- Top of page
- Subjects and Methods
We screened the GJB2 gene for mutations in 534 (108 multiplex and 426 simplex) probands with non-syndromic sensorineural deafness, who were ascertained through the only residential school for the deaf in Mongolia, and in 217 hearing controls. Twenty different alleles, including four novel changes, were identified. Biallelic GJB2 mutations were found in 4.5% of the deaf probands (8.3% in multiplex, 3.5% in simplex). The most common mutations were c.IVS1 + 1G > A (c.-3201G > A) and c.235delC with allele frequencies of 3.5% and 1.5%, respectively. The c.IVS1 + 1G > A mutation appears to have diverse origins based on associated multiple haplotypes. The p.V27I and p.E114G variants were frequently detected in both deaf probands and hearing controls. The p.E114G variant was always in cis with the p.V27I variant. Although in vitro experiments using Xenopus oocytes have suggested that p.[V27I;E114G] disturbs the gap junction function of Cx26, the equal distribution of this complex allele in both deaf probands and hearing controls makes it a less likely cause of profound congenital deafness. We found a lower frequency of assortative mating (37.5%) and decreased genetic fitness (62%) of the deaf in Mongolia as compared to the western populations, which provides an explanation for lower frequency of GJB2 deafness in Mongolia.
- Top of page
- Subjects and Methods
Mutations in the gap junction β-2 (GJB2) gene, encoding the connexin 26 (Cx26) protein, account for up to 50% of non-syndromic autosomal recessive deafness in some populations (Estivill et al., 1998; Rabionet et al., 2000). Although more than 150 DNA alterations in GJB2 have been reported, a significant variation in the frequency and distribution of the mutations has been observed in different populations (The Human Gene Mutation Database –http://www.hgmd.cf.ac.uk; FINDbase worldwide –http://www.findbase.org). The c.35delG mutation accounts for up to 70% of the pathogenic alleles in Caucasian populations (Denoyelle et al., 1997; Estivill et al., 1998; Kelley et al., 1998; Rabionet et al., 2000) with carrier frequencies of 2.8% in Southern Europe,1.3% in Northern Europe (Gasparini et al., 2000), and 2.5% in the Midwestern U.S. (Green et al., 1999). Another GJB2 mutation, c.167delT, has a 4% carrier frequency among Ashkenazi Jews (Morell et al., 1998) and accounts for 40% of the pathologic alleles in the deaf population of Israel (Sobe et al., 2000). These two mutations are absent or exceptionally low in East Asian populations, but another single base pair deletion, c.235delC, is relatively frequent (Fuse et al., 1999; Abe et al., 2000; Kudo et al., 2000; Liu et al., 2002). A nonsense mutation, p.W24X was found to be frequent in India and in European Gypsies (Minarik et al., 2003; RamShankar et al., 2003). A single founder has been described for each of these four population specific common mutations (Morell et al., 1998; Van Laer et al., 2001; RamShankar et al., 2003; Yan et al., 2003). Furthermore, two variants (p.V27I and p.E114G), believed to be polymorphisms, were found to be very common in East Asian countries such as Japan, Korea, and Thailand (Fuse et al., 1999; Abe et al., 2000; Kudo et al., 2000; Park et al., 2000; Wattanasirichaigoon et al., 2004).
We had proposed earlier that the current high prevalence of GJB2 mutations in many but not all populations is the consequence of relaxed selection and assortative mating among the deaf (Nance et al., 2000). It seems plausible to assume that in previous millennia the genetic fitness of individuals with deafness must have been very low, possibly approaching zero. With the discovery that the deaf are educable, the development of sign language, and the establishment of schools for the deaf during the past 2–3 centuries there has been a remarkable improvement in the social, educational, and economic circumstances of the deaf that has been associated with a well documented increase in their fertility (Rose, 1975). The spread of sign language and residential schools for the deaf was accompanied by the onset of intense assortative mating among the deaf. This pattern of marriages would be expected to cause a selective amplification of the commonest form of recessive deafness in a population. We have recently shown that in the U.S., non-complementary matings which can only produce deaf children have increased by a factor of more than five in the past 100 years in association with a statistically significant linear increase in the frequency of pathologic GJB2 mutations (Arnos et al., 2008).
