EVOLUTION & DEVELOPMENT 8:4, 370 –377 (2006)

Molecular evolution of fibrillar collagen in chordates, with implications

for the evolution of vertebrate skeletons and chordate phylogeny

Hiroshi Wada,,a,b,c Makiko Okuyama,a,b Nori Satoh,d and Shicui Zhange

a
Graduate School of Life and Environmental Sciences, University of Tsukuba, Tsukuba 305-8572, Japan
b
Seto Marine Biological Laboratory, FSERC, Kyoto University, 459 Shirahama, Kyoto 649-2211, Japan
c
PRESTO, JST 4-1-8 Honcho, Kawaguchi 332-0012, Japan
d
Department of Zoology, Graduate School of Sciences, Kyoto University, Kyoto 606-8502, Japan
e
Department of Marine Biology, Ocean University of Qingdao, Qingdao 266003, China
Author for correspondence (email: hwada@biol.tsukuba.ac.jp)

SUMMARY Vertebrates have seven types of fibrillar
collagens that are encoded by 11 genes. Types I, V, and
XXIV collagens are components of mineralized bone, whereas
types II, XI, and XXVII collagens are components of cartilage.
In this study, we traced the molecular evolutionary history of
chordate collagen genes and examined how gene dupli-
cations gave rise to the collagen genes used for skeletons.
Our analyses of deuterostome collagen genes, including one
amphioxus gene that we identified in this study, suggest that
the common ancestors of deuterostomes possessed three
fibrillar collagen genes. Expression analyses of chordate
fibrillar collagen genes suggest that in the ancestors of
chordates, fibrillar collagen was co-opted to the formation of
the notochord sheath independently in three clades. Our
results also imply that co-option of collagen genes to cartilage
occurred in clade A (col2A1), clade B (col11A1, 11A2), and
clade C (COL27A1). Similarly, some fibrillar collagen genes
have been co-opted for mineralized bone independently from
clade A genes (col1A1, 1A2, 5A2), clade B genes (col5A1),
and clade C genes (COL24A1). These frequent co-options for
notochord, cartilage, and mineralized bone must have been
accompanied by the rapid evolution of cis-regulatory elements
for transcription. In addition, we found that one of the ascidian
fibrillar collagen genes possesses an amino acid insertion at
the identical site of the C-terminal noncollagenous domain in
vertebrate fibrillar collagen genes. This observation raises a
suspicion about the relatively well-accepted phylogeny of the
close relationship between amphioxus and vertebrates.

INTRODUCTION genes, types XXIV and XXVII, are expressed in mineralized

Gene duplication is one of the major driving forces that en-
ables living creatures to evolve novel structures (Ohno 1970;
Carroll et al. 2001). It is now widely accepted that vertebrates
experience two rounds of genome duplication (Furlong and
Holland 2002), as exemplified by their four Hox clusters,
whereas most invertebrates possess only a single Hox cluster.
It is likely that duplicated genes have contributed to the in-
novation of vertebrate-specific structures, such as a neural
crest and a complicated central nervous system (Wada
2001a, b).
Gene duplications also have important roles in the evolu-
tion of skeletons, which is one of the most important verte-
brate characteristics. Vertebrates have two types of skeletons:
cartilage and mineralized bone, whose major components are
different types of collagen: type II and type XI collagens in
cartilage, and type I and type V collagens in mineralized bone
(Miller 1984; Vuorio and de Crombrugghe 1990; Boot-Hand-
ford and Tuckwell 2003). Two recently identified collagen
bone and cartilage, respectively (Boot-Handford et al. 2003;
Koch et al. 2003; Pace et al. 2003). Therefore, gene duplica-
tions, through which several repertoires of collagens evolved,
were essential for the evolution of vertebrate skeletons. In
addition, during the early stages of chordate evolution, col-
lagen genes performed another important role in the evolu-
tion of the notochord, one of the characteristic features of
chordates. The notochord is a skeletal device that provides
support for the body along the anterior–posterior axis, and its
collagen sheath confers its mechanical strength.
Collagens are found in a wide variety of multi-cellular ani-
mals and are major components of extracellular matrices (re-
viewed by Boot-Handford and Tuckwell 2003). All collagen
molecules are composed of three polypeptide chains with a
characteristic Gly-X-Y repeat sequence that forms a triple
helical structure. Based on their supramolecular structure,
collagens can be classified into two classes: fibril-forming
(fibrillar) collagens and non-fibril-forming collagens (Vuorio
and de Crombrugghe 1990). Fibrillar collagen molecules are

370 & 2006 The Author(s)

