METHODS

Subjects

Familial cases of non-syndromic hearing loss were collected with the only criterion of having at least two affected subjects. Our collection procedure did not cause any other bias, such as preferential selection of large pedigrees or compatibility with maternal inheritance. A total of 649 unrelated Spanish families were enrolled in the study. After getting informed consent, peripheral blood samples were obtained from all participating family members, and DNA extraction was performed by standard procedures.

Mutation detection

Screening for the 1555A>G mutation was carried out by PCR amplification of a 339 bp DNA fragment containing the mutation site, followed by digestion with restriction endonuclease HaeIII, as described previously.¹⁴ In the wild type allele, digestion results in two fragments of 216 bp and 123 bp. The mutation specifically creates a novel restriction site, and so digestion results in three fragments (216 bp, 93 bp, and 30 bp).

Quantification of the mutation load

The proportion of mutant copies was quantified by detection of fluorescently labelled PCR products separated by capillary electrophoresis. A 359 bp DNA fragment was amplified with hot start FastStart Taq DNA polymerase (Roche) using 75 ng of DNA from the subject as template. Primer sequences were: upper primer 5′-AGACGTTAGGTCAAGGTG-3′; lower primer, 5′-GTTTAGCTCAGAGCGGTC-3′. The upper primer was fluorescently labelled at 5′ with TET. PCR was carried out in a final volume of 15 μl with the following conditions: an enzyme activation step, at 95°C for six minutes; 10 cycles of denaturation at 94°C for 15 seconds, annealing at 50°C for 15 seconds, and extension at 72°C for 30 seconds; 13 cycles of denaturation at 89°C for 15 seconds, annealing at 50°C for 15 seconds, and extension at 72°C for 30 seconds; plus a final extension step at 72°C for 10 minutes. Subsequently, 8 μl of the PCR product were digested at 37°C for two hours with a large excess (10 units) of restriction endonuclease Alw26I (Fermentas), in a final volume of 10 μl. Three microlitre aliquots of digestion were subjected to capillary electrophoresis in an ABI Prism 310 Genetic Analyzer (Applied Biosystems) according to the recommendations of the manufacturer. By using this technique, two fluorescent products can be detected in a given sample, wild type (127 bp) and/or mutant (359 bp, since the 1555A>G mutation destroys the restriction site for Alw26I). In heteroplasmic subjects, the proportion of mutant copies was estimated from the peak areas of the fragments, in quantification experiments from three independent PCR amplifications. The quantification results were consistent with those previously observed in agarose gels after HaeIII digestion.

RESULTS AND DISCUSSION

At least one subject from each of the 649 collected families was tested for the presence of the mitochondrial 1555A>G mutation, the result being positive in 105 families (16%). In the positive cases, a search for mutation carriers was performed on all the remaining participating relatives. This screening showed heteroplasmy for the mutation in 19 subjects from six unrelated families, which also included 12 subjects with the mutation in homoplasmy (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Pedigrees of the six Spanish families segregating the 1555A>G mutation in heteroplasmy. A question mark inside a symbol is used to represent subjects whose clinical status could not be ascertained. Age (in years) and audiograms are shown below or to the right of subject symbols. Hearing level (in dB) is plotted versus sound frequency (in Hz). Since the hearing loss was sensorineural in all cases, only results for air conduction are depicted. Circles, right ear; crosses, left ear. For each subject, the proportion (%) of mutant copies (mean of three independent experiments (standard deviation)), estimated from DNA from peripheral blood, is indicated below the audiogram.

All of the mutation carriers in the families with heteroplasmy were studied clinically. There were records of treatment with aminoglycoside antibiotics in only two subjects (S138 I.2 and S160 I.2, streptomycin). Other environmental factors were excluded as causes of hearing loss in all the subjects. No syndromic features were found. Conductive hearing loss was ruled out by otoscopic examination, tympanometry with acoustic reflex testing, and use of the tuning fork tests. Pure tone audiometry, testing for air and bone conduction, confirmed that the hearing loss was bilateral and sensorineural in all affected subjects. Audiograms for air conduction are shown in fig 1. There were no vestibular symptoms except in patient S141 I.2, who reported episodes of positional vertigo. Patients S141 I.2, S141 II.3, and S338 II.1 reported bilateral tinnitus.

