Учебники / Genetic Hearing Loss Willems 2004

Figure 2 Coomassie-stained gels of mouse (a,b) and chick (c,d) tectorial membrane proteins separated by SDS-PAGE on 8.25% gels under nonreducing (NR) and reducing (R) conditions. The collagenase-sensitive components of the mouse tectorial membrane are indicated by small arrowheads. Collagens are not present in the chick tectorial membrane. Mouse a-tectorin (a, lane a) generates three bands on reduction, the high-, medium-, and low-molecular-mass mouse tectorins (H, M, and L, lane b). Chick a-tectorin (a, lane c) generates six bands on reduction (a1–a6, lane d). The difference in mobility of nonreduced mouse and chick a-tectorin (lanes a and c) is most likely due to a di erence in glycosylation. Mouse and chick h-tectorin have a similar apparent molecular mass under nonreducing conditions (h, lanes a and c), and comigrate with a-tectorin subunits on reduction (L and h, lanes b and d).

Antibodies raised to both the chick and the mouse tectorin ‘‘subunits’’ prepared by gel electrophoresis (1,8,9) all react intensely with the tectorial and otolithic membranes, but most show little, if any, cross-reactivity outside the inner ear. An antiserum directed against HMM mouse a-tectorin and an antibody raised to the chick a1-tectorin subunit stain the mucus layer in the olfactory epithelium of the nose (9). An antiserum directed against the combined a5 and a6 chick tectorin subunits stains the epithelial glands supplying the respiratory epithelia of the avian nose (8). These studies suggest that the tectorins have a very restricted tissue distribution, and that a-tectorin may have properties in common with the mucins.

IV. CLONING AND MOLECULAR STRUCTURE

OF THE TECTORINS

Chick and mouse tectorin cDNAs were obtained by screening cochlear cDNA libraries with antibodies and DNA probes (3,10,11). Southern blot analysis indicates the chick and mouse tectorins are the products of singlecopy genes. The deduced amino acid sequences for the tectorins from these two species have f74% similarity (73% for chick and mouse a-tectorin, 75% for chick and mouse h-tectorin) and predict virtually identical proteins in which all the major features are conserved. A composite DNA sequence for human a-tectorin derived from genomic DNA predicts a protein with 95% identity to mouse a-tectorin (12), also with all features conserved.

The cDNA for mouse a-tectorin predicts a large, modular protein of 239 kDa with an N-terminal domain that has similarity to the G1 domain of entactin (nidogen), a large central region composed of three full (D1–D3) and two partial (D0 and D4) von Willebrand factor (vWF) type D repeats, and a C-terminal region forming a zona pellucida (ZP) domain (Fig. 3a). The cDNA for mouse h-tectorin encodes a small, 36-kDa protein that is predicted to form a single zona pellucida domain (Fig. 3b). The predicted protein sequences for the chick, mouse, and human tectorins all encode proteins with hydrophobic N-terminal signal sequences characteristic of secreted proteins, and hydrophobic C-termini characteristic of proteins that are linked to the lipid bilayer via a glycosyl-phosphatidylinositol (GPI) anchor (13). The predicted sites for the addition of the C-terminal GPI anchor are preceded by extended or tetra-basic consensus sites for cleavage by furin-like endoproteinases (14,15). GPI anchors act as signals for targeting proteins to the apical membranes of polarized epithelial cells (16), so these sequence data indicate that the tectorins are most likely to be synthesized as GPI-linked, membrane-bound precursors, targeted to the apical surface of the sensory epithelia in the ear by the lipid tail, and released from the membrane either while on route to or when at the apical membrane.

The N-terminal amino acid sequence data obtained from gel-purified tectorin bands from mouse and chick (3) allow one to determine where the a-tectorin sequence from these species is cleaved to produce the various a-tectorin subunits that are observed on SDS gels under reducing conditions (Fig. 3a). This analysis indicates mouse a-tectorin is cleaved at two sites to produce the HMM, MMM, and LMM mouse a-tectorin subunits that correspond approximately to the central region containing the vWF type D repeats, the C-terminal ZP domain, and the N-terminal entactin G1-like domain, respectively. In chick, the same sites are used but the central vWF repeat region appears to be cleaved at two other points to produce the additional subunits, chick a5and chick a6-tectorin. The sites at which the a-

Figure 3 Diagram illustrating the predicted domain structures for (a) a- and (b) h- tectorin, and how a-tectorin (a) is cleaved to produce the LMM, HMM, and MMM fragments in the mouse, and the a1–a6 fragments in the chick. Predicted molecular masses (in Da) are indicated for each fragment. ENT = entactin G1-like domain; D = vWF type D repeats; ZP = zona pellucida domain; black boxes = hydrophobic N- and C-termini; hatched boxes indicate regions with no similarity to other proteins.

tectorin sequence in mouse and chick is cleaved to produce these subunits are not specific for any known endoproteinases, and the extent to which the fragments observed on reduction are bona fide subunits and a consequence of active posttranslational processing or a consequence of proteolytic damage at susceptible sites remains to be determined.

The N-terminal region of a-tectorin shares sequence similarity with a part of the G1 domain of entactin (nidogen), a component of the basal lamina that is known to interact with laminin, type IV collagen, perlecan, and fibulin (17). These interactions mostly involve the G2 and G3 domains of entactin, although there is some evidence from surface plasmon resonance

assays that the G1 domain of nidogen-1 interacts with fibulin-2 (17). It is hard to ascribe a function to the N-terminal entactin G1-like domain of a- tectorin, although a large deletion in this region (18) has a profound influence on the stability of the molecule or its ability to be incorporated into the matrix (see Sec. VII).

The vWF type D repeats of a-tectorin are cysteine-rich domains characteristic of a number of molecules including the mucins (19). vWF is produced by endothelial cells and megakaryocytes and is involved in hemostasis, acting as a carrier for procoagulant factor VIII and mediating platelet adhesion (20). It forms disulfide-linked dimers that can then assemble into large disulfide-linked multimers, and this multimerization process is thought to be dependent on the protein disulfide isomerase activity of vicinal cysteines present in the D1 and D2 domains of pre-pro vWF (21). Vicinal cysteines are present in the D1 and the D4 domains of a-tectorin and, although there is no evidence that a-tectorin forms disulfide cross-linked multimers, these vicinal cysteines may be involved in the assembly of the tectorin subunits.

The ZP domains of a- and h-tectorin are common to a number of other proteins, including ZP1, ZP2, and ZP3, the three major proteins of the zona pellucida, the extracellular matrix that surrounds the unfertilized egg. ZP domains are approximately 260 amino acids long, with a conserved pattern of hydrophobic, polar, and turn-forming residues and have eight highly conserved cysteine residues (22). The ZP domain is a module found in many proteins that either can or do form filamentous-based extracellular structures and is usually located at or toward the C-termini of these proteins (23). In addition to ZP1, ZP2, and ZP3, these proteins include GP2, a protein that forms a mesh-like matrix in the pancreatic acinar ducts (24), uromodulin, a major component of urine (25) that can form urinary casts under pathological conditions (26), and nompA, a protein that forms part of the dendritic cap of the insect mechanoreceptor (27). The ZP domains of a- and h-tectorin may enable these proteins to form the filamentous-based matrix of the tectorial membrane.

The vWF type D-repeats of a-tectorin show highest similarity with the vWF type D repeats of a pig sperm membrane protein known as zonadhesin that has been shown to bind to the zona pellucida in vitro (28), and may be involved in sperm-egg interactions in vivo. Zonadhesin is processed during sperm capacitation to leave a membrane-bound extracellular domain that is composed exclusively of vWF repeats with an arrangement that is identical to that found in the central core of a-tectorin.

These sequence similarities have led to the suggestion (3) that a- and h-tectorin may each form homomeric filaments via their ZP domains. These two di erent filament types could then interact with one another via the zon-

adhesin-like region of a-tectorin, possibly in conjunction with the entactin G1-like domain. Alternatively, a- and h-tectorin could form heteromeric (a/ h) filaments via their ZP domains. These filaments could also self-assemble into larger structures via interactions between the ZP domains and the zonadhesin-like domain of a-tectorin.

V.EXPRESSION OF TECTORIN mRNAs IN THE INNER EAR

Northern blot analysis of poly(A)+ mRNA from eight di erent tissues of the 2-3 day postnatal mouse indicates that the tectorins are expressed at high levels only in the cochlea, with probes for a- and h-tectorin hybridising to single major bands of 7.5 and 2.7 kb, respectively (3). Tectorin splicing has not been systematically investigated. However, there is evidence for the use of an alternative 3V splice site in Tecta (3,12). Also in the rat cochlea, a mouse h-tectorin cDNA probe hybridizes to two bands of 2.9 and 2.6 kb (29), indicating h-tectorin splice variants may exist in this species.

In situ hybridization has been used to study the spatial and temporal aspects of tectorin mRNA expression during the development of the mouse inner ear (30). Expression of both tectorins can be detected in the basal end of the cochlear duct as early as embryonic day (E) 12.5. By E14.5 expression is detected throughout the length of the cochlea. By birth two distinct but partially overlapping expression patterns are observed for the two tectorin mRNAs (Fig. 4). a-Tectorin mRNA is expressed throughout the greater epithelial ridge and in the immature Hensen’s cells that lie immediately lateral to the developing organ of Corti (Fig. 4a–c). h-Tectorin is expressed within a narrow region of the greater epithelial ridge that lies close to the inner hair cells, by the pillar cells that separate the rows of inner and outer hair cells, and by Deiter’s cells surrounding the outer hair cells, with expression levels being greatest in the third, outermost row of Deiter’s cells (Fig. 4d–f ). Expression of a-tectorin mRNA in the cochlea as revealed by in situ hybridization decreases rapidly during the first week after birth, and can no longer be detected by P15. h-Tectorin mRNA expression can be detected by in situ hybridization in the cochlea at P15, but not at P21. A similar postnatal decrease in tectorin mRNA expression has been shown using both in situ hybridization and Northern blot analysis in the cochlea of the rat (29), where expression is no longer detectable for either tectorin by P15. A quantitative RT-PCR study confirms there is a dramatic developmental decrease in a- tectorin mRNA levels in the mouse cochlea between P3 and P15, but suggests low-level expression may persist until P67 (31). These data indicate that the tectorin mRNAs are only expressed at high levels transiently during the development of the cochlea, and imply that the tectorins may be stable proteins with very long half-lives.

Figure 4 Distribution of mRNA for a- (a–c) and h- (d–f ) tectorin mRNA in the early postnatal mouse cochlea revealed by in situ hybridization of apical-coil cochlear cultures with DIG-labeled probes (a,b,d,e), and cochlear sections with 35S- labeled probes (c,f ). For a-tectorin (a–c), two bands of expression are observed, one encompassing the entire greater epithelial ridge (1), and the other in the immature Hensen’s cells that flank the lateral edge of the organ of Corti (2). For h-tectorin (d–f ), three bands are seen, one in the region of the greater epithelial ridge (1) adjacent to the inner hair cells, one in the pillar cells (2), and one in the third row of Deiter’s cells

(3). In parts c and f, G = greater epithelial ridge; H = Hensen’s cells; P = pillar cells; D = 3rd-row Deiters’ cell; large arrowhead indicates the position of the inner hair cell, and small arrowheads indicate the positions of the three rows of outer hair cells. Bars = 200 Am (a,d), 50 Am (b,c,e,f ). (Adapted from Ref. 30 with permission from the publisher.)

a- and h-tectorin are also expressed in the maculae of the saccule and utricle (30). Expression of a-tectorin mRNA is detected in the maculae of the mouse inner ear by E12.5, and expression of h-tectorin mRNA by E14.5. h- Tectorin mRNA expression is always restricted to the maculae, and within the maculae it is restricted to the striolar regions. The expression of h-tectorin mRNA persists within the striolar region until at least 5 months after birth (the latest stage examined thus far). a-Tectorin mRNA is also expressed in the epithelia that surround the sensory maculae and continues to be expressed in these flanking regions for the first 2 weeks after birth, after expression has decreased within the maculae. As in the cochlea, a-tectorin mRNA expression can no longer be detected in the vestibular organs by 3 weeks after birth. Tectorin expression has not been detected in any of the ampullary organs of the semicircular canals.

VI. MUTATIONS IN TECTA CAUSING HUMAN

HEREDITARY DEAFNESS

The mouse gene for a-tectorin, Tecta, maps to mouse chromosome 9, and the human TECTA gene maps to a syntenic region of human chromosome 11 (32), within the interval for autosomal dominant nonsyndromic hearing loss DFNA8/12 (33–35). Tectb maps to mouse chromosome 19, and TECTB maps to a region of synteny on human 10q15 (36). TECTB is not a candidate for any of the deafness loci that have been reported thus far, but mutations in TECTA have been shown to cause autosomal, nonsyndromic deafness in a number of familes. To date six families have been identified with mutations in TECTA (12,37–41); the data from these studies are summarized in Table 1. Mutations in TECTA cause both dominant (DFNA) and recessive (DFNB) forms of deafness. The dominant mutations are at conserved residues in both the ZP (12,41) and zonadhesin-like domains (38,39) of a-tectorin. In the Belgian family there are two closely spaced missense mutations, both of which are at residues conserved within the most closely related members of the ZP domain family. It is not yet known whether one of these is a rare polymorphism or whether the two mutations interact synergistically to produce the phenotype. In the Swedish family, there is considerable phenotypic variation that may possibly be caused by modifying genes.

The various dominant mutations lead to a spectrum of phenotypes. It was originally suggested (38) that mutations in the zonadhesin-like domain of a-tectorin lead to preor postlingual, progressive, high-frequency hearing loss, and mutations in the ZP domain cause prelingual, stable, midfrequency hearing impairments. However, the discovery of a family with a dominant mutation in a conserved cysteine residue in the ZP domain of a-tectorin that

is associated with a progressive midfrequency hearing loss (41) suggests the situation may be more complex. Currently it appears that mutations at cysteine residues may underlie the progressive forms of hearing loss. Inter and intrachain disulfide bonds may be essential for the long-term stability of a- tectorin, and it is interesting that the mutation in the French family involves one of the two vicinal cysteines in the vWF D4 repeat that is thought to have protein disulfide isomerase activity.

The recessive mutation (DFNB21) is a G-to-A transition at the intron 9 donor splice site that is predicted to cause skipping of the preceding exon and introduce a premature stop codon in the vWF D2 repeat (40). The mutation causes severe to profound deafness across all frequencies, is prelingual in onset, and may essentially be a null mutation. Heterozygous carriers in the DFNB21 family have apparently normal hearing, indicating that a reduction in the a-tectorin content of the tectorial membrane as a consequence of reduced gene dosage may have little e ect on hearing, and lending support to the suggestion (40) that the DFNA mutations have a dominant negative e ect. However, it is not yet known at what level the tectorin content of the tectorial membrane is regulated. Halving the gene dose may not necessarily lead to a reduction in the amount of protein incorporated.

VII. HEARING IN A Tecta MUTANT MOUSE

Transgenic mice homozygous for a targeted deletion in the entactin (nido- gen)-like G l domain of a-tectorin (TectaDENT/DENT mice) have tectorial

membranes that lack all known noncollagenous components—a-tectorin, h- tectorin, and otogelin, are composed solely of randomly organized collagen fibrils, and are completely detached from the surface of the organ of Corti and the spiral limbus (18). The tectorial membranes of mice heterozygous for this deletion are apparently normal, indicating the deletion does not act as a dominant negative. These observations indicate that a-tectorin is probably a major component and organizer of the noncollagenous striated sheet matrix of the tectorial membrane. The organ of Corti in the Tecta mice is normal, apart from the loss of a tectorial membrane (Fig. 5), and the structure and orientation of the hair bundles are una ected.

The basilar membrane in the high-frequency basal coil of the cochlea in these Tecta mutant mice with detached tectorial membranes is still tuned, but 35 dB less sensitive than the basilar membrane of a wild-type mouse. Cochlear microphonics and compound action potentials can still be recorded at high sound levels, indicating the hair bundles can be driven by fluid coupling in the absence of a tectorial membrane. However, the cochlear microphonic potentials di er in phase and symmetry relative to those recorded from wild-type mice, so the mechanical feedback from the

Figure 5 Cochlear sections from wild-type (a) and TectaDENT/DENT (b) mice il-

lustrating how the deletion in Tecta causes detachment of the tectorial membrane (TM) from the organ of Corti (C) and the spiral limbus (SL). In the TectaDENT/DENT

mice the tectorial membrane is associated with Reissner’s membrane (R). Bar = 100 Am.

electromotile hair cells to the basilar membrane will be largely ine ective, as it will be delivered with reduced gain and at an inappropriate time. At low

frequencies, the CAP threshold in these TectaDENT/DENT mice is reduced by as much as 80 dB. As TectaDENT/DENT mice are e ectively functional a-

tectorin null mutant mice, they may provide a model for DFNB21. A recent in vitro study has shown that mutations in ZP2 equivalent to the Y1870C and C1837G mutations in a-tectorin prevent or severely reduce incorporation of the mutant proteins into the zona pellucida of mouse eggs (42). However, a real understanding of the e ect the various TECTA mutations have on the structure and function of the tectorial membrane will have to await the production of knock-in mice bearing the same mutations in Tecta.

ACKNOWLEDGMENTS

Work in the authors’ laboratories is supported by the Wellcome Trust (Grant ref: 057410/Z/99/Z), Defeating Deafness, The Flemish Fonds voor Wetenschappelijk onderzoek (FWO), and the University of Antwerp.

REFERENCES

1.Goodyear RJ, Richardson GP. Extracellular matrices associated with the apical surfaces of sensory epithelia in the inner ear: molecular and structural diversity. J Neurobiol 2002; 53:212–227.

2.Thalmann I, Thallinger G, Crouch EC, Comegys TH, Barret N, Thalmann R.