(Received for publication, December 4, 1996, and in revised form, January 13, 1997)
From the School of Biological Sciences, University of
Sussex, Falmer, Brighton, BN1 9QG, United Kingdom and the
§ Department of Molecular Biology and Biochemistry,
University of Leeds, Leeds LS2 9JT, United Kingdom
The cDNA and derived amino acid sequences for
the two major non-collagenous proteins of the mouse tectorial membrane,
- and
-tectorin, are presented. The cDNA for
-tectorin
predicts a protein of 239,034 Da with 33 potential
N-glycosylation sites, and that of
-tectorin a smaller
protein of 36,074 Da with 4 consensus N-glycosylation
sites. Southern and Northern blot analysis indicate
- and
-tectorin are single copy genes only expressed in the inner ear, and
in situ hybridization shows they are expressed by cells
both in and surrounding the mechanosensory epithelia. Both sequences
terminate with a hydrophobic COOH terminus preceded by a potential
endoproteinase cleavage site suggesting the tectorins are synthesized
as glycosylphosphatidylinositol-linked, membrane bound precursors,
targeted to the apical surface of the inner ear epithelia by the lipid
and proteolytically released into the extracellular compartment. The
mouse
-tectorin sequence contains a single zona pellucida domain,
whereas
-tectorin is composed of three distinct modules: an
NH2-terminal region similar to part of the entactin G1
domain, a large central segment with three full and two partial von
Willebrand factor type D repeats, and a carboxyl-terminal region which,
like
-tectorin, contains a single zona pellucida domain. The
central, high molecular mass region of
-tectorin containing the von
Willebrand factor type D repeats has homology with zonadhesin, a sperm
membrane protein that binds to the zona pellucida. These results
indicate the two major non-collagenous proteins of the tectorial
membrane are similar to components of the sperm-egg adhesion system,
and, as such may interact in the same manner.
The sensory epithelia of the inner ear contain hair and supporting cells, and the apical surfaces of these epithelia are covered by extracellular matrices which transmit forces to the mechanosensitive stereociliary bundles of the hair cells. In the ampullae of the semi-circular canals, a cupula sits on top of each crista; in the saccule and utricule, the maculae are covered by an otolithic membrane; in the cochlea, a tectorial membrane lies over the surface of the organ of Corti. The structures of these three types of matrices, the cupula, the otolithic membrane, and the tectorial membrane, are complex and not identical, and each may be tailored to provide the optimal stimulus for the type of hair cell present in each organ.
The tectorial membrane is the best characterized of these matrices. In
the chick it is a dense fibrillogranular matrix (1) composed of two
major glycoproteins, - and
-tectorin (2, 3). Chick
-tectorin
is a large (Mr = 196,000) disulfide cross-linked complex composed of six polypeptides, the
1- to
6-tectorins, whereas chick
-tectorin is considerably smaller
(Mr = 43,000) and not covalently associated with
-tectorin. The cDNA for chick
-tectorin has been recently
cloned (3) and analysis of the sequence indicates
-tectorin is
related to the pancreatic zymogen granule protein GP2, the urinary
glycoprotein uromodulin (Tamm-Horsfall protein), and two components of
the extracellular matrix that surrounds the unfertilized egg, the zona
pellucida proteins ZP2 and ZP3.
The mammalian tectorial membrane is considerably more complex than that
of the bird, containing polypeptides that react with antibodies to
types II, V, and IX collagen (4, 5) and a number of
collagenase-insensitive glycoproteins that account for up to 50% of
the tectorial membrane protein (4). Under reducing conditions these
non-collagenous proteins of the mouse tectorial membrane form three
broad, diffuse bands on SDS gels with peak masses of 173, 60, and 45 kDa which have been referred to as the high, medium, and low molecular
mass tectorins (HMM,1 MMM, and LMM
tectorin, Ref 6). The polydisperse behavior of the mouse tectorins
observed on SDS gels may result partially from glycosylation
heterogeneity. For example, mouse HMM tectorin is sulfated and
following treatment with endo--galactosidase it is converted from a
diffuse smear to a sharp band with an Mr of
160,000 (4, 6). Under nonreducing conditions most of the
non-collagenous mouse tectorial membrane protein forms a large (Mr > 240,000), disulfide cross-linked complex
which may be homologous to chick
-tectorin. A proportion of the
protein in the LMM tectorin fraction is not covalently associated with
the other tectorins (4) and may represent the murine homologue of chick
-tectorin.
The studies of Kronester-Frei (7) originally described the presence of two major fibril systems in the mammalian tectorial membrane; the Type A protofibrils, straight unbranched filaments organized in bundles that run predominantly radially across the tectorial membrane, and the Type B protofibrils, described as branched, coiled, irregular diameter fibrils that could apparently exist in two states of hydration, with the highly hydrated form comprising the matrix within which the Type A protofibrils were found, and the weakly hydrated form forming the covernet fibrils, the marginal net, Hensen's stripe, and the limbal undersurface of the tectorial membrane. The Type A protofibrils are entirely degraded by bacterial collagenase (8), and can be labeled by antibodies directed against Types II and IX collagen (9). The matrix within which the Type A protofibrils are embedded is resistant to degradation by bacterial collagenase and fixation in the presence of tannic acid reveals it is composed of two types of fine, 7-9-nm diameter filament; a light and a dark staining type that are linked to one another by staggered cross-bridges and arranged in sheets with the filaments lying within the plane of each sheet (8). The alternating arrangement of the two fibril types in these sheets gives the matrix a striated appearance, and these sheets roll up to form the thicker, hydrated Type B protofibrils described by Kronester-Frei (7).
