A novel carbonic anhydrase from the giant clam Tridacna
gigas contains two carbonic anhydrase domains
William Leggat
1,2
, Ross Dixon
1
, Said Saleh
1
and David Yellowlees
1
1 Biochemistry and Molecular Biology, James Cook University, Townsville, Queensland, Australia
2 Centre for Marine Studies, University of Queensland, Queensland, Australia
Carbonic anhydrase (CA; EC 4.2.1.1) catalyses the
hydration of CO
2
to HCO
3
–
, is ubiquitous amongst all
living organisms and fulfils a variety of metabolic roles
[1]. Currently there are five evolutionarily distinct CA
gene families (a, b, c, d and e) [1,2] and it is generally
believed that all animal CAs belong to the a-CA fam-
ily. a-CAs are characterized by 36 amino acids found
around the active site [3]. Of these, 15 are conserved in
all active CAs, suggesting that they are required for
CA activity [3].
To date 15 a-CA or a-CA-like proteins have been
identified in mammals. These can be divided into five
broad subgroups, the cytosolic CAs (CA I, CA II,
CA III, CA VII and CA XIII), mitochondrial CAs

anhydrase (CA) in the giant clam, Tridacna gigas. CA plays an important
role in the movement of inorganic carbon (C
i
) from the surrounding sea-
water to the symbiotic algae that are found within the clam’s tissue. One of
these isoforms is a glycoprotein which is significantly larger (70 kDa) than
any previously reported from animals (generally between 28 and 52 kDa).
This a-family CA contains two complete carbonic anhydrase domains
within the one protein, accounting for its large size; dual domain CAs have
previously only been reported from two algal species. The protein contains
a leader sequence, an N-terminal CA domain and a C-terminal CA
domain. The two CA domains have relatively little identity at the amino
acid level (29%). The genomic sequence spans in excess of 17 kb and con-
tains at least 12 introns and 13 exons. A number of these introns are in
positions that are only found in the membrane attached ⁄ secreted CAs. This
fact, along with phylogenetic analysis, suggests that this protein represents
the second example of a membrane attached invertebrate CA and it con-
tains a dual domain structure unique amongst all animal CAs characterized
to date.
Abbreviations
CA, carbonic anhydrase; C
i
, inorganic carbon; GPI, glycosylphosphatidylinositol.
FEBS Journal 272 (2005) 3297–3305 ª 2005 FEBS 3297
consolidated into two separate groups based upon
sequence similarity [3]. Of the membrane-associated
CAs, CA IX, XII and XIV contain integral trans-
membrane domains while CA IV is membrane atta-
ched through a glycosylphosphatidylinositol (GPI)
anchor.

found intercellularly within the clam tissue [9,10]. It
has been demonstrated that these alga can supply
up to 100% of the clam’s energy requirements [11].
In addition T. gigas contains two other CA isoforms,
one of 32 kDa and one of approximately 200 kDa
[9,10].
Here we present further characterization of the
70 kDa CA isoform from T. gigas, including cDNA
sequence indicating that it codes for a unique animal
CA containing two putative CA catalytic domains
within the one protein. Each domain contains those
residues thought to be essential for CA activity.
Results
tgCA cDNA sequence, deduced amino acid
sequence and domains
Only cDNA sequence for the first 1438 bp of the
70 kDa CA could be obtained from both the gill and
mantle cDNA libraries; both sequences were identical.
In both cases the sequence terminated in an identical
position (1438 bp after the start codon), suggesting that
the mRNA secondary structure prevented complete sec-
ond strand synthesis. For this reason cDNA sequence
was also obtained using RT-PCR using a higher than
normal temperature for second strand synthesis. Using
RT-PCR the 3¢ end of the cDNA was obtained yielding
an open reading frame of 1803 bp encoding 600 amino
acids and a protein of 66.7 kDa (Fig. 1). This is flanked
by a 58 bp 5¢-UTR and an 89 bp 3¢-UTR. Although no
polyadenylation signal was found a poly A tail was
sequenced. The translated protein sequence (tgCA) was