Journal compilation & 2006 Blackwell Publishing Ltd.
Wada et al.
produced as procollagens, which contain noncollagenous do-
mains in both the N- and C-termini. The C-terminus noncol-
lagenous domain is essential for the formation of the triple
helix by zipping up the three polypeptides in the C-to-N di-
rection (Vuorio and de Crombrugghe 1990; Boot-Handford
and Tuckwell 2003). In addition, the C-terminus noncolla-
genous domain has an important role in selecting an appro-
priate partner for trimerization (Lees et al. 1997). After being
secreted to the extracellular space, the noncollagenous pro-
peptides of both ends are cleaved by specific proteinases.
Boot-Handford et al. (2003) and Aouacheria et al. (2004)
presented molecular phylogenetic analyses of the fibrillar col-
lagens, including those from invertebrates. Their analyses are
consistent with the conclusion that vertebrate fibrillar colla-
gens are classified into three clades. However, they disagreed
on two points. First, regarding the timing of the gene dupli-
cations that gave rise to the three clades, Boot-Handford et al.
(2003) suggested that the three clades appeared in the
common ancestors of vertebrates, whereas Aouacheria et al.
(2004) stressed that the gene duplications predate the diver-
gence of chordates, and that most invertebrates possessed fi-
brillar collagens of the three clades. Another disagreement is
over the timing of when the chain selection sequence in the C-
terminus noncollagenous domain evolved in the fibrillar col-
lagen genes. Originally, Boot-Handford and Tuckwell (2003)
proposed that the chain selection sequence evolved in the an-
cestral vertebrates, because only the vertebrate fibrillar colla-
gen genes possess the long stretches of the chain selection
sequence. As the clade C genes recently discovered in verte-
brates have short sequences, Boot-Handford et al. (2003)
suggested that the selection sequence has become shorter in
the clade C genes. Conversely, Aouacheria et al. (2004) found
that one of the ascidian fibrillar collagen genes has a long-
chain selection sequence. Therefore, they proposed that the
appearance of the longer chain selection sequence is not a rare
genomic event; rather, it is ‘‘the result of a long evolutionary
process.’’
In order to settle these issues, and to understand the link
between the evolutionary history of fibrillar collagen genes
and the evolution of skeletons in chordates, we further analy-
zed the molecular evolution of deuterostome fibrillar collagen
genes, including one gene from amphioxus that we identified
in this study.
MATERIALS AND METHODS
Isolation of fibrillar collagen genes
Basic local alignment search tool (BLAST) (Altschul et al. 1997)
was used to search for fibrillar collagen genes in the draft sequence
of the genome of Ciona intestinalis (Dehal et al. 2002) using an
amino acid sequence of the C-terminus noncollagenous domain of
human COL2A1 as the query. Four independent genetic loci were
Evolution of chordate collagen genes 371
found to contain sequences homologous to the C-terminus non-
collagenous domain. Here, the four Ciona fibrillar collagen genes
were designated as CiFCol1–4 (Ciona fibrillar collagen), which
correspond to Ci759, Ci301, Ci606, and Ci916 in Aouacheria et al.
(2004), respectively. Complete sequences of the C. intestinalis fi-
brillar collagen genes were recovered from the database, except for
CiFCol3, whose nucleotide sequence of the C-terminus noncolla-
genous domain was determined by using cDNA clone from the
Ciona Gene Collection (Satou et al. 2002), as it was not available
from the database. CiFCol1 is included in Scaffold 10, and the
corresponding cDNA sequence is found in the Ciona cDNA da-
tabase (Cluster 03573; Satou et al. 2002). CiFCol2 is located in
Scaffold 142, and the corresponding cDNA sequence could be re-
covered from the Ciona cDNA database (Cluster 00299; Satou et
al. 2002). CiFCol3 and CiFCol4 are included in Scaffolds 222
(Cluster 02162) and 116 (Cluster 05368), respectively. As several
exons were misassigned in the gene model of the database, we have
reconstructed the amino acid sequence referring the conserved
exon–intron structures (Aouacheria et al. 2004) and expressed se-
quence tag (EST) sequence data when available.
For amphioxus, a single fibrillar collagen gene was recovered
from either the EST data set (Panopoulou et al. 2003) or the draft
genome sequence available from NCBI (http://www.ncbi.nlm.nih.
gov/BLAST/mmtrace.shtml), and was designated BfFCol1. We
used the nucleotide sequence from the EST database (Cluster
00172) in the phylogenetic analyses.
Phylogenetic analyses
Phylogenetic analyses of the amino acid sequences of fibrillar col-
lagen genes were performed using TreePuzzle 5.0 (Schmidt et al.
2002), and confidence values for each node were calculated using
the quartet puzzling (QP) maximum likelihood (ML) analysis. The
following options were applied: exact estimation of parameters, g
distribution model for inter-site rate variation, and use of the VT
model (Muller and Vingron 2000) for amino acid substitution. Two
hundred and fifty-six amino acid sites of C-terminus noncollagen-
ous domain were used for the construction of the tree in Fig. 1A.
Seven hundred and two amino acid sites of the triple helix domain,
together with the 256 amino acid sites of the C-terminus noncol-
lagenous domain, were used for the construction of the tree in Fig.
1B. Sequence alignments are available upon request from H. W.
In situ hybridization
Adult specimens of C. intestinalis were collected in Shirahama,
Japan, and artificial fertilization was performed to obtain embry-
onic specimens for in situ hybridization. In situ hybridization of
Ciona embryos was performed following Yasuo and Satoh (1994).
Adult specimens of Pacific amphioxus, Branchiostoma belcheri,
were collected in Qingdao, China. Amphioxus embryos were col-
lected as described by Yasui et al. (1998a), and fixed for in situ
hybridization. We followed the methods of Yasui et al. (1998b) for
amphioxus in situ hybridization.
Probes for in situ hybridization were produced using respective
cDNA clones of four Ciona fibrillar collagen genes. For the probe
for the fibrillar collagen gene of B. belcheri, we amplified an
orthologous gene by RT-PCR from an adult specimen, referring to
the nucleotide sequence of BfFCol1. The primers used for ampli-
372 EVOLUTION & DEVELOPMENT Vol. 8, No. 4, July^August 2006