To further our understanding of the way in which the tectorial membrane
matrix is formed and functions in the process of mechanotransduction, we have now deduced the primary structure of the mouse tectorins. The
data indicate that the HMM, MMM, and LMM tectorins observed on SDS gels
are mainly derived from a single large mouse -tectorin sequence,
with a smaller sequence, that of mouse
-tectorin, contributing partially to the LMM tectorin fraction. The ways in which these two
proteins may interact via their different domains to form the observed
structure of the non-collagenous matrix of the tectorial membrane are
discussed.
Mouse and chick
tectorial membranes were collected by dissection in phosphate-buffered
saline containing a mixture of protease inhibitors (1 mM
phenylmethylsulfonyl fluoride, 2 mM benzamidine, 10 µM pepstatin, and 1 µg/ml leupeptin), washed with the
same buffer containing 0.1% Triton X-100 and stored frozen at
70 °C until sufficient numbers had been collected. To prepare the
mouse tectorins for sequencing, approximately 550 tectorial membranes were washed with 50 mM sodium acetate, pH 5.8, containing
the protease inhibitor mixture, and digested overnight at 37 °C with 5 milliunits of endo-
-galactosidase (ICN Biomedicals, Thame, United
Kingdom) in a 60-µl volume of the same buffer. Membranes were then
washed 3 times with 100 mM Tris-HCl, pH 7.4, 0.05% Triton X-100 containing protease inhibitors and incubated with 0.2 mg/ml bacterial collagenase (Sigma, Poole, United Kingdom) in a 100-µl volume of 50 mM Tris-HCl, pH 7.2, 5 mM
CaCl2 for 4 h at 37 °C. Following collagenase
digestion, membranes were washed 5 times with 100 mM
Tris-HCl, pH 7.4, containing 0.05% Triton X-100 and protease
inhibitors, and then solubilized by boiling in a 200-µl volume of
0.4% SDS, 2%
-mercaptoethanol, 10 mM sodium phosphate, pH 7.2, for 2 min. The sample was cooled to room temperature, and 80 µl of 5 times concentrated digestion buffer (2.5% CHAPS, 50 mM EDTA, 100 mM sodium phosphate, pH 7.2, 50 mM NaN3), 120 µl of H2O and 0.4 unit of N-glycosidase F (Boehringer Mannheim, Lewes, United
Kingdom) added. After digestion with N-glycosidase F for
3 h at 37 °C, the sample was dialyzed against cold
H2O overnight, lyophilized, and resuspended in 40 µl of
reducing SDS-PAGE sample buffer. For the chick tectorins, approximately
800 tectorial membranes were washed once with high salt buffer (1 M NaCl, 10 mM Tris-HCl, pH 7.4, 0.1% Triton
X-100), twice with low salt buffer (10 mM Tris-HCl, pH 7.4, 0.1% Triton X-100), and finally with H2O. All solutions
contained the protease inhibitor mixture. Membranes were then boiled in
nonreducing SDS-PAGE sample buffer and run on 7.5% preparative
mini-gels (10). The chick
- and
-tectorin bands were cut out from
the gels and the protein recovered by electroelution. Electroeluted
samples were dialyzed against cold H2O and lyophilized.
Lyophilized
-tectorin samples were redissolved in reducing SDS-PAGE
sample buffer. Total mouse tectorin and chick
-tectorin samples were
run on 8.25% mini-gels (using material from 100 to 200 tectorial
membranes per lane) and transferred to polyvinylidene difluoride
membranes using semidry electroblotting. The membranes were washed with
H2O, stained briefly with 0.005% sulforhodamine B in 30%
methanol, 0.2% acetic acid, rinsed in H2O, and dried in a
dessicator. The stained tectorin bands were identified, cut out from
the blots, and sequenced using standard procedures on a 477A
liquid-pulse instrument (PE Applied Biosystems) and a Beckman LF 3000 automated NH2-terminal sequencer.
Differential analysis of reverse
transcriptase (RT)-mediated PCR products obtained from a region of the
cochlea that is known to produce the tectorial membrane and from one
that is not involved in tectorial membrane production was used to
identify products potentially derived from tectorin cDNAs. The
combined greater and lesser epithelial ridges (GLER) which are known to
produce the tectorial membrane, and the stria vascularii which are not involved in tectorial membrane synthesis were dissected from the cochleae of 100 2-3-day-old postnatal Swiss CD1 mice (Charles River,
United Kingdom) in Hepes-buffered Hanks' balanced salt solution, pH
7.0, collected separately, and snap frozen in liquid nitrogen.