secreted ⁄ membrane-associated and the CA-related pro-
teins (Fig. 3). Both domains of the clam CA group
with the membrane-associated CAs.
Intron ⁄ exon mapping
There are a number of intron ⁄ exon locations that are
specific for the various CA classes [3,12]. With this in
mind the introns for tgCA were mapped to further
characterize the two CA domains. The genomic
sequence of tgCA, between positions 101 and 1810 of
the cDNA, was amplified in a number of PCR reac-
tions spanning in excess of 17 kb. These sequences
included the complete coding sequence for the gene
between positions 101 and 1810. Twelve introns and
13 exons were found in this region, five in n-tgCA, six
Fig. 2. Alignment of n-tgCA and c-tgCA with other CAs showing intron position and conserved motifs. The 15 amino acids thought to be
required for CA activity are indicated (#), while cysteine residues involved in disulfide bonding in CA IV (two disulfide bonds) and CA VI, XII,
XIV (one disulfide bond) are indicated (Ù) above the alignment. Numbers below the alignment indicate the intron number while intron posi-
tions are represented by: ( ⁄ ) intron between amino acids, (\) intron located after the first codon position of the following amino acid, (+)
intron located after the second codon position of the following amino acid, and (*) represents no intron present. Residues shared by more
than 50% of the CAs examined are shaded. The alignment was performed using
CLUSTALW. Note that hCA1 contains an alanine at position
122 rather then the conserved Val122, however, the consensus for the CA-1 isoform from vertebrates is valine [3]. hCA, human CA; NCBI
accession numbers in brackets, hCA1 (NM_001738), hCA2 (NM_000067), hCA3 (NM_005181), hCA4 (NM_000717), hCA5 (NM_001739),
hCA6 (NM_001215), hCA7 (NM_005182), hCA8 (NM_004056), hCA12 (NP_001209), hCA14 (NP_036245).
W. Leggat et al. A dual domain carbonic anhydrase
FEBS Journal 272 (2005) 3297–3305 ª 2005 FEBS 3299
in c-tgCA and one between the two domains (Fig. 2).
Unfortunately, despite repeated attempts it was not
possible to obtain genomic sequence corresponding to
the cDNA sequence prior to position 101 of the cDNA

CLUSTALW, bootstrapped 1000 times and the
trees constructed using maximum likelihood.
hCA, human CA; ce, Caenorhabditis
elegans;Dr,Drosophila melanogaster;ae,
Anthopleura elegantissma; CAH1 from the
alga Clamydomonas reinhardtii was used as
an outgroup.
Table 1. Exon ⁄ intron junctions of tgCA. No genomic sequence was obtained before the position corresponding to base pair 101 in the cDNA
sequence. Numbering of the cDNA sequence begins at the first codon position of the first in-frame methionine. The exact intron size is
given where known; estimates were made from the size of PCR products. Uppercase letters indicate coding sequence, while lowercase
indicate intron sequence.
Intron cDNA codon position preceding intron (bp) Intron size (bp) 5¢ Splice donor 3¢ Splice acceptor
1265  1040 CACGG gtaaac ttccag TGGTA
2417  956
a
TTGAG gtaggt ttgtag GTACA
3510  1200 TTGAG gtgggt ttctag ATCGA
4583  1830 TAAAA gttagt ttctag ATGGA
5744  2280 CTCAG gtatat tttcag CTTGC
6868  2000 TACAG gttggg ttacag CTCAA
7910  1380
a
TCAAG gtatgt ttacag GAGTG
8 1093  950 TACAA gtactt ctacag TCCAA
9 1242  1100 TCGAG gtactg tttcag CTACA
10 1335  1370 TTGAA gtaagt tttcag ATCGG
11 1399  1300,  1200
a
TAAAG gtatgt tttcag ATGCC
12 1581 347 GTCAG gtaagt ttccag TTGGT

