eral unrelated Turkish progenitors supports the evidencefor disequilibrium (Fig.1C). Assortative mating, when cou-pled with relaxed selection and stochastic events, may beanother potential cause for linkage disequilibrium. In a fewgenerations, these mechanisms could have amplified hap-lotypes that were present in the population when assortativemating among the deaf first became common in Turkey.

Although we lack data on secular trends in the fertilityand mating structure of the deaf in Turkey, marriagesamong the deaf are clearly not exceptional in this countryas they are, for example, in India. Thus, the suggestive ev-idence for disequilibrium between the 35delG mutationand closely linked markers could either indicate a foundereffect or possibly the amplification by assortative matingof several independent mutations, which happened to occuron chromosomes carrying the same closely linked poly-morphic markers. More extensive haplotypic data couldhelp resolve these possibilities.

Acknowledgements This work was partially supported by NIHgrants KO8HDO1172–01A1 and RO1DCO2530–0-AB to A.P. andW.E.N. We thank the Turkish families for their kind participationin this study, and Ayșegül Ülker, Esin Bozbaș, Zeliha Gümüș, andKadir Tekin for their help with the family data. We also thank Bar-bara Landa for her technical assistance and Dr. Robert Morell forhis help in the determination of the allele sizes.

References

Carrasquillo MM, Zlotogora J, Barges S, Chakravarti A (1997)Two different connexin 26 mutations in an inbred kindred seg-regating non-syndromic recessive deafness: implications forgenetic studies in isolated populations. Hum Mol Genet 6:2163–2172

388

Table 1 Results of typingwith three microsatellite mark-ers and two SNPs in the Turk-ish and American populations

aP<0.05 for carrier chromo-somes vs control chromosomesin the Turkish populationbP<0.005 for carrier chromo-somes vs control chromosomesin the American populationcP<0.001 for carrier chromo-somes vs control chromosomesin the American population

Markers Alleles Frequency in 35delG chromosomes Control chromosomes(bp) from deaf families

Turkish American Turkish American

D13S250 n=11 n=64 n=208 n=861 (285) 0.454 0.406 0.668 0.5562 (289) 0.545 0.563 0.327 0.4423 (293) 0.016 0.005

D13S175 n=11 n=64 n=226 n=881 (99) 0.0232 (101) 0.077 0.026 0.0453 (103) 0.062 0.230 0.2614 (105)a,b 0.818 0.766 0.491 0.4665 (107) 0.015 0.079 0.0456 (109) 0.182 0.030 0.079 0.0687 (111) 0.0228 (113) 0.030 0.071 0.0919 (117) 0.015

D13S141 n=11 n=66 n=222 n=881 (117) 0.013 0.0572 (119) 0.0053 (121) 0.005 0.0114 (125) 0.364 0.150 0.540 0.4775 (127)c 0.636 0.820 0.387 0.4436 (129) 0.030 0.050 0.011

SNPs n=11 n=32 n=42 n=30

CGP-C-50237 T-A 1.000 1.000 0.880 0.966

CGP-C-53447 G-G 0.071 0.033G-A 0.048

Cohn ES, Kelley PM (1999) Clinical phenotype and mutations inconnexin 26 (DFNB1/GJB2), the most common cause of child-hood hearing loss. Am J Med Genet 89:130–136

Denoyelle F, Weil D, Maw MA, Wilcox SA, Lench NJ, Allen-Powell DR, Osborn AH, Dahl HH, Middleton A, HousemanMJ, Dode C, Marlin S, Boulila-ElGaied A, Grati M, Ayadi H,BenArab S, Bitoun P, Lina-Granade G, Godet J, Mustapha M,Loiselet J, El-Zir E, Aubois A, Joannard A, Levilliers J,Garabedian EN, Mueller RF, Gardner RJM, Petit C (1997)Prelingual deafness: high prevalence of a 30delG mutation inthe connexin 26 gene. Hum Mol Genet 6:2173–2177

Gasparini P, Rabionet R, Barbujani G, Melchionda S, Petersen M,Brondum-Nielsen K, Metspalu A, Oitmaa E, Pisano M, FortinaP, Zelante L, Estivill X (2000) High carrier frequency of the35delG deafness mutation in European populations. Eur J HumGenet 8:19–23

Kudo T, Ikeda K, Kure S, Matsubara Y, Oshima T, Watanabe K,Kawase T, Narisawa K, Takasaka T (2000) Novel mutations inthe connexin 26 gene responsible for childhood deafness in theJapanese population. Am J Med Genet 90:141–145

Morell RJ, Kim HJ, Hood LJ, Goforth L, Friderici K, Fisher R,Van Camp G, Berlin CI, Oddoux C, Ostrer H, Keats B, Fried-man TB (1998) Mutations in the connexin 26 gene (GJB2)among Ashkenazi Jews with nonsyndromic recessive deafness.N Engl J Med 339:1500–1505

Nance WE, Liu XZ, Pandya A (2000) Relation between choice ofpartner and high frequency of connexin-26 deafness. Lancet356:500–501

Storm K, Willocx S, Flothmann K, Van Camp G (1999) Determi-nation of the carrier frequency of the common GJB2 (connexin26) 35delG mutation in the Belgian population using as easyand reliable screening method. Hum Mutat 14:263–266

Zelante L, Gasparini P, Estivill X, Melchionda S, D’Agruma L,Govea N, Mila M, Monica MD, Lutfi J, Shohat M, MansfieldE, Delgrosso K, Rappaport E, Surrey S, Fortina P (1997) Con-nexin 26 mutations associated with the most common form ofnon-syndromic neurosensory autosomal recessive deafness(DFNB1) in Mediterraneans. Hum Mol Genet 6:1605–1609

389