Poly(A)+ RNA was isolated by a one-step oligo(dT)-cellulose
selection method (Quick Prep Micro kit, Pharmacia, St. Albans, United
Kingdom) and double stranded cDNA was synthesized from 2 µg of
poly(A)+ RNA in a final volume of 20 µl using a
commercial kit (Boehringer Mannheim). Degenerate forward and reverse
oligonucleotide primers based on the NH2-terminal amino
acid sequences of the chick 4- and
5-tectorins (
4S1,
ATGGCNTCN(CT)TNTA(TC)CCNTT;
4R1, AANGG(GA)TANA(GA)NGANGCCAT;
5S1, GTNACNGCNAA(AG)AA(TC)GA(AG)GA;
5R1,
TC(TC)TC(AG)TT(TC)TTNGCNGTNAC) were used to isolate candidate
-tectorin RT-PCR products. A degenerate sense primer based on
NH2-terminal sequence from one of the mouse tectorin bands
(mtm4S1, GA(AG)CA(TC)ACNCCNAA(TC)AA(AG)GC), and a degenerate
antisense primer (
TDEGR, GGNGTNGCCCA(AG)CA(AG)TT) based on an
amino acid sequence (CWATPS) conserved in chick
-tectorin and mouse
-tectorin were used to isolate a
-tectorin specific RT-PCR
product. Separate PCR reactions using 1-µl aliquots of the two
cDNA populations and 50 pmol of each primer in a 50-µl reaction
volume (as described in Ref. 3) were hot started and followed by 25 cycles of 50 °C for 15 s, 72 °C for 1 min, and 94 °C for
15 s. The reactions ended with a 10-min incubation at 72 °C.
PCR products were analyzed by agarose gel electrophoresis in 1 × Tris borate-EDTA buffer and products specific to the GLER were isolated
using GeneClean (Stratech Scientific, Luton, United Kingdom).
GLER samples were collected from the cochleae of
approximately 500 2-3-day-old postnatal Swiss CD1 mice as described
above. Total RNA was isolated using guanidinium thiocyanate extraction (11) followed by cesium trifluoroacetate gradient centrifugation (12)
and poly(A)+ mRNA selected by oligo(dT)-cellulose
chromatography (Quick Prep Micro kit, Pharmacia). A directionally
cloned, oligo(dT)-primed cDNA library was constructed using
commercial kits (Stratagene, Cambridge, United Kingdom). First strand
cDNA synthesis was primed with an oligo(dT) primer containing a 5
XhoI site, the cDNA was then double stranded,
EcoRI adaptors were added and the cDNA digested with
XhoI. The cDNA was directionally ligated into
UniZap
XR arms and packaged in vitro. A randomly primed cDNA
library was constructed in a similar manner, with the following
modifications; poly(A)+ RNA was isolated directly from the
GLER of 300 2-3-day-old postnatal mouse cochleae with a Quick Prep
Micro kit, first strand cDNA synthesis was primed with a mixture of
random hexamers and an antisense primer designed to the 5
end of clone
A2 (GAAATGGAGGCGTAGTGCTG) and, after addition of EcoRI
adaptors, the double stranded cDNA was ligated into the
EcoRI site of
Zap II. Libraries were screened at high
density with 32P-labeled DNA probes. Positive plaques were
rescreened to purity and converted to plasmids by phage rescue.
Genomic DNA was isolated from neonatal mouse brain by a modification of the method of Blin and Stafford (13, 14). Aliquots of DNA (10 µg) were digested to completion with EcoRI, PstI, SacI, or KpnI in the supplied buffers (Promega Ltd., Southampton, United Kingdom), electrophoresed on a 0.7% agarose gel in 1 × Tris borate-EDTA buffer, transferred to nylon membrane (Hybond N, Amersham International, Little Chalfont, United Kingdom) (15), and covalently bound to the dried membrane by UV cross-linking. Hybridization to 32P-labeled DNA probes and washing to high stringency were carried out according to the manufacturer's instructions.
Poly(A)+ RNA was isolated from cochlear, brain, heart, liver, kidney, intestine, lung, and skin tissues of 2-3-day-old postnatal mice as described for RT-PCR. Aliquots (3 µg) were electrophoresed on a 1% agarose-formaldehyde gel (14) and transferred to nylon membrane as described above. Hybridization to 32P-labeled DNA probes and washing to high stringency were carried out as described by Angst et al. (16).
Probe Preparation and LabelingProbes for Southern and
Northern blotting were generated by PCR or restriction enzyme digestion
of cDNA clones. Probe fragments were separated by gel
electrophoresis, recovered with GeneClean or Gelex resin (Stratech
Scientific), and random primer labeled (17) with
[-32P]dCTP using a MegaPrime labeling kit (Amersham).
Unincorporated label was removed on Sephadex G-50 columns (Pharmacia)
and the probes were denatured with alkali before use.
Double stranded plasmid DNA minipreps (Promega) of cDNA clones were sequenced by the method of Sanger et al. (18) using T7 DNA polymerase (Pharmacia) and a combination of sequencing templates from nested deletions (19) and templates primed with synthetic oligonucleotides. Analysis of DNA and derived amino acid sequences was performed using the DNA Star package (DNA Star, London, United Kingdom). DNA and protein sequence data bases (GenBank, EMBL, DDBJ, and PDB) were searched using the BLAST network service at the National Center for Biotechnology Information (20).
