METHODS

Subjects

A Dutch family with autosomal dominant hearing loss, W99–060, was ascertained (fig 1). Pure tone audiograms were obtained as described by Kemperman et al.¹⁰ The study was approved by the local ethics committee, and written informed consent was obtained from each participating individual.

Amplification of the ACTG1 gene

DNA was isolated from blood samples using a salting out procedure.¹¹ Primers for amplification of exons and exon–intron boundaries of the P4HB, MAFG, CARD14, SLC26A11, and DKFZp434D1428 genes are available on request. Primers for amplification of the six exons and splice sites of the ACTG1 gene were designed using the published genomic sequence (NCBI accession No, M19283). The primers used (Invitrogen), annealing temperatures, and MgCl₂ concentrations are listed in table 1. Amplification was done in a PTC200 thermo cycler (MJResearch). Cycling conditions were five minutes denaturation at 95°C, followed by 30 cycles of 30 seconds at 95°C, 30 seconds annealing, and one minute elongation at 72°C.

Sequence analysis

Polymerase chain reaction (PCR) fragments were purified on Qiaquick spin columns (Qiagen). Sequencing reactions were done with the ABI PRISM Big Dye Terminator cycle sequencing V2.0 ready reaction kit. For the analysis we used the ABI PRISM 3700 DNA analyser (Applied Biosystems). The sequencing primers were the same as those used for amplification of the exons. For sequencing of exon 6 two additional internal primers were synthesised: 5′-TGTACATATCTTTGAGTTC-3′ and 5′-CAAGCCACAGACTTGTCTTC-3′.

Amplification-refractory mutation system analysis

Screening of the c.833C>T mutation in exon 5 in 300 unrelated control individuals was carried out by amplification-refractory mutation system (ARMS) analysis. We designed ARMS primers for both the wild-type and mutant allele—wild-type reverse primer, 5′-ACGTCACACTTCATGATGGAGTTGAAAG-3′; mutant reverse primer, 5′-ACGTCACACTTCATGATGGAGTTGAAAA-3′. As a forward primer, the exon 4 forward primer was used. Amplification was done as described above with an annealing temperature of 58°C and an MgCl₂ concentration of 2.0 mM on a PTC200 thermo cycler (MJResearch).

Molecular modelling

The mutation was analysed using the DGMUT command¹² in WHAT IF.¹³ DGMUT scanned 400 high resolution x ray structures for stretches of five residues with an isoleucine in the middle, which superimpose well on the five residues centred on threonine in the known actin structure (1NM1). For every hit, the isoleucine rotamer was displayed on screen. Especially for large flexible side chains, such an analysis yields a distribution of equally likely rotamers, and energy functions are required to choose the best one. In the case of the Thr278Ile mutation, however, only one possibility was found.

RESULTS

Genetic analyses

Recently, the genetic defect causing hearing loss in a Dutch family with dominant progressive hearing loss was localised to the DFNA20/26 region at chromosome 17q25.3. The critical interval in the present family was between the marker D17S784 and 17qter.¹⁰ The hearing loss in the family is comparable to that in the family described by DeWan et al,⁹ in which the critical region is flanked by the markers D17S784 and D17S928. In order to identify the causative gene for DFNA20/26, candidate genes were selected from the interval between D17S784 and D17S928 comprising 6 cM (2.4 Mb). This region contains 62 known genes and more than 100 expressed sequence tags (ETS). Causative mutations in the protein coding regions and intron–exon boundaries of the genes P4HB, MAFG, CARD14, SLC26A11, and DKFZp434D1428 were excluded by sequence analyses. Recently, the ACTG1 gene encoding γ 1 actin was relocated in the critical region.¹⁴ Owing to the important function of actin in hair cell structure, this gene was considered a major candidate and the exons and exon–intron boundaries were amplified for an affected individual (IV:7) and sequenced. Several changes were found when compared with the published sequence of the gene (NCBI accession No, M19283). Eight nucleotide changes were found including the known polymorphism Tyr306Tyr (ENTREZ SNP database 1139405). In exon 1 the c.-55G>A substitution was homozygously present, as was the g.IVS3+13C>A substitution. The heterozygous changes g.IVS1+135C>T, g.IVS1+151G>C, and g.IVS2-38C>T are not located within the splice site consensus sequences, are not predicted to create new splice sites, and are also not located within the branchpoint.¹⁵ The detected g.IVS5+54 delG does not co-segregate with the disease. In exon 5 the nucleotide transition c.833 C>T was detected. The putative effect of this mutation is the substitution of isoleucine for threonine at position 278 of the protein (Thr278Ile) (fig 2).

An ARMS PCR was developed to test for the presence of the c.833C>T mutation in control individuals and to analyse the segregation of the mutation in the family. The mutation co-segregates with the disease (fig 1) but is not present in individual IV:2 who was already suggested to be a phenocopy by linkage analysis.¹⁰ The mutation was not present in 300 white controls, of whom 200 were of Dutch origin. Moreover, residue Thr278 is conserved in almost all actin isoforms throughout eukaryotes including yeast (fig 3). These results exclude the mutation being a common polymorphism and suggest that it is causative for the disease.

Molecular modelling

The effect of the Thr278Ile mutation was investigated using a known actin structure from Dictyostelium discoideum, actin 1 (PDB entry 1NM1).¹⁶ Contrary to known vertebrate actins, this structure was solved at high resolution (1.8 Å), while it is also highly similar to human γ actin (95% sequence identity, fig 3). As the residues around Thr278 are also conserved, the mutation could be analysed directly in the known structure, without building a model for human actin first. Using WHAT IF,¹³ we found that an isoleucine at position 278, located in helix 9 (residues 274–282) of the protein,¹⁷ can adopt only one possible side chain conformation or rotamer (fig 4). In this conformation, its δ carbon comes within 2.4 Å of the ε carbon of methionine 313 in helix 11 (residues 309–320). This is 1.4 Å shorter than the sum of the Van der Waals radii. Owing to the tight packing inside the hydrophobic core, Met313 has no space to move away from this strong clash. It has been shown that clashes of this type often lead to reorganisation of entire secondary structure elements,¹⁸ in this case helices 9 and 11. A movement of either helix is likely to affect actin polymerisation: helix 9 provides structural support for the hydrophobic loop 262–274 which was shown to be involved in polymerisation,¹⁹ while helix 11 is attached to the ATP binding P3 loop.¹⁶

Additional evidence that the Thr278Ile mutation cannot be accommodated without affecting the function comes from another member of the actin multigene family in D discoideum (Swissprot accession No, P07828): this protein actually has an isoleucine at position 278, which is compensated by a Thr162Ala exchange that creates the space required by the additional methyl group in Ile 278. In the mutant reported here, this space is not available.

DISCUSSION

We have shown here that the c.833C>T mutation in the gene encoding cytoplasmic γ-1-actin (ACTG1) is involved in dominant sensorineural hearing loss DFNA20/26. The c.833C>T mutation in ACTG1 is the first germline mutation described in a cytoplasmic actin leading to disease.

The absence of the nucleotide change in 300 control individuals, together with structural analysis, indicates that the change is the disease causing mutation. Mutation analysis in four additional families linked to the DFNA20/26 locus might reveal additional mutations in the ACTG1 gene, thereby providing definite proof for ACTG1 as a hearing loss gene. Alternatively, this analysis might point towards a second gene in the region.

Actins are ubiquitous proteins in eukaryotes and are involved in diverse cellular functions including muscle contraction, cell motility, cell adhesion, and cell shape.²⁰ In vertebrates, six different isoforms—three α, one β, and two γ—are distinguished. The α isoforms and one γ isoform are muscle specific. The β and γ isoforms are the non-muscle cytoplasmic isoforms.²¹ Actin monomers (G-actin) polymerise to form filaments (F-actin). The affected residue Thr278 is invariant in all human actin isoforms and is highly conserved in other eukaryotes. By using a known actin structure, the effect of the amino acid substitution can be predicted to cause a reorganisation of the helices 9 and 11 in subdomain III of the actin molecule owing to an increase in size of the side chain at position 278 in the protein. Helix 11 is attached to the P3 loop containing residues 302–308 that form one side of the adenine pocket²² (reviewed by Hennessey et al²³).

Because actin monomers are stabilised by ATP binding under physiological conditions and unbound actin is rapidly degraded,²⁴ impairment of ATP binding by the Thr278Ile mutation might lead to a reduced availability of G-actin for polymerisation into filaments. Helix 9 is attached at its N-terminal side to loop 262–274, which has been shown to play a critical role in polymerisation of actin into filaments.¹⁹ At its C-terminal side helix 9 is attached to loop 282–288. Residues in this loop make close contact with profilin, an actin monomer binding protein that facilitates the addition of these monomers to the plus end of F-actin.²⁵ Reorganisation of these loops because of the Thr278Ile mutation can therefore be predicted to influence actin polymerisation. In addition to the localised effects, the mutation might have long range effects on the structure of other regions of the actin molecule as is shown for other amino acid substitutions in actin by Drummond et al.²⁶