We screened a large representative population from Mongolia, including deaf probands and control subjects, for GJB2 mutations to determine the importance of Cx26 deafness in this population and to investigate possible differences in the distribution of the mutant alleles. We also obtained data on the mating characteristics and genetic fitness of the deaf in Mongolia in order to test the hypothesis that the high frequency of GJB2 mutations in many but not all populations is the consequence of relaxed selection and assortative mating.
Subjects and Methods
- Top of page
- Subjects and Methods
We performed GJB2 mutation screening on a total of 534 deaf probands with non-syndromic severe to profound sensorineural hearing loss, who were ascertained in Ulaanbaatar at the only residential school for the deaf in Mongolia. The probands and available family members were clinically evaluated by one of the authors for a history of deafness, family history, and physical examination. Individuals with a syndromic form of deafness were excluded. An audiogram was obtained on most patients during these examinations. We identified 108 probands with at least one hearing impaired relative (multiplex probands), and 426 who were the only deaf member in their family (simplex probands). Blood spots from 217 controls were obtained during blood donation following a brief interview in which hearing difficulties and family history of hearing loss were specifically questioned. Individuals with a history of hearing loss or positive family history were excluded. A signed informed consent was obtained from all participants.
To find the carrier frequency of the c.IVS1 + 1G > A (c.-3201G > A) mutation in Turkish speaking populations, who once shared the same land with Mongolians and later migrated to the Near East and Anatolia (Cinnioglu et al., 2004), 206 Anatolian and 101 Azeri Turks with normal hearing were screened. Three Anatolian Turks with homozygous c.IVS1 + 1G>A (c.-3201G>A) mutation had been previously identified with DNA sequencing using the same protocol described below. These probands were among the more than 500 unrelated Turkish subjects who were evaluated for the genetic causes of deafness at Ankara University School of Medicine as a part of our ongoing research program. These three homozygotes along with seven Mongolian homozygotes for the same mutation were genotyped for seven SNPs flanking c.IVS1 + 1G>A (c.-3201G>A) to evaluate the genealogy of this mutation. Thirty-four chromosomes from Anatolian Turkish subjects with normal hearing were typed for the same SNPs in order to find allele frequencies in the general population.
DNA was extracted from peripheral blood or blood spots using standard protocols. The coding exon (exon 2) of GJB2 was PCR amplified using forward (F4) 5′GCT TAC CCA GAC TCA GAG AAG 3′ and reverse (R1′) 5′ CTT AAT CTA ACA ACT GGG CAA TGC 3′ primers. Both forward and reverse internal primers, forward (F4′) 5′CTG TCC TAG CTA TGT TCC 3′ and reverse (R1) 5′ TCA GCA CGG GTT GCC TCA ATC 3′, were used for each sample in cycle sequencing reactions. The DNA sequencing was performed on an ABI 377 automated sequencer. Three other primers (F5: 5′ GCT GCA AGA ACG TGT GCT ACG ATC 3′; F2: 5′ GGA CAT CGA GGA GAT CAA AAC C 3′; R3: 5′ GAA GAC GTA CAT GAA GGC GGC 3′) were used to obtain smaller size PCR products when the initial primers failed to amplify the entire exon. The c.IVS1 + 1G>A (-3201G>A) mutation in intron 1 of GJB2 was screened using direct sequencing or a PCR-RFLP protocol. Primers used were; forward 5′-CCC TCC GTA ACT TTC CCA GT-3′ and reverse 5′-CCA AGG ACG TGT GTT GGT C-3′. The c.IVS1 + 1G>A (c.-3201G>A) mutation removes the single HphI recognition site, which was used for RFLP. Most of the samples had been screened for mitochondrial mutation m.1555GA>G, the results of which were published previously (Pandya et al., 1999), and 37 students were found to be positive. Detected DNA changes in the GJB2 gene were denoted according to recommendations of the Human Genome Variation Society using the NCBI Reference Sequence NC_000013.10 (den Dunnen & Antonarakis 2000).
All samples that remained heterozygous for one pathologic GJB2 allele were screened for the DeltaGJB6-D13S1830 (342 kb) deletion using a previously reported protocol (Pandya et al., 2003). Known homozygous and heterozygous samples were used as positive controls.
In order to distinguish relative localisation of the p.V27I and p.E114G alterations, 32 deaf and 16 hearing control samples that were found to have one p.V27I and one p.E114G changes were PCR amplified using F4 and R1′ primers and cycle sequenced using allele specific primers for the p.V27I change. Allele specific primers were: forward p.V27I wild type 5′ GGA AAG ATC TGG CTC ACC G 3′; and p.V27I mutation specific 5′ GGA AAG ATC TGG CTC ACC A 3′. Results were obtained using an ABI 377 automated sequencer.