A B
Sea urchin (Para) Col6 Human COL11A2
85
100
Ascidian CiFCol2 Ascidian CiFCol2

Clade B
Human COL11A2 70 Human COL11A1

Clade B
83 99
Human COL5A3 Human COL5A1

56 Ascidian CiFCol4
Human COL11A1
96
71 Ascidian CiFCol3
Human COL5A1

Clade C
92 Human COL27A1
Sea urchin (Para) Col7 98
76
Human COL24A1
Ascidian CiFCol4
Sea urchin (Str) Col1

Clade C
51 Ascidian 99
CiFCol3
Sea urchin (Str) Col2
62 Human COL24A1
68
Human COL1A1
95
Human COL27A1
Human COL1A2

Clade A
Ascidian CiFCol1
Mouse col3a1
Sea urchin (Str) Col1
84 Human COL5A2
52 0.1
Sea urchin (Str) Col2
79 Ascidian CiFCol1
87
Sea urchin (Para) Col5
Human COL2A1
Amphioxus BfCol1
Clade A

Mouse col3a1

58
Human COL1A1

Human COL1A2
51

0.1 Human COL5A2

Human COL2A1

Fig. 1. Molecular phylogenetic trees of the deuterostome fibrillar collagen genes. The confidence values for each node were calculated using
quartet puzzling in TreePuzzle 5.0 (Schmidt et al. 2002). (A) Phylogenetic tree is constructed based on amino acid sequences of C-terminal
noncollagenous domain. (B) The tree is constructed from a triple helix domain and a C-terminus domain. As amino acid sequences and
exon–intron structures of the triple helix region are not available for some genes, we could not include all the genes in this analysis.

fication were as follows: forward 5 0 -CGGTCCCAACGGTCCC
GCTGGAATCCCCGG-3 0 , and reverse 50 -GCCGAATTCTTGG
TCTGGTCTGCCGATGTC-3 0 . The nucleotide sequence of this
collagen gene (designated BbFCol1, Acc. No. AB193827) was de-
termined and it was confirmed to be a type A fibrillar collagen gene
of B. belcheri, orthologous to BfFCol1.
RESULTS
Molecular evolution of fibrillar collagen in
chordates
Although Boot-Handford et al. (2003) and Aouacheria et al.
(2004) presented molecular phylogenetic analyses of fibrillar
collagen genes, including those from invertebrates, the results
are not consistent with respect to the timing of the gene du-
plications that gave rise to the three clades. This may be be-
cause substitutions are saturated when all invertebrates were
included in the analyses. Therefore, in this study, we focused
on the evolutionary history of collagens in deuterostomes,
and tried to understand the link between collagen molecular
evolutionary history and the evolution of skeletons and the
notochord. Together with the amphioxus fibrillar collagen
recovered from the EST database (Panopoulou et al. 2003),
molecular phylogenetic analyses of deuterostome fibrillar col-
lagen were performed using 256 amino acid sites of the C-
terminus noncollagenous domain.
Although resolution of the resultant tree is not satisfacto-
ry, the tree supports that deuterostome fibrillar collagen genes
might be classified into three groups (clades A–C; Fig. 1A).
The amphioxus fibrillar collagen gene (BfFCol1) is closely re-
lated to the vertebrate clade A genes. Together with one of the
ascidian genes (CiFCol1) and three sea urchin genes (StrCol1–
2 and ParaCol5), they form a monophyletic unit (87% QP
Wada et al. Evolution of chordate collagen genes 373

