13
Characterization of Carbonic Anhydrases from Riftia pachyptila, a Symbiotic Invertebrate from Deep-Sea Hydrothermal Vents Marie-Ce ´ cile De Cian, 1 * Xavier Bailly, 1 Julia Morales, 2 Jean-Marc Strub, 3 Alain Van Dorsselaer, 3 and Franc ¸ ois H. Lallier 1 1 Equipe Ecophysiologie, CNRS-UPMC UMR 7127 CEOBM, Station Biologique, Roscoff Cedex, France 2 Equipe Cycle Cellulaire et De ´veloppement, CNRS-UPMC UMR 7127 CEOBM, Station Biologique, Roscoff Cedex, France 3 Laboratoire de Spectrome ´trie de Masse Bio-Organique, CNRS-ULP UMR 7509, ECPM, Strasbourg Cedex, France ABSTRACT The symbiotic hydrothermal vent tubeworm Riftia pachyptila needs to supply its internal bacterial symbionts with carbon dioxide, their inorganic carbon source. Our aim in this study was to characterize the carbonic anhydrase (CA) involved in CO 2 transport and conversion at various steps in the plume and the symbiotic tissue, the trophosome. A complete 1209 kb cDNA has been sequenced from the trophosome and identified as a putative -CA based on BLAST analysis and the similarities of total deduced amino-acid sequence with those from the GenBank database. In the plume, the putative CA sequence obtained from cDNA li- brary screening was 90% identical to the tropho- some CA, except in the first 77 nucleotides down- stream from the initiation site identified on trophosome CA. A phylogenetic analysis showed that the annelidan Riftia CA (CARp) emerges clus- tered with invertebrate CAs, the arthropodan Dro- sophila CA and the cnidarian Anthopleura CA. This invertebrate cluster appeared as a sister group of the cluster comprising mitochondrial and cytosolic isoforms in vertebrates: CAV, CAI II and III, and CAVII. However, amino acid sequence alignment showed that Riftia CA was closer to cytosolic CA than to mitochondrial CA. Combined biochemical approaches revealed two cytosolic CAs with differ- ent molecular weights and pI’s in the plume and the trophosome, and the occurrence of a membrane- bound CA isoform in addition to the cytosolic one in the trophosome. The physiologic roles of cytosolic CA in both tissues and supplementary membrane- bound CA isoform in the trophosome in the optimiza- tion of CO 2 transport and conversion are discussed. Proteins 2003;51:327–339. © 2003 Wiley-Liss, Inc. Key words: symbiosis; carbonic anhydrase; cDNA sequence; phylogenetic analysis; MALDI- TOF; cytosolic and membrane-bound isoforms INTRODUCTION Carbonic anhydrases (CAs) are key enzymes for the maintenance of homeostasis in a wide range of organisms. These zinc-containing enzymes catalyze the reversible hydration of CO 2 into bicarbonate and a proton. 1 CAs are presently ordered in three distinct classes of gene families: , , and , corresponding schematically to animal, plant, and bacterial CA. 2 The occurrence of a new type of CA without any significant identity to the three classes, has been recently demonstrated in the marine diatom Thalas- siosira weissflogii by Roberts and coworkers. 3 The -CA group constitutes a multigenic family coding for cytosolic (CA-I, CA-II, CA-III, CA-VII, CA-VIII), mitochondrial (CA- V), membrane-associated (CA-IV, CA-IX), or secreted (CA- VI) isoforms. 4 Depending on their subcellular localization, -CAs are involved in different physiologic processes, including acid– base homeostasis, CO 2 and ion transport, respiration 5 or calcification, 6 reproduction, and neurosen- sitive processes. 7 Historically, the first discovered CA isoforms were -CAs from human red blood cells. 8,9 How- ever, in the last 20 years, a remarkable number of -CAs have been described from nonmammalian organisms, in- cluding zebrafish, 10 flounder, 11 oysters, 12 crabs, 13 sea anemones, 14 and viruses. 15 CA activity was also demon- strated in autotrophic organisms such as plants (- CA) 16,17 or bacteria (-CA), 18 in which CA appears directly involved in carbon supply for the Rubisco (ribulose 1,5- biphosphate carboxylase) enzyme responsible for carbon fixation. Deep-sea hydrothermal vent environments are consid- ered extreme given the high pressure and temperature conditions, the toxicity of surrounding chemicals, and the total lack of phototrophic production for animal nutrition. The ability of hydrothermal vent communities to thrive and survive under such conditions is mainly due to molecu- lar, biochemical, and physiologic adaptations that enable the organisms to maintain vital functions. This study investigates CA in the model organism Riftia pachyptila, a hydrothermal vent annelid. In this worm, adaptation to extreme conditions led to the emergence of Contract grant sponsor: French Research Department; Contract grant sponsor: DORSALE *Correspondence to: Marie-Ce ´cile De Cian, Ecophysiologie, Station Biologique, Place Georges Teissier, BP 74, 29682 Roscoff Cedex, France. E-mail : [email protected] Received 1 July 2002; Accepted 5 September 2002 PROTEINS: Structure, Function, and Genetics 51:327–339 (2003) © 2003 WILEY-LISS, INC.

Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

Embed Size (px)

Citation preview

Page 1: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

Characterization of Carbonic Anhydrases from Riftiapachyptila, a Symbiotic Invertebrate from Deep-SeaHydrothermal VentsMarie-Cecile De Cian,1* Xavier Bailly,1 Julia Morales,2 Jean-Marc Strub,3 Alain Van Dorsselaer,3

and Francois H. Lallier1

1Equipe Ecophysiologie, CNRS-UPMC UMR 7127 CEOBM, Station Biologique, Roscoff Cedex, France2Equipe Cycle Cellulaire et Developpement, CNRS-UPMC UMR 7127 CEOBM, Station Biologique, Roscoff Cedex, France3Laboratoire de Spectrometrie de Masse Bio-Organique, CNRS-ULP UMR 7509, ECPM, Strasbourg Cedex, France

ABSTRACT The symbiotic hydrothermal venttubeworm Riftia pachyptila needs to supply itsinternal bacterial symbionts with carbon dioxide,their inorganic carbon source. Our aim in this studywas to characterize the carbonic anhydrase (CA)involved in CO2 transport and conversion at varioussteps in the plume and the symbiotic tissue, thetrophosome. A complete 1209 kb cDNA has beensequenced from the trophosome and identified as aputative �-CA based on BLAST analysis and thesimilarities of total deduced amino-acid sequencewith those from the GenBank database. In the plume,the putative CA sequence obtained from cDNA li-brary screening was 90% identical to the tropho-some CA, except in the first 77 nucleotides down-stream from the initiation site identified ontrophosome CA. A phylogenetic analysis showedthat the annelidan Riftia CA (CARp) emerges clus-tered with invertebrate CAs, the arthropodan Dro-sophila CA and the cnidarian Anthopleura CA. Thisinvertebrate cluster appeared as a sister group ofthe cluster comprising mitochondrial and cytosolicisoforms in vertebrates: CAV, CAI II and III, andCAVII. However, amino acid sequence alignmentshowed that Riftia CA was closer to cytosolic CAthan to mitochondrial CA. Combined biochemicalapproaches revealed two cytosolic CAs with differ-ent molecular weights and pI’s in the plume and thetrophosome, and the occurrence of a membrane-bound CA isoform in addition to the cytosolic one inthe trophosome. The physiologic roles of cytosolicCA in both tissues and supplementary membrane-bound CA isoform in the trophosome in the optimiza-tion of CO2 transport and conversion are discussed.Proteins 2003;51:327–339. © 2003 Wiley-Liss, Inc.

