Clin Genet
2000: 58: 270 – 279
Printed in Ireland. All rights reser
6ed
Developmental Biology: Frontiers for Clinical Genetics
Collagens: building blocks at the end of the
development line
Section Editor: Roderick R McInnes
e-mail: mcinnes@sickkids.on.ca
Byers PH. Collagens: building blocks at the end of the development
line.
Clin Genet 2000: 58: 270 – 279. © Munksgaard, 2000
Collagens are some of the major building blocks of the vertebrate
body. In addition to their structural role, they are important for cell
guidance during development and for maintaining tissue integrity. In
their absence, phenotypes range from lethal to mild. These studies
demonstrate that collaens,in their rich array, play important roles in
development and are significant elements in reading the developmental
code.
PH Byers
Departments of Pathology and Medicine,
University of Washington, Seattle, WA
98195, USA
Key words: basement membranes – bone
– cartilage – collagen – development –
genes – mutations – procollagen
Corresponding author: PH Byers, Depart-
ment of Pathology, University of Washing-
ton, Seattle, WA 98195, USA. Tel:
+1 206
543 4206; fax:
+1 206 616 1899; e-mail:
pbyers@u.washington.edu
Received 6 July 2000, revised and ac-
cepted for publication 19 July 2000
Our physical appearance is the end result of the
initiation of a complex developmental program,
the cellular memory of the instructions from that
program, and the cell-, tissue-, and organ-specific
expression of a set of morphogenetic molecules. In
rare instances, the entire morphogenetic program
goes awry presumably because the early program-
ming events in development are miscued (EB
White’s ‘Stuart Little’ being one of the best docu-
mented examples, although still poorly understood
at the molecular level almost half a century after
the first report). Sometimes, these instances of
poor planning lead to rearrangements that can
only be seen with detailed studies of internal
anatomy (such as situs in
6ersus) while others are
more easily appreciated by simple observation (ab-
sence or duplication events, such as forms of ectro-
dactyly or polydactyly). These disorders flow from
fundamental errors in the specification of the body
plan and can be attributed to early molecular
events that occur at specific times in embryonic
development. Although these morphogenetic er-
rors are easy to spot, they are probably the mildest
of the consequences of mutations that affect the
expression or structure of this set of early deter-
mining molecules, the most severe rarely coming to
light because of the great biological efficiency of
our surveillance system for developmental errors.
Collagens: nature’s building blocks
In the real world of dysmorphology, syndromol-
ogy, and genetics there is increasing recognition of
disorders that result from mutations in molecules
that function to organize the developmental role.
At the same time, it is clear that ‘end molecule’
failure leads to some of the more common disor-
ders in which body form is distorted. The distinc-
tive phenotypes resulting from mutations that
affect the amount or structure of type I collagen
(osteogenesis imperfecta [OI] in all its forms) or
type II collagen (with a phenotypic range from
achondrogenesis to Stickler syndrome) are striking
examples (1). Here, the basic plan is laid out, but
the building materials are not delivered appropri-
ately or are unusable because of design flaws.
In this context, the most fundamental of all the
building blocks available are the collagens, until
just the last couple of decades considered to be
some of the least interesting biological molecules
because of their long-term stability. The reputation
of these dynamic, multi-functional molecules was
not enhanced when one of the pioneers of modern
270
Collagen building blocks
molecular biology hoped (perhaps this is apoc-
ryphal) that the structure of DNA would not be
as dull as the triple helix of collagen. ‘Collagen’
has, however, survived and now presents a fasci-
nating tale of evolutionary (almost revolutionary)
adaptation to fill a large number of both general
and highly specialized niches in morphogenesis
and tissue integrity (2).
Most mutations in collagen genes act in a
dominant
manner,
probably
because
the
molecules they form are trimers and incorpora-
tion of a single abnormal chain interferes with
the assembly of three chains in the molecule, the
folding of the triple helix, or interactions with
other molecules in the matrix (3). There are rare
recessively inherited disorders that result from
mutations in fibrillar collagen genes. Recessive
variants of some disorders, dystrophic epidermol-
ysis bullosa that results from mutations in type
VII collagen, can result from homozygosity, or
compound heterozygosity, for functional null alle-
les in collagen genes (4). These experiments of
nature and the deliberate (or fortuitous) inactiva-
tion of collagen genes in model organisms
provide the few instances in which the develop-
mental effects of the absence of these molecules
can be assessed without the interference of the
abnormal molecule.
More than a molecule: collagen is a family of genes
with characteristic features
‘Collagen’ is now known to be encoded by a fam-
ily of at least 30 genes (see Table 1). The protein
products are diverse in size, structure, distribu-
tion, and abundance. Each collagen gene encodes
a precursor chain, known as pro
a chains, of spe-
cified number and type, which contain sequences
that direct them to assemble with specific part-
ners into molecules that contain three chains –
some are homotrimers and some are hetero-
trimers. In aggregate, collagens represent the
most abundant proteins of the body and, for the
most part, function as extracellular building
blocks. Some, notably type XIII and type XVII,
are transmembrane molecules, and others, type
XV and type XVIII, house other functional
molecules – endostatins – that appear to regulate
vasculogenesis and are released from the parent
molecule by proteolysis (5, 6). Thus, it should
come as no surprise that mutations in many of
these genes lead to phenotypes characterized by
altered shape, although not altered form. That is,
the basic instructions for laying down the body
plan are uncompromised; it is only the final con-
struction units that are either present in low
abundance or modified in shape. However, it is
good to keep in mind that some collagens have
other functional components and mutations in
them may have quite different effects on, for ex-
ample, vessel formation.
The biosynthetic pathway is complex
Collagen biosynthesis involves one of the richest
processing pathways of all proteins. Almost all
collagen genes have dozens to more than 100 ex-
ons. The COL
7
A
1
gene has 118 exons, the largest
number of all known vertebrate genes, while the
COL
10
A
1
gene has only two, making it the most
intron poor of all the collagen genes. The fibrillar
collagen genes have between 51 and 66 exons.
The complexity of these genes and the deleterious
consequences of exon-skipping and of the use of
cryptic splice sites provide a rich substrate for
mutations. The mature mRNAs are translated on
membrane-bound polysomes and the precursor
chains are inserted into the lumen of the rough
endoplasmic reticulum. During and after transla-
tion, most prolyl residues that precede glycine
residues in the triple-helical domain are hydroxy-
lated, as are a variable number of lysyl residues,
the number depends on the collagen type. Chains
associate through regions in the carboxyl-terminal
propeptide, which also specifies binding partners,
and triple helix propagates from that end toward
the amino-terminal end of the molecule. Prolyl
4-hydroxylase, a tetramer of two
a-chains and
two
b-chains that are identical to protein disulfide
isomerase (PDI), HSP47 (a collagen-specific chap-
erone), and several other proteins chaperone the
association
and
folding.
Intact
precursor
molecules are secreted, or in the case of type XIII
and type XVII remain membrane-bound through
transmembrane components, are processed prote-
olytically in the extracellular space, and form
fibrillar or meshwork complexes as their final
structure.
Mutations: the common language of dysfunction
For the large part, the phenotypes that result
from mutations in collagen genes reflect the tissue
and organ distribution of the expression of these
genes (Tables 1 – 3). The developmental role of
collagens can be difficult to define when muta-
tions induce additional phenotypes. It is probably
only in the presence of homozygosity for ‘null’
mutations that the role of a particular collagen,
modifying enzyme, or chaperone can be assessed
because of the ability of the organism to adapt to
structural alterations.
271
Byers
Table 1. Collagen genes and their disorders
Gene
Chromosomal lo-
Collagen type
Protein
Disorders
cation
17q21.31-q22.05
pro
a1(I)
Osteogenesis imperfecta
I
COL1A1
Ehlers–Danlos syndrome type VIIA
7q22.1
pro
a2(I)
Osteogenesis imperfecta
COL1A2
Ehlers–Danlos syndrome type VIIB
Ehlers–Danlos syndrome type II
12q13.11-q13.2
pro
a1(II)
II
Stickler syndrome, type I
COL2A1
Wagner syndrome type II
Spondylepiphyseal dysplasia congenita
Kniest dysplasia
Hypochondrogenesis
Achondrogenesis type II
Spondylo-metaphyseal-epiphyseal dysplasia (SMED), Strudwick type
2q31
pro
a1(III)
III
Ehlers–Danlos syndrome type IV
COL3A1
Ehlers–Danlos syndrome type III (?)
13q34
pro
a1(IV)
IV
COL4A1
COL4A2
13q34
pro
a2(IV)
2q36-q37
pro
a3(IV)
COL4A3
Alport syndrome, recessive
COL4A4
2q36-q37
pro
a4(IV)
Alport syndrome, recessive
Xq22
pro
a5(IV)
Alport syndrome, X-linked
COL4A5
Xq22
pro
a6(IV)
Alport syndrome, X-linked Leiomyomatosis
COL4A6
9q34.2-q34.3
pro
a1(V)
Ehlers–Danlos syndrome type I
V
COL5A1
Ehlers–Danlos syndrome type II
COL5A2
2q31
pro
a2(V)
Ehlers–Danlos syndrome type I
Not mapped
pro
a3(V)
COL5A3
COL6A1
21q22.3
pro
a1(VI)
Bethlem myopathy
VI
21q22.3
pro
a2(VI)
Bethlem myopathy
COL6A2
2q37
pro
a3(VI)
Bethlem myopathy
COL6A3
3p21.3
pro
a1(VII)
Epidermolysis bullosa, recessive dystrophic
VII
COL7A1
Epidermolysis bullosa, dominant dystrophic
Epidermolysis bullosa, pretibial
3q12-q13.1
pro
a1(VIII)
VIII
COL8A1
COL8A2
1p34.4-p32.3
pro
a2(VIII)
6q13
pro
a1(IX)
Multiple epiphyseal dysplasia
IX
COL9A1
1p33-p32.2
pro
a2(IX)
COL9A2
Multiple epiphyseal dysplasia, type II
20q13.3
pro
a3(IX)
Multiple epiphyseal dysplasia, type III
COL9A3
6q21-q22.3
pro
a1(X)
X
Metaphyseal chondrodysplasia, Schmid type
COL10A1
Spondylometaphysealdysplasia, Japanese type
1p21
pro
a1(XI)
XI
Stickler syndrome, type III
COL11A1
Marshall syndrome
6p21.3
pro
a2(XI)
Stickler syndrome, type II
COL11A2
Otospondylomegaepiphyseal dysplasia (OSMED)
Weissenbacher-Zweymuller syndrome
Non-syndromic deafness (DFNA13)
6
XII
pro
a1(XII)
COL12A1
10q22
pro
a1(XIII)
XIII
COL13A1
COL14A1
8q23
pro
a1(XIV)
XIV
COL15A1
XV
9q21-q22
pro
a1(XV)
1p34
pro
a1(XVI)
XVI
COL16A1
COL17A1
10q24.3
pro
a1(XVII)
Epidermolysis bullosa, generalized atrophic benign
XVII
21q22.3
pro
a1(XVIII)
XVIII
COL18A1
6q12-q14
pro
a1(XIX)
COL19A1
XIX
272
Collagen building blocks
Table 2. Collagen types, chain composition, and tissue distribution
Chains
Molecules
Collagen type
Tissue distribution
Fibrillar collagens
a1(I), a2(I)
a1(I)
2
a2(I)
I
Ubiquitous in hard and soft tissues, major protein of bone, skin.
a1(I)
3
Uncommon, found in some tumors, amniotic fluid cells
II
a1(II)
a1(II)
3
Cartilage, vitreous, intervertebral disk
See also type XI
III
a1(III)
a1(III)
3
Soft tissues and hollow organs
V
a1(V), a2(V), a3(V)
a1(V)
2
a2(V)
Soft tissues, placental, vessels, chorion
a1(V)a2(V)a3(V)
a2(V) can substitute for the a2(XI) chain in vitreous
See also type XI
a1(XI), a2(XI)
a1(XI)a2(XI)a1(II)
XI
Cartilage
a1(XI)a2(V)a1(II)
Vitreous
Basement membrane collagens
a1(IV), a2(IV)
a1(IV)
2
a2(IV)
IV
Basement membranes
a3(IV), a4(IV)
Others uncertain
a5(IV), a6(IV)
Fibril-associated collagens with interrupted triple helices (FACIT)
a1(IX), a2(IX), a3(IX)
IX
a1(IX),a2(IX)a3(IX)
Cartilage, vitreous
a1(XII)
XII
a1(XII)
3
Soft tissues
a1(XIV)
a1(XIV)
3
XIV
Soft tissues
Meshwork-forming collagens
a1(VIII), a2(VIII)
a1(VIII)
2
a2(VIII)
VIII
Cornea, endothelium
X
a1(X)
a1(X)
3
Hypertrophic zone of the growth plate
Anchoring-fibril collagen
a1(VII)
a1(VII)
3
VII
Anchoring fibrils, dermal epidermal junction
Microfibril-forming collagens
a1(VI), a2(VI), a3(VI)
a1(VI)a2(VI)a3(VI)
Microfibrils in soft tissues and cartilage
VI
Transmembrane collagens
a1(XIII)
XIII
a1(XIII)
3
Cell surfaces, epithelial cells
a1(XVII)
a1(XVII)
3
Epidermal cell surfaces
XVII
Endostatin forming collagens
a1(XV)
XV
a1(XV)
3
Endothelial cells
a1(XVIII)
a1(XVIII)
3
XVIII
Endothelial cells
Others
a1(XVI)
XVI
a1(XVI)
3
Ubiquitous
a1(XIX)
a1(XIX)
3
Ubiquitous
XIX
Type I collagen genes
Type I collagen is the major protein of bone, skin,
tendons and ligaments, blood vessel walls, and
other connective tissues except cartilage. The ma-
jor phenotypes that are known to result from mu-
tations in the COL
1
A
1
and COL
1
A
2
genes are
forms of OI (1), Ehlers – Danlos syndrome (EDS)
type VII (7, 8), an uncommon form of EDS type II
(9), and rare disorders of blood vessel integrity
(10). With rare exception, these disorders are in-
herited in an autosomal dominant fashion and
result either from haploinsufficiency mutations
(11 – 14) or mutations that act in a dominant nega-
tive fashion (1). Survival of individuals ho-
mozygous for ‘null’ mutations in the COL
1
A
2
gene
(moderately severe OI and a rare form of EDS
type II) demonstrate that this is not an essential
gene (9). In contrast, homozygosity for COL
1
A
1
null mutations is lethal in the mouse, which serves
to demonstrate the essential nature of the gene
product (15, 16). These mice die at embryonic day
11 because of loss of integrity of the vascular
system under hydrostatic pressure. With the excep-
tion of some alteration in branching during lung
morphogenesis, the early embryos have no major
form alterations. Their cultured cells do, however,
have multiple alterations of proteins in the extra-
cellular matrix, which may contribute to the lack
of tissue integrity (17).
The COL
1
A
1
gene encodes the pro
a1(I) chains
of type I procollagen, at least two of which are
needed to make a functional molecule. Molecules
cannot accommodate more than a single pro
a2(I)
chain (encoded by the COL
1
A
2
gene). Ho-
mozygosity for COL
1
A
2
non-functional alleles
273
Byers
274
Table
3.
Animal
models,
human
disorders
and
type
of
mutations
in
collagen
genes
Gene
Animal
models
Human
disorders
Type
of
mutation
Type
of
mutation
Homozygous
null
(−/−
)
Heterogyzous
Homozygous
null
(−/−
)
Transgenic
animals
(over
Missense
mutations;
structural
alterations
Heterozygous
null
(+/−
)
null
(+/−
)
expression
or
missense
mutations)
COL1A1
Embryonic
lethal
at
day
11
Wide
range
of
phenotype
from
lethal
to
mild
OI
type
I
OI
type
I
‘like’
Not
seen
Range
from
lethal
to
moderate
in
the
mouse
because
of
phenotype
osteogenesis
imperfecta
vascular
rupture
COL1A2
OI
type
III
Very
mild
os-
Wide
range
of
phenotype
from
lethal
to
mild
NA
OI
type
III,
EDS
type
I/
II
?mild
osteopenia
teopenia
osteogenesis
imperfecta
COL2A1
Neonatal
or
late
fetal
lethal
‘Stickler-like’
Range
from
severe
lethal
to
Wide
range
of
phenotype
from
achondrogene-
Not
seen
Stickler
syndrome
sis
type
II
to
spondyloepiphy-seal
dysplasias
mild
chondrodysplasia
COL3A1
Early
death
from
arterial
rupture
Normal
NA
Not
seen
EDS
type
IV
EDS
type
IV
COL4A1
Point
mutations
are
embryonic
Not
known
Not
known
Not
known
lethal
in
C
.
elegans
COL4A2
Not
known
Not
known
Not
known
COL4A3
Homozygosity
produces
autosomal
recessive
Alport
syndrome
COL4A4
Homozygosity
produces
autosomal
recessive
Alport
syndrome
COL4A5
Hemizygosity
produces
Alport
Hemizygosity
produces
Alport
syndrome
syndrome
COL4A6
Combined
with
loss
of
COL4A5
produces
Alport
syndrome
and
enteric
leiomyomatosis
COL5A1
EDS
type
I/
II
EDS
type
I/
II
COL5A2
Deletion
of
exon
6
(contains
EDS
type
I/
II
N-propeptide
cleavage
site)
produces
EDS
type
I/
II
like
picture
in
homozygous
mice
COL6A1
Early
onset
severe
myopathy
Bethlem
myopathy
COL6A2
Bethlem
myopathy
COL6A3
Bethlem
myopathy
COL7A1
Recessive
dystrophic
Normal
Recessive
dystrophic
epi-
Dominant
dystrophic
epidermolysis
bullosa
Normal
epidermolysis
bullosa
dermolysis
bullosa
COL9A1
Late
onset
mild
degenerative
Multiple
epiphyseal
dysplasia
Normal
joint
disease
COL9A2
Multiple
epiphyseal
dysplasia
COL9A3
Multiple
epiphyseal
dysplasia
COL10A
COL11A1
Perinatal
lethal
in
the
cho
/
cho
Stickler
and
Marshall
syndromes
Normal
Stickler
and
Marshall
syndromes
mouse
COL11A2
Stickler
syndrome
COL17A1
Generalized
benign
epidermolysis
bullosa
Collagen building blocks
produces much milder phenotypes of OI (18) or
EDS (9). Thus, although the COL
1
A
2
gene
product provides some plasticity for the type I
procollagen molecules (particularly, it seems, in
permitting normal bone mineralization and normal
vascular and skin integrity), it is not essential for
survival. In contrast, the absence of COL
1
A
1
gene
products means that no type I procollagen
molecules can be made. Under these circumstances
all mechanical integrity of tissues is lost. This phe-
notype would be expected in a quarter of all preg-
nancies initiated by two parents, both of whom
have OI type I (which results from heterozygosity
for COL
1
A
1
null mutations).
Type II collagen and other collagens expressed in cartilage
Cartilage is home to several collagens: types II, IX,
X, and XI. Type II collagen is the most abundant
collagen in the matrix of cartilage, in the early
anlage of developing bone, and the vitreous of the
eye. Type IX and type XI collagens have similar
patterns of expression and distribution. They form
heterotypic (multiple types of collagen) fibrils with
type II collagen. Because of its appearance early in
the formation of endochondral bone, it seemed
likely that both the form and the growth of the
bones would be dependent on the presence of type
II collagen. In mice homozygous for non-expres-
sion COL
2
A
1
alleles, membranous and periosteal
bone form normally but long bones are rudimen-
tary, lack marrow cavities and the only mineralized
bone is formed from the periosteum. The mice
have no palate; their ribs are small and not prop-
erly mineralized so that the thorax is small. Alveoli
are not distended and the animals do not survive.
Calvarial mineralization is normal, although the
shape is altered (19, 20). In these animals the
notochord is not resorbed and vertebral body
structure is altered (21). It is thought, surprisingly,
that most of these alterations, even the failure of
palate formation and rotation, arise because of
changes in the mechanical properties of tissues, not
specification of the plan of the tissue (21).
Type II collagen is a homotrimer. In addition,
the pro
a1(II) chain is included with the proa1(XI)
and pro
a2(XI) chain in type XI collagen. Muta-
tions that alter sequences in the triple-helical do-
main of the pro
a1(II) chain interfere with the
normal helix formation of 7/8 molecules made by
the cell and many of those alter secretion (1). The
perinatal lethal disorder achondrogenesis type II
results from heterozygosity for mutations in the
triple-helical domain and the phenotype resembles
the homozygous COL
2
A
1
null mouse, probably
because there are virtually no normal molecules
and the abnormal molecules, although poorly
secreted, interfere with fibrillogenesis. The pheno-
type of the heterozygous null mutation in humans,
Stickler syndrome (1), resembles the heterozygous
null mouse.
Type IX and type XI collagens interact with type
II collagen in the cartilage matrix to form het-
erotypic fibrils (that is fibrils that contain all three
molecules). Although both are present in much
lower amounts than type II collagen, the effects of
homozygosity for null mutations (premature termi-
nation codons) in the Col
11
a
1
gene in the cho/cho
mouse are lethal as a consequence of abnormalities
in the cartilage of limbs, ribs, the mandible and
trachea (22). Mutations in the COL
11
A
1
gene in
people give rise to a variant of Stickler syndrome,
Marshall syndrome, or an intermediate phenotype
(23). This variation may depend on the outcome of
splice site mutations. Homozygosity for a null
Col
9
a
1
gene in mice yields a minimal phenotype of
late onset mild degenerative joint disease (24), even
though the protein product is required for assem-
bly of the entire type IX collagen protein (25).
Type X collagen is found almost exclusively in
the hypertrophic zone of the growth plate. Mice
homozygous for inactivation mutants of the
Col
10
a
1
gene have variable phenotypes with about
10% dying in the perinatal period and more by
early adult life. These animals all have significant
growth plate compression as well as hematopoietic
alterations (26). They resemble, to some extent, the
mice produced by a dominant interference trans-
gene (27). In people, dominantly inherited muta-
tions result in Schmid metaphyseal chondro-
dysplasia with short stature and bowed bones. No
homozygous nulls are known but most of the
human mutations may act as heterozygous nulls in
that they likely interfere with chain association or
with multimer formation (28 – 30). Almost all the
mutations in the COL
10
A
1
gene are found in the
domain that encodes the chain association domain.
The relationship between the skeletal and hemato-
poietic findings is not clear at this point, but sug-
gests that the development of the latter may
depend on certain signals from an intact marrow.
Type III collagen
The homozygosity for a Col
3
a
1
null mutation in
the mouse results in a perinatal lethal phenotype
with vascular rupture (31). These mice have very
thin vascular walls and thin skin. As type III
collagen is expressed early in the development of
these tissues, one proposal has been that the
molecules provide a part of the scaffolding on
which the mature organ is built. The published
275
Byers
mutations in the COL
3
A
1
gene all alter the se-
quence of the chains (1, 32). A small number of
premature termination mutations in the COL
3
A
1
gene are known and also result in an EDS type IV
phenotype (vascular, bowel, uterine rupture with
early death), similar to the effects of heterozygous
dominant mutations (U Schwarze and PH Byers,
unpublished).