Preparation of Antipeptide SeraThree synthetic peptides
for -tectorin (m
25-38, RELMYPFWQNDTRT;
m
742-755, TCPERPEYLEIDIN; m
2000-2014,
KDNTIGIEENGVSLT) and one for
-tectorin (m
226-240,
PTDETVLVHENGKDH) based on the derived amino acid sequences of
- and
-tectorin were synthesized and purified by high performance liquid
chromatography. Peptides were coupled to bovine serum albumin using
glutaraldehyde (21) and dialyzed against phosphate-buffered saline.
Conjugates (500 µg) were suspended in Freund's complete adjuvant and
injected subcutaneously into rabbits. The immunization was repeated a
further three times with Freund's incomplete adjuvant at 4-week
intervals. Antipeptide antibodies were affinity isolated using the
respective peptide coupled to either cyanogen bromide-activated Sepharose 4B or Affi-Gel 15.
Mouse tectorial membranes were solubilized in reducing SDS-PAGE sample buffer, separated on 7.5% polyacrylamide gels, and transferred to nitrocellulose membranes using semi-dry blotting. Nitrocellulose blots were preblocked for 1 h with 3% dried milk powder, 10% horse serum in Tris-HCl buffered salt solution containing 0.05% Tween 20 and incubated overnight in the same solution containing the affinity isolated antipeptide antibodies at concentrations of 5 or 50 µg/ml. After washing, bound antibodies were labeled with biotinylated anti-rabbit Ig (1:1000 dilution) (Dako, High Wycombe, United Kingdom), followed by alkaline phosphatase-conjugated streptavidin (1:1000) (Vector Laboratories, Peterborough, United Kingdom). Alkaline phosphatase activity was visualized with 0.05 mg/ml bromochloroindolyl phosphate, 0.1 mg/ml nitro blue tetrazolium in alkaline phosphatase buffer (100 mM NaCl, 5 mM MgCl2, 100 mM Tris-HCl, pH 9.5).
In Situ HybridizationInner ears from 2-day-old postnatal
mice were immersion fixed in ice-cold 4% paraformaldehyde in
phosphate-buffered saline for 1 h prior to overnight fixation in
4% paraformaldehyde in phosphate-buffered saline containing 30%
sucrose. Tissues were embedded in agarose and cryosectioned at a
thickness of 20 µm. Aliquots (10 µg) of plasmid minipreps (Promega)
of the - and
-tectorin cDNA clones A1 and B3 were linearized
with either EcoRI or XhoI (Promega, United
Kingdom)). Antisense RNA was transcribed from EcoRI cut
clones using T7 RNA polymerase and sense RNA was transcribed from
XhoI cut clones using T3 RNA polymerase. In situ hybridization, washing and autoradiography were carried out as described by Goodyear et al. (22).
NH2-terminal amino acid sequence data for
the six chick -tectorins and four mouse tectorin bands are presented
in Tables I and II. The sequences for the
chick
2- and
3-tectorin bands are identical suggesting chick
3-tectorin is either a truncated or differentially glycosylated form
of chick
2-tectorin. The sequences show no similarity to derived
amino acid sequences or proteins in the current data bases (GenBank,
EMBL, DDBJ, and PDB) except for the sequence of the mtm4 band which has
86% identity with the predicted NH2 terminus of mature
chick
-tectorin (3). The mouse tectorin bands mtm2 and mtm3, which
have Mr of 40,000 and 34,000, respectively, have
almost identical NH2-terminal sequences, suggesting mtm2 is
an incompletely deglycosylated form of mtm3 and that both tectorin
bands are derived from the same protein. Furthermore, identity between
the sequence YPFW in mtm3 and chick
4-tectorin suggests that the
mtm3 sequence is derived from the mouse homologue of chick
4-tectorin. The first nine amino acids of the mtm1 band and chick
1-tectorin also share 67% identity suggesting mtm1 is the mouse
homologue of chick
1-tectorin.
|
|
Assuming that the different mouse -tectorins
were derived from a large precursor protein, RT-PCR was performed on
poly(A)+ RNA from the GLER and stria vascularis with
combinations of forward and reverse primers based on the
NH2-terminal amino acid data. Using one primer pair based
on the NH2-terminal amino acid sequences of the chick
4
and
5-tectorins (
5S1 and
4R1), a GLER-specific product of 300 bp was amplified which, while smaller than expected, was not amplified
from the stria vascularis cDNA and hybridized to a
cochlear-specific mRNA of about 7.5 kb on a Northern blot (data not
shown). This GLER-specific PCR product was used to screen 4.5 × 105 plaque-forming units of the oligo(dT)-primed mouse
cochlear cDNA library. Thirty-two positive clones were identified,
of which two, A1 and A2, were selected for further study. Clone A1 is
encoded by a single gene as judged by Southern blotting and, like the GLER specific PCR product, hybridizes to a cochlear-specific mRNA of approximately 7.5 kb on a Northern blot (Fig. 1,
a and c). Sequence analysis of A1 and A2
suggested that they were derived from the same mRNA, but both
lacked the 5
end. Clones extending further 5
were obtained by
screening 1.0 × 105 plaque-forming units of the
randomly primed mouse cochlear cDNA library with a PCR product from
the 5
end of clone A2. Twenty-four further positives were identified
of which one, A3, was found to encode the 5
end of the mRNA. Clone
A1 was completely sequenced on both strands; A2 was sequenced on both
strands where it extended 5
of A1, and clone A3 where it extended 5
of A2. Other regions were sequenced on one strand only (Fig.
2a). Sequences were assembled into a
composite cDNA sequence of 7340 bp containing a large open reading
frame (Fig. 2, a and b). A perfect match between
the derived amino acid sequence and the NH2-terminal amino
acid sequence of mtm3 confirms the cDNA encodes
-tectorin.
Complete identity was also found between the NH2-terminal
sequence of chick
1-tectorin and the derived mouse
-tectorin
sequence, but the NH2-terminal sequence of mtm1 only
matches the derived sequence in 6 out of 9 positions.
RT-PCR on poly(A)+ RNA from the GLER using degenerate
primers for -tectorin generated a 570-bp product (data not shown),
which was of the size predicted from the chick
-tectorin sequence
data. The product was used to screen 1.0 × 105
plaque-forming units of the randomly primed mouse cochlear cDNA library and 9 positive clones were identified and rescreened to purity.
Two of these, B1 and B2, were selected for further study. Clone B2 is
encoded by a single gene as judged by Southern blotting and hybridizes
to a cochlear-specific mRNA of approximately 2.7 kb (Fig. 1,
b and d). Sequence analysis of B1 and B2
suggested these were partial cDNA clones. Screening 1.0 × 105 plaque-forming units of the oligo(dT)-primed mouse
cochlear cDNA library with clone B2 identified a further 20 clones.
One of these, clone B3, was found to be full-length. Clone B2 was
completely sequenced on both strands, and B1 and B3 were sequenced on
both strands where they did not overlap with B2 (Fig. 2c).
Overlapping regions were sequenced on one strand only. The DNA
sequences were assembled into a composite cDNA sequence of 2745 bp
with a single open reading frame (Fig. 2d). The derived
amino acid sequence matches the NH2-terminal sequence of
mtm4 in 6 out of 7 positions and has 75.3% overall sequence identity
with chick
-tectorin, confirming the cDNA sequence encodes the
mouse homologue of chick
-tectorin.
The -tectorin
cDNA sequence (Fig. 2b) has a 5
-untranslated region of
267 bp, a single open reading frame of 6453 bp, and a 3
-untranslated
region of 620 bp containing two consensus polyadenylation signals, the
second of which is followed by a poly(A) tail. The open reading frame
is initiated at base 268 by the second ATG codon in the cDNA
sequence. This codon matches the consensus initiation codon (23) by the
presence of a purine at
3 and a cytosine at
4, whereas the first
ATG codon, at position 121, matches only at the
3 position and is
also followed by three in-frame stop codons at positions 154, 184, and
227. The
-tectorin cDNA sequence (Fig. 2d) has a
5
-untranslated region of 130 bases followed by an open reading frame
of 963 bp, which is initiated at base 131 by the first ATG codon in the
sequence. This codon matches the consensus initiation codon (23) by the
presence of a purine at
3 and a cytosine at
2. The 1652-bp
3
-untranslated region of
-tectorin is unusually long and contains
two potential polyadenylation signals. The second of these is used and
is followed 24 bases downstream by a poly(A) tail.
The open reading frame of -tectorin encodes a polypeptide of 2150 amino acids which has a calculated molecular mass of 239,034 Da and
contains 33 potential N-glycosylation sites (Fig.
2b), whereas that of
-tectorin encodes a much smaller
polypeptide of only 320 amino acids with a calculated molecular mass of
36,074 Da and 4 potential N-glycosylation sites (Fig.
2d). A number of features are common to these two derived
amino acid sequences. Both start with hydrophobic signal sequences, as
expected for secreted molecules (24). Identity between the
NH2-terminal amino acid sequence of mtm2/3 and residues
25-31 of the derived amino acid sequence of
-tectorin implies the
signal peptide is cleaved between amino acids 24 and 25. The presence
of an alanine at position 22 conforms to the
3 position of the signal
peptide cleavage site (25), however, the proline at position 24 does
not fit the consensus
1 position. Similarly, for
-tectorin,
identity between the NH2-terminal amino acid sequence for
mtm4 and residues 18-24 of the derived amino acid sequence implies
that the signal peptide is cleaved between amino acids 17 and 18 to
generate the mature NH2 terminus of
-tectorin. Alanine
residues at positions 15 and 17 in the
-tectorin sequence conform
perfectly to the
3,
1 rule of von Heijne (25) for signal peptide
cleavage. The derived amino acid sequences of both
- and
-tectorin terminate with sequences of predominantly hydrophobic
residues (Fig. 2, b and d) characteristic of
proteins that are membrane bound via a glycosylphosphatidylinositol
tail (26-28). The method of Kodukula et al. (29) predicts
that the asparagine at position 2086 in
-tectorin is the most likely
acceptor for the glycosylphosphatidylinositol anchor, but does not
provide a clear candidate residue in the case of
-tectorin. These
hydrophobic COOH termini are both preceded a short distance upstream by
tetrabasic motifs (Fig. 2, b and d) that are
characteristic of endoproteinase cleavage sites (30, 31) and it seems
likely that, as previously suggested for chick
-tectorin (3), mouse
- and
-tectorin are both synthesized as lipid-linked,
membrane-bound precursors, targeted to the apical surface of the
cochlear epithelium by the lipid (32) and then released into the
extracellular compartment by the action of an endoproteinase.
Comparing the
cDNA sequences of the three -tectorin clones identified a 15-bp
sequence present in clone A1, but absent in A2, which encodes five
amino acids, PLAPS (Fig. 3a). PCR of mouse genomic DNA with primers immediately adjacent to this sequence amplified a product of 1.5 kbp (Fig. 3b), suggesting this
sequence may be an alternatively spliced exon flanked by approximately 1.4 kbp of intron sequences. RT-PCR of mouse GLER cDNA amplified products of 77 and 62 bp corresponding to both possible splice variants
and suggests the variant lacking the 15-bp exon is the more common form
(Fig. 3c). Splicing this exon into the mRNA changes Ser1659 to an Arg residue, creating a new dibasic sequence,
Lys-Arg1658-1659. Additional faint bands of 101 and 104 bp
are also observed (Fig. 3c) suggesting there are additional
splice variants in this region.
Sequence Similarities
Similarity searches of the current data
bases suggest -tectorin is composed of three distinct modules, each
with homology to different proteins. The first 219 amino acids of the
NH2-terminal region of
-tectorin show 24.9% similarity
to a part of the first globular domain (G1) of entactin/nidogen (Fig.
4a) (33, 34). A 39-amino acid stretch
separates this NH2-terminal domain from the central domain
which is 1528 amino acids long and is composed of three full repeats
and two partial repeats homologous to the D domains of prepro-von
Willebrand factor (vWF), zonadhesin, and the intestinal mucin muc2
(Fig. 4b) (35-37). The D domains are rich in cysteine, and
the positions of the cysteines within the individual domains of
-tectorin match well, with 28 out of 37 positions being fully
conserved (Fig. 4b). At the carboxyl end of the
-tectorin
sequence there is a stretch of 255 amino acids which exhibits
similarity to the zona pellucida domains (38) of uromodulin (39) and
GP2 (40) (Fig. 4c, Table III). This region of
-tectorin is also similar to other members of the zona pellucida
domain family, and, most significantly, to chick and mouse
-tectorin
(Fig. 4c, Table III). Mouse
-tectorin contains a single
zona pellucida domain (Fig. 4c) and all the major features of chick
-tectorin are conserved in the mouse sequence, including the four consensus N-glycosylation sites, the 12 cysteine
residues, and the extended basic sequence preceding the hydrophobic
tail.
|
Antibodies were
raised to peptide sequences in the three different modules of the
-tectorin sequence and used to determine whether the HMM, MMM, and
LMM tectorins observed on reducing SDS gels are derived from these
different regions (Fig. 5). Antibodies to a peptide
sequence (m
742-755) in the second full vWF type D
repeat of
-tectorin detect a broad smear extending from the top of
the separating gel down to the 170-kDa region of the gel corresponding
to HMM tectorin (Fig. 5, lane 1). Antibodies to a sequence
(m
2000-2014) in the zona pellucida domain react with a
band of 60 kDa corresponding to MMM tectorin, and additional
polydisperse material in the 210-135-kDa region of the gel (Fig. 5,
lane 2). Antibodies to a peptide sequence (m
25-38) at the predicted NH2 terminus
of the module with similarity to the G1 domain of entactin, recognize a
broad diffuse band in the 42-55-kDa region of the gel (LMM tectorin)
and a faint band of 205 kDa (Fig. 5, lane 3). An antibody
raised to a peptide in the derived amino acid sequence of
-tectorin
recognizes a single sharp band of 45 kDa (Fig. 5, lane
4).
Expression Patterns of
In situ hybridization was used to study the
distribution of - and
-tectorin mRNAs in the inner ear of the
2-day postnatal mouse. This stage of development was chosen as it is
known to be a period when the tectorial membrane is being produced. In the cochlea,
-tectorin mRNA is expressed on both sides of the organ of Corti. It is expressed in the pseudostratified cells of the
greater epithelial ridge on the modiolar side of the inner hair cells,
and in the immature Hensen's cells that lie alongside the outermost
row of outer hair cells (Fig. 6a). Expression
of
-tectorin is also found in the immature Hensen's cells and in the greater epithelial ridge. However, in comparison to
-tectorin,
-tectorin is expressed within a more restricted zone of the greater epithelial ridge that lies adjacent to the inner hair cells of the
organ of Corti (Fig. 6b). In addition,
-tectorin is
expressed by the pillar cells that lie between the inner and outer hair cells within the organ of Corti, and the resultant pattern observed with
-tectorin antisense probes is one of three stripes arrayed across the organ (Fig. 6b). In the saccule and utricule,
-tectorin is expressed at low levels within the sensory macula, but
strongly expressed in the transitional zone around the periphery of the macula and in a region that is producing the accessory membrane, a
structure that connects the otolithic membrane to the roof of the organ
(Fig. 6c). In contrast,
-tectorin is expressed in a restricted region of the macula called the striola (Fig.
6d), a region where there is known to be a high density of
vestibular Type I hair cells (41).