support). Therefore, these genes probably arose from a single
gene in the deuterostome ancestor. The invertebrate counter-
parts of the vertebrate clade B and clade C genes are not quite
as obvious from this tree, although an affinity between CiF-
Col3 and clade C genes of vertebrates is weakly supported. In
order to assign invertebrate counterparts of clade B and clade
C genes, we have performed molecular phylogenetic analyses
by adding amino acid sequences of a triple helix region. The
triple helix region is repetitive sequence of three amino acids
(Gly-X-Y), and thus it is not easy to align sequences confi-
dently. However, as Aouacheria et al. (2004) indicated, by
reference to an exon–intron structure, we could overcome this
problem, and could produce a reliable alignment. As amino
acid sequences and/or exon–intron structures of the triple he-
lix region are not available for some genes, we could not
include all the genes in this analysis. The resultant tree not
only supported the monophyly of the clade A but also suc-
cessfully assigned ascidian counterparts of clade B genes and
clade C genes (Fig. 1B). The tree supports the affinity of
vertebrate clade B genes with CiFCol2, and a close relation-
ship between vertebrate clade C genes and CiFCol3–4 genes is
also supported. Therefore, it is likely that the ancestors of
ascidians and vertebrates possessed three fibrillar collagen
genes. As this is an unrooted tree, we could not exclude the
possibility that two sea urchin genes (ParaCol6 and ParaCol7)
that cannot be included in this analysis form a monophyletic
group with StrCol1 and StrCol2, and it branched off from the
basal node of the three clades of chordate fibrillar collagen
genes. If this is the case, gene duplications for three clades
occurred in the ancestral chordates after they split from the
echinoderm lineage. However, we think it unlikely because
both ParaCol6 and ParaCol7 show affinities with respective
ascidian genes of clade B and clade C (Fig. 1A), which are
supported by a certain level of confidence values (85% and
76%, respectively). Therefore, the present result indicates that
the gene duplications that gave rise to three clades of fibrillar
collagen genes predate the divergence between vertebrates
and ascidians, and it is more likely that the gene duplication
predates the phylogenetic split between echinoderms and
chordates.
Regarding the chain selection sequence, we found that the
amphioxus clade A gene possesses the short type of chain
selection sequence, as in other nonchordate invertebrates, in-
cluding sea urchins (Fig. 2). Aouacheria et al. (2004) have
already reported that the ascidian clade A gene (CiFCol1) has
a long-chain selection sequence, whereas the rest of the genes
(CiFCol2-4) possess short sequences. Therefore, it is likely
that the long-chain selection sequence evolved convergently in
the two clades; that is, before the ascidian–vertebrate split
in clade A, and after the ascidian–vertebrate split in clade B
(Fig. 4). The implications of this result for chordate phylogeny
are discussed below.
Expression of the fibrillar collagen genes
To trace the evolutionary history of the co-option of fibrillar
collagen genes for vertebrate skeletons, the expression
patterns of the fibrillar collagen genes of ascidians and
amphioxus were analyzed. Expression of CiFCol1 was re-
stricted to the posterior part of the larval body, in the muscle
cells, mesenchyme, posterior nerve cord, posterior endoder-
mal strand, and notochord (Fig. 3, A and B). CiFCol2 showed
an expression pattern similar to that of CiFCol1 in the pos-
terior part of the larva, although the expression in the neural
tube was restricted to the floorplate (Fig. 3, C and D). CiF-
Col3 and CiFCol4 showed a similar expression in the noto-