Type IV collagen
Basement membrane molecules come early and stay
late. They are among the first matrix molecules
synthesized by embryonic cells and the same
molecules are important for isolation of organ-
forming buds from endodermal cells and in the
separation of tissue formed from different cell
pools. These molecules persist in the major base-
ment membranes of the kidney, lung, skin, and
other regions in which cells of one origin are sepa-
rated from those of other origins. In mammals,
there are six type IV collagen genes that occur in
three pairs, each oriented in a head-to-head fashion
and sharing a bifunctional promotor, no doubt
having evolved from a single gene duplication/in-
version event followed by duplication and disper-
sion of the two-gene structure. The two genes
expressed ubiquitously in these tissues are the
COL
4
A
1
and COL
4
A
2
genes, located on chromo-
some 13. No human mutations in these genes are
known and no knockout mice appear to have been
generated. The only clue to the fate of mutations in
these genes is offered in Caenorhabditis elegans. In
C. elegans, mutations in both the COL
4
A
1
and
COL
4
A
2
equivalent genes often have a lethal out-
come during embryogenesis, although others may
permit development to proceed (33 – 35). In hu-
mans, it is likely that most mutations in these genes
result in phenotypes that do not survive early
embryogenesis.
The COL
4
A
3
, COL
4
A
4
, COL
4
A
5
, and COL
4
A
6
genes all appear to be expressed in kidney, either in
heterotrimers that contain a single chain each of the
COL
4
A
3
, COL
4
A
4
, and COL
4
A
5
products or in
trimers that contain two products of the COL
4
A
5
and a single chain encoded by the COL
4
A
6
gene.
Homozygosity for knockout/null mutations in the
murine Col
4
A
3
gene results in a progressive
glomerular disease (36) and cochlear abnormalities
(37), similar to those seen in people with autosomal
recessive Alport syndrome. The most striking effect
of mutations in the type IV collagen genes is the
combined Alport syndrome and diffuse esophageal
leiomyomatosis seen with deletions that extend
from intron 2 of the COL
4
A
6
gene into the
COL
4
A
5
gene (38). It is not clear if there is a gene
involved in the control of smooth muscle growth
located in the large intron 2 or if the presence of
both gene products is important for that function
(39, 40).
Type VI collagen
The type VI collagen genes, COL
6
A
1
, COL
6
A
2
,
and COL
6
A
3
, encode a collagenous protein that
forms a portion of the epimysium and is involved in
connections with many of the proteins of the muscle
fiber. Mutations in any of the three of these genes,
the protein products of which contribute to a single
heterotrimer, result in Bethlem myopathy, an un-
common form of muscle disease characterized by
muscular dystrophy and joint contracture (41 – 43).
Homozygosity for a targeted disruption in the
murine Col
6
a
1
gene results in no clinical phenotype
but histological evidence of myopathy (44). Type VI
collagen is a heterotrimers of all three chains and
the
a1(VI) chain is essential for molecular assembly.
Type VII and type XVII collagens
Mutations in the COL
7
A
1
gene, the encoded
product of which forms anchoring fibrils at the
dermal – epidermal
junction,
produce
different
forms of dystrophic epidermolysis bullosa (45).
Targeted inactivation of the murine Col
7
a
1
gene
results in the recessive dystrophic epidermolysis
bullosa picture, essentially unchanged from that in
humans (46). Mutations in another collagen gene
expressed by basal keratinocytes that functions as a
transmembrane protein, COL
17
A
1
, result in a
much milder form of bullous skin disease (47).
Genotype to phenotype: a well-trodden but poorly
mapped pathway
Dramatic effects on development are seen, to date,
only with null mutations in a small number of
genes: Col
1
a
1
in mice and type IV collagen genes in
C. elegans. Mutations in some of the genes of the
post-translational pathway, notably the prolyl 4-hy-
droxylase gene, may also be an embryonic lethal, at
least when examined in C. elegans (48). It is clear
that, while there are complex interactions among
collagens and other proteins, many of these are
important only at the final stages of growth and
have relatively little importance in the translation of
other developmental signals.
The tissue-specific distribution of gene expression
is one key to understanding the phenotypic effects
of mutations but it remains difficult to see
how mutations that either alter the amount of a
chain synthesized or alter the structure in a very
subtle fashion give rise to the diverse phenotypes
276
Collagen building blocks
represented by dominant interference (dominant
negative) or recessive mutations in collagen genes.
These pathways are most extensively explored for a
few of the fibrillar collagens, type I collagen and
bone disorders, and type II collagen for disorders
of cartilage growth and structure.
The complexity of pathogenesis for mutations in
type I collagen starts in the nucleus
Most known mutations in the COL
1
A
1
and
COL
1
A
2
genes give rise to forms of brittle bone
disease, OI (49, 1). These phenotypes range from a
subtle increase in the risk of fractures (a few during
an individual’s lifetime) to intrauterine or perinatal
lethality (50). The mildest of these forms (OI type
I) results from failure to synthesize the products of
one COL
1
A
1
allele (11 – 14, 51). The mutations
that result in this phenotype are largely premature
termination codons along the length of this 51
exon gene. Three mechanisms are involved: frame
shifts due to small insertions or deletions, point
mutations that create termination codons, and
splice site mutations that result in inclusion of
additional nucleotides or deletion of nucleotides
due to alternate splicing such that new termination
codons result at or downstream from the muta-
tional event. These products are either extremely
unstable or are segregated into the nucleus away
from the protein translation machinery and secre-
tory pathway.
Goes on to the secretory pathway
In the case of mutations that alter the sequence of
the protein, but allow full-length, or nearly full-
length, chains to be synthesized, the cell brings a
very different armamentarium to bear in its attack
on these purveyors of molecular chaos. The consti-
tutive pro
a chains of type I procollagen (and of
other fibrillar collagens) are synthesized on ribo-
somes bound to the membrane of the rough endo-
plasmic
reticulum
and
inserted
through
the
membrane during elongation. Because chain asso-
ciation is determined by sequences in the carboxyl-
terminal 200 residues of these 1400 residue long
chains, the chains lie quietly in the lumen – but
not alone. The modifying prolyl and lysyl hydroxy-
lases, hydroxylysyl glycosidases, and additional
chaperone proteins – probably BiP, GRP78, prolyl
cis – trans isomerase, and HSP47 accompany these
chains in their travels. Once synthesis is completed,
the carboxyl-terminal regions fold, expose interact-
ing surfaces, and chains assemble into appropriate
combinations. By itself, the correct combination of
chains with their mates is a remarkable testament
to the specificity of interactions lent by only a few
residues of a protein backbone. In some cells, the
constituent chains of three or more types of colla-
gen are synthesized simultaneously with few if any
errors occurring in the correct association of the
half a dozen or more different gene products.
