382
11 Extracellular Structural and Secretory Proteins
about 1000 amino acids. The triple helix contains
cysteine residues, which can form disulphide
bridges between the monomers, as well as several
putative cell-binding sites with the sequence ArgGly-Asp and N-glycosylation sites with the
sequence Asn-X-Thr. The N-terminal domains of
the chains a1(VI) and a2(VI) contain one
sequence, and the large a3(VI) chains have eight
sequences of 190-210 amino acids, which
resemble the type-A domains of the Willebrand
factor and, like the latter, also bind collagen I.
Thus, collagen VI apparently has the function of
a bridge between cells and collagen fibres [11, 20,
50]. The molecules of collagen VII consist of
three identical subunits and occur in connective
tissue as disulphide-linked dimers [83]. The shortchain collagen VIII is found in the majority of
endothelial cells and, in particular, in the thick
basement membrane of the cornea (Descemet's
membrane). The molecule is a heterotrimer with
the composition (a1)2a2. The a chain is made up
of a short triple helix of 454-457 amino acids and
two terminal, non-helical domains of 117 and 173
amino acids [85].
Collagens IX and X are deposited in various
cartilaginous structures formed from fibrils of collagens II and XI. Collagen X, which is closely
related to type VIII, is expressed only at mineralization sites in hypertrophic chondrocytes; it is a
homotrimer of short a chains which have triple
helices with eight interruptions; the sequences of
the chicken and bovine chains agree in 67 % of
their 674 amino acids [127]. Collagen IX is a
heterotrimer of three different a chains of
72-115 kDa and these include three triple-helix
domains, COL1-COL3, and four flexible, nonhelical domains, NC1-NC4 (the numbering
beginning at the C-terminus). a2(IX) bears a
covalently bound glycosaminoglycan on domain
NC3. Collagen IX alone is unable to form fibrils,
but it is covalently bound with collagen II of cartilage fibrils via lysine-derived cross-links. In this
way, the collagen fibrils of the cartilage are completely surrounded by glycosaminoglycan, and
this contributes to the high hydration state of the
cartilage matrix. Two forms of the a1(IX) chain
arise by alternative splicing. The cartilaginous
form bears a large, globular domain, NC4, of
266 amino acids at the N-terminus; this extends
beyond the fibril surface and can interact with
other components of the matrix. The form found
in the cornea and in the vitreous humour has a
much shorter N-terminus [84, 140]. In addition to
collagen IX, there are other fibril-associated collagens with interrupted triple helices (FACITs),
e.g. type XII in the tendons of chicken embryos,
type XIII in the endothelium of the human
umbilical vein, and type XIV in calf tendon and
skin [28, 43, 99].
The genes of the fibril-forming collagens I-III
have 5-8 exons in the region of the N-propeptide and 44 exons in the triple-helix region,
i.e. they have more interruptions than most other
genes. Because correct splicing apparently requires a minimal intron length of about 80 bp, these
genes with pro-collagen-coding sequences of less
than 5 kb often occupy more than 40 kb of chromosomal DNA. Most of the exons have a length
of 54 bp, and some have derived values like 45
(= 54-9), 108 (= 2 X 54) or 99 (= 108-9) bp.
One exon of 54 bp encodes 18 amino acids or six
G-X-Y triplets. Exons of 54 bp also predominate
in collagens V and XI. The characteristic gene
structure of the fibril-forming collagens has led to
the concept that the genes of all collagens arose
by the multiplication of a primordial sequence of
54 bp. However, further collagen genes that have
been sequenced in the meantime do not fit this
54-bp hypothesis: the pro-collagen genes of
type VI mostly have exons of 63 bp in addition to
some with 27,36,45,54 or 90 bp, i.e. multiples of
a 9-bp unit. The human and murine a1(IV) genes
contain 52 and 47 exons, respectively, with very
variable lengths between 27 and 192 bp. The
complete triple helix in the a1(VIII) and
a2(VIII) genes of mammals and the a1(X) gene
of the chicken is encoded by a single large exon
[16,50,68, 117, 133].
It is difficult to put together a detailed picture
of the evolution of vertebrate collagens. Collagens of types I and II have been identified in
agnathans, elasmobranchs and teleosts, and in
some cases have been partially characterized; little is known about other fish collagens. The agnathans, which are at the root of vertebrate evolution, possess a cartilaginous skeleton, the matrix
of which contains chondroitin-6-sulphate as the
glycan component, as in the higher vertebrates.
However, the main protein components in the
lamprey are not collagen and elastin as in other
vertebrates; in this case collagen is subsidiary to a
newly discovered structural protein, lamprin,
which makes up 44-51 % of the dry weight. This
protein differs in its amino acid spectrum from
both vertebrate collagen and elastin. It consists of
over 80 % of the non-polar amino acids proline,
glycine, alanine, valine and leucine, and contains
6.3 % tyrosine and 4.2 % histidine. Lamprin is
cross-linked into fibres but the linking mechanism
is not known; desmosin and isodesmosin are in
11 Extracellular Structural and Secretory Proteins
about 1000 amino acids. The triple helix contains
cysteine residues, which can form disulphide
bridges between the monomers, as well as several
putative cell-binding sites with the sequence ArgGly-Asp and N-glycosylation sites with the
sequence Asn-X-Thr. The N-terminal domains of
the chains a1(VI) and a2(VI) contain one
sequence, and the large a3(VI) chains have eight
sequences of 190-210 amino acids, which
resemble the type-A domains of the Willebrand
factor and, like the latter, also bind collagen I.