A Human COL2A1 HIWFGETING GFHFSYG--D DNLAPNTA-N VQMTFLRLLS TEGSQNITYH

Mouse col3a1 HVWFGESMNG GFQFSYG--P PDLPEDVV-D VQLAFLRLLS SRASQNITYH
Human COL1A1 HVWFGESMTD GFQFEYG--G QGSDPADV-A IQLTFLRLMS TEASQNITYH
Human COL1A2 HVWLGETINA GSQFEYN--V EGVTSKEM-A TQLAFMRLLA NYASQNITYH
Human COL5A2 PVWYGLDMNR GSQFAYG--D HQSP-NTA-I TQMTFLRLLS KEASQNITYI
Amphioxus BfFCol1 HVWFGDSMKG GYHFTYT--A H--------E IQMQFLRLSS TGARQNITYQ
Ascidian CiFCol1 HVYYGEEMAT GSTVRYDPSD EDRVPHADYT SQLTFLRLLS RQVKQHVTFY
Sea urchin (Str) Col2 GSRYITEM-G LEKFSYE--A S--------E VQLTFLRLLS TKAHQNVTYH
Sea urchin (Para) Col5 GHRYISEMNG LEKFTYD--A S--------E VQVTFLRLLS NKAHQNVTYH
Sea urchin (Str) Col1 KRAFFSSMHG GDKFAYI--E D--------S TQMTFLRLLS TSARQTVTYF Fig. 2. Amino acid insertions in
the C-terminus noncollagenous
B Human COL11A1 GSWFSEFKRG -KLLSYL--D VEGNSIN--M VQMTFLKLLT ASARQNFTYH domain of fibrillar collagen
Human COL5A1 GSWFSEFKRG -KLLSYV--D AEGNPVG--V VQMTFLRLLS ASAHQNVTYH genes of deuterostomes. Verte-
Human COL5A13 GGWYSTFRRG -KKFSYV--D ADGSPVN--V VQLNFLKLLS ATARQNFTYS brate genes in clade A and clade
Mouse col11a2 ---------- ---FSYV--D SEGSPVG--V VQLTFLRLLS VSAHQDVSYP B possess six to seven insertions
Ascidian CiFCol2 GDWYSSYRLG -EKFDYN--T T--------I PQYNFLRLLS SHARQRFTYK in the chain selection sequence,
Sea urchin (Para) Col6 GEWFLENLRG -KEFSYQ--G G--------I VQLTFLRLLS SKATQKFTYN whereas clade C genes do not
(boxed). All invertebrate genes,
C Human COL24A1 ---------- -KLEF-G--V G--------K VQMNFLHLLS SEATHIITIH except for the ascidian clade A
Human COL27A1 ---------- -KVEF-A--I S--------R VQMNFLHLLS SEVTQHITIH gene (CiFCol1), lack the in-
Ascidian CiFCol3 GARVGEGRPG -VNGDSL--- ---------T SQLRFLRLQS AEAEQVISFQ sertion. The chain selection se-
Ascidian CiFCol4 YLWWRRYKGG -HMINYS--A G--------K VQLKILRYFS KFAVQKFKFK
quence (Lees et al., 1997) is un-
Sea urchin (Para) Col7 TDWFSGWNNG -FEFEYE--A N--------V VQLRFLHMKS NIATQLFTFN
derlined.
374 EVOLUTION & DEVELOPMENT Vol. 8, No. 4, July^August 2006
chord (Fig. 3, E and F). CiFCol4 was expressed in the ad-
hesive palp (Fig. 3F). Expression of BbFCol1 was observed in
the notochord and floorplate cells of the neural tube (Fig.
3G). Notably, floorplate expression of BbFCol1 was not ob-
served in the cerebral vesicle (Fig. 3H), which is comparable
to the vertebrate brain (Williams and Holland 1998; Wada
and Satoh 2001), although notochord expression extended to
the anterior tip.
DISCUSSION
Molecular evolution of fibrillar collagen genes
The molecular evolutionary history of fibrillar collagen genes
in deuterostomes is summarized in Fig. 4. The present results
Fig. 3. Expression pattern of fibril-
lar collagen genes in protochor-
dates. (A–F) Expression pattern of
fibrillar collagen in a Ciona tailbud
embryo. CiFCol1 (A, B) and CiF-
Col2 (C, D) are expressed in
the nerve cord, notochord, muscle,
mesenchyme, and endodermal
strand. Note that the expression
of CiFCol2 in the neural tube is
restricted to the floorplate (D),
whereas CiFCol1 is expressed in
the dorsal part of the neural tube
as well (B). nt, neural tube.
Expression of CiFCol3 (E) and
CiFCol4 (F) is observed in the
notochord, and CiFCol4 is also
expressed in pulp (arrow in F). (G,
H) Expression of BbFCol1 is ob-
served in the notochord and floor-
plate cells of the neural tube. The
anterior tip of floorplate expres-
sion is at the level of the posterior
end of the cerebral vesicle (arrow-
head in F), whereas notochord ex-
pression extends to the anterior
tip.
provide strong support for the idea that the gene duplications
for three clades of fibrillar collagen genes predate divergence
of chordate groups. Although we cannot be completely cer-
tain whether the gene duplications for the three clades predate
the divergence between chordates and echinoderms, ParaCol6
and ParaCol7 show affinities with ascidian counterparts in
clade B and clade C. Therefore, we prefer to interpret our
results as suggesting that the ancestral deuterostomes pos-
sessed three fibrillar collagen genes, which are the origin of the
three clades of vertebrate fibrillar collagen genes.
Based on the evolutionary scheme of fibrillar collagens in
deuterostomes, we mapped functional innovations of chor-
date fibrillar collagens. In the lineage leading to chordates,
fibrillar collagens contributed to the evolution of one of the
most important chordate characteristics: the notochord. As
Wada et al. Evolution of chordate collagen genes 375

ol1
l1

l2

Se an C ol3

l4
5
FC

l7
Co

rch FCo

rch FCo
l2/
rch ol1

l6

FC

Co
Bf
iF

Co

Co
C

Ci

Ci

i
Am ian C

in
us

in

in

in
i an

i an
iox

A2

Co 1

A1

As A1
1

r ch
A2

As 1

As 3
A

i
l3A
l1A

l5A

l2A

l5A

l5A
ci d

ci d

ci d

ci d
au
l11

l11

l24

l27
ph
au

au

au
l1

As
Co
Co

Co

Co

Co

Co
Co

Co

Co

Co
Se

Se

Se
Fig. 4. Schematic illustration of
the molecular evolution of chor-
date fibrillar collagen genes. The
insertion of clade A collagen
Clade A Clade B Clade C
Co-option to Notochord genes of vertebrates and as-
cidians might support a phylo-
Co-option to bone
genetic affinity between these
Co-option to cartilage two taxa. Co-option to cartilage
and mineralized bone occurred
Long chain selection sequnce
independently in three clades.