Substitutions for glycine, exon-skipping, or
small in-frame deletions or insertions within the
triple-helical domains of these chains do not inter-
fere with association, but do bring helix propaga-
tion to a halt, awaiting sufficient energy to
overcome the barrier or re-initiating helix forma-
tion. These molecules remain bound to the modify-
ing enzymes and their rate of secretion is
substantially slowed. These molecules remain in
the ER for extended periods. Many are degraded
by mechanisms that are not yet clear, but a small
number can still traverse with secretory pathway,
disregarding the many checkpoints on the lookout
for misfolded molecules, only to exit the cell and
continue their misadventures by interfering with
fibrillogenesis or with, in the case of bone, mineral-
ization. The only protection from this rude intru-
sion on order is the intracellular retention and
possibly degradation of the abnormal molecules.
Unfortunately for the organism, these processes
are not entirely successful, so that some abnormal
molecules are secreted and their deleterious effects
prevail.
Mutations that alter sequences in the carboxyl-
terminal region of chains can ablate chain associa-
tion, have no effect, or alter the rate of association.
Those that do not permit chain association act as
‘null’ alleles so that the phenotype would be ex-
pected to be mild. Those that alter the rate of
association activate the cellular ‘stress’ protein re-
sponse and stimulate the synthesis of several chap-
erone proteins which then appear either to
facilitate chain association or interfere with the
process (52 – 54). Although generally inefficient,
some of these molecules are secreted. The pheno-
types tend to be severe, but it is not clear whether
this is true for all such mutations or is biased by
ascertainment.
And ends in the extracellular matrix
Still lacking is a clear step-by-step understanding
of how mutations create the specific phenotype. It
is all well and good to say that too little normal
molecules create a mild phenotype while the pres-
ence of abnormal molecules creates a more severe
picture. After all, although generally true, this is
not always the case. As a rule, for type I collagen
mutations, the phenotype reflects the nature of the
mutation, the location of the mutation in the
277
Byers
molecule, the effect of the mutation on the chain,
and, for substitutions for glycine residues, the nature
of the substituting amino acid. It appears that all
mutations that alter the sequence within the triple-
helical domain of the chain interfere with folding
and, to a greater or lesser extent, decrease secretion.
However, it may well be the secreted molecules that
are the major purveyors of damage. These molecules
are less efficiently incorporated into fibrils and when
present interfere with mineralization. For other
fibrillar collagens, where mineralization is not an
issue, e.g. type III, V, and XI collagens, no doubt
the interaction with other molecular building blocks
disrupts function and structure.
‘Building blocks’ are not the only story
Although perceived largely as the ‘Legos
®
’ of the
skeleton and other tissues, it has become abundantly
clear that collagens have far more diverse functions.
During early development and regeneration of tis-
sue, collagens provide one of the pathways that
direct cell motility and organization. Transmem-
brane molecules, particularly integrins, bind classes
of collagen in a molecular-specific fashion so that
cells can adhere specifically to them. When certain
collagens are not made at all, these interactions
disappear, the ability to produce appropriate cell
guidance and thus to encounter the next set of
signals is lost. These effects are brought to light only
with the ability to inactivate collagen genes, perhaps
because the organism has many adaptive tools
available when structures are altered. Thus, colla-
gens of different types have important roles in
determining cell interactions, cell motility, and tissue
organization. They are important in development,
both from the structural point of view and in
allowing the cells that make them or use them access
to new signal molecules. Mutations in these
molecules that permit survival illustrate the adapt-
ability of the organism and represent the physiolog-
ical response to alterations in a system that has been
tens of thousands of generations in the making.
References
1. Kuivaniemi H, Tromp G, Prockop DJ. Mutations in fibril-
lar collagens (types I, II, III, and XI), fibril-associated
collagen (type IX), and network-forming collagen (type X)
cause a spectrum of diseases of bone, cartilage, and blood
vessels. Hum Mutat 1997: 9: 300 – 315.
2. Brown JC, Timpl R. The collagen superfamily. Int Arch
Allergy Immunol 1995: 107: 484 – 490.
3. Engel J, Prockop DJ. The zipper-like folding of collagen
triple helices and the effects of mutations that disrupt the
zipper. Annu Rev Biophys Chem 1991: 20: 137 – 152.
4. Hovnanian A, Rochat A, Bodemer C et al. Characterization
of 18 new mutations in COL
7A1 in recessive dystrophic
epidermolysis bullosa provides evidence for distinct molec-
ular mechanisms underlying defective anchoring fibril for-
mation. Am J Hum Genet 1997: 61: 599 – 610.
5. Saarela J, Ylikarppa R, Rehn M, Purmonen S, Pihlajaniemi
T. Complete primary structure of two variant forms of
human type XVIII collagen and tissue-specific differences in
the expression of the corresponding transcripts. Matrix Biol
1998: 16: 319 – 328.
6. Sasaki T, Fukai N, Mann K, Gohring W, Olsen BR, Timpl
R. Structure, function and tissue forms of the C-terminal
globular domain of collagen XVIII containing the angio-
genesis inhibitor endostatin. EMBO J 1998: 17: 4249 – 4256.
7. Byers PH, Duvic M, Atkinson M et al. Ehlers – Danlos
syndrome type VIIA and VIIB result from splice-junction
mutations or genomic deletions that involve exon 6 in the
COL
1A1 and COL1A2 genes of type I collagen. Am J Med
Genet 1997: 72: 94 – 105.
8. Giunta C, Superti-Furga A, Spranger S, Cole WG, Stein-
mann B. Ehlers – Danlos syndrome type VII: clinical fea-
tures and molecular defects. J Bone Joint Surg Am 1999: 81:
225 – 238.
9. Hata R, Kurata S, Shinkai H. Existence of malfunctioning
pro alpha
a2(I) collagen genes in a patient with a proa2(I)-
chain-defective variant of Ehlers – Danlos syndrome. Eur J
Biochem 1988: 174: 231 – 237.
10. Mayer SA, Rubin BS, Starman BJ, Byers PH. Spontaneous
multivessel cervical artery dissection in a patient with a
substitution of alanine for glycine (G13A) in the
a1 (I) chain
of type I collagen. Neurology 1996: 47: 552 – 556.
11. Willing MC, Deschenes SP, Scott DA, Byers PH, Slayton
RL, Pitts SH, Arikat H et al. Osteogenesis imperfecta type
I: molecular heterogeneity for COL
1A1 null alleles of type
I collagen. Am J Hum Genet 1994: 55: 638 – 647.
12. Redford-Badwal DA, Stover ML, Valli M, McKinstry MB,
Rowe DW. Nuclear retention of COL
1A1 messenger RNA
identifies null alleles causing mild osteogenesis imperfecta.
J Clin Invest 1996: 97: 1035 – 1040.
13. Willing MC, Deschenes SP, Slayton RL, Roberts EJ. Pre-
mature chain termination is a unifying mechanism for
COL
1A1 null alleles in osteogenesis imperfecta type I cell
strains. Am J Hum Genet 1996: 59: 799 – 809.
14. Korkko J, Ala-Kokko L, De Paepe A, Nuytinck L, Earley
J, Prockop DJ. Analysis of the COL
1A1 and COL1A2 genes
by PCR amplification and scanning by conformation-sensi-
tive gel electrophoresis identifies only COL
1A1 mutations in
15 patients with osteogenesis imperfecta type I: identifica-
tion of common sequences of null-allele mutations. Am J
Hum Genet 1998: 62: 98 – 110.
15. Schnieke A, Harbers K, Jaenisch R. Embryonic lethal
mutation in mice induced by retrovirus insertion into the
a1(I) collagen gene. Nature 1983: 304: 315–320.
16. Harbers K, Kuehn M, Delius H, Jaenisch R. Insertion of
retrovirus into the first intron of
a1(I) collagen gene to
embryonic lethal mutation in mice. Proc Natl Acad Sci
USA 1984: 81: 1504 – 1508.
17. Iruela-Arispe ML, Vernon RB, Wu H, Jaenisch R, Sage
EH. Type I collagen-deficient Mov-13 mice do not retain
SPARC in the extracellular matrix: implications for fibrob-
last function. Dev Dyn 1996: 207: 171 – 183.
18. Pihlajaniemi T, Dickson LA, Pope FM, Korhonen VR,
Nicholls A, Prockop DJ, Myers JC. Osteogenesis imper-
fecta: cloning of a pro-
a2(I) collagen gene with a frameshift
mutation. J Biol Chem 1984: 259: 12941 – 12944.
19. Li SW, Prockop DJ, Helminen H et al. Transgenic mice
with targeted inactivation of the Col
2a1 gene for collagen
II develop a skeleton with membranous and periosteal bone
but no endochondral bone. Genes Dev 1995: 9: 2821 – 2830.
20. Talts JF, Pfeifer A, Hofmann F. Endochondral ossification
is dependent on the mechanical properties of cartilage tissue
278
Collagen building blocks
and on intracellular signals in chondrocytes. Ann NY Acad
Sci 1998: 857: 74 – 85.
21. Aszodi A, Chan D, Hunziker E, Bateman JF, Fassler R.
Collagen II is essential for the removal of the notochord and
the formation of intervertebral discs. J Cell Biol 1998: 143:
1399 – 1412.
22. Li Y, Lacerda DA, Warman ML et al. A fibrillar collagen
gene, Col
11a1, is essential for skeletal morphogenesis [see
comments]. Cell 1995: 80: 423 – 430.
23. Annunen S, Kr J, Czarny M et al. Splicing mutations of
54-bp exons in the COL
11A1 gene cause Marshall Syn-
drome, but other mutations cause overlapping Marshall/
Stickler phenotypes. Am J Hum Genet 1999: 65: 974 – 983.
24. Fassler R, Schnegelsberg PN, Dausman J et al. Mice lacking
a1(IX) collagen develop noninflammatory degenerative
joint disease. Proc Natl Acad Sci USA 1994: 91: 5070 – 5074.
25. Hagg R, Hedbom E, Mollers U, Aszodi A, Fassler R,
Bruckner P. Absence of the
a1(IX) chain leads to a func-
tional knock-out of the entire collagen IX protein in mice.
J Biol Chem 1997: 272: 20650 – 20654.
26. Gress CJ, Jacenko O. Growth plate compressions and
altered hematopoiesis in collagen X null mice. J Cell Biol
2000: 149: 983 – 993.
27. Jacenko O, LuValle PA, Olsen BR. Spondylometaphyseal
dysplasia in mice carrying a dominant negative mutation in
a matrix protein specific for cartilage-to-bone transition.
Nature 1993: 365: 56 – 61.
28. Wallis GA, Rash B, Sweetman WA et al. Amino acid
substitutions of conserved residues in the carboxyl-terminal
domain of the
a1(X) chain of type X collagen occur in two
unrelated families with metaphyseal chondrodysplasia type
Schmid. Am J Hum Genet 1994: 54: 169 – 178.
29. Chan D, Cole WG, Rogers JG, Bateman JF. Type X
collagen multimer assembly in
6itro is prevented by a Gly618
to Val mutation in the
a1(X) NC1 domain resulting in
Schmid metaphyseal chondrodysplasia. J Biol Chem 1995:
270: 4558 – 4562.
30. Wallis GA, Rash B, Sykes B et al. Mutations within the gene
encoding the
a1(X) chain of type X collagen (COL10A1)
cause metaphyseal chondrodysplasia type Schmid but not
several other forms of metaphyseal chondrodysplasia. J Med
Genet 1996: 33: 450 – 457.