As expected, the pattern of transmission of the 1555A>G mutation was consistent with maternal inheritance in all the families (fig 1). The proportion of mutant copies was determined in the 19 heteroplasmic subjects and ranged from 3.75–96.60% (fig 1). The mutation load in the offspring of a heteroplasmic mother was highly variable. For instance, subject S338 I.2 (52.14% of mutant copies) has two sons with the mutant allele in homoplasmy within our detection limits, and another son with values close to homoplasmy for the wild type allele (3.84% of mutant copies). A wide variation is also observed in the offspring of subject S138 I.2 (fig 1). In addition, pedigrees S160 and S068 have some relevant characteristics. Subject S160 I.2, homoplasmic for the mutant allele within our detection limits, has two heteroplasmic daughters (94.74% and 96.15% of mutant copies, respectively). This result indicates that subject S160 I.2 keeps some wild type copies at least in the germline. In peripheral blood, the wild type allele would have been lost or would be in a proportion small enough to go undetected. As regards pedigree S068, heteroplasmic subject II.5 has two sibs (a brother and sister) who are apparently homoplasmic for the wild type allele. However, both her brother (II.1) and a son of her sister (III.1) are affected by bilateral sensorineural hearing loss, more severe in the high frequencies. Two hypotheses may explain these data. First, I.2, the mother of subjects II.1, II.3, and II.5, would have carried the 1555A>G mutation in heteroplasmy. The mutation would remain in II.5, but would have been lost in II.1 and II.3. If this were the case, the hearing loss in subjects II.1 and III.1 would have a cause different from the 1555A>G mutation, which is a plausible explanation given the genetic heterogeneity of non-syndromic hearing loss. It should be taken into account, however, that the characteristics of the hearing loss in subjects II.1 and III.1 closely resemble those of other members of the family, but it is also true that high frequency hearing loss is the most common type. The second hypothesis concerns the possibility that peripheral blood from subjects II.1, II.3, and III.1 contained a very small, undetected, proportion of mutant copies (apparent homoplasmy) or none at all (real homoplasmy). However, the mutation load in the inner ear would be large enough to be pathogenic in subjects II.1 and III.1. Were this the case, its implications would be relevant for genetic diagnosis (see below).

We investigated the effect of heteroplasmy on the severity of the hearing loss. Subjects carrying less than 20% of mutant copies were asymptomatic (S138 II.1 and II.2, S297 III.2, S338 II.3 and II.4), or had a mild hearing loss (subject S297 IV.1, with a U shaped audiogram). Conversely, the remaining 13 subjects, with percentages of mutant copies between 52.14% and 96.60%, suffered from hearing loss. Two of them, monozygotic twins from family S141 (II.1 and II.2) with 70.93% and 64.48% of mutant copies, respectively, had a mild hearing loss for high frequencies. Five others (S068 II.5 and S138 II.3, II.4, II.5, and II.6), with percentages of mutant copies between 69.78% and 96.60%, had normal hearing for low and middle frequencies, but suffered from moderate or severe hearing loss for high frequencies. In the remaining six cases, the audiogram shape was sloping, affecting both middle and high frequencies (S141 II.3, S160 II.1, and S338 I.2; mutant copies ranging from 52.14% to 94.74%) or all the frequencies (S138 I.2, S141 I.2, and S160 II.3; mutant copies ranging from 61.03% to 96.15%). In this last group, subject S138 I.2 had a history of treatment with streptomycin. The study of 10 subjects from families S068, S160, and S338, carrying the mutation in homoplasmy, showed that eight of them had hearing losses which were more severe than those of their heteroplasmic relatives (only one homoplasmic subject, S160 I.2, had a history of treatment with streptomycin). The two remaining cases (S160 II.5 and II.6) were asymptomatic, but it should be considered that they are younger than their four affected sibs, and may be below the age of onset.

A statistical analysis of our data for all the six families showed significant correlation of the mutation load with the hearing thresholds, for all the frequencies (125–8000 Hz range) (fig 2A), and for only the high frequencies (2000–8000 Hz range) (fig 2B). However, these results must be interpreted cautiously, as indicated by intrafamilial analysis. In three families (S068, S160, S338), the severity of the hearing loss clearly correlates with the mutation load. Regarding family S141, the proportion of mutant copies influences the severity, but this seems to be modulated also by age. Conversely, no apparent correlation is observed between the severity of the hearing loss and the mutation load in four sibs from family S138 with mutant copies ranging from 69.78% to 96.60% (II.3–II.6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Statistical correlation analysis of mutation load (proportion of mutant allele) and hearing thresholds. Data from a total of 27 subjects from the six families were included in the analysis (the two subjects with a record of treatment with aminoglycoside antibiotics were excluded). Spearman rank correlation tests were performed using GraphPad InStat version 3.05 for Windows 95 (GraphPad Software, San Diego, California, USA). r_s, Spearman correlation coefficient. (A) Pure tone average (PTA) of hearing thresholds for all the frequencies (125–8000 Hz range) versus mutation load. (B) Pure tone average (PTA) of hearing thresholds for the high frequencies (2–8 kHz range) versus mutation load.

In subjects with only high frequency hearing loss, it was difficult to ascertain their age of onset, since frequently they were not aware of their hearing loss. However, a majority of cases with the mutation in heteroplasmy reported that the hearing loss first manifested in adulthood (between 17 and 50 years of age). In contrast, in eight out of 10 subjects carrying the mutation in homoplasmy, onset was in early childhood (between 1 and 5 years of age).

The study of the genotype-phenotype correlation in subjects carrying the 1555A>G mutation in homoplasmy, which are the vast majority of the cases reported so far, has shown considerable heterogeneity in age of onset, evolution, severity, and other audiological features of the hearing loss resulting from this mutation.^4,5 This variability has been attributed to the influence of both environmental and genetic factors. Undoubtedly, aminoglycoside antibiotics induce a severe worsening of the hearing loss in mutation carriers. In addition, there is in vitro evidence of the influence of the nuclear background in modulating the phenotype caused by the 1555A>G mutation.^18,19 Also, the hypothesis of the existence of nuclear genes acting as modifiers of mitochondrial hearing loss has recently received strong support.^20–23 The existence of a not negligible percentage of cases with heteroplasmy (5.7% in our sample of 105 families with the 1555A>G mutation) adds more complexity to the picture. Our study of a set of 19 heteroplasmic subjects illustrates the difficulties in extracting general principles from the analysis of the genotype-phenotype correlation regarding this mutation. First, among our heteroplasmic cases, most of the subjects carrying less than 20% of mutant copies were asymptomatic, whereas all of the subjects with mutation loads higher than 52% suffered from hearing loss. This suggests that there is a threshold in mutation load for manifestation of clinical symptoms. However, it is also known that there exist subjects homoplasmic for 1555A>G, who are asymptomatic (for example, subjects S160 II.5 and II.6). Second, statistical analysis of our data indicates a significant correlation of the the severity of the hearing loss with the mutation load when considering the six families altogether. However, when studying the families separately, this correlation is confirmed in three of them and excluded in another (family S138). This situation may be because of intrafamilial differences in the nuclear background modulating the phenotype, and/or individual variability in mutation load in peripheral blood and inner ear. In fact, it has been reported that the level of heteroplasmy for a given mutation can vary among different tissues within the same person.²⁴

The conclusions of our study are relevant for genetic diagnosis of mitochondrial mutations that are responsible for non-syndromic hearing loss. The estimations of mutation load obtained from mitochondrial DNA from peripheral blood may not always reflect accurately the real situation in the inner ear. In extreme cases, the mutation load may be pathogenic in the inner ear and remain undetectable in blood. Therefore, in large families with several affected subjects and a clear maternal inheritance of the disorder, several probands from different branches in the pedigree should be tested before excluding the presence of the mutation. This issue may be critical for prevention of aminoglycoside ototoxicity in subjects whose carrier status would go unnoticed.

Although the last few years have witnessed great advances in the understanding of mitochondrial pathogenesis, there are many important issues that remain unsolved, such as the basis of tissue specificity and the mechanisms by which a heteroplasmic mutation segregates and is fixed. Further investigation of these matters is needed to improve genetic counselling regarding the 1555A>G mutation.