The functional consequences of the p.V27I and p.E114G mutations were investigated by comparing the channel activity of wild type and mutant proteins using an in vitro expression system composed of paired Xenopus oocytes as reported previously (White et al., 1998). Homozygous samples for the p.V27I and p.E114G mutations were used for obtaining and cloning constructs.
To determine the fitness and frequency of assortative mating among the deaf, a survey of 158 deaf adults living in Ulaanbaatar City and two subprovinces in Arhangai and Suhebataar was conducted. Measures of fertility and fitness were obtained from 162 deaf adults who were 35 years of age or older and their 151 hearing siblings. Genetic fitness can refer to the reproductive success of a group, an individual, a genotype or specific phenotype, and can have important socioeconomic and behavioural, as well as biological determinants. Fertility is usually measured by the number of live-born offspring except for traits associated with childhood mortality or sterility where the average number of grandchildren may be more appropriate. Estimates of the genetic fitness are obtained by determining the ratio of the average fertility of cases with matched controls selected to remove potential sources of bias. For genetic studies, affected and normal siblings are often compared because they are members of the same chronologic cohort, have the same expected age, and are closely matched for other genes, ethnic background, religion, parental socioeconomic status, and home environment.
- Top of page
- Subjects and Methods
Twenty different alleles were identified in the study participants including four (p.W3X, p.W172C, c.290_295delACCGGAinsCCCG, and p.V226D) previously unreported changes and three complex alleles that contains more than one previously reported change, c.[79G>A;341A>G] (p.[V27I;E114G]), c.[-3201G>A;79G>A], and c.[-3201G>A;79G>A;341A>G] (Table 1). The observed genotypes of the detected mutations and polymorphisms in deaf probands and hearing controls are shown in Table 2. Three novel sequence changes, p.W3X, c.290_295delACCGGAinsCCCG, and p.W172C, were paired with previously reported pathogenic mutations in the deaf probands.
Table 1. Observed alleles in 534 deaf probands and 217 hearing controls from Mongolia.
|Alleles||Deaf (frequency)||Control (frequency)||Reference|
|c.-3201G>A (c.IVS1 + 1G>A)||-||36(0.034)||NA||Denoyelle et al., 1999|
|c.[-3201G>A;79G>A]1||-||4(0.004)||NA||Complex allele is reported in this study|
|c.[-3201G>A;79G>A;341A>G]||-||2(0.002)||NA||Complexe allele is reported in this study|
|c.35delG||p.G12fs||2(0.002)||0||Zelante et al., 1997|
|c.35insG||p.V13fs||1(0.001)||0||Estivill et al., 1998|
|c.79G>A||p.V27I||164(0.153)||75(0.173)||Kelley et al., 1998|
|c.[79G>A;341A>G]||p.[V27I;E114G]||114(0.106)||55(0.127)||Fuse et al., 1999; Abe et al., 2000 (as separate changes)|
|c.109G>A||p.V37I||6(0.006)||0||Kelley et al., 1998|
|c.235delC||p.L79fs||16(0.015)||2(0.005)||Fuse et al., 1999; Abe et al., 2000|
|c.299_300delAT||p.H100fs||4(0.004)||2(0.005)||Abe et al., 2000|
|c.312_325delGAAGTTCATCAAGG||p.K105fs||1(0.001)||0||Denoyelle et al., 1997|
|c.368C>A||p.T123N||2(0.002)||0||Park et al., 2000|
|c.380G>A||p.R127H||1(0.001)||0||Estivill et al., 1998|
|c.427C>T||p.R143W||1(0.001)||0||Brobby et al., 1998|
|c.457G>A||p.V153I||3(0.003)||1(0.005)||Marlin et al., 2001|
|c.608T>C||p.I203T||5(0.005)||1(0.002)||Abe et al., 2000, Kudo et al., 2000|
|TOTAL|| ||1068||434|| |
Table 2. Distribution of genotypes in deaf students and hearing controls.