Key words: symbiosis; carbonic anhydrase; cDNAsequence; phylogenetic analysis; MALDI-TOF; cytosolic and membrane-boundisoforms

INTRODUCTION

Carbonic anhydrases (CAs) are key enzymes for themaintenance of homeostasis in a wide range of organisms.These zinc-containing enzymes catalyze the reversible

hydration of CO2 into bicarbonate and a proton.1 CAs arepresently ordered in three distinct classes of gene families:�, �, and �, corresponding schematically to animal, plant,and bacterial CA.2 The occurrence of a new type of CAwithout any significant identity to the three classes, hasbeen recently demonstrated in the marine diatom Thalas-siosira weissflogii by Roberts and coworkers.3 The �-CAgroup constitutes a multigenic family coding for cytosolic(CA-I, CA-II, CA-III, CA-VII, CA-VIII), mitochondrial (CA-V), membrane-associated (CA-IV, CA-IX), or secreted (CA-VI) isoforms.4 Depending on their subcellular localization,�-CAs are involved in different physiologic processes,including acid–base homeostasis, CO2 and ion transport,respiration5 or calcification,6 reproduction, and neurosen-sitive processes.7 Historically, the first discovered CAisoforms were �-CAs from human red blood cells.8,9 How-ever, in the last 20 years, a remarkable number of �-CAshave been described from nonmammalian organisms, in-cluding zebrafish,10 flounder,11 oysters,12 crabs,13 seaanemones,14 and viruses.15 CA activity was also demon-strated in autotrophic organisms such as plants (�-CA)16,17 or bacteria (�-CA),18 in which CA appears directlyinvolved in carbon supply for the Rubisco (ribulose 1,5-biphosphate carboxylase) enzyme responsible for carbonfixation.

Deep-sea hydrothermal vent environments are consid-ered extreme given the high pressure and temperatureconditions, the toxicity of surrounding chemicals, and thetotal lack of phototrophic production for animal nutrition.The ability of hydrothermal vent communities to thriveand survive under such conditions is mainly due to molecu-lar, biochemical, and physiologic adaptations that enablethe organisms to maintain vital functions.

This study investigates CA in the model organism Riftiapachyptila, a hydrothermal vent annelid. In this worm,adaptation to extreme conditions led to the emergence of

Contract grant sponsor: French Research Department; Contractgrant sponsor: DORSALE

*Correspondence to: Marie-Cecile De Cian, Ecophysiologie, StationBiologique, Place Georges Teissier, BP 74, 29682 Roscoff Cedex,France. E-mail : [email protected]

Received 1 July 2002; Accepted 5 September 2002

PROTEINS: Structure, Function, and Genetics 51:327–339 (2003)

© 2003 WILEY-LISS, INC.

Page 2: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

unique features, including an unusual mode of animalnutrition that relies on chemoautotrophic bacterial symbi-onts, which are maintained within specialized cells calledbacteriocytes, located in an internal organ, the tropho-some. The symbionts generate energy by oxidizing reducedsulfur expelled from hydrothermal vents, and requiremolecular CO2 for fixation into organic carbon, which isthen used by the host.19,20 Indeed, this symbiosis involvesa net inorganic carbon uptake in the presence of hydrogensulfide.21 Because within R. pachyptila the symbioticbacteria are remotely located relative to the surroundingmedium, the transport of CO2 is a stepwise process.Inorganic carbon (CO2) is first acquired from the environ-ment by diffusion across a branchial organ, the plume.22

CO2 then accumulates in the body fluids23 and is trans-ported to trophosome cells mainly in the form of bicarbon-ate.24 At the level of the bacteriocytes, bicarbonates arepresumably converted into CO2 again, because bacterialsymbionts express Rubisco enzyme form II,25 which can-not use HCO3

� as a carbon source.The physiologic necessity for CO2 conversion, both at the

environment–branchial plume interface (CO2 to HCO3�)

and at the body fluid–bacteriocyte interface (HCO3� to

CO2),26,27 suggested the existence of CA activity in R.pachyptila. The first evidence for a CA activity in R.pachyptila in plume and trophosome tissues was given byKochevar and coworkers,28 and was recently confirmed byphysiologic and immunