Báo cáo khoa học: Structural and functional evidence for a singular repertoire of BMP receptor signal transducing proteins in the lophotrochozoan Crassostrea gigas suggests a shared ancestral BMP/activin pathway docx

Structural and functional evidence for a singular
repertoire of BMP receptor signal transducing proteins in
the lophotrochozoan Crassostrea gigas suggests a shared
ancestral BMP/activin pathway
Amaury Herpin
1,2
, Christophe Lelong
2
, Thomas Becker
1
, Frederic Rosa
3
, Pascal Favrel
2
and Charles Cunningham
1
1 Sars International Centre for Marine Molecular Biology, High Technology Centre, Bergen, Norway
2 Laboratoire de Biologie et Biotechnologies Marines, IBFA, Universite
´
de Caen Basse-Normandie, IFREMER UMR 100, Physiologie et
Ecophysiologie des mollusques marins, Caen, France
3 U 368 INSERM, Ecole Normale Supe
´
rieure, Paris, France
The genes governing mesoderm specification have been
extensively studied in vertebrates, arthropods and nem-
atodes. The latter two phyla belong to the ecdysozoan
clade but little is understood of these molecules in the
other major protostomal clade, the lophotrochozoa.
An increasing amount of comparative data from
ecdysozoans as well as from vertebrates suggests

that many of the proteins involved in mesodermal
Keywords
Crassostrea gigas; zebrafish; BMP;
TGF-beta; early embyogenesis
Correspondence
A. Herpin, University of Wuerzburg,
Physiological Chemistry I, Am Hubland,
97074 Wuerzburg, Germany
Fax: +49 931888 4150
Tel: +49 931888 4165
E-mail: amaury.herpin@biozentrum.
uni-wuerzburg.de
(Received 15 April 2005, accepted 12 May
2005)
doi:10.1111/j.1742-4658.2005.04761.x
The transforming growth factor b (TGF-b) superfamily includes bone mor-
phogenetic proteins, activins and TGF-b sensu stricto (s.s). These ligands,
which transduce their signal through a heteromeric complex of type I and
type II receptors, have been shown to play a key role in numerous biologi-
cal processes including early embryonic development in both deuterostomes
and ecdyzozoans. Lophochotrozoans, the third major group of bilaterian
animals, have remained in the background of the molecular survey of
metazoan development. We report the cloning and functional study of the
central part of the BMP pathway machinery in the bivalve mollusc Cras-
sostrea gigas (Cg-BMPR1 type I receptor and Cg-TGFbsfR2 type II recep-
tor), showing an unusual functional mode of signal transduction for this
superfamily. The use of the zebrafish embryo as a reporter organism
revealed that Cg-BMPR1, Cg-TGFbsfR2, Cg-ALR I, an activin Type I
receptor or their dominant negative acting truncated forms, when over-
expressed during gastrulation, resulted in a range of phenotypes displaying

severe disturbance of anterioposterior patterning, due to strong modula-
tions of ventrolateral mesoderm patterning. The results suggest that
Cg-BMPR1, and to a certain degree Cg-TGFbsfR2 proteins, function in
C. gigas in a similar way to their zebrafish orthologues. Finally, based on
phylogenetic analyses, we propose an evolutionary model within the com-
plete TGF-b superfamily. Thus, evidence provided by this study argues for
a possible conserved endomesoderm ⁄ ectomesoderm inductive mechanism in
spiralians through an ancestral BMP ⁄ activin pathway in which the singu-
lar, promiscuous and probably unique Cg-TGFbsfR2 would be the shared
type II receptor interface for both BMP and activin ligands.
Abbreviations
BMP, bone morphogenetic protein; BMPR2, type II BMP receptors; TGF-b, transforming growth factor b.
3424 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
patterning are highly conserved with respect to both
structure and function, regardless of diversity and evo-
lution of body plans [1–5].
The transforming growth factor b (TGF-b) super-
family, which includes bone morphogenetic proteins
(BMPs), activin and activin-like proteins such as nodal
and their receptors, has been implicated in multiple
processes during animal development. Members of the
TGF-b superfamily transduce signals through hetero-
meric complexes of ligand specific type I and II ser-
ine ⁄ threonine kinase receptors [6]. Type II receptors
are capable of binding ligand dimers alone, while type
I receptors can only bind ligands in cooperation with
type II receptors. Ligand binding induces the forma-
tion of a heterotetracomplex in which the two type II
receptors unidirectionally transphosphorylate a dimer
of type I receptors. Activated type I receptors in turn

catalyse the phosphorylation of receptor substrates,
the Smads. Smad family members were originally iden-
tified through genetic screens in flies (mad Drosophila
mutants) and worms (sma Caenorhabditis mutants).
These move to the nucleus to associate with transcrip-
tional coactivators and regulate the transcription of
target genes [7]. While this pathway is conserved for
most TGF-b superfamily ligands, including BMPs and
activin, nodal binds the activin specific type I receptor
and the cripto coreceptor to stimulate downstream
responses [8,9]. In the absence of cripto, the type I act-
ivin receptor can mediate signal transduction stimula-
ted by activin but not nodal. Mutations in the gene
encoding the mouse type IB activin receptor, ActRIB,
as well as the ActRIIA ⁄ ActRIIB double mutants, dis-
play gastrulation defective phenotypes resembling
those of mouse nodal mutants [10–12].
In Drosophila, decapentaplegic (DPP), screw and a
third BMP ligand Gbb appear to share a common set
of receptors that include the type II receptor punt and
the type I receptors thick veins and saxophone
(reviewed in [13]). The activin type I receptor baboon
also signals in conjunction with punt, though the acti-
vin pathway appears to have little influence on pattern-
ing [14]. While punt appears most closely related to the
vertebrate type II activin receptors, another receptor
(wishful thinking) has been identified that is homolog-
ous to the vertebrate type II BMP receptors (BMPR2).
Vertebrate BMPR2 receptors are the only ones that
bind BMP ligands exclusively, and in Drosophila phe-

notypes arising from mutations in the gene encoding
wishful thinking, wit , suggest a role for this protein in
synapse regulation and ⁄ or maintenance [15,16].
As part of an ongoing project to understand the role
of the TGF-b superfamily ligands, their receptors and
signal transduction pathways in the lophotrochozoan
bivalve mollusc Crassotrea gigas, we report the cloning
and functional study of the central part of the BMP
pathway (the Cg-BMPR1 type I receptor and
Cg-TGFbsfR2 type II receptor). This shows probably
the most ancestral and unusual functional mode of sig-
nal transduction for this superfamily, with a duplicate
extracellular ligand binding domain TGFbsfR2 type II
homologous receptor displaying a unique extracellular
structure. Because technical limitations relative to our
model make direct functional studies difficult, we have
tested whether Cg-BMPR1 and Cg-TGFbsfR2 mole-
cules can function in the context of a vertebrate TGF-b
superfamily signalling pathway by overexpressing them
during zebrafish early embryogenesis. The molecular
nature of dorsoventral and anteroposterior patterning
in molluscs is discussed, in the context of Cg-BMPR1
and Cg-TGFbsfR2 expression patterns during C. gigas
early development.
One piece of evidence from this study suggests that
the molecular mechanisms controlling mesodermal pat-
terning across all bilateria may be conserved through a
complete, original and functional BMP ⁄ activin path-
way in lophotrochozoans, for which a singular and
promiscuous type II receptor would be the shared

interface for both BMP and activin ligands.
Results
Type I and II TGFb superfamily receptor ortho-
logues from C. gigas
Four full length cDNA clones were obtained that
encode orthologues of three type I and one type II
TGFb superfamily receptor(s) from the oyster C. gigas.
Clones encoding a type 1 activin-like receptor (Cg-
ALR1: accession number AJ309316) as well as a
TGFb sensu stricto type I-like receptor (Cg-TGFbR1:
accession number CAD66433) have been described
previously [17,18]. These clones will not be discussed
in detail here, apart from within the phylogenetic and
functional (Cg-ALR1) analyses of the receptor family.
The characteristics of the two remaining cDNA clones
and the proteins their sequences infer are discussed
below.
A full length 1907 base pair cDNA clone containing
an open reading frame encoding 534 amino acids was
isolated from a C. gigas mantle edge library. The pre-
dicted protein contained a number of features charac-
teristic of BMP type 1 receptors [19]. The protein,
named Cg-BMPR1, comprises a leader peptide, an
extracellular domain containing 10 cysteines whose
positions are conserved in comparison to those of
vertebrate BMP type I receptors, and a CCX
(5)
CN
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3425

cysteine knot preceding the transmembrane region
(Fig. 1A). A glycine-serine domain (GS box) did not
follow the canonical SGSGSGLP consensus sequence
but rather was encoded by a SSGCGSGPP motif. The
remaining intracellular catalytic domains are highly
conserved. Membership of Cg-BMPR1 to the BMP
type 1 receptor subfamily is clearly established by the
sequence of the L45 loop kinase domain. The
Cg-BMPR1 L45 loop sequence differs from the canon-
ical sequence (A
SDIKGT ⁄ NGSW) by only a single
residue (underlined). This motif plays an important
role across phyla in determining the specificity of type
I receptors for Smad proteins [20]. The gene and
inferred protein sequence of Cg-BMPR1 has been
lodged in the GenBank database with the accession
number AJ577293.
The second full length cDNA clone encoded a pro-
tein with an expected length of 1174 amino acids. The
inferred protein sequence bore most resemblance to
TGF-b superfamily type 2 receptors and was thus
named Cg-TGFbsfR2. Interestingly, the extracellular
domain of Cg-TGFbsfR2 is structurally divergent from
all other type II receptors that have been described
previously. Uniquely, it contains two extracellular
domains that we have named C1 and C2. C1 defines
the domain closest to the amino terminal end and C2
defines the domain closest to the cell membrane. A
comparison of the inferred amino acid sequence of the
C1 and C2 domains reveals that only the approximate

spacing of the 10 cysteine residues is conserved
(Fig. 1B). Both domains contain a typical BMP ⁄ activin
type II receptor CCCX
(4)
CN cysteine knot at their
N-terminal ends (Fig. 1B). Phylogenetic analyses com-
paring the extracellular domains of BMP ⁄ activin type
II receptors, C. gigas C1 and C2 regions, as well as
sequences from sponge receptors for which such extra-
cellular dupli ⁄ triplications are observed, showed that
Cg-TGFbsfR2 C1 and C2 domains were not clustering
together but were branching at the root defining BMP
and activin type II receptor clades (Fig. 2A). This
characteristic was also shared with the duplicated
domains of the sponge Ephydatia fluviatilis ALK-6
type II receptor (Fig. 2A). The C1 and C2 domains
were not directly adjacent but were joined by a linker
sequence. The intracellular kinase domain conforms to
the canonical sequence of serine ⁄ threonine protein kin-
ase domains seen for these receptors, and exhibits a
singularly long C-terminal extension similar to many
BMP type 2 receptors [21].
A more general phylogenetic tree was generated
using a conserved kinase cytoplasmic protein sequence
of all four C. gigas receptors together with selected
protostome and deuterostome TGF-b superfamily
receptor orthologues (Fig. 2B). Cg-BMPR1 and Cg-
TGFbsfR2 cluster reliably with BMP type I and II
receptors, respectively, and are closely associated with
the Drosophila orthologues thick vein and wishful think-

ing, respectively. Cg-ALR1 and Cg-TGFbR1 were
most closely related to activin and TGF-b type I recep-
tors, respectively.
The intron–exon organization of the genes encoding
Cg-BMPR1 and Cg-TGFbsfR2 is shown in Fig. 2C.
The serine ⁄ threonine kinase domain in both proteins is
encoded by two exons equivalent to kinase subdomains
X and XI [22]. In addition, the GS box and L45 loop
of Cg-BMPR1, as well as the C-terminal extension
of Cg-TGFbsfR2, are encoded by unique exons. The
C1 and C2 domains of the extracellular region of
Cg-TGFbsfR2 are each encoded by one or two exons.
Interestingly these two domains are separated by a
short linker encoded by its own exon (Figs 1B and
2B). Both genes show high levels of phase conservation
in comparison to other oyster TGFb superfamily re-
ceptors as well as orthologous receptors from other
species (data not shown).
Expression patterns of Cg-BMPR1 and
Cg-TGFbsfR2 in adult tissues, during early
embryogenesis and larval development
The early origin and high degree of conservation of
TGF-b signalling protein orthologues during animal
evolution from radiata to highly evolved bilateria sug-
gest that they are involved in key biological processes
common to most metazoans [23]. To gain insight into
a possible physiological role of Cg-BMPR1 and
Cg-TGFbsfR2, temporal gene expression patterns in
early larval developmental stages and adult tissues
were investigated using real time quantitative PCR

(Fig. 3). mRNAs from adult tissues (haemocytes, man-
tle edge, adductor muscle, digestive tract, gills, heart
and labial palp) including female gonads (oocytes),
and from various stages of embryonic and larval devel-
opment (blastula, gastrula, trochophore larvae, D lar-
vae, 7 and 14 days post fertilization larvae, pediveliger
larvae and metamorphosing larvae) were used as sam-
ples. Although Cg-BMPR1 and Cg-TGFbsfR2 tran-
scripts were ubiquitously expressed at reasonable levels
in all adult tissues, interestingly Cg-BMPR1 transcripts
were always expressed about 10-fold more than Cg-
TGFbsfR2 basal levels (respectively around 0.1 and
0.01 copies per 1 copy of GAPDH; Fig. 3B). Remark-
ably, this propensity is even more pronounced (up to
100 fold) when considering embryonic and larval
development (Fig. 3A). Then, Cg-BMPR1 was around
10-fold more expressed during early development
BMP/activin pathway in Crassostrea gigas A. Herpin et al.
3426 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
A
B
Fig. 1. Deduced amino acid sequence of
Cg-BMPR1 and Cg-TGFbsfR2. (A) The
implied amino acid sequence of Cg-BMPR1
contains a leader peptide shown in italics.
Cysteine residues characteristic of the TGFb
superfamily type I receptors are in bold and
underlined, and the cysteine knot is boxed.
Also boxed are the transmembrane domain,
the ATP binding site, the L45 loop and the

serine ⁄ threonine kinase domain. (B) The
implied amino acid sequence of Cg-
TGFbsfR2 contains a leader peptide
shown in italics. Two extracellular domains
were present in Cg-TGFbsfR2. The first (C1)
contained 10 cysteine residues whose spa-
cing was characteristic of TGFb superfamily
type II receptors. These are in bold and
underlined, and the cysteine knot is boxed.
The second extracellular domain (C2) also
contained 10 cysteines and these are also
shown in bold and the cysteine knot is
boxed. The C1 and C2 domains appeared to
be joined by a ‘linker’ sequence. Also boxed
are the transmembrane domain and the
serine ⁄ threonine kinase domain.
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3427
Daf-1 C. elegans
Crassostrea gigas C2 domain
ALK-6 E. fluviatilis C2 domain
Crassostrea gigas C1 domain
Wit D. melanogaster
Punt D. melanogaster
ActR-2b H. sapiens
ActR-2b D. rerio
ActR-2b C. auratus
BMPR-2 X. laevis
BMPR-2 H. sapiens
Daf-4 C. elegans

ALK-6 E. fluviatilis C1 domain
ALK-4 E. fluviatilis C1 domain
ALK-4 E. fluviatilis C2 domain
s
r
o
t
pece
r
I
e
p
y
T
s
r
o
t
p
e
c
er II

e
p
y
T
Activin
BMP
100

100
100
92
100
94
79
74
88
77
69
76
64
Daf-4 C. elegans
ActR-2b D. rerio
ALK-6 E. fluviatilis
ALK-1 E. fluviatilis
C32D5.2(actr) C. elegans
Crassostrea gigas
Cg-BMPR-1
Crassostrea gigas
Wit D. melanogaster
ALK-4 E. fluviatilis
Cg-ALR1
Crassostrea gigas
Saxophone D. melanogaster
ALK-8 D. rerio
Acvrl-1 D. rerio
ALK-1 H. sapiens
Hr-BMPR H. roretzi
Cg-Tβ

R-1 Crassostrea gigas
Baboon (AtR-I) D. melanogaster
TβR1 H. sapiens
TARAM D. rerio
STKR1 T. rubripes
Thick vein D. melanogaster
Punt D. melanogaster
TGFR B. pahangi
STPK A. caninum
ALK-2 E. fluviatilis
ActR-2b C. auratus
TβR-II H. sapiens
TβR-II G. gallus
TβR-IIa X. laevis
BMPR-2 X. laevis
BMPR-2 H. sapiens
ActR-2b H. sapiens
Daf-1 C. elegans
BMP-RIa H. sapiens
BMP-RIa D. rerio
BMP-RIb D. rerio
Raf D. melanogaster (out group)
B-raf H. sapiens (out group)
s
ro
t
p
e
c
e

r

I
e
pyT
s
ro
t
p
e
c
e
r
I
I
e
p
yT
Activin
TGF-
Activin
BMP
BMP
TGF-
100
99
100
100
100
100