|Genotypes|| ||Deaf probands||Controls|
|Two pathologic alleles1||c.[-3201G>A]+[-3201G>A]||5||1||4||NA|
|Total (frequency)||24 (0.045)||9 (0.083)||15 (0.035)||-|
|One pathologic allele1||c.[-3201G>A]+[=]||6||2||4||NA|
|Total (frequency)||29 (0.054)||6 (0.055)||23 (0.054)||4 (0.018)|
|Two polymorphic alleles||c.[79G>A]+[79G>A]||12||3||9||5|
|Total (frequency)||52 (0.097)||12 (0.110)||40 (0.094)||15 (0.069)|
|One polymorphic allele||c.[79G>A]+[=]||104||19||85||56|
|c.[79G>A;341A>G]+[=]||66 (0.123)||12 (0.111)||54 (0.127)||44 (0.203)|
|Total (frequency)||172 (0.322)||31 (0.287)||141 (0.331)||102 (0.470)|
|Total number of study subjects||534||108||426||217|
Only 24 of 534 deaf probands were found to carry two pathogenic mutations (frequency 0.045, with a 95% confidence interval of 0.028–0.062). Of these, 0.083 were from multiplex and 0.035 from simplex families. The carrier frequencies of both c.235delC and c. 299_300delAT were 0.009 in 217 control subjects. DeltaGJB6-D13S1830, a 342 kb deletion in the GJB6 gene, was not detected in any of the heterozygous samples.
Eight samples (seven deaf individuals and one hearing control) were homozygous for both p.V27I and p.E114G alterations, demonstrating that these two mutations occurred in cis. Although the p.V27I change was observed several times either in the homozygous or heterozygous state without the p.E114G variant, p.E114G was never detected alone in our samples (Table 3). All studied samples that were initially found to be heterozygous for both p.V27I and p.E114G by sequencing were later shown to be p.[V27I;E114G]+[=] using allele specific sequencing (Fig. 1). These results suggest that the p.E114G alteration is always on the same chromosome that carries the p.V27I change, and haplotype analysis confirmed the absence of p.E114G without p.V27I in the Mongolian population (Table 3).
Table 3A. Joint distribution of the p.V27I and p.E114G alterations among deaf probands and hearing controls.
Table 3B. Frequencies of haplotypes created with p.V27I and p.E114G alterations.
|Subjects||V27I−;E114G− (frequency)||V27I+;E114G− (frequency)||V27I+;E114G+ (frequency)||V27I−;E114G+ (frequency)|
|Deaf Probands||784 (0.734)||169 (0.158)||115 (0.108)||0|
|Hearing Controls||304 (0.700)||75 (0.173)||55 (0.127)||0|
Figure 1. Upper panel: Sequencing of the region from the PCR product of regular primers (F4-R1′) and using F4′ primer in Sanger sequencing. Lower panel: Sequencing results of a person who is heterozygous for c.79G>A (p.V27I) and c.341A>G(p.E114G) using c.79G (wild type) specific and c.79A (mutant) specific forward primers.
Download figure to PowerPoint
The results of in vitro functional studies on the Xenopus oocyte system demonstrated that the double mutant allele with p.V27I and p.E114G in cis, significantly impaired the function of the gap junctions (Fig. 2).
Figure 2. Results of the functional studies on the single p.V27I and double p.[V27I;E114G] alleles in Xenopus oocyte system. The mean intercellular conductance of p.[V27I;E114G] is <5% of wild type (p < 0.05). uS = microSiemens, the unit of electrical conductance derived by dividing the gap junctional current (in nanoamperes) by the voltage difference between the cells (in millivolts).
Download figure to PowerPoint
Genotyping of seven SNPs showed that different haplotypes (at least six) were associated with the c.IVS1 + 1G>A (c.-3201G>A) mutation in Mongolians. However, three Turkish homozygotes showed a single conserved haplotype, which suggests a single common ancestor with an intervening population bottleneck in the Turkish branch (Table 4). Two heterozygotes in 206 Anatolian, and one in 101 Azeri Turks were detected. The overall carrier frequency of c.IVS1 + 1G>A (c.-3201G>A) in two Turkish populations was estimated to be three in 307 (0.0097).