one peptide, one from the green alga Dunaliella salina
contains two a-CA domains while the second from the
red alga Porphyridium purpureum contains two b-CA
repeats.
tgCA contains a 17 amino acid leader sequence, an
N-terminal CA domain of 272 amino acids and a
C-terminal domain of 311 amino acids. The cDNA
encodes a protein of 66.7 kDa, when the leader
sequence is removed the mature molecular mass is
reduced to 64.8 kDa, this is similar to the 62 kDa
molecular weight obtained for the deglycosylated pro-
tein determined by SDS ⁄ PAGE [9]. Both CA domains
of tgCA contain all residues thought to be required for
CA activity [3] suggesting that both domains are cata-
lytically active. In addition, both domains contain a
histidine residue (His87 in n-tgCA, His363 in c-tgCA)
conserved with His64 in human CA2 (Fig. 2). This his-
tidine residue has been found to act as a proton shuttle
in CO
2
hydration in high activity CAs (reviewed in
[14,15]) supporting the notion that both CA domains
within this protein are catalytically functional. How-
ever this will have to be confirmed experimentally
through either mutational studies or the use of select-
ive inhibitors.
The predicted cleavage position of the leader
sequence between Ala17 and Ala18 produces a mature
N-terminal protein sequence which is identical for the
first 21 amino acids to that obtained from N-terminal

result in the release of tgCA from crude gill homogen-
ate, suggesting that it is not GPI-anchored and is
instead an integral membrane protein. However analy-
sis of the predicted protein sequence presented here
does not seem to support this hypothesis, as evidence
of a putative transmembrane domain region is ambigu-
ous (data not shown). In addition the DPGI database
predicts that a GPI anchor may be present at Gly577
(it must be stated that the big-PI Predictor did not
identify a GPI-anchor for this protein). How tgCA is
associated with the membrane is still unknown and
requires further research, although the lack of a hydro-
phobic domain suggests that, as with human CA4, it is
attached through a GPI-anchor.
Perhaps surprisingly there is very little identity
between the two CA domains of this protein at either
the coding (48%) or amino acid level (29%). This level
of identity is similar to that seen when comparing the
different domains of tgCA to human CAs. Of the ver-
tebrate and invertebrate CAs used for the tree con-
struction (Fig. 3), n-tgCA had the greatest identity at
the amino acid level with human CA2 and human
CA7 (32% for both) while c-tgCA was most similar to
human CA1 and human CA7 (28% identity for both).
The lowest identity was with the alga C. reinhardtii
CAH1 gene (13% n-tgCA and 16% c-tgCA).
Despite this greater identity with human cytosolic
CAs, phylogenetic analysis suggests that both the
N- and C-terminal domains belong to the secreted ⁄
membrane-associated CA group (Fig. 3). Despite low

shows a similar pattern in the presence of 5% 2-merca-
ptoethanol [16].
While four of the six cysteines in tgCA are implica-
ted in intrachain disulfide bonds, evidence suggests
that at least one of the remaining two cysteines forms
a disulfide bond with another tgCA subunit making a
dimer of the 70 kDa protein. Gel filtration experiments
had previously suggested that tgCA exists as a dimer,
with a native weight of approximately 141 kDa [9].
When a gel purified fraction of similar estimated mole-
cular mass (145 kDa) is separated by SDS ⁄ PAGE, the
70 kDa band is found in addition to the original
145 kDa protein. Upon addition of high concentra-
tions of the reducing agent 2-mercaptoethanol this
band ( 145 kDa) disappears (Fig. 4B) supporting the
conclusion that this represents a dimer. The gel filtra-
tion results in combination with reduction of the
145 kDa protein to the 70 kDa isoform suggest that
there are interchain disulfide bonds between two
70 kDa subunits.
Analysis of the genomic sequence data, where differ-
ent intron sequences have been obtained (Table 1),
Southern blots (data not shown) and protein two-
dimensional gels all suggest that tgCA is a multicopy
gene. Despite this no sequence differences were
observed in the coding sequence or intron position
where different copies, evidenced by different intron
sequences, were obtained. Given this it seems reason-
able then to use the combined genomic data, even if it
does not represent one gene, for analysis of intron ⁄