100
100
98
100
96
88
87
83
91
86
69
86
86
62
87
95
77
82
71
79
95
78
66
87
87
58
62
78
97
A

B
Fig. 2.
BMP/activin pathway in Crassostrea gigas A. Herpin et al.
3428 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
(1.2 Cg-BMPR1 copies for 1 of GAPDH in oocytes)
when compared to adult tissues. Cg-BMPR1 expres-
sion steadily dropped during early and larval develop-
ment. After D larvae stage and up to metamorphosis,
expression levels returned to the basal adult state
(0.1–0.15 copies relative to GAPDH). Although Cg-
TGFbsfR2 transcripts were only detected at moderate
levels, two peaks of expression were observed, the first
during gastrulation and the second just before meta-
morphosis. In all cases, Cg-TGFbsrR2 average expres-
sion level is around 10-fold lower than Cg-BMPR1
when referring to adult tissues.
Cg-BMPR1 transduces a ventralizing signal
during zebrafish mesoderm induction
To determine whether Cg-BMPR1 and Cg-TGFbsfR2
are able to function in a manner similar to their ortho-
logues, we employed the zebrafish embryo as a ‘repor-
ter organism’. Specifically, we wished to analyse how
expression of these two molecules was able to perturb
the TGFb superfamily ligand–receptor signalling path-
way during zebrafish early development. Examples of
the range and severity of the phenotypes recorded in
the following experiments are shown in Fig. 4.
Injection of 5–200 pg per embryo of full length
Cg-BMPR1 transcript produced a range of ventralized
embryos (Figs 4A and 5A). Whole mount in situ

hybridization showed that tbx6 expression was expan-
ded towards the dorsal part of the embryo while gsc
expression was slightly reduced and not seen ectopi-
cally (Fig. 6B1–4). When mRNA encoding a truncated
version of Cg- BMPR1 (DN-Cg-BMPR1) was injected,
a range of dorsalized embryos was observed at increas-
ing concentrations (Figs 4B and 5B). The tbx6 and gsc
expression patterns were congruent with these observa-
Cg-ALR1
(6Kb, 6 introns)
011
11
11
1
2
0101 01
Cg-TGFβR1
(10,5Kb, 9 introns)
Cg-BMPR1
(10Kb, 10 introns)
01
21
1
2
2
101
Cg-TGFβsfR2
(>15Kb, 9 introns)
0
1

0
C1 C2
Linker
11
TM
Domain
1
0
1
GS
Box
XXI
I
Extracellular Domain
L45 loop
TM
Domain
GS

1
TGF
β
type 1 receptor
prototype
C
Fig. 2. (A) Phylogenetic relationship of the extracellular domain of TGFb superfamily type II receptors. Sequences used for the alignment of
extracellular parts were truncated to strictly embed the 10 conserved cysteines upstream of the characteristic activin ⁄ BMP cysteine knot
CCCX
(4)
C. Split duplicated extracellular domains are reported as C1 and C2 domains from the N-terminal part of the protein. This tree was

generated by using
CLUSTAL X. From this alignment a distance-based phylogenetic tree was constructed using the minimum evolution method
of the
PAUP package. The percentage recovery of the branch in 1000 bootstrap replications is indicated. ActR2b Carassius auratus
(ABB58749)Daf-1 Caenohabditis elegans (P20792), Daf-4 Caenohabditis elegans (P50488), Cg-TGFbsfR2 C. gigas (CAD20574), ActR-2b Danio
rerio (NP_571285), Punt Drosophila melanogaster (AAC41566), Wishful thinking D. melanogaster (AAF60175), ALK-4 Ephydatia fluvatilis
(AB026827), ALK-6 E. fluvatilis (AB026829), ActR-2b Homo sapiens (NP_001097), BMPR-2 H. sapiens (NP_001195), BMPR-2 Xenopus laevis
(AAB39883). (B) Phylogenetic tree showing the relationship of Cg-ALR1, Cg-BMPR1, Cg-TGFbR1 and Cg-TGFbsfR2 to other TGFb superfamily
ligand receptors. This tree was generated by using the alignment in
CLUSTAL X. From this alignment a distance-based phylogenetic tree
was constructed using the minimum evolution method of the
PAUP package. The percentage recovery of the branch in 1000 bootstrap repli-
cations is indicated. STPK Ancylostoma caninum (AAL06642), TGFR Brugia pahangi (ACC47801), C32D5.2(Actr) Caenohabditis ele-
gans (NP_495271), Daf-1 Caenohabditis elegans (P20792), Daf-4 Caenohabditis elegans (P50488), ActR2b Carassius auratus (ABB58749),
Cg-ALR1 Crassostrea gigas (AJ309316), Cg-TbR1 C. gigas (AJ544074), Cg-BMPR1 C. gigas (CAE11917), Cg-TGFbsf2 C. gigas (CAD20574),
ActR-2b Danio rerio (NP_571285), Acvrl-1 Danio rerio (AAM53074), ALK-8 Danio rerio (NP_571420), BMP-RIa Danio rerio (BAA32748), BMP-
RIb Danio rerio (BAA76408), TARAM Danio rerio (NP_571065), Baboon (Atr-I) Drosophila melanogaster (A55921), Punt D. melanogaster
(AAC41566), Saxophone D. melanogaster (I45712), Thick vein D. melanogaster (XP_079689), Wishful thinking D. melanogaster (AAF60175),
ALK-1 Ephydatia fluvatilis (BAA82601), ALK-2 E. fluvatilis (BAA82602), ALK-4 E. fluvatilis (AB026827), ALK-6 E. fluvatilis (AB026829), TbR-II
Gallus gallus (I50429), HrBMPR Halocynthia roretzi (BAB87725), ActR-2b Homo sapiens (NP_001097), ALK-1 H. sapiens (CAA80255),
BMP-RIa H. sapiens (NP_004320), BMPR-2 H. sapiens (NP_001195), TbR1 H. sapiens (P36897), TbR-II H. sapiens (P37173), STKR1 Takifugu
rubripes (AAC34382), BMPR-2 X. laevis (AAB39883), TbR-IIa X. laevis (AAG40577). Outgroups: Raf D. melanogaster (X07181), B-raf H. sap-
iens (M95712). (C) Exon structure and domain organization of TGFb superfamily type I and II receptor genes. The intron phase (0, 1 or 2) is
indicated above each intron–exon boundary. Boxes I, X and XI are representative of kinase subdomains [22]. Extracellular and transmem-
brane (TM) domains are also shown. The L45 loop and the GS box are specific to type I receptors. The type II receptor contains two extra-
cellular domains (C1 and C2) joined by a linker sequence. C1 and the linker are encoded by single exons, C2 by two exons.
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3429
tions: tbx6 was considerably repressed while gsc was
up-regulated and ectopically expressed throughout the

embryo (Fig. 6C1–4).
Expression of Cg-ALR1 results in both posterior
and anterior defects in zebrafish embryos
Expression of Cg-ALR1 resulted in a dose-dependent
range of anterior defects, including in some embryos a
lack of otic vesicles (Fig. 7A,B). These defects were
always combined with mild posterior defects. In addi-
tion, a significant fraction of the Cg-ALR1 injected
embryos (between 5 and 10% depending on the
mRNA concentration) displayed a bifida chordata
phenotype in combination with severe anterior defects
(Fig. 7A,Bc). Cg-ALR1 expression resulted in an
expression of gsc in ventral regions of the embryo
(Fig. 6D3,D4). The expression domain of tbx6 was
restricted to the ventral regions and fragmented at the
gastrula stage (Fig. 6D1,D2).
When mRNA encoding a truncated version of
Cg-ALRI (DN-Cg-ALR1) was injected at a range of
2–400 pg per embryo, posterior structure defects were
observed in a dose-dependent manner (Fig. 7C,D).
The tbx6 expression pattern was restricted to the
ventral side (Fig. 6E1,E2) while gsc expression in the
dorsal mesoderm was almost completely abolished
(Fig. 6E3,E4).
Cg-TGFbsfR2 transduces a dorsalizing signal
during zebrafish mesoderm induction
When injected at concentrations of between 10 and
200 pg per embryo, Cg-TGFbsfR2 induced dorsaliza-
tion in a concentration-dependent manner (Figs 4B
and 8A). Expression of tbx6 was dramatically repres-

sed, even if in some cases its expression at the mar-
gin of the blastoderm was expanded (Fig. 9A1,A2).
Expression of gsc was clearly expanded in all cases
(Fig. 9A3,A4). mRNA encoding a truncated Cg-
TGFbsfR2 (DN-Cg-TGFbsfR2) was generated by
inserting a stop codon at the C-terminal side of the
transmembrane domain. This protein included both
extracellular C1 and C2 domains as well as the
Cg-BMPR1
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
s
e
t
ycoO
a
lu
tsa
l
B
alurtsaG
ero
hpo
cor

T
.L
e
a
vr
a
L

D
fpd7
regile
V
.
L

f
p
d
4
1
.L
r
e
g
ile
V
r
e
g
i

l
ev
i
d
e
P
.
L
a
teM
m
h
p
r
o
o
.L

g
n
i
s

1RPMB-
g
C fo se
ipoc fo rebmun
HDPAG fo ypoc rep
Cg-TGFβsfR2
0

0.001
0.002
0.003
0.004
0.005
0.006
0.007
s
e
ty
c
o
O
a
l
u
ts
a
l
B
al
u
rts
a
G
er
ohp
o
c
o

rT
.
L
e
a
vr
a
L

D
.L
r
e
gil
e
V
f
pd
7

f
p
d
4
1
.
L
re
g
i

l
eV
r
eg
i
l
e
v
i
d
e
P
.L
a
teM
m
h
p
ro
o
.
L
gn
i
s
FGT-gC fo seipoc fo rebmun ββ
2
Rfs
HDPA
G

fo
yp
oc rep
0
0.05
0.1
0.15
0.2
0.25
HMEPAMDGG HeLP

1R
P
M
B-
gC
fo seipoc fo rebm
un
H
DPAG fo ypoc rep
Cg-BMPR1
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04

HMEPAMDGGHeLP
F
GT
-gC
fo
seipoc

fo rebmun
ββ
2Rfs
HDPA
G
f
o y
po
c
r
e
p
Cg-TGFβsfR2
A
B
Fig. 3. Differential Cg-BMPRI and Cg-TGFbsfR2 expression patterns during early development (A) and in adult tissues (B) measured by real
time quantitative RT-PCR. Each value is the mean ± SE of three animals (tissues) or the mean of a pool of embryos or larva (L) from one
spawn. ME, mantle edge; DG, digestive gland; LP, labial palps; PAM, posterior adductor muscle; G, gills; He, heart; H, haemocytes. The rel-
ative level of receptor expression was calculated for one copy of the GAPDH housekeeping gene by using the following formula: N ¼
1 · 2(Ct GAPDH – Ct target).
BMP/activin pathway in Crassostrea gigas A. Herpin et al.
3430 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
transmembrane region. Ectopic expression of this

protein led to phenotypes comparable to those
obtained after injection of truncated Cg-BMPR1
(Figs 8B and 4B). At concentrations of up to 100 pg
mRNA per embryo, the range of dorsalization
observed (Figs 4B and 8B) was comparable with that
obtained after Cg-TGFbsfR2 injection. D4 and D5
phenotypes were observed solely with concentrations
of 200 pg mRNA per embryo (Figs 4B and 8B). As
was observed with full length Cg-TGFbsfR2 mRNA,
when the truncated protein was expressed tbx6 was
repressed even if in some cases its expression at
the margin of the blastoderm was expanded
(Fig. 9B1,B2). In all cases gsc expression was clearly
expanded (Fig. 9B3,B4).
The C1 and C2 domains of TGFbsfR2 have different
ligand binding properties
Although the C1 binding domain of Cg-TGFbsfR2
clustered with its BMP type II Drosophila wit homo-
logous receptor (Fig. 2A), phylogenetic analyses
suggested that both C1 and C2 were unspecified
(Fig. 2A). To clarify the binding properties of each of
these domains we generated synthetic mRNA encoding
only the C1 or C2 domains. Expression of only the C1
domain resulted in the dorsalization of embryos in a
concentration-dependent manner (Fig. 8C). Similarly,
tbx6 and gsc expression (Fig. 9C1–4) were repressed
and expanded, respectively, through dorso–ventral ter-
ritories. Although the majority of embryos expressing
the C2 domain exhibited weakly dorsalized phenotypes
(Figs 8D and 4C), a sizeable minority were weakly

ventralized and showed the bifida chordata phenotype
in a manner similar to those obtained after Cg-Tolloid,
a C. gigas Tolloid-like orthologue; (A. Herpin et al.,
unpublished results), injections but with additional
anterior defects. It is not clear from the tbx6 and gsc
expression patterns observed in these embryos whether
they are dorsalized or ventralized (Fig. 9D1–4).
Ventralized embryos
24 Hpf
A
Dorsalized embryos
24 Hpf
B
Fig. 4. The range of zebrafish phenotypes observed on overexpres-
sion of TGF-b superfamily receptors. (A) Examples showing the ven-
tralized phenotypes obtained after overexpression of Cg-BMPR1.
These phenotypes ranged from the least (Vt1) to the most severe
(Vt3). (B) Examples showing the dorsalized phenotypes obtained
after overexpression of intact and truncated Cg-TGFbsfR2 and its C1
and C2 domains and truncated Cg-BMPR1. These phenotypes ran-
ged from the least (D1) to the most severe (D5).
0
25
50
75
100
2 10 50 100 200
D4
D5
D3

D2
D1
Normal
DN-Cg-BMPR1
n=195
n=209
n=167
n=114 n=182
pg/embryo
Proportion
of
embryos
(%)
Phenotypes
Vt3
Vt2
Vt1
Normal
5 50 100 150 200
0
25
50
75
100
Cg-BMPR1
n=77
n=96
n=80
n=73 n=101
pg/embryo

Proportion
of
embryos
(%)
Phenotypes
A
B
Fig. 5. Histograms showing the phenotype distribution after over-
expression of (A) Cg-BMPR1 and (B) truncated Cg-BMPR1 (DN-
Cg-BMPR1). The proportion of embryos showing an individual
phenotype is indicated by colour. The number of embryos injected
for each concentration of mRNA is indicated above each bar of
the histograms.
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3431
Discussion
We have described the cloning and functional analyses
of three TGFb superfamily type I and one type II
receptor orthologues. These are the first molecules of
this kind to be identified in lophotrochozoans. Below
we discuss some of the questions that arise from our
experiments and their analysis.
Did the evolution of TGF-b superfamily receptors
occur episodically or gradually?
Phylogenetic analysis of TGFb superfamily receptors
shows them to be clearly divided into two major clus-
ters, containing either type I or type II receptors. Each
cluster is further divided into individual clades contain-
ing TGF-b sensu strico (s.s.), activin or BMP receptors.
This observation is congruent, structurally, with the

subfamilies already defined by the ligands and suggests
a concerted evolution between ligands and receptors
[23]. According to the phylogenetic tree shown in
Fig. 2B, type (I or II) and subtype (TGF-b s.s., activin
or BMP) duplications that gave rise to all known types
and subtypes, would predate the divergence between
parazoans–eumetazoans and protostomes–deutero-
stomes for types and most subtypes, respectively. In
addition, although the observation of a second extra-
cellular domain in the Cg-TGFbsfR2 receptor is
unique among protostome and deuterostome type II
receptors, multiple extracellular domains are observed
80% epiboly
tbx6
(ND gC )1R-PMB-(ND gC )1RLA-
gC 1R-PMB-
80% epiboly
goosecoid
B3
B4
E3 E4
B1
B2
E2
E1
C1
C2
gC 1RLA-
D2
D1

D4D3
C3 C4
lortnoC
A1 A2
A3
A4
Fig. 6. In situ hybridization of zebrafish embryos using the ventro-lateral mesoderm marker tbx6 and the dorsal mesoderm marker goosecoid
at 80% epiboly. Two examples are shown for each group and each marker. Changes in the localization of the tbx6 expression pattern is
highlighted. DN (dominant negative) indicates that truncated receptor was overexpressed in these experiments.
BMP/activin pathway in Crassostrea gigas A. Herpin et al.
3432 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
among sponge (parazoan) molecules [24]. At this time,
the basic receptor repertoire may have already consis-
ted of BMP and activin type I and II receptors [24].
The type II TGFb s.s. receptor has only been identified
in deuterostomes. It may therefore have been lost
during protostome evolution or alternatively have been
acquired during the gene explosion that occurred prior
to the emergence of the chordates. Finally, detailed
phylogenetic analyses of duplicated extracellular
domains (Fig. 2A) observed in sponges and C. gigas
receptors showed them to be more closely related to
other extracellular domains than to their duplicated
counterpart, suggesting a very early duplication event,
probably before the one that gave rise to BMP and
activin subtypes.
The gene organization of TGF-b superfamily type I
and II receptors suggests evolution by exon
shuffling
Type I and II TGF-b superfamily receptor gene organ-

ization across protostomes and deuterostomes reveals
that domain distribution among exons is conserved.
Usually, intron boundaries do not interrupt functional
domains. Indeed, for both type I and II receptors,
both the phase and position of the exon boundaries of
the core kinase domain are highly conserved suggesting
that these genes were derived from a common ances-
tral kinase gene that encoded sections X and XI of the
receptor serine ⁄ threonine kinase domain [22]. Another
striking point is the phase conservation of the intron–
exon boundaries that lie either side of the exon enco-
ding the extracellular domain. This lends support to
the theory that the ligand specificity of these receptors
may have been achieved by exon shuffling.
Cg-BMPR1 and Cg-TGFbsfR2 transcripts are
maternally supplied and make the oyster early
embryo susceptible to respond to a BMP/activin
signal
In many animal species members of the TGF-b super-
family of growth factors play a crucial role in specific
developmental events [25,26]. During early embryo-
genesis both Cg-BMPR1 and, to a certain degree,
Cg-TFGbsfR2 show an apparent accumulation in
C
A1
A2
A3
A4
P1
P2

P3
Bc
A
DN-Cg-ALR1
0
25
50
75
100
dead
A4
A3
A2
A1
Normal
10 50 100 200 400
pg/embryo
n=177
n=250
n=207
n=197 n=218
D
Phenotypes
Proportion
of
embryos
(%)
10 50 100 200 400
0
25

50
75
100
P3
P2
P1
Normal
Bc
Cg-ALR1
n=99
n=123
n=108
n=99 n=118
pg/embryo
B
Proportion
of
embryos
(%)
Phenotypes
Fig. 7. Distribution of phenotypes after over-
expression of Cg-ALR1 and its truncated
form DN-Cg-ALR1. (A) A dose-dependent
range of anterior defects (P1-3) combined
with mild posterior defects was observed
after overexpression of Cg-ALR1. A signifi-
cant fraction of the embryos also displayed
a bifida chordata (Bc) phenotype in combina-
tion with severe anterior defects. (B) Histo-
gram showing the proportion of embryos

displaying each phenotype after injection
with Cg-ALR1. The number of embryos
injected for each experiment is also indica-
ted. (C) A dose-dependent range of range of
posterior defects (A1-4) was observed after
overexpression of truncated Cg-ALR1
(DN-Cg-ALR1). (D) Histogram showing the
proportion of embryos displaying each
phenotype after injection with DN-Cg-ALR1.
The number of embryos injected for each
experiment is also indicated.
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3433
unfertilized oocytes (up to 10-fold more than the aver-
age level for Cg-BMPR1). This demonstrates that
newly fertilized eggs could have the aptitude to respond
to a BMP-like signal before zygotic transcription initi-
ation. After mid-blastula transition, while Cg-BMPR1
transcripts are steadily decreasing until metamorphosis,
two peaks of Cg-TFGbsfR2 mRNA occur at the time
of the two major organomorphogenetic events arising
during lophotrochozoan larval development, namely
early morphogenesis (gastrula) and metamorphosis
[27]. Such expression patterns, especially during early
embryogenic development, are consistent with that of
several serine ⁄ threonine kinase receptors [28–30] imply-
ing a way in which Cg-BMPR1 and Cg-TFGbsfR2
may be involved as morphogens for early development
in C. gigas. If we consider the whole repertoire of
TGF-b pathway components including ligands and re-

ceptors so far identified in C. gigas (reviewed in [23]),
all of the receptors (activin type I receptor [17], TGF-b
type I sensu stricto receptor [18] and now BMP type I
receptor) are without exception expressed in a synex-
pression group throughout early development. On the
other hand mGDF [1], a C. gigas BMP2 ligand ortho-
logue, like Cg-TFGbsfR2 is specifically highly
expressed at a later stage, mainly prior to or during
metamorphosis (Fig. 3A). Taken together, these com-
plementary expression fluctuations between type I and
II receptors and ligands probably reflect a relatively lax
system for which receptor ⁄ receptor, as well as lig-
and ⁄ receptor, interactions would be flexible across
TFG-b subfamilies [14,31,32]. Hence, ligands/receptors
versatility and type I/type II receptors variegation,
leading in some cases to cross talk between TGF-b sub-
family pathways, are probably answers to compensate
for the relatively low number of primary interacting
components observed in protostome species when
compared to their deuterostome counterparts [23].
Cg-BMPR1 partially ventralizes zebrafish embryos
during gastrulation by transducing a BMP-like
ventralizing signal in the mesoderm
Expression of Cg-BMPR1 resulted in moderate vent-
ralization of zebrafish embryos with the loss of anter-
ior structures together with slightly enlarged posterior
somites. This phenotype is similar to those induced by
Cg-Tolloid ⁄ tolloid ⁄ xolloid overexpression in fish and
frog (A. Herpin et al., unpublished results; [33,34]).
This phenotype also resembles that of the zebrafish

chordin loss of function mutant [35] and suggests that
the mesoderm may be partially ventralized. This hypo-
thesis was supported by analysis of tbx6 and gsc
expression by the gastrula. Expression of tbx6, which
specifies ventral-type mesoderm [36], was expanded
towards the dorsal region, showing that Cg-BMPR1
induced ventral mesoderm in zebrafish. In contrast, the
gsc expression pattern showed the dorsal mesoderm
to be more diffuse and clearly redistributed towards
Bc
D5
D4
D3
D2
D1
Normal
10 50 100 200 300
0
25
50
75
100
DN-Cg-TGFβsfR2C2
D
n=152
n=216
n=170
n=216 n=148
pg/embryo
Phenotypes

Proportion
of
embryos
(%)
0
25
50
75
100
10 50 75 100 200
D4
D5
D3
D2
D1
Normal
Cg-TGFβsfR2
A
n=176
n=97
n=123
n=110 n=155
pg/embryo
Proportion
of
embryos
(%)
Phenotypes
0
25

50
75
100
10 50 75 100 200
D4
D5
D3
D2
D1
Normal
DN-Cg-TGFβsfR2
B
n=183
n=362
n=300
n=225 n=364
pg/embryo
Proportion
of
embryos
(%)
10 50 100 200 300
0
25
50
75
100
D4
D5
D3

D2
D1
Normal
DN-Cg-TGFβsfR2C1
C
n=145
n=182
n=147
n=165 n=117
pg/embryo
Proportion
of
embryos
(%)
Phenotypes
Fig. 8. Histograms showing the proportion of embryos displaying
each phenotype after injection with (A) Cg-TGFbsfR2 mRNA, (B)
DN-Cg-TGFbsfR2 mRNA, (C) DN-Cg-TGFbsf2RC1 domain mRNA,
and (D) DN-Cg-TGFbsfR2 C2 domain mRNA. The number of
embryos injected for each experiment is also indicated.
BMP/activin pathway in Crassostrea gigas A. Herpin et al.
3434 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
ventral regions. Unsurprisingly, when Cg-BMPR1 was
overexpressed, a similar outcome to that observed with
previously characterized oyster Cg-Tolloid was seen,
although the phenotypes were not usually so severe
(A. Herpin et al., unpublished results). This probably
reflects the fact that, even in the presence of an abnor-
mally high concentration of type I BMP recep-
tor, enhanced signalling is limited by the endogenous

ligand concentration. Expression of truncated Cg-
BMPR1 predictably produced a range of dose-depend-
ent dorsalized phenotypes, as the dominant negative
BMP type I receptor probably sequestered BMP lig-
ands. The phenotypes associated with the abrogation
of BMP pathway signalling were comparable to those
obtained in BMP2-deficient zebrafish [37]. Moreover,
the pattern of tbx6 and gsc expression indicated that
the observed dorsalization was due to inappropriate
expansion of the dorsal mesoderm with a concomitant
reduction in the ventral mesoderm.
These observations suggest that Cg-BMPR1 is cap-
able of functioning in a similar way to its zebrafish
orthologues by promoting ventral mesoderm induction
and repressing dorsal mesoderm formation during
gastrulation.
Cg-ALRI interferes with endogenous BMP/activin
pathways during zebrafish gastrulation
Expression of full length Cg-ALRI in zebrafish resul-
ted in embryos with disruption of the anteroposterior
N
D
( gC FGT- β 2C-2R-fs )
ND
( gC FGT- β 1C-2R-fs )
gC FGT- β 2R-fs
D2
D1
D3 D4
C2