Table 4. Genotypes of seven flanking SNPs, p.V27I, and p.E114G associated with c.IVS1 + 1G>A (c.-3201G>A).
|Frequency of A* in controls||0.94||0.94||0.87||0.7||0.97||0.59||0.99||0.97||0.7||0.94|
|Physical distance from||8592||6557||3535||3273||1084||785||0||1229||2797||4800|
| IVS1 + 1G>A (bp)|| || || |
|Telomeric side||Centomeric side|
As shown in Table 5, among the 127 deaf individuals who were married, a frequency of 0.375 selected deaf partners. In Table 6, data on the fertility of 162 deaf adults and 151 hearing siblings are summarised. The reduced overall fertility of the deaf was attributed both to a lower frequency of marriage as well as the reduced fecundity of those who were married, yielding an overall genetic fitness of 0.62.
Table 5. Marital status of 158 deaf adults in Mongolia.
|Status||Males N = 87||Females N = 71||Total N = 158|
| Deaf × Hearing||0.46||0.55||0.50|
| Deaf × Deaf||0.28||0.34||0.30|
Table 6. Genetic fitness of the deaf in Mongolia.
| ||Deaf Adults||Hearing Siblings|
|Males N = 90||Females N = 72||Total N = 162||Males N = 54||Females N = 97||Total N = 151|
|Genetic Fitness||0.62||0.62||0.62|| || || |
- Top of page
- Subjects and Methods
With a population density of 4.3 per square mile, the 2.9 million citizens of Mongolia inhabit the most sparsely populated country in the world. Approximately 30% of the population is nomadic or semi-nomadic. The majority are of Mongols of Halh ethnicity, although at least a dozen other Mongol minorities are represented as well as about 6% who are of Turkic extraction including the Kazakhs, Tuvans, Urianhai and Hoton ethnic groups. Our sample is estimated to include at least half of the student-age deaf population of Mongolia.
Our results clearly demonstrate that the prevalence of the pathologic GJB2 mutations is much lower in the deaf population of Mongolia compared to many Caucasian populations from Western Europe and the U.S. (Denoyelle et al., 1997; Estivill et al., 1998; Kelley et al., 1998; Rabionet et al., 2000; Pandya et al., 2003). Even among the multiplex probands, in whom the deafness is most likely due to a genetic etiology, the frequency of biallelic GJB2 mutations was only 8.3%. We have previously suggested that the high frequency of GJB2 deafness in many populations is at least in part the consequence of the combined effects of relaxed genetic selection against deafness, which coincided with the introduction of sign language and the establishment of residential schools for the deaf in many western populations, along with the appearance of intense assortative mating among the deaf (Nance et al., 2000). We have shown by computer simulation that this mechanism would specifically increase the frequency of the commonest gene for recessive deafness in a population and could have doubled the frequency of GJB2 mutations in 200 years (Nance & Kearsey, 2004). More recently, we have provided direct evidence for such changes in a comparative genetic analysis of data on marriages among the deaf collected more than a century apart in the U.S., as well as secular trends in the incidence of GJB2 deafness in a contemporary sample of deaf probands (Arnos et al., 2008). Although the School for the Deaf in Ulaanbaatar was established early in the 20th Century, sign language was not used until its introduction by an American Peace Corps worker in 1995. Prior to that time, the communication skills of deaf adults were generally quite poor. Because there is no “deaf culture” in Mongolia or any long tradition of intermarriage among the deaf, we expected that the fitness, assortative mating, and frequency of GJB2 deafness would all be low.
To test the hypothesis that these demographic features of the deaf population of Mongolia may contribute to the low frequency of GJB2 deafness in that country, we estimated the frequency of assortative mating among the deaf as well as their genetic fitness in a sample of deaf adults and their hearing siblings as shown in Tables 5 and 6. In the U.S., the frequency of assortative mating among the deaf is at least 0.86 and their genetic fitness had increased to about 0.85 by 1970 and may be higher today (Schein & Marcas, 1974).