The results describe the isolation and characterization of
cDNA clones for mouse - and
-tectorin. The cDNA for
-tectorin encodes a large protein of 239 kDa and that of
-tectorin a smaller protein of 36 kDa, and together these two
proteins can account for the HMM, MMM, and LMM tectorins observed with
SDS-PAGE. The genes encoding these proteins are expressed at high
levels in the developing inner ear, and not in a number of other
tissues of the early postnatal mouse, indicating the tectorins may be matrix molecules unique to the inner ear. Sequence data show both proteins are probably synthesized as
glycosylphosphatidylinositol-linked membrane bound precursors which may
be targeted to the apical surface of the inner ear epithelia by the
lipid and released by endoproteolytic activity. The two proteins share
a common module, the zona pellucida domain, which may enable them to
form filaments, and
-tectorin contains additional modules which
could allow it to interact with
-tectorin. Mouse and chick
-tectorin share 75% overall sequence identity, both the
NH2-terminal sequence data for the chick
-tectorins
presented in this study and preliminary data from chick
-tectorin
cDNA clones2 indicate the chick and
mouse
-tectorins are conserved at a similar level, and the derived
amino acid sequence of a human brain expressed sequence tag (accession
number R84585[GenBank]) is 92% identical to mouse
-tectorin. The tectorins
are therefore highly conserved and may be fundamentally important in
the process of mechanotransduction.
Matches between the NH2-terminal sequence data for the
mouse and chick tectorins and the derived amino acid sequence of mouse -tectorin, together with data from Western blots with antipeptide antibodies, indicate that the HMM, MMM, and LMM tectorins are generated
by proteolytic cleavage of
-tectorin, with
-tectorin also
contributing to the LMM tectorin band. The domain structure of
-tectorin and way in which the sequence may be cleaved are shown in
Fig. 7a. The NH2-terminal amino
acid sequences of mtm1 and chick
1-tectorin match residues
328-336/7 of the derived
-tectorin sequence in 6/9 and 10/10
positions, respectively, that of chick
2/3 tectorin partially
matches residues 1659-1668 in 7/10 positions, and that of mtm4 is
identical to the predicted mature NH2 terminus of the
-tectorin sequence. Polypeptides with these NH2 termini
would have predicted masses of 147,360, 47,020, and 34,092 Da,
respectively (Fig. 7a), values which are close to those
observed for mouse tectorins following enzymatic deglycosylation with a
combination of endo-
-galactosidase and N-glycosidase F. Western blotting with antibodies to synthetic peptides based on different regions of the predicted
-tectorin sequence provides further evidence for this scheme. Antibodies to a sequence in the
second full vWF type D repeat react with HMM tectorin, antibodies to a
sequence in the ZP domain stain MMM tectorin, and antibodies to the
predicted mature NH2 terminus of
-tectorin stain LMM
tectorin. Good partial matches for the NH2-terminal amino
acid sequence data of the chick
5- and
6-tectorins (75 and 67%,
respectively) are also found within the region of the predicted mouse
sequence that gives rise to mouse HMM tectorin according to the scheme described above. Cleavage at these sites in chicken
-tectorin may
explain why the polypeptide core of the largest chick
-polypeptide, chick
1-tectorin, is some 60 kDa smaller than that of the largest mouse
-tectorin.3
Although the predicted sequence of mouse -tectorin can account for
the three major tectorins observed on SDS gels under reducing conditions, it is not clear whether the three
-tectorins are genuine
subunits generated by active processing or simply the result of
proteolysis occurring between intrachain disulfide bonds. Furthermore,
the way in which this sequence appears to be processed does not clearly
divide the protein into the three different modules defined by sequence
analysis as the predicted NH2 termini of HMM and MMM
tectorin both occur within the vWF type D repeats. The presence of a
splice variant introducing a dibasic, potential endoproteinase cleavage
site close to the theoretical start of the MMM tectorin polypeptide
within the D4 repeat indicates this site may be of functional
significance, but it should be noted that there are also many dibasic
sites throughout the predicted sequence and none lie immediately
upstream of the theoretical NH2 terminus of HMM tectorin.
However, the splice variant without the dibasic site upstream of the
potential NH2 terminus of MMM tectorin is the major species
present and this may explain why the antibodies to the peptide sequence
in the zona pellucida domain react with both MMM tectorin and higher
molecular mass material.
While it is unclear whether -tectorin is processed or slowly subject
to site-specific proteolytic damage in vivo, the three distinct modules may play different roles in organizing the structure of the non-collagenous matrix of the mammalian tectorial membrane, in
conjunction with
-tectorin. The predicted sequence of mouse
-tectorin, like that of
-tectorin, contains a single zona
pellucida domain, a 260-amino acid module with 8 strictly conserved
cysteine residues (38). This is the only common feature shared by a
number of different proteins (Fig. 7b) all of which either
can or do form filament based matrices or gels (42-48), and it has
been suggested that it may be the element that enables them to form
filaments (3). The presence of this domain in both of the two major
components of the filament based non-collagenous matrix of the mouse
tectorial membrane suggests that these two proteins may either
self-associate via their respective zona pellucida domains to form
homomeric filaments, or interact with each other via their zona
pellucida domains to form heteromeric filaments. The presence of two
distinct filament types, a light and a dark staining filament, in the
collagenase-insensitive, striated sheet matrix of the tectorial
membrane (8) immediately suggests
-tectorin may form one filament
type and
-tectorin the other. If the two molecules self-associate to
form homomeric filaments via their zona pellucida domains, then it is
conceivable that the two filament types interact with one another to
form the striated sheet matrix via either the LMM or HMM
-tectorin modules. In this respect it is interesting to note that the central, HMM domain of
-tectorin containing the 2 partial and three full vWF
type D repeats shares its highest homology with zonadhesin, a recently
described sperm membrane protein that binds to the zona pellucida in a
species specific manner (49, 36). Zonadhesin is a transmembrane protein
with 3 types of extracellular domain, an NH2-terminal
domain with no homology to other proteins, a
threonine/serine/proline-rich mucin type domain, and a membrane
proximal domain with 1 partial and 4 full vWF type D repeats (36) (Fig.