C1
B2B1
ND
FGT-gC( β fs )2R-
A1 A2
A3
B3
B4
80% epiboly
goosecoid
80% epiboly
tbx6
lortnoC
1
2
3
4
A4
C3 C4
Fig. 9. In situ hybridization of zebrafish embryos using the ventro-lateral mesoderm marker tbx6 and the dorsal mesoderm marker goosecoid
at 80% epiboly. Two examples are shown for each group and each marker. Changes in the localization of the tbx6 expression pattern is
highlighted. DN (dominant negative) indicates that truncated receptor was overexpressed in these experiments.
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3435
axis. In contrast to overexpression of either endo-
genous TbRI [38] or mouse ActRIA and ActRIB [39] in
zebrafish, Cg-ALRI expression did not induce a dupli-
cation of anterior structures but rather a bifida chor-
data phenotype where the posterior part of the
notochord is divided around the yolk. This phenotype

is probably due to a delay in blastopore closure during
gastrulation. Cg-ALRI expression induced gsc expres-
sion ectopically towards ventral regions while repress-
ing tbx6 expression. Expression of truncated Cg-ALRI
generally resulted in dorsalized phenotypes combined
with some anterior defects and analysis of the expres-
sion of the mesodermal markers showed repression of
both the dorsal and ventrolateral mesoderm at gastrula
stage. These observations are in agreement with those
of Thisse and colleagues following overexpression
of the activin specific competitive inhibitor, antivin
[40,41].
Expression of Cg-TGF bsfR2 in zebrafish embryos
While the structure and function of oyster type I BMP
and activin receptors appears to be conserved in com-
parison to those of other invertebrates and vertebrates,
the structure of the oyster type II receptor was more
surprising. Cg-TGFbsfR2 possess two extracellular
domains joined by a linker sequence, an observation
so far unique to type II TGFb superfamily receptors
from protostomes or deuterostomes, although it is a
characteristic shared with sponges for which, among
seven receptors cloned in Ephydatia fluviatilis, five pos-
sess a single ‘conventional’ domain, one a duplicated
and another one a triplicated extracellular domain [24].
Unfortunately, our experiments do not clarify whether
this receptor is active or not in its full length form.
Expression of both the wild type and dominant negat-
ive forms of Cg-TGFbsfR2 give rise to the same phe-
notype and modified tbx6 ⁄ gsc marker expression

patterns. We believe that this is probably due to the
fact that the oyster type II receptor is incompatible
with the zebrafish TGF-b superfamily type I receptors.
The observation that expression of the truncated oys-
ter type II receptor is able to disrupt mesoderm forma-
tion certainly suggests that this receptor is capable of
ligand binding. More surprising is the apparent differ-
ential ligand affinity of the C1 and C2 extracellular
domains. Expression of C1 led to the apparent seques-
tration of BMP-like ligands, resulting in the produc-
tion of dorsalized phenotypes. Injection of relatively
high concentrations of mRNA encoding the C2
domain, however, resulted in embryos with a combina-
tion of features of both dorsalized and ventralized
phenotypes. This is suggestive of an oyster type II
receptor able to interact with both BMPs and activins
via its double extracellular ligand binding domains.
Such promiscuous behaviour has already been des-
cribed for the Drosophila sax [42] and Caenohabditis
elegans Daf-1 proteins [43], but these two receptors
are conventional single extracellular receptors. It is of
course feasible that TGFbsfR2 acts similarly to the
TGF-b silencing pseudoreceptor bambi [44]; however,
the highly conserved cytoplasmic domains suggests
that its encoding gene is still under evolutionary pres-
sure to maintain its integrity. This hitherto single
C. gigas type II receptor may represent an undefined
evolutionary stage in TGF-b ⁄ activin ⁄ BMP pathway
signal transduction involving a pleiotropic and promis-
cuous receptor.

Evidence for the conservation of the mechanism
of mesoderm induction in molluscs and across
bilaterians
Although Lartillot and colleagues have observed for
a number of developmental genes involved in antero-
posterior patterning that their spatial and temporal
expression pattern is highly conserved across bilate-
rians [5,45], the molecular mechanisms in lophotrocho-
zoans remain unclear. Unequal cleaving spiralian adult
mesoderm is derived from two split territories. The
endomesoderm arises from the fourth quartet micro-
mere of the D quadrant, the mesentoblast [46]; subse-
quently, the ectomesoderm is derived from the second
and ⁄ or third quartets of micromeres (reviewed in [47]).
For the first time in a lophotrochozoan we have shown
that two key members of the BMP pathway
Cg-BMPR1 and the recently characterized Cg-Tolloid
(A. Herpin et al. unpublished results) are present and
expressed as a synexpression group throughout early
development as well as in adult tissues of C. gigas.
Moreover, using the zebrafish embryo as a reporter
system we have demonstrated that these two pathway
components, as well as the activin-like type I receptor
Cg-ALRI, are able to regulate mesoderm induction
during gastrulation and early development stages,
despite the considerable evolutionary distance between
spiralians and chordates. The results of these func-
tional studies when combined with our expression ana-
lyses and phylogenetic data, would suggest a conserved
mechanism for both ectomesoderm and mesendoderm

induction in spiralians in comparison with ecdysozoans
and protostomes. Further support for the conservation
of such an inductive mechanism has been provided by
Lambert and Nagy [48] who have shown asymmetric
inheritance of mRNA encoding a DPP orthologue dur-
ing embryonic cleavage by Ilyanassa obsolete, a gastro-
BMP/activin pathway in Crassostrea gigas A. Herpin et al.
3436 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
pod mollusc. These events produce a complex pattern
of mRNA localization in which the DPP transcripts,
after being distributed diffusely in the cytoplasm,
become localized in all four of the macromere cells at
the eight-cell stage. These transcripts then segregate
into the second and third quartet of micromeres but
decay in the second quartet cells, resulting in a specific
localization in the third quartet cells as well as in
macromere 3D. Interestingly, the third quartet cells of
spiralians, as well as macromere 3D, have been shown
through cell lineage studies to give rise to the endomes-
oderm and ectomesoderm [47]. Together with the
evidence provided by this study, this argues for a poss-
ible conserved endomesoderm ⁄ ectomesoderm inductive
mechanism in spiralians through an ancestral BMP ⁄
activin pathway, in which the singular and promiscu-
ous Cg-TGFbsfR2 could be the shared type II receptor
interface for both BMP and activin ligands. This hypo-
thesis is supported by the fact that different methods,
such as RT-PCR with degenerate primers directed
against the highly conserved intracellular kinase
domain of these receptors, and low stringency screen-

ing of cDNA, genomic and BAC libraries (10 times
coverage) using different probes, failed to isolate any
new type I or II receptors. Cg-TGFbsfR2 would repre-
sent the only (TGF-b) type II receptor encoded within
the C. gigas genome.
Experimental procedures
Cloning and sequencing of TGFb superfamily
receptor cDNAs
We previously reported the cloning of two TGF-b super-
family receptors, a type I activin-like receptor (Cg- ALR1)
[17] as well as another type I TGF-b s.s. receptor
(Cg-TGFbRI) from C. gigas [18]. cDNAs encoding novel
type I receptors and one type II receptor were cloned from
C. gigas in an identical manner. Gene-specific primers were
then used to isolate a full length cDNA from an oyster
mantle edge library. The resulting clones were designated
Cg-BMPR1 and Cg-TGFbsfR2.
Phylogenetic analysis
Sequences encoding TGF b superfamily receptor proteins
were chosen to represent a range of protostomes and deut-
erostomes in phylogenetic analyses. Preference was given to
sequences for which functional data were available. The
sequences were aligned using clustal x version 1.81 and
by manual inspection. Alignments used for phylogenetic
relationships of the cytoplasmic domain were carried out,
making use of the whole receptor intracellular part amputa-
ted of its highly variable C-terminal domain after the
terminal conserved arginine of the cytoplasmic serine ⁄
threonine kinase domain. Sequences for phylogenetic rela-
tionships of the extracellular part were truncated to strictly

take into consideration the 10 conserved cysteines upstream
of the characteristic cysteine knot CCCX
(4)
C. From these
alignments, distance-based phylogenetic trees were con-
structed using the minimum evolution method of the paup
package version 4.0b4a. One thousand bootstrap trials were
run using the neighbor-joining algorithm for each node.
Isolation and characterization of Cg-BMPR1 and
Cg-TGFbsfR2 TGFb genes
A C. gigas genomic library was constructed in kDASH II
(Stratagene, La Jolla, CA, USA). A total of 1.8 · 10
6
inde-
pendent clones were recovered. After amplification to
4.5 · 10
10
pfuÆmL
)1
, a total of 2.5 · 10
5
recombinant phag-
es were screened at high stringency with digoxigenin-11-
dUTP labelled probes encoding full length TGF-b super-
family receptor cDNAs. Unique clones containing each of
the two novel receptor genes were sequenced and the
intron–exon structures of the genes determined by compar-
ing the genomic sequence with the cDNA sequence.
Quantification of Cg -BMPR1 and Cg-TGFbsfR2
receptor transcripts