31. Liu X, Wu H, Byrne M, Krane S, Jaenisch R. Type III
collagen is crucial for collagen I fibrillogenesis and for
normal cardiovascular development. Proc Natl Acad Sci
USA 1997: 94: 1852 – 1856.
32. Pepin M, Schwarze U, Superti-Furga A, Byers PH. Clinical
and genetic features of Ehlers – Danlos syndrome type IV,
the vascular type. New Engl J Med 2000: 342: 673 – 680.
33. Sibley MH, Johnson JJ, Mello CC, Kramer JM. Genetic
identification, sequence, and alternative splicing of the
Caenorhabditis elegans
a2(IV) collagen gene. J Cell Biol
1993: 123: 255 – 264.
34. Sibley MH, Graham PL, von Mende N, Kramer JM.
Mutations in the
a2(IV) basement membrane collagen gene
of Caenorhabditis elegans produce phenotypes of differing
severities. EMBO J 1994: 13: 3278 – 3285.
35. Gupta MC, Graham PL, Kramer JM. Characterization of
a1(IV) collagen mutations in Caenorhabditis elegans and the
effects of
a1 and a2(IV) mutations on type IV collagen
distribution. J Cell Biol 1997: 137: 1185 – 1196.
36. Cosgrove D, Meehan DT, Grunkemeyer JA et al. Collagen
COL
4A3 knockout: a mouse model for autosomal Alport
syndrome. Genes Dev 1996: 10: 2981 – 2992.
37. Cosgrove D, Samuelson G, Meehan DT et al. Ultrastruc-
tural, physiological, and molecular defects in the inner ear
of a gene-knockout mouse model for autosomal Alport
syndrome. Hear Res 1998: 121: 84 – 98.
38. Zhang X, Zhou J, Reeders ST, Tryggvason K. Structure of
the human type IV collagen COL
4A6 gene, which is mutated
in Alport syndrome-associated leiomyomatosis. Genomics
1996: 33: 473 – 479.
39. Heidet L, Cohen-Solal L, Boye E, Thorner P, Kemper MJ,
David A, Larget Piet L et al. Novel COL
4A5/COL4A6
deletions and further characterization of the diffuse leiomy-
omatosis-Alport syndrome (DL-AS) locus define the DL
critical region. Cytogenet Cell Genet 1997: 78: 240 – 246.
40. Thorner P, Heidet L, Moreno Merlo F, Edwards V, Antig-
nac C, Gubler MC. Diffuse leiomyomatosis of the esopha-
gus: disorder of cell-matrix interaction? Pediatr Dev Pathol
1998: 1: 543 – 549.
41. Jobsis GJ, Bolhuis PA, Boers JM et al. Genetic localization
of Bethlem myopathy. Neurology 1996: 46: 779 – 782.
42. Jobsis GJ, Keizers H, Vreijling JP, de Visser M, Speer MC,
Wolterman RA, Baas F et al. Type VI collagen mutations
in Bethlem myopathy, an autosomal dominant myopathy
with contractures. Nat Genet 1996: 14: 113 – 115.
43. Speer MC, Tandan R, Rao PN et al. Evidence for locus
heterogeneity in the Bethlem myopathy and linkage to 2q37.
Hum Mol Genet 1996: 5: 1043 – 1046.
44. Bonaldo P, Braghetta P, Zanetti M, Piccolo S, Volpin D,
Bressan GM. Collagen VI deficiency induces early onset
myopathy in the mouse: an animal model for Bethlem
myopathy. Hum Mol Genet 1998: 7: 2135 – 2140.
45. Christiano AM, Hoffman GG, Zhang X et al. Strategy for
identification of sequence variants in COL
7A1 and a novel
2-bp deletion mutation in recessive dystrophic epidermolysis
bullosa. Hum Mutat 1997: 10: 408 – 414.
46. Heinonen S, Mannikko M, Klement JF, Whitaker-Menezes
D, Murphy GF, Uitto J. Targeted inactivation of the type
VII collagen gene (Col
7a1) in mice results in severe blistering
phenotype: a model for recessive dystrophic epidermolysis
bullosa. J Cell Sci 1999: 112: 3641 – 3648.
47. Pulkkinen L, Uitto J. Mutation analysis and molecular
genetics of epidermolysis bullosa. Matrix Biol 1999: 18:
29 – 42.
48. Winter AD, Page AP. Prolyl 4-hydroxylase is an essential
procollagen-modifying enzyme required for exoskeleton
formation and the maintenance of body shape in the
nematode Caenorhabditis elegans. Mol Cell Biol 2000: 20:
4084 – 4093.
49. Byers PH. Osteogenesis imperfecta. In: Royce PM, Stein-
mann B, eds. Connectivie Tissue and its Heritable Disorders.
Molecular, Genetic and Medical Aspects. New York: Wiley-
Liss, 1993: 317 – 350.
50. Sillence DO, Senn A, Danks DM. Genetic heterogeneity in
osteogenesis imperfecta. J Med Genet 1979: 16: 101 – 116.
51. Willing MC, Pruchno CJ, Byers PH. Molecular heterogene-
ity in osteogenesis imperfecta type I. Am J Med Genet 1993:
45: 223 – 227.
52. Chessler SD, Byers PH. BiP binds type I procollagen pro
chains with mutations in the carboxyl-terminal propeptide
synthesized by cells from patients with osteogenesis imper-
fecta. J Biol Chem 1993: 268: 18226 – 18233.
53. Chessler SD, Wallis GA, Byers PH. Mutations in the
carboxyl-terminal propeptide of the pro
a1(I) chain of type
I collagen result in defective chain association and produce
lethal osteogenesis imperfecta. J Biol Chem 1993: 268:
18218 – 18225.
54. Lamande SR, Chessler SD, Golub SB et al. Endoplasmic
reticulum-mediated quality control of type I collagen pro-
duction by cells from osteogenesis imperfecta patients with
mutations in the pro
a1(I) chain carboxyl-terminal propep-
tide which impair subunit assembly. J Biol Chem 1995: 270:
8642 – 8649.
279