The dual domain structure of tgCA could have arisen
through one of two mechanisms, either the fusion of
two separate CA genes or a duplication of a single gene
followed by a fusion event. If this protein arose through
duplication and fusion event, and given the poor iden-
tity between the two CA domains (29%), the duplication
event must be old, thereby allowing time for the two
domains to diverge. This low identity between domains
is especially striking when compared to other duplicated
domain CA proteins, 52 and 72% identity for D. salina
and P. purpureum, respectively [17,18]. Furthermore
there are similar examples of non-CA duplicated domain
proteins from invertebrates. Phosphagen kinases from a
number of bivalves [19–21], sea anemones [22] and sea
urchins [23] have been shown to contain bi- or tripartite
repeat domains. In all of these examples, identity
between the domains is in excess of 60%. For each pro-
tein it has been concluded that the dual domain struc-
ture arose through gene duplication of one gene and
subsequent fusion. Where genomic sequence is available
[20,22] this is supported by the presence of an intron
between the two domains. Similarly in tgCA an intron is
found between the two domains. Given the low homol-
ogy between the two domains of tgCA in comparison to
other duplicated domain proteins it is not possible
to exclude the possibility that this protein has arisen
through the fusion of two different CA genes rather
than a duplication event.
Given the unique structure of this CA protein it
would be interesting to know if both domains display

Pharmacia, Piscataway, NJ, USA) which was then further
separated on an 8–18% SDS ⁄ PAGE gradient gel using the
manufacturer’s protocol (Pharmacia, Cat # 18-1038-63).
Gels were visualized using Sypro-Ruby (Molecular Probes,
Eugene, OR, USA).
Purification of RNA and cDNA library construction
Total RNA was prepared from T. gigas mantle and gill
tissue. Fresh tissue (1.3 g) was snap frozen in liquid nitro-
gen and ground in a mortar and pestle. Total RNA was
prepared from the tissue using cesium chloride [24].
mRNA was then purified from total RNA using the
QuickPrepÒ mRNA Purification Kit (Pharmacia). The gill
cDNA was synthesized for rapid amplification of cDNA
ends by the polymerase chain reaction (RACE-PCR) using
the Clontech
TM
cDNA Amplification Kit. The mantle lib-
rary was initially synthesized as a phage library in k-ZapII
(Stratagene, La Jolla, CA, USA). It was used as a
template for RACE-PCR using specific primers for the
adaptors.
Clam CA primers were designed to previously known
cDNA sequence of the 70 kDa CA isoform from T. gigas
[25] and to N-terminal amino acid sequence [9] (Fig. 1).
From the derived sequence further primers were designed
to amplify the remaining portion of the cDNA.
Products were also amplified using RT-PCR. mRNA was
purified as previously described and first strand synthesis
performed using Omniscript
TM

the Centre for Biological Sequence Analysis database [28].
Potential GPI-anchor sites were examined using the DPGI
database (http://129.194.185.165/dgpi/DGPI_demo_en.html)
and the big-PI Predictor ( />gpi_server.html) [29].
Isolation of genomic DNA
Genomic DNA was isolated from T. gigas sperm. Spawning
was induced by the injection of approximately 5 mL of a
2mm serotonin solution into the gonads. The sperm was
collected from the water and centrifuged (1000 g for
5 min). Sperm (1 mL packed cell volume) was diluted with
11 mL proteinase K solution [50 mm Tris ⁄ HCl pH 7.5,
20 mm EDTA, 100 mm NaCl, 1% (w ⁄ v) SDS, 100 lgÆmL
)1
proteinase K] and the sample was incubated overnight at
55 °C. The solution was centrifuged (1000 g for 15 min at
4 °C), the supernatant removed, mixed with 10 mL of
ultra-pure phenol (Sigma, St Louis, MO, USA) and then
equilibrated with 4 mL of TE (10 mm Tris pH 8.0, 1 mm
EDTA) buffer. After adding an equal volume of chloro-
form, the solution was left overnight. The solution was
again centrifuged (1000 g for 15 min at 4 °C) and the aque-
ous phase removed. This was re-extracted twice with chlo-
roform and the DNA precipitated with 0.1 volume sodium
acetate (3 m, pH 5.2) and 2.5 volume 100% (v ⁄ v) ethanol.
After precipitation the DNA was spooled and resuspended
in TE buffer.
Genomic sequencing
The intron ⁄ exon structure of the 70 kDa CA was mapped
using a series of sequence specific primers obtained from the
cDNA sequence that bracketed possible intron positions [3].