all the fibrillar collagens of ascidian and amphioxus from the
three clades are expressed in the notochord, the present results
suggest that fibrillar collagen genes were co-opted for the
notochord in the three clades independently, although this
idea should be confirmed by more robust evidences about the
phylogenetic positions of echinoderm fibrillar collagen genes.
The mechanical strength provided by the fibrillar collagens
must have been essential for the evolution of a tadpole-like
body plan, because without a notochord, a tadpole cannot
swim by beating its muscular tail. This function of fibrillar
collagen in the notochord sheath was conferred by type II and
type XI collagens, as these two genes are major components
of the vertebrate notochord sheath (Miller 1984; Vuorio and
de Crombrugghe 1990; Boot-Handford and Tuckwell 2003).
Interestingly, we can observe similar independent co-op-
tions of fibrillar collagens in the evolutionary stage of ances-
tral vertebrates. The gene duplications that gave rise to
multiple genes in vertebrates in each clade probably occurred
through two rounds of genome-wide duplication in ancestral
vertebrates. These genome duplications gave rise to four Hox
clusters. Five collagen genes of clade A are linked to Hox
clusters (as COL5A2 and 3A1 are located next to each other,
linked with the HoxD cluster, five genes link four Hox clus-
ters) (Bailey et al. 1997; Boot-Handford and Tuckwell 2003).
Type II collagen is a main component of cartilage, and type
XI and type XXVII are also included. Therefore, the co-op-
tion of collagen genes to cartilage occurred three times inde-
pendently in clade A (COL2A1), clade B (COL11A1, 11A2),
and clade C (COL27A1). In the other type of skeleton,
mineralized bone, type I collagens are included as a main
component, together with type V and type XXIV as minor
components. Therefore, similar to the case of cartilage, three
independent co-options occurred in the evolutionary history
of mineralized bone: three type A genes (COL1A1, 1A2, 5A2),
one type B gene (COL5A1), and one type C gene (COL24A1)
were co-opted for mineralized bone. It would be intriguing to
determine how these functional linkages (four genes for car-
tilage and five genes for mineralized bone) evolved in genes
with different phylogenetic histories. In this respect, it is note-
worthy that the expression of col2a1 and col11a2 in cartilage
is controlled by the same transcription factor: Sox9 (Bell et al.
1997; Lefebvre et al. 1997; Liu et al. 2000). This observation
may raise a possibility that Sox9 controls the ancestral col-
lagen genes before the gene duplications occurred, and that
descendant genes inherit regulation by Sox9. However, Sox9
and fibrillar collagen genes do not show overlapping expres-
sion either in ascidian or amphioxus (H. Wada, manuscript in
prep.). Therefore, we think it more likely that the transcrip-
tional network for cartilage differentiation is built up via a
molecular evolutionary history in which several fibrillar col-
lagen genes independently acquired cis elements controlled by
cartilage-specific transcriptional factors, such as Sox9. These
observations of the independent co-option of fibrillar collagen
genes for the notochord, cartilage, and mineralized bone are
consistent with evidence that the cis-regulatory element can
evolve faster than has previously been thought to be the case
(Stone and Wray 2001; Wray et al. 2003).
Some of the Ciona fibrillar collagen genes (CiFCol1-2) may
have acquired additional functions in the larval muscle, neural
tube, and endoderm. CiFCol4 also appears to have acquired a
376 EVOLUTION & DEVELOPMENT Vol. 8, No. 4, July^August 2006
new role in palp formation. It is also likely that the four
fibrillar collagen genes play additional roles during the adult
stage. Although we do not know how the four types of fi-
brillar collagens of Ciona are coordinated in collagen fibers,
similar expression patterns between CiFCol1 and CiFCol2
and between CiFCol3 and CiFCol4 suggest functional link-
ages between these genes.
Convergent insertion of vertebrate-specific amino
acid residues and implications for chordate
phylogeny
Boot-Handford and Tuckwell (2003) found that vertebrate
fibrillar collagen genes possess specific insertions of six to
seven amino acid residues in the C-terminus noncollagenous
domain. As these amino acids are found in the region that is
involved in the selection of proper partners during trimerizat-
ion (Lees et al. 1997), the insertion of these amino acid res-
idues has been suggested to be advantageous for vertebrates
that have multiple collagen genes. However, we found that
although one of the ascidian collagen genes (CiFCol1) in-
cluded in clade A possesses an amino acid insertion at the site
(Aouacheria et al. 2004), the amphioxus clade A gene lacks
the insertion. As only vertebrate genes possess the insertion in
clade B, it is likely that these residues were inserted conver-
gently in the two clades, that is, before the ascidian–vertebrate