Another interesting observation noted in this study is the invariable occurrence of the p.E114G change in coupling with the p.V27I substitution. These two variants were initially reported as independent common polymorphisms in Asian populations (Kelley et al., 1998; Fuse et al., 1999; Abe et al., 2000; Kudo et al., 2000; Park et al., 2000). Kudo et al. (2000) used these polymorphisms to create haplotypes and detected six chromosomes with both changes. They also reported six chromosomes with only p.E114G without p.V27I. It has also been reported in previous studies that p.V27I and p.E114G can be present independently. Of these two, the former is much more common (Tang et al., 2006). In the Mongolian population eight examples of individuals homozygous for p.[V27I;E144G] were observed. The in vitro data on channel function suggest a pathogenic role for the double mutant allele, but our population data from normal hearing controls included one double homozygote, and a very similar distribution of allele frequencies in the deaf probands and hearing controls. Previous studies have also reported at least 20 control subjects homozygous for the p.[V27I;E114G] allele (Park et al., 2000; Wattanasirichaigoon et al., 2004; Tang et al., 2006). Nevertheless, in the absence of audiologic testing, we cannot exclude sub-clinical hearing loss in our single doubly homozygous control subject. Therefore, despite the data from deaf probands in Mongolia and in vitro studies suggesting a pathogenic role for the p.[V27I;E114G] allele, in aggregate, the genetic data, especially its homozygosity in a hearing control argues against a causative role of p.[V27I;E114G] for fully penetrant, early onset severe to profound deafness. It will be interesting to see if this allele plays a role in mild or slight hearing loss, similar to the reported phenotype of p.V37I (Dahl et al., 2006). It is conceivable that some p.[V27I;E114G] alleles are pathologic because of an as yet unidentified regulatory mutation outside of exons. If this were the case, we might expect to find the allele located on a single or limited number of extended haplotypes in populations in which there has been a recent expansion in the number of pathologic GJB2 alleles. The data in Table 3 permit estimation of the frequency of the four possible haplotypes involving the p.V27I and p.E114G substitutions on the 1068 proband and 434 control chromosomes. There was no significant difference in the distribution of haplotypes in the deaf and control subjects. The absence of the V27I-;E114G+ haplotype in both samples is entirely consistent with the observed joint distribution of genotypes for the two individual polymorphisms.
We have identified three novel mutations in GJB2. The p.W3X alteration was predicted to result in severe truncation of the protein. The c.290_295delACCGGAinsCCCG mutation was predicted to result in a stop codon due to a frame shift after substitution of three amino acids starting at codon 97. The p.W172C change is located in the second extracellular domain of Cx26. Its effect on the protein was assessed using the PolyPhen website (http://genetics.bwh.harvard.edu/pph/) where a PSIC (Position-Specific Independent Counts) score difference between W and C was found to be 2.578. Based on this difference p.W172C was considered to be probably damaging. The p.V226D change is located in the 3rd intracellular domain. PolyPhen gave a PSIC score of 2.33 for this missense alteration, which was also considered to be probably damaging.
The most common pathogenic GJB2 allele we identified in Mongolia is c.IVS1 + 1G>A (c.-3201G>A). This mutation impairs proper splicing between non-coding exon 1 and coding exon 2 of GJB2. It was first described in a Caucasian population (Denoyelle et al., 1999), in which it accounted for at least 0.75% of the mutant alleles in a deaf population (Snoeckx et al., 2005a). It was subsequently reported in many other populations of the world, including Egypt (Snoeckx et al., 2005b), India (RamShankar et al., 2003), Turkey (Sirmaci et al., 2006), and Iran (Najmabadi et al., 2005). Since mutation analysis of the non-coding exon was not included in all reported studies, it is difficult to assess the worldwide distribution of this mutation. We provide evidence in the present study that this mutation has a carrier frequency of approximately 0.01 in two Turkish populations, which suggests that it is widespread in the Near East as well as in Central Asia. The haplotype analysis in Mongolian samples shows different haplotypes associated with c.IVS1 + 1G>A (c.-3201G>A), suggesting multiple origins in the Mongolian population. However, a single haplotype was inferred from three Turkish samples that were homozygous for the mutation, which also appears to be a common haplotype in Mongolia. Since Turks and Mongolians shared a long history in Central Asia, it is possible that c.IVS1 + 1G>A (c.-3201G>A) in the Turkish population has its origin in Central Asia.
In summary, we report that the prevalance of biallelic GJB2 mutations is only 0.045 among the deaf in Mongolia. A lower frequency of assortative mating and decreased genetic fitness of the deaf in Mongolia as compared to the western populations provides an explanation for the lower frequency of GJB2 deafness in Mongolia. The most common GJB2 mutation in Mongolia is c.IVS1 + 1G>A (c.-3201G>A), which appears to have diverse origins based on multiple associated haplotypes. The p.V27I and p.E114G variants are frequent in both deaf probands and hearing controls and the p.E114G variant is always in cis with the p.V27I variant. Although our in vitro experiments suggest a pathogenetic role for the complex allele, the equal distribution of p.[V27I;E114G] in deaf probands and hearing controls makes it a less likely cause of profound congenital deafness.