Thus, collagen VI apparently has the function of
a bridge between cells and collagen fibres [11, 20,
50]. The molecules of collagen VII consist of
three identical subunits and occur in connective
tissue as disulphide-linked dimers [83]. The shortchain collagen VIII is found in the majority of
endothelial cells and, in particular, in the thick
basement membrane of the cornea (Descemet's
membrane). The molecule is a heterotrimer with
the composition (a1)2a2. The a chain is made up
of a short triple helix of 454-457 amino acids and
two terminal, non-helical domains of 117 and 173
amino acids [85].
Collagens IX and X are deposited in various
cartilaginous structures formed from fibrils of collagens II and XI. Collagen X, which is closely
related to type VIII, is expressed only at mineralization sites in hypertrophic chondrocytes; it is a
homotrimer of short a chains which have triple
helices with eight interruptions; the sequences of
the chicken and bovine chains agree in 67 % of
their 674 amino acids [127]. Collagen IX is a
heterotrimer of three different a chains of
72-115 kDa and these include three triple-helix
domains, COL1-COL3, and four flexible, nonhelical domains, NC1-NC4 (the numbering
beginning at the C-terminus). a2(IX) bears a
covalently bound glycosaminoglycan on domain
NC3. Collagen IX alone is unable to form fibrils,
but it is covalently bound with collagen II of cartilage fibrils via lysine-derived cross-links. In this
way, the collagen fibrils of the cartilage are completely surrounded by glycosaminoglycan, and
this contributes to the high hydration state of the
cartilage matrix. Two forms of the a1(IX) chain
arise by alternative splicing. The cartilaginous
form bears a large, globular domain, NC4, of
266 amino acids at the N-terminus; this extends
beyond the fibril surface and can interact with
other components of the matrix. The form found
in the cornea and in the vitreous humour has a
much shorter N-terminus [84, 140]. In addition to
collagen IX, there are other fibril-associated collagens with interrupted triple helices (FACITs),
e.g. type XII in the tendons of chicken embryos,
type XIII in the endothelium of the human
umbilical vein, and type XIV in calf tendon and
skin [28, 43, 99].
The genes of the fibril-forming collagens I-III
have 5-8 exons in the region of the N-propeptide and 44 exons in the triple-helix region,
i.e. they have more interruptions than most other
genes. Because correct splicing apparently requires a minimal intron length of about 80 bp, these
genes with pro-collagen-coding sequences of less
than 5 kb often occupy more than 40 kb of chromosomal DNA. Most of the exons have a length
of 54 bp, and some have derived values like 45
(= 54-9), 108 (= 2 X 54) or 99 (= 108-9) bp.
One exon of 54 bp encodes 18 amino acids or six
G-X-Y triplets. Exons of 54 bp also predominate
in collagens V and XI. The characteristic gene
structure of the fibril-forming collagens has led to
the concept that the genes of all collagens arose
by the multiplication of a primordial sequence of
54 bp. However, further collagen genes that have
been sequenced in the meantime do not fit this
54-bp hypothesis: the pro-collagen genes of
type VI mostly have exons of 63 bp in addition to
some with 27,36,45,54 or 90 bp, i.e. multiples of
a 9-bp unit. The human and murine a1(IV) genes
contain 52 and 47 exons, respectively, with very
variable lengths between 27 and 192 bp. The
complete triple helix in the a1(VIII) and
a2(VIII) genes of mammals and the a1(X) gene
of the chicken is encoded by a single large exon
[16,50,68, 117, 133].
It is difficult to put together a detailed picture
of the evolution of vertebrate collagens. Collagens of types I and II have been identified in
agnathans, elasmobranchs and teleosts, and in
some cases have been partially characterized; little is known about other fish collagens. The agnathans, which are at the root of vertebrate evolution, possess a cartilaginous skeleton, the matrix
of which contains chondroitin-6-sulphate as the
glycan component, as in the higher vertebrates.
However, the main protein components in the
lamprey are not collagen and elastin as in other
vertebrates; in this case collagen is subsidiary to a
newly discovered structural protein, lamprin,
which makes up 44-51 % of the dry weight. This
protein differs in its amino acid spectrum from
both vertebrate collagen and elastin. It consists of
over 80 % of the non-polar amino acids proline,
glycine, alanine, valine and leucine, and contains
6.3 % tyrosine and 4.2 % histidine. Lamprin is
cross-linked into fibres but the linking mechanism
is not known; desmosin and isodesmosin are in