Because technical limitations relative to oyster embryos and
larvae make whole-mount in situ hybridizations less reliable
than RT-PCR [17], quantitative real time RT-PCR was the
best alternative. The larval developmental stage pattern
and tissue distribution of Cg-BMPR1 and Cg-TGFbsfR2
mRNA was investigated by quantitative RT-PCR using an i-
Cycler (Bio-Rad, Hercules, CA, USA). Total RNA from var-
ious developmental stages and adult tissues was isolated
using Tri-Reagent (Sigma-Aldrich, Munich, Germany)
according to the manufacturer’s instructions. After treatment
for 20 min at 37 °C with 1 U of DNase I (Sigma) to prevent
genomic DNA contamination, 1 lg of total RNA was
reversed transcribed using 1 lg of random hexanucleotide
primers (Promega, Madison, WI, USA), 0.5 mm dNTPs and
200 U M-MLV reverse transcriptase (Promega) at 37 °C for
1 h in the appropriate buffer. The reaction was stopped by
incubation at 70 °C for 10 min. iQ
TM
SYBR Green supermix
PCR kit (Bio-Rad) was used for real time monitoring of
amplification (5 ng of template cDNA, 40 cycles: 95 °C for
15 s, 60 °C for 15 s) with 250 nm of the following primers:
QsBMPR1, 5¢-AGCTTGCCCCCAACCTC-3¢; QaBMPR1,
5¢-ATGGTCTCTGCGGGTTGA-3¢; QsTGFbsfR2, 5¢-GCC
AGATCCCAAATTAGTGC-3¢; QaTGFbsfR2, 5¢-TGAA
ACCACAGCCTCAGTTG-3¢, where ‘s’ and ‘a’ indicate
sense and antisense primers, respectively. Accurate amplifica-
tion of the target amplicon was checked by performing melt-
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3437

ing curve analysis. Using QsGAPDH (5¢-TTCTCTT
GCCCCTCTTGC-3¢) and QaGAPDH (5¢-CGCCCAATCC
TTGTTGCTT-3¢) primers, a parallel amplification of oyster
GAPDH transcript (EMBL CG548886) was carried out to
normalize the expression data of the receptor transcripts.
The relative level of receptor expression was calculated for 1
copy of the GAPDH housekeeping gene by using the formula
N ¼ 1 · 2(Ct GAPDH – Ct target).
Zebrafish maintenance and preparation of eggs
TAB zebrafish were reared and maintained on a light ⁄ dark
cycle essentially as described by Westerfield [49]. All experi-
ments and fish maintenance comply with current Norwe-
gian law. Synchronized eggs were obtained by mixing
previously separated female and male fish at a ratio of
2 : 1. Eggs were then collected as described by Culp et al.
[50]. Self-digested Pronase E (Sigma) at 0.5 mgÆmL
)1
was
used to remove the egg chorion. The reaction was stopped
by removing the enzyme and transferring the eggs into Hol-
tfreter’s solution (85 mm NaCl, 1 mm KCl, 1 mm CaCl
2
and 3.5 mm NaHCO
3
) and rinsing.
Microinjections
Initially, full length cDNAs encoding TGFb superfamily
receptors were subcloned into the plasmid pRN3 [51].
Receptor cDNAs for use as dominant-negative controls
were obtained by inserting a stop codon either immediately

up- or downstream of the sequence encoding the transmem-
brane domain. Capped synthetic mRNAs encoding both
full length and truncated proteins were transcribed in vitro
using the Stratagene mCAP RNA capping kit and linea-
rized plasmids as templates. The capped mRNAs were dis-
solved in RNAse-free 0.2 m KCl containing 0.5% (w ⁄ v)
phenol red to monitor injection. mRNA or plasmid DNA
was injected into the cytoplasm of a single-cell embryo.
After injection, embryos were transferred into 24-well tissue
culture plates containing Holtfreter’s solution (5–7 embryos
per well), incubated at 28 °C for up to 24 h and then exam-
ined by microscopy for phenotype analysis. Unfertilized
eggs were removed after 4 h. Alternatively, embryos were
fixed for in situ hybridization at 80% epiboly. Uninjected
and control embryos injected with 0.2 m KCl were analysed
under the same conditions.
Whole-mount in situ hybridization
Antisense probes were transcribed in vitro and labelled
using digoxigenin-labelled UTP (Roche, Indianapolis, IN,
USA). The anteroaxial goosecoid mesodermal marker [52],
and the ventrolateral tbx6 mesodermal marker [36] were
used as probes. Whole-mount staining of embryos was
performed as described by Rissi et al. [53].
Acknowledgements
This work was performed with the aid of grants from
the Norwegian Research Council, Basse-Normandie
Regional Council, France, and the French-Norwegian
Foundation for Scientific, Technical and Industrial
Research. We would like to acknowledge the help pro-
vided by the Zebrafish facility technical staff at the

Sars Centre. We would also like to thank Prof Patrick
Lemaire (Laboratoire de Ge
´
ne
´
tique et Physiologie du
De
´
veloppement, IBDM, CNRS ⁄ INSERM ⁄ Universite
´
de la Me
´
diterrane
´
e ⁄ AP de Marseille) for provision of
the pRN3 plasmid. We acknowledge J. Laisney and
A. H. Hansen for their technical assistance. We also
thank Prof Manfred Schartl (Physiological Chemistry
I, University of Wuerzburg, Germany) for his com-
ments and suggestions on the manuscript.
References
1 Lelong C, Mathieu M & Favrel P (2000) Structure and
expression of mGDF, a new member of the transform-
ing growth factor-beta superfamily in the bivalve mol-
lusc Crassostrea gigas. Eur J Biochem 267, 3986–3993.
2 Lelong C, Mathieu M & Favrel P (2001) Identification
of new bone morphogenetic protein-related members in
invertebrates. Biochimie 83, 423–426.
3 Arendt D, Technau U & Wittbrodt J (2001) Evolution
of the bilaterian larval foregut. Nature 409, 81–85.

4 Hinman VF & Degnan BM (2002) Mox homeobox
expression in muscle lineage of the gastropod Haliotis
asinina: evidence for a conserved role in bilaterian myo-
genesis. Dev Genes Evol 212, 141–144.
5 Lartillot N, Le Gouar M & Adoutte A (2002) Expres-
sion patterns of fork head and goosecoid homologues in
the mollusc Patella vulgata supports the ancestry of the
anterior mesendoderm across Bilateria. Dev Genes Evol
212, 551–561.
6 ten Dijke P, Miyazono K & Heldin CH (1996) Signaling
via hetero-oligomeric complexes of type I and type II
serine ⁄ threonine kinase receptors. Curr Opin Cell Biol 8,
139–145.
7 Massague J & Wotton D (2000) Transcriptional control
by the TGF-beta ⁄ Smad signaling system. EMBO J 19,
1745–1754.
8 Schier AF & Shen MM (2000) Nodal signalling in verte-
brate development. Nature 403, 385–389.
9 Whitman M (2001) Nodal signaling in early vertebrate
embryos: themes and variations. Dev Cell 1, 605–617.
10 Gu Z, Nomura M, Simpson BB, Lei H, Feijen A, van
den Eijnden-van Raaij J, Donahoe PK & Li E (1998)
The type I activin receptor ActRIB is required for egg
cylinder organization and gastrulation in the mouse.
Genes Dev 12, 844–857.
BMP/activin pathway in Crassostrea gigas A. Herpin et al.
3438 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
11 Gu Z, Reynolds EM, Song J, Lei H, Feijen A., Yu L,
He W, MacLaughlin DT, van den Eijnden-van Raaij J,
Donahoe PK et al. (1999) The type I serine ⁄ threonine

kinase receptor ActRIA (ALK2) is required for gastrula-
tion of the mouse embryo. Development 126, 2551–2561.
12 Song J, Oh SP, Schrewe H, Nomura M, Lei H, Okano
M, Gridley T & Li E (1999) The type II activin recep-
tors are essential for egg cylinder growth, gastrulation,
and rostral head development in mice. Dev Biol 213,
157–169.
13 Raftery LA & Sutherland DJ (1999) TGF-beta family
signal transduction in Drosophila development: from
Mad to Smads. Dev Biol 210 , 251–268.
14 Brummel T, Abdollah S, Haerry TE, Shimell MJ, Merr-
iam J, Raftery L, Wrana JL & O’Connor MB (1999)
The Drosophila activin receptor baboon signals through
dSmad2 and controls cell proliferation but not pattern-
ing during larval development. Genes Dev 13, 98–111.
15 Aberle H, Haghighi AP, Fetter RD, McCabe BD,
Magalhaes TR & Goodman CS (2002) wishful thinking
encodes a BMP type II receptor that regulates synaptic
growth in Drosophila. Neuron 33, 545–558.
16 Marques G, Musacchio M, Shimell MJ, Wunnenberg-
Stapleton K, Cho KW & O’Connor MB (1997) Produc-
tion of a DPP activity gradient in the early Drosophila
embryo through the opposing actions of the SOG and
TLD proteins. Cell 91, 417–426.
17 Herpin A, Favrel P & Cunningham C (2002) Gene
structure and expression of cg-ALR1, a type I activin-
like receptor from the bivalve mollusc Crassostrea gigas.
Gene 301, 21–30.
18 Herpin A, Lelong C, Becker T, Rosa FM, Favrel P &
Cunningham C (2005) Structural and functional evi-