7b). During sperm maturation the first two domains of
zonadhesin are lost and the membrane proximal domain with the vWF
repeats is cleaved into two, disulfide cross-linked polypeptides with
Mr of 105,000 and 45,000 which as a complex have
the ability to bind to the zona pellucida. Not only is the homology
between the vWF type D repeats of zonadhesin and
-tectorin high
(26%), but the order of these repeats is the same in both
-tectorin
and zonadhesin, whereas it is different in vWF and the mucins.
Homomeric filaments of
-tectorin and
-tectorin may therefore
interact with one another via the HMM, zonadhesin-like module of
-tectorin, either alone or in conjunction with LMM tectorin.
Although the LMM module shows similarity with nidogen, an organizer of
the extracellular matrix (50, 51), the region of identity has no known
function.
However, there are alternative possibilities for how the two proteins
could interact. For example, -tectorin molecules could form
homomeric filaments via the vWF type D domains of HMM tectorin, with
these filaments then interacting via their zona pellucida domains with
homomeric
-tectorin filaments. Individual pro-vWF subunits form
dimers between adjacent COOH termini, and the dimers then oligomerize
to form multimers via the NH2 termini of their D3 domains.
These vWF multimers appear as long, thin filaments in rotary shadowed
preparations (52, 53). Multimerization of vWF is dependent on vicinal
cysteines present in the D1 and D2 domains of the vWF pro-sequence that
are thought to have protein disulfide isomerase activity and be
responsible for interchain disulfide bonding (54). Vicinal cysteines
are also present in the D1 and D4 domains of
-tectorin
(Cys458 and Cys461 in D1, Cys1619
and Cys1622 in D4, see Fig. 4) and may have similar
catalytic activity and be involved in some aspect of tectorial membrane
matrix assembly, although this may not be necessarily the
oligomerization of
-tectorin to form filaments as there is no
evidence for the presence of covalently linked
-tectorin multimers
in the tectorial membrane. A third potential model for the organization
of
- and
-tectorin in the matrix would be one in which the two
proteins interact via their ZP domains to form heteromeric filaments
like the zona pellucida proteins, ZP2 and ZP3 (47). These in turn could
bind to one another via the HMM module of
-tectorin, again either with or without the participation of the LMM module.
In situ hybridization shows that there are groups of cells
in the inner ear that express either - or
-tectorin, or both molecules simultaneously. While this suggests
- and
-tectorin can
probably form homomeric filaments, it does not rule out the possibility
they form heteromeric filaments in those areas where they are
co-expressed. The patterns of
- and
-tectorin expression observed
in the inner ear are more complex than expected and do not obviously
correlate with the regional variations in matrix structure that have
been described in previous morphological studies (7, 55, 8). For
example, the immature Hensen's cells are thought to produce the
marginal band, a region of densely packed matrix, and the cells in the
greater epithelial ridge lying adjacent to the inner hair cells form
the midbody of the tectorial membrane where the matrix is loosely
packed (7), yet both of these cell groups co-express
- and
-tectorin. Likewise the limbal undersurface of the tectorial
membrane and Hensen's stripe are regions of densely packed matrix, and
yet the former is most likely to be produced by the cells of the
greater epithelial ridge lying next to the limbus that are expressing
only
-tectorin, and the latter could be produced by the pillar cells
that are expressing only
-tectorin. Regional differences in matrix
structure may be important for eliciting the correct response from
different types of hair cells. For example, as in the chick,
-tectorin expression in the otolithic maculae is restricted to the
striolar region, an area in the mammal where a high density of type I
hair cells is found (41). While differences in the ratio of
- and
-tectorin expressed in any one region may be one factor that
influences matrix structure, variations in their glycosylation
patterns, or the
-tectorin splice forms that are used could also be
important determinants. Modulation of the relative expression of
-
and
-tectorin with respect to time during development could also
play a role in shaping the structure of the tectorial membrane.
The results of this present study provide a complete molecular characterization of the two major non-collagenous glycoproteins of the mammalian tectorial membrane, an extracellular matrix essential for the perception of sound. These data should now allow the structure and properties of this matrix to be manipulated both selectively and non-invasively and reveal how the tectorial membrane influences frequency tuning in the cochlea.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) X99805[GenBank] (-tectorin) and X99806[GenBank] (
-tectorin).
We thank C. Malenczak for technical assistance, Chris Kowalczyk of the Sussex Center for Neuroscience for providing the mouse tectorin NH2-terminal sequence data, and both Ian Russell and Richard Goodyear for critical discussions and comments on the manuscript.