split in clade A, and after the ascidian–vertebrate split in clade
B. If this were the case, the absence of the amino acid residues
in amphioxus gene BfFCol1 supports the idea that vertebrates
are more closely related to ascidians than to amphioxus.
Aouacheria et al. (2004) pointed out that the amino acid se-
quences surrounding the chain selection sequence are diver-
gent, and they suggested that evolution of the long-chain
selection sequence is not a rare genomic event. However, it
should also be noted that the amino acid sequences of inver-
tebrate fibrillar collagens are confidently aligned around this
region, and all invertebrate genes possess an identical number
of amino acids (Boot-Handford and Tuckwell 2003). In ad-
dition, the insertion site does not correspond to an exon–
intron boundary. Therefore, we think that although amino
acid sequences are divergent around the region, the insertion
of amino acid residues is in fact a rare genomic event, which
can be a phylogenetic index. Although our result prefers a
close relationship between ascidians and vertebrates, it may
not be strong enough evidence alone to support this conclu-
sion. However, there is some evidence that supports the
phylogenetic affinity between ascidians and vertebrates. Given
that 18S rDNA molecular phylogenetic studies support the
traditional phylogenetic framework of the amphioxus–verte-
brate affinity in chordates (Wada and Satoh 1994; Turbeville
et al. 1994), this phylogeny has been well accepted. In con-
trast, Oda et al. (2002) have argued in support of the ascidian–
vertebrate affinity, based on cadherin gene structure. Recent-
ly, Philippe et al. (2005) and Blair and Hedges (2005) reached
the same conclusion from their multi-gene analyses. In addi-
tion, Wada (2001a, b) has pointed out that some of the genes
involved in neural development show similar expression pat-
terns in ascidians and vertebrates, but are distinct from their
amphioxus counterparts. Compared with amphioxus, uro-
chordates may have accumulated more autapomorphic char-
acters, as exemplified by fast embryogenesis, lack of feeding in
the tadpole-larva stage, and the presence of a tunic whose
main component is cellulose (Satoh 1994). The absence of
segmentation may be linked to the former two autapomorp-
hies. Owing to these autapomorphies, ascidians seem to be
more primitive superficially. However, we should note that as
long as they are autapomorphic, these characters are phylo-
genetically uninformative. The 18S rDNA molecular phylo-
geny fails to recover the monophyly of chordates, although
the affinity between amphioxus and vertebrates is always
supported (Wada and Satoh 1994; Turbeville et al. 1994;
Bromham and Degnan 1999). Even combined with 28S
rDNA sequences, only weak support is obtained for chordate
monophyly (Winchell et al. 2002). The relatively long branch-
es leading to the ascidian lineage might make it difficult to
infer the phylogenetic relationship of ascidians from rDNA
sequence data (Blair and Hedges 2005). Rather, several lines
of recent evidences support the affinity between ascidians and
vertebrates.
Ackowledgments
We thank Koji Hotta and Hiroki Takahashi for helpful discussions
about Ciona collagen genes. This work is supported by Toray Science
and Technology Grant, Narishige Science Award, and Grants-in-Aid
from the ministry of Education, Science, Sports and Culture, Japan
(Nos. 17018018 and 16027226) to H. W.
REFERENCES
Altschul, S. F., et al. (1997). Gapped BLAST and PSI-BLAST: a new
generation of protein database search programs. Nucleic Acids Res. 25:
3389–3402.
Aouacheria, A., Cluzel, C., Lethias, C., Gouy, M., Garrone, R., and Ex-
posito, J. Y. (2004). Invertebrate data predict an early emergence of
vertebrate fibrillar collagen clades and an anti-incest model. J. Biol.
Chem. 279: 47711–47719.
Bailey, W. J., Kim, J., Wagner, G. P., and Ruddle, F. H. (1997). Phylo-
genetic reconstruction of vertebrate Hox cluster duplications. Mol. Biol.
Evol. 14: 843–853.
Bell, D. M., et al. (1997). SOX9 directly regulates the type-II collagen gene.
Nat. Genet. 16: 174–178.
Blair, J. E., and Hedges, S. B. (2005). Molecular phylogeny and divergence
tomes of deuterostomes animals. Mol. Biol. Evol. 22: 2275–2284.
Boot-Handford, R. P., and Tuckwell, D. S. (2003). Fibrillar collagen: the
key to vertebrate evolution? A tale of molecular incest. BioEssays 25:
142–151.
Boot-Handford, R. P., Tuckwell, D. S., Plumb, D. A., Rock, C. F., and
Poulsom, R. (2003). A novel and highly conserved collagen
(pro(alpha)1(XXVII)) with a unique expression pattern and unusual
molecular characteristics establishes a new clade within the vertebrate
fibrillar collagen family. J. Biol. Chem. 278: 31067–31077.
Wada et al.
Bromham, L. D., and Degnan, B. M. (1999). Hemichordates and deuteros-
tome evolution: robust molecular phylogenetic support for a hemichor-
date 1 echinoderm clade. Evol. Dev. 1: 166–171.
Carroll, S. B., Greiner, J. K., and Weatherbee, S. D. (2001). From DNA to
