Methods and results

Fifty one subjects with non-syndromic hearing loss were ascertained through genetic counselling and audiological medicine clinics at Great Ormond Street Hospital, London and from referrals forGJB2 testing to the North Thames (East) Regional Diagnostic Laboratory, London. The ethnic background of the cases was approximately 75% white, but included cases from the Indian subcontinent, such as India, Pakistan, and Sri Lanka. The 51 cases included the unaffected parents of a deaf child who was initially unavailable for mutation testing. Twelve of the cases had previously identified base changes and were used as positive controls (table 1) and 18 of the remainder had already been tested for the 35delG mutation. All cases had congenital sensorineural hearing loss (mostly severe to profound) and were examined by a clinical geneticist to exclude syndromic causes of deafness.

Two sets of primers were used for amplification of the entire coding region, and the GJB2 sequence was taken from GenEmbl accession number M86849. Two overlapping fragments were amplified using previously published primers4 Cx26AF 5′-TCTTTTCCAGAGCAAAC CGC-3′ with Cx26AR 5′-GACACGAAGAT CAGCTGCAG-3′ (287 bp) and Cx26BF 5′-GGGCAATGCGTTAAACTGGC-3′ with Cx26BR 5′-CCAGGCTGCAAGAACGTG TG-3′ (522 bp). PCR conditions were 32 cycles of denaturation at 96°C for one minute, annealing at 60°C for one minute, and extension at 72°C for one minute. PCR amplification was carried out oil free, owing to the sensitivity of DHPLC to mineral oil.

DHPLC, unlike SSCP, uses double stranded DNA. As heteroduplexes form weaker interactions with the hydrophobic column matrix owing to mismatch pairing of the DNA when partially denatured, these are eluted from the column sooner than homoduplexes during reverse phase ion exchange HPLC. This allows identification of mutations because of the difference in elution peak patterns when using the Transgenomic WAVE^TM DNA Fragment Analysis System.19 Optimal column temperature and acetonitrile gradient for each fragment was calculated for the two fragment sequences using the WAVEMaker^TM program (conditions available on request). All samples were run both unspiked (only patient PCR product present) and spiked (sample mixed in 1:1 ratio with wild type PCR product), so that homozygote alleles would form heteroduplexes and be detected. It was also essential to run the samples unspiked to ensure that the patient's sample had amplified. The samples and a normal control (sequenced to check for changes) were amplified in microtitre plates. For heteroduplex creation, 10 μl of PCR product (“unspiked”) was heat denatured and allowed to cool slowly using the following conditions: 95°C for two minutes, 90°C for two minutes, 85°C for two minutes, and so on in 5°C decrements until 25°C was reached. Five μl of each sample was mixed with 5 μl of normal control PCR product (“spiked”) and also subjected to heteroduplex formation. A wild type control was included in each run. Spiked and unspiked DHPLC analysis was carried out for the first and second overlapping fragments, giving a total of four runs for each sample. All samples were directly sequenced using BigDye^TMTerminator Cycle Sequencing kit (PE Applied Biosystems) and run on an ABI 377 DNA sequencer. PCR amplification used primers Cx263F (5′-GGTCCTGTGTTGTGTGCATTC-3′) with Cx26R (5′-CTGGGCAATGCGTTAAACTG GC-3′) and the products were purified using Microspin S400 columns (Amersham Pharmacia).

Fifty one samples were tested, of which 12 had known nucleotide changes (table 1). DHPLC analysis of the two overlapping fragments of the gene detected all 12 positive control base changes shown in table 1 in their relevant amplicons, as well as eight additional altered elution peaks in cases with no known mutation. The results of sequencing these eight cases can be seen in table 2.

One of these changes was identified in a person from a large Sri Lankan family and contained an unusual double nucleotide substitution, c→t at 493, R165W, on one allele, and a previously described sequence variant g→a at 457, V153I, on both alleles.

The two W77X heterozygotes were the consanguineous parents of an affected child, who was subsequently confirmed to be a W77X homozygote by restriction enzyme digestion. No mutant homozygotes were detected in any of the unspiked samples.

Discussion

Severe to profound congenital deafness is a common genetic disorder affecting 1:1000 births. With an increasing demand forGJB2 mutation detection by audiological physicians, geneticists, and families, a rapid and accurate screening service for the variety of mutations prevalent in our multi-ethnic society will be essential in regional DNA diagnostic laboratories. In this cohort of multi-ethnic deaf subjects, DHPLC identified the 12 known mutations among the samples and also detected sequence changes in eight cases where no mutation had previously been identified. No other nucleotide variants were found upon subsequent sequence analysis of each sample. Our screening strategy would not have detected the IVS+1g→a mutation, which occurs at the donor splice site in the first intron, between exon 1 (non-coding) and exon 2, which contains all of the coding sequence of the gene, as this would be in a separate PCR amplicon. As expected in a recessive disorder, no homozygote mutations were identified in samples from this cohort that were not spiked with wild type DNA, as no heteroduplexes were formed.