tion and inhibition. J Exp Biol 205, 591–602.
7 Seron TJ, Hill J & Linser PJ (2004) A GPI-linked car-
bonic anhydrase expressed in the larval mosquito mid-
gut. J Exp Biol 207, 4559–4572.
8 Miyamoto H, Miyashita T, Okushima M, Nakano S,
Morita T & Matsushiro A (1996) A carbonic anhydrase
from the nacreous layer in oyster pearls. Proc Natl Acad
Sci USA 93, 9657–9660.
9 Baillie B & Yellowlees D (1998) Characterization and
function of carbonic anhydrase in the zooxanthellae-
giant clam symbiosis. Proc R Soc Lond B 265, 465–473.
A dual domain carbonic anhydrase W. Leggat et al.
3304 FEBS Journal 272 (2005) 3297–3305 ª 2005 FEBS
10 Leggat W, Marendy EM, Baillie B, Whitney SM,
Ludwig M, Badger MR & Yellowlees D (2002) Dinofla-
gellate symbioses: strategies and adaptations for the
acquisition and fixation of inorganic carbon. Funct
Plant Biol 29, 309–322.
11 Fisher CR, Fitt WK & Trench RK (1985) Photosynth-
esis and respiration in Tridacna gigas as a function of
irradiance and size. Biol Bull 169, 230–245.
12 Jiang W & Gupta D (1999) Structure of the carbonic
anhydrase VI (CA6) gene: evidence for two distinct
groups within the a-CA gene family. Biochem J 344,
385–390.
13 Mount SM (1982) A catalog of splice junction
sequences. Nucleic Acids Res 10, 459–472.
14 Lindskog S (1997) Structure and mechanism of carbonic
anhydrase. Pharmacol Ther 74, 1–20.
15 Supuran CT, Scozzafava A & Casini A (2003) Carbonic

cDNA-derived amino acid sequence of two-domain
arginine kinase from the sea anemone Anthopleura japo-
nicus. Biochem J 328, 301–306.
23 Wothe DD, Charbonneau H & Shapiro BM (1990) The
phosphocreatine shuttle of sea urchin sperm: flagellar
creatine kinase resulted from a gene triplication. Proc
Natl Acad Sci USA 87, 5203–5207.
24 Sambrook J, Fritsch EF & Manniatis T (1989) Mole-
cular Cloning, 2nd edn. Cold Spring Harbor Laboratory
Press, Cold Spring Harbour, NY.
25 Baillie BK, (1995) Inorganic carbon acquisition and utili-
sation in the giant clam symbiosis. PhD Thesis, James
Cook University of North Queensland, Australia.
26 Thompson JD, Higgins DG & Gibson TJ (1994) CLUS-
TAL W: improving the sensitivity of progressive multi-
ple sequence alignment through sequence weighting,
position-specific gap penalties and weight matrix choice.
Nucleic Acids Res 22, 4673–4680.
27 Felsenstein J (1989) P
HYLIP – Phylogeny Inference Pack-
age Version 3.2. Cladistics, 5, 164–166.
28 Bendtsten JD, Nielsen H, von Heijne G & Brunak S
(2004) Improved prediction of signal peptides: SignalP
3.0. J Mol Biol 340, 783–795.
29 Eisenhaber B, Bork P & Eisenhaber F (1999) Prediction
of potential GPI-modification sites in proprotein
sequences. J Mol Biol 292, 741–758.
W. Leggat et al. A dual domain carbonic anhydrase
FEBS Journal 272 (2005) 3297–3305 ª 2005 FEBS 3305