Diversity. Blackwell, Malden.
Dehal, P., et al. (2002). The draft genome of Ciona intestinalis: insights into
chordate and vertebrate origins. Science 298: 2157–2167.
Furlong, R. F., and Holland, P. W. H. (2002). Were vertebrates octoploid?
Philos. Trans. R. Soc. London B 357: 531–544.
Koch, M., et al. (2003). Collagen XXIV, a vertebrate fibrillar collagen with
structural features of invertebrate collagens: selective expression in de-
veloping cornea and bone. J. Biol. Chem. 278: 43236–43244.
Lees, J. F., Tasab, M., and Bulleid, N. J. (1997). Identification of the mo-
lecular recognition sequence which determines the type-specific assembly
of procollagen. EMBO J. 16: 908–916.
Lefebvre, V., Huang, W., Harley, V. R., Goodfellow, P. N., and de
Crombrugghe, B. (1997). SOX9 is a potent activator of the chondrocyte-
specific enhancer of the pro alpha1(II) collagen gene. Mol. Cell. Biol. 17:
2336–2346.
Liu, Y., Li, H., Tanaka, K., Tsumaki, N., and Yamada, Y. (2000). Iden-
tification of an enhancer sequence within the first intron required for
cartilage-specific transcription of the alpha2(XI) collagen gene. J. Biol.
Chem. 275: 12712–12718.
Miller, E. J. (1984). Chemistry of the collagens and their distribution. In K.
A. Piez and A. H. Reddi (eds.). Extracellular Matrix Biochemistry. Else-
vier, New York.
Muller, T., and Vingron, M. (2000). Modeling aminoacid replacement. J.
Comput. Biol. 7: 761–776.
Oda, H., Wada, H., Tagawa, K., Akiyama-Oda, Y., Satoh, N., Zhang, S.,
and Tsukita, S. (2002). A novel amphioxus cadherin that localizes to
epithelial adherens junctions has an unusual domain organization with
implication for chordate phylogeny. Evol. Dev. 4: 426–434.
Ohno, S. (1970). Evolution by Gene Duplication. Springer-Verlag, New
York.
Pace, J. M., Corrado, M., Missero, C., and Byers, P. H. (2003). Identifi-
cation, characterization and expression analysis of a new fibrillar colla-
gen gene COL27A1. Matrix Biol. 22: 3–14.
Panopoulou, G., et al. (2003). New evidence for genome-wide duplications
at the origin of vertebrates using an amphioxus gene set and completed
animal genomes. Genome Res. 13: 1056–1066.
Philippe, H., Lartillot, N., and Brinkmann, H. (2005). Multigene
analyses of bilaterian animals corroborate the monophyly of
Ecdysozoa, Lophotrochozoa, and Protostomia. Mol. Biol. Evol. 22:
1246–1253.
Evolution of chordate collagen genes 377
Satoh, N. (1994). Developmental Biology of Ascidian. Cambridge University
Press, Cambridge.
Satou, Y., et al. (2002). A cDNA resource from the basal chordate Ciona
intestinalis. Genesis 33: 153–154.
Schmidt, H. A., Strimmer, K., Vingron, M., and von Haeseler, A. (2002).
TREE-PUZZLE: maximum likelihood phylogenetic analysis using quar-
tets and parallel computing. Bioinformatics 18: 502–504.
Stone, J. R., and Wray, G. A. (2001). Rapid evolution of cis-regulatory
sequences via local point mutations. Mol. Biol. Evol. 18: 1764–1770.
Turbeville, J. M., Schulz, J. R., and Raff, R. A. (1994). Deuterostome
phylogeny and the sister group of the chordates: evidence from molecules
and morphology. Mol. Biol. Evol. 11: 648–655.
Vuorio, E., and de Crombrugghe, B. (1990). The family of collagen genes.
Annu. Rev. Biochem. 59: 837–872.
Wada, H. (2001a). Origin and evolution of the neural crest: a hypothetical
reconstruction of its evolutionary history. Dev. Growth Differ. 43: 509–
520.
Wada, H. (2001b). The origin of the neural crest and insights into evolution
of the vertebrate face. In H. Sawada, H. Yokosawa, and C. C. Lambert
(eds.). The Biology of Ascidians. Springer-verlag, Tokyo.
Wada, H., and Satoh, N. (2001). Patterning the protochordate neural tube.
Curr. Opin. Neurobiol. 11: 16–21.
Wada, H., and Satoh, N. (1994). Details of the evolutionary history from
invertebrates to vertebrates, as deduced from the sequences of 18S
rDNA. Proc. Natl. Acad. Sci. USA 91: 1801–1804.
Williams, N. A., and Holland, P. W. H. (1998). Molecular evolution of the
brain of chordates. Brain Behav. Evol. 52: 177–185.
Winchell, C. J., Sullivan, J., Cameron, C. B., Swalla, B. J., and Mallat, J.
(2002). Evaluating hypotheses of deuterostome phylogeny and chordate
evolution with new LSU and SSU ribosomal DNA data. Mol. Biol. Evol.
19: 762–776.
Wray, G. A., Hahn, M. W., Abouheif, E., Balhoff, J. P., Pizer, M., Rock-
man, M. V., and Romano, L. A. (2003). The evolution of transcriptional
regulation in eukaryotes. Mol. Biol. Evol. 20: 1377–1419.
Yasui, K., Tabata, S., Ueki, T., Uemura, M., and Zhang, S. (1998a). Early
development of the peripheral nervous system in a lancelet species. J.
Comp. Neurol. 393: 415–425.
Yasui, K., Zhang, S., Uemura, M., Aizawa, S., and Ueki, T. (1998b). Ex-
pression of a twist-related gene, Bbtwist, during the development of a
lancelet species and its relation to cephalochordate anterior structures.
Dev. Biol. 195: 49–59.
Yasuo, H., and Satoh, N. (1994). An ascidian homolog of the mouse Bra-
chyury (T) gene is expressed exclusively in notochord cells at fate re-
stricted stage. Dev. Growth Differ. 36: 9–18.