dences for a type 1 TGF-beta sensu stricto receptor in
the lophotrochozoan Crassostrea gigas suggest con-
served molecular mechanisms controlling mesodermal
patterning across Bilateria. Mech Dev 122, 695–705.
19 ten Dijke P, Yamashita H, Sampath TK, Reddi AH,
Estevez M, Riddle DL, Ichijo H, Heldin CH & Miya-
zono K (1994) Identification of type I receptors for
osteogenic protein-1 and bone morphogenetic protein-4.
J Biol Chem 269, 16985–16988.
20 Chen YG, Hata A, Lo RS, Wotton D, Shi Y, Pavletich
N & Massague J (1998) Determinants of specificity in
TGF-beta signal transduction. Genes Dev 12, 2144–
2152.
21 Rosenzweig BL, Imamura T, Okadome T, Cox GN,
Yamashita H, ten Dijke P, Heldin CH & Miyazono K
(1995) Cloning and characterization of a human type II
receptor for bone morphogenetic proteins. Proc Natl
Acad Sci USA 92, 7632–7636.
22 Hanks SK, Quinn AM & Hunter T (1988) The protein
kinase family: conserved features and deduced phylo-
geny of the catalytic domains. Science 241, 42–52.
23 Herpin A, Lelong C & Favrel P (2004) Transforming
growth factor-beta-related proteins: an ancestral and
widespread superfamily of cytokines in metazoans. Dev
Comp Immunol 28, 461–485.
24 Suga H, Ono K & Miyata T (1999) Multiple TGF-beta
receptor related genes in sponge and ancient gene
duplications before the parazoan-eumetazoan split.
FEBS Lett 453, 346–350.
25 Schier AF (2001) Axis formation and patterning in

zebrafish. Curr Opin Genet Dev 11, 393–404.
26 Schier AF (2003) Nodal signaling in vertebrate develop-
ment. Annu Rev Cell Dev Biol 19, 589–621.
27 Freeman GAL (1992) Evolutionary implications of the
mode of D quadrant specification in coelomates with
spiral cleavage. J Evol Biol 5 , 205–247.
28 Roelen BA, Van Eijk MJ, Van Rooijen MA, Bevers
MM, Larson JH, Lewin HA & Mummery CL (1998)
Molecular cloning, genetic mapping, and developmental
expression of a bovine transforming growth factor beta
(TGF-beta) type I receptor. Mol Reprod Dev 49, 1–9.
29 Yelick PC, Abduljabbar TS & Stashenko P (1998)
zALK-8, a novel type I serine ⁄ threonine kinase recep-
tor, is expressed throughout early zebrafish develop-
ment. Dev Dyn 211, 352–361.
30 Aoki TO, David NB, Minchiotti G, Saint-Etienne L,
Dickmeis T, Persico GM, Strahle U, Mourrain P &
Rosa FM (2002) Molecular integration of casanova in
the Nodal signalling pathway controlling endoderm for-
mation. Development 129, 275–286.
31 de Iongh vRU, Chen Y, Kokkinos MI & McAvoy JW
(2004) BMP and activin receptor eypression in lens
development. Mol Vis 10, 566–576.
32 Krishna S, Maduzia LL & Padgett RW (1999) Specifi-
city of TGFbeta signaling is conferred by distinct type I
receptors and their associated SMAD proteins in Cae-
norhabditis elegans. Development 126, 251–260.
33 Goodman SA, Albano R, Wardle FC, Matthews G,
Tannahill D & Dale L (1998) BMP1-related metallopro-
teinases promote the development of ventral mesoderm

in early Xenopus embryos. Dev Biol 195, 144–157.
34 Piccolo S, Sasai Y, Lu B & De Robertis EM (1996)
Dorsoventral patterning in Xenopus: inhibition of ven-
tral signals by direct binding of chordin to BMP-4. Cell
86, 589–598.
35 Hammerschmidt M, Pelegri F, Mullins MC, Kane DA,
van Eeden FJ, Granato M, Brand M, Furutani-Seiki M,
Haffter P & Heisenberg CP et al. (1996) dino and mer-
cedes, two genes regulating dorsal development in the
zebrafish embryo. Development 123, 95–102.
36 Hug B, Walter V & Grunwald DJ (1997) tbx6, a Bra-
chyury-related gene expressed by ventral mesendodermal
precursors in the zebrafish embryo. Dev Biol 183, 61–73.
37 Kishimoto Y, Lee KH, Zon L, Hammerschmidt M &
Schulte-Merker S (1997) The molecular nature of zebra-
FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS 3439
A. Herpin et al. BMP/activin pathway in Crassostrea gigas
fish swirl: BMP2 function is essential during early dor-
soventral patterning. Development 124, 4457–4466.
38 Renucci A, Lemarchandel V & Rosa F (1996) An acti-
vated form of type I serine ⁄ threonine kinase receptor
TARAM-A reveals a specific signalling pathway
involved in fish head organiser formation. Development
122, 3735–3743.
39 de Vries CJ, de Boer J, Joore J, Strahle U, van Ach-
terberg TA, Huylebroeck D, Verschueren K, Miyaz-
ono K, van den Eijnden-van Raaij AJ & Zivkovic D
(1996) Active complex formation of type I and type
II activin and TGF beta receptors in vivo as studied
by overexpression in zebrafish embryos. Mech Dev 54,

225–236.
40 Thisse C & Thisse B (1999) Antivin, a novel and divergent
member of the TGFbeta superfamily, negatively regulates
mesoderm induction. Development 126, 229–240.
41 Thisse B, Wright CV & Thisse C (2000) Activin- and
Nodal-related factors control antero-posterior pattern-
ing of the zebrafish embryo. Nature 403, 425–428.
42 Singer MA, Penton A, Twombly V, Hoffmann FM &
Gelbart WM (1997) Signaling through both type I DPP
receptors is required for anterior-posterior patterning of
the entire Drosophila wing. Development 124, 79–89.
43 Georgi LL, Albert PS & Riddle DL (1990) daf-1, a
C. elegans gene controlling dauer larva development,
encodes a novel receptor protein kinase. Cell 61, 635–
645.
44 Onichtchouk D, Chen YG, Dosch R, Gawantka V,
Delius H, Massague J & Niehrs C (1999) Silencing of
TGF-beta signalling by the pseudoreceptor BAMBI.
Nature 401, 480–485.
45 Lartillot N, Lespinet O, Vervoort M & Adoutte A
(2002) Expression pattern of Brachyury in the mollusc
Patella vulgata suggests a conserved role in the estab-
lishment of the AP axis in Bilateria. Development 129,
1411–1421.
46 Lillie FR (1895) The embryology of the Unionidae.
J Morph 10, 1–100.
47 Boyer BC, Henry JQ & Martindale MQ (1996) Dual
origins of mesoderm in a basal spiralian: cell lineage
analyses in the polyclad turbellarian Hoploplana inqui-
lina. Dev Biol 179, 329–338.

48 Lambert JD & Nagy LM (2002) Asymmetric inheritance
of centrosomally localized mRNAs during embryonic
cleavages. Nature 420, 682–686.
49 Westerfield M (1989) The Zebrafish Book. University of
Oregon Press, Eugene, OR.
50 Culp P, Nusslein-Volhard C & Hopkins N (1991) High-
frequency germ-line transmission of plasmid DNA
sequences injected into fertilized zebrafish eggs. Proc
Natl Acad Sci USA 88, 7953–7957.
51 Lemaire P, Garrett N & Gurdon JB (1995) Expression
cloning of Siamois, a Xenopus homeobox gene
expressed in dorsal-vegetal cells of blastulae and able to
induce a complete secondary axis. Cell 81, 85–94.
52 Thisse C, Thisse B, Halpern ME & Postlethwait JH
(1994) Goosecoid expression in neurectoderm and
mesendoderm is disrupted in zebrafish cyclops gastrulas.
Dev Biol 164, 420–429.
53 Rissi M, Wittbrodt J, Delot E, Naegeli M & Rosa FM
(1995) Zebrafish Radar: a new member of the TGF-beta
superfamily defines dorsal regions of the neural plate
and the embryonic retina. Mech Dev 49, 223–234.
3440 FEBS Journal 272 (2005) 3424–3440 ª 2005 FEBS
BMP/activin pathway in Crassostrea gigas A. Herpin et al.