Recent comparisons of SSCP versus DHPLC in other genes have shown DHPLC to be superior in almost every study, with mutation detection rates of 92-100%.15-18 Moreover, it has many advantages over conventional gel based mutation detection techniques, as it is easy to use, gives consistent results, and allows analysis and storage of data on computer. With its microtitre plate based technology, which can be combined with a PCR robot, there is also the potential to allow higher throughput screening. This may become necessary given the prevalence of congenital deafness in the population and the increasing awareness of the role of GJB2 as an aetiological factor by families and those caring for large numbers of deaf children in primary care.

As with other disorders such as cystic fibrosis, the prevalence of a common mutation (35delG) decreases significantly outside the white population. Approximately half the white subjects screened were already known to be 35delG negative, and DHPLC yielded only two new mutation positives (one 35delG homozygote and one 35delG/W24X compound heterozygote), both in cases which had not previously been tested for 35delG. However, the remainder of the new mutations identified were in patients from the Indian subcontinent, and none of these mutations were 35delG.

Examination of the previously undetected mutations that DHPLC identified, with regard to ethnic origins of the patients, has yielded interesting data. The W77X truncating mutation was detected in two families, one a Pakistani family in which the proband was a W77X homozygote and in another Bangladeshi subject, who was also homozygous for the mutation. This mutation has been described previously in Pakistani and Indian kindreds suggesting that it may be common in the Indian subcontinent. We also identified two hearing impaired sibs from India who were heterozygous for the W24X truncating mutation, which has also been described in Pakistani and Indian families.4 7A second mutation in the GJB2 coding region has not been identified in these sisters by DHPLC or by sequencing. It is possible that they have an unidentified mutation in the untranslated regions or promoter of the gene, or even in another deafness gene. Finally, we have identified a homozygous Q124X mutation in a person of Indian origin, which has also been described in a patient from the same ethnic background.7 The discovery of these previously reported mutations suggests that a number of common mutations may be enriched in families from the Indian subcontinent, although without population studies it is not possible to be sure whether the prevalence in this ethnic group is statistically higher. Haplotyping of DNA from different families with these mutations may indicate whether they are likely to be founder mutations or not.

Our DHPLC and sequence analysis has shown an unusual combination of sequence variants in a large Sri Lankan pedigree. The proband had mild/moderate hearing loss and was shown to have a R165W amino acid variant on one chromosome and V153I on both chromosomes. His mildly affected father was shown to be heterozygous for both changes on the same allele. It is possible that the R165W variant is a mild dominant mutation in this family. R165W has not been previously reported as a disease causing mutation and was not detected in forty normal controls. The arginine residue at position 165 is conserved in the highly homologous human Cx30,31, and 32 genes, and also in the mouse GJB2 gene, and occurs in the second extracellular domain of the protein.

V153I has been reported in a heterozygous state in hearing subjects (Hilbert et al. The connexin 26(GJB2) deafness homepage athttp://www.iro.es/cx26deaf.html) and we have detected it heterozygously in two out of 186 normal controls. Further investigation has now shown that a profoundly deaf woman from this ethnic group, who married into the family, is homozygous for V153I. However, we have found V153I to be segregating within this community, which operates a dowry and caste system, and we cannot be certain that the variant is the cause of her deafness and not a homozygous non-pathogenic polymorphism. Valine to isoleucine is a conservative amino acid change. V27I and V37I are other amino acid changes involving a valine to isoleucine substitution and both have been identified heterozygously in normal controls.6 V27I was found to be enriched in the normal Japanese population (39%),9 suggesting that it occurs too frequently to be a deafness causing mutation, whereas V37I is thought to be pathogenic as it has been found on both alleles of a deaf patient.20 However, it is extremely difficult to determine the effect of these three sequence variants when all occur in the normal population. Further functional and genetic studies are under way in this family, in order to determine the pathogenicity of both V153I and R165W.

In the UK, many deaf children attend primary care based clinics under the care of community paediatricians and audiologists. With completion of the first draft of the sequence of the Human Genome, awareness of the role of genetics in hearing loss is becoming more widely appreciated throughout the health care system. Thus, the demand for molecular diagnosis will increase. Genetic services will need to respond by providing rapid, sensitive, semiautomated mutation screening of the GJB2 gene.