The consequences of this type of effect for actin function are difficult to predict. That the Thr278Ile mutation leads to function impairment is supported by studies on actin mutated in either helix 9 or helix 11. The Arg312His mutation in the cardiac actin gene ACTC is causative for dominantly inherited dilated cardiomyopathy.²⁷ The Arg312Ala mutation in yeast ACT1 leads to reduced viability,²³ and the Glu316Lys mutation in the Drosophila Act88F actin leads to reduced stability of the protein in vitro, muscles with slightly disrupted ultrastructure, and altered kinetics of force generation by muscle fibres.^28,29 Glu316 forms hydrogen bonds with residues 312 and 275 (helix 9).²⁹ The amino acid substitution Asn280Lys in human ACTA1 leads to dominant nemaline myopathy.³⁰

Although cytoplasmic β and γ actins are ubiquitously expressed, the only symptom in the present family is sensorineural hearing loss. This points towards actin based structures specific for the inner ear as the site of the defect. The most conspicuous cochlear structures containing actin filaments are in the hair cells. They include the stereocilia, which contain a rigid dense core of actin filaments, the cuticular plate into which stereocilia are anchored, consisting of a gel-like actin network, and a ring of parallel actin filaments connected to the zonula adherens.³¹ Although β actin is the predominant cytoskeletal isoform in several mouse tissues,³² in chicken auditory hair cells γ actin is the predominant actin isoform, present in all three conspicuous actin based structures.²¹ β Actin is specifically sorted to the actin core filament bundles of the stereocilia.²¹ The actin filaments of the stereocilia core are continuously renewed at the tip through the addition of G-actin. The filament bundles “treadmill” towards the base of the stereocilia.³³ This process might be disturbed by a reduced amount of available G-actin. This would imply that the mutation results in haploinsufficiency. Impaired polymerisation through a rearrangement of the loops 262–274 and 282–288 might also affect the process of filament renewal in stereocilia. Whether the cuticular plate and the zonula adherens belt are subject to a comparable fast renewal of actin filaments remains to be determined.

Besides an effect on stability or incorporation of the mutated actin monomers in filaments, the mutation also might have an effect through a subtle change in the structure of the molecule. Upon incorporation of mutant G-actin into a filament, this structural change might lead to functional impairment of the whole filament. Such dominant negative effects are shown to be caused by mutated subunits in collagen fibres and keratin filaments.^34,35

The hearing loss in the present family shows a very rapid progression. Deterioration is almost complete within the first three decades of life.¹⁰ The high frequencies are clearly affected most within these first three decades, which might be explained by the fact that the rigidity of the stereocilia is highest in the basal region of the cochlea³¹ and therefore will suffer most from impaired stereocilia structure. The highly progressive nature of the hearing loss suggests that initially the affected structures are functioning normally but that over several decades a combination of a disbalanced renewal of actin filaments and the effects of aging lead to an early deterioration of the system. Progressive retinitis pigmentosa in patients with Usher syndrome caused by myosin VIIa (MYO7A) mutations might well be an example of the same type of pathogenesis. Photoreceptor disks are constantly renewed and this process is affected by MYO7A mutations.³⁶

It is noteworthy that the causative genes in the three other known types of dominant sensorineural hearing loss with rapid progression in the first three decades of life (DFNA1, DFNA17, and DFNA22) encode proteins that are functionally related to actin (diaphanous, non-muscle myosin heavy chain A and myosin VI).^37–39 The diaphanous gene (DFNA1), for example, plays a role in the regulation of actin polymerisation.⁴⁰

Audiograms in the present DFNA20/26 patients from about 30 years of age onwards are reminiscent of the hearing loss in (young) patients with USH1.⁴¹ Mouse models for the causative genes for USH1, myosin VIIA (MYO7A), cadherin 23 (CDH23), protocadherin 15 (PCDH15), and SANS show disorganised hair bundles (reviewed by Ahmed et al⁴²). Several histological and biochemical studies indicate that these proteins—together with the F-actin bundling protein harmonin, mutated in USH1C—are part of a functional complex anchored to the actin core of stereocilia and responsible for the cohesion of these hair cell protrusions.^43–45 The similarities observed between DFNA20/26 and the above mentioned non-syndromic and syndromic types of hearing loss support the hypothesis that the Thr278Ile mutation in γ actin 1 leads to stereocilial dysfunction. Our data suggest that the gene encoding β actin is a good candidate for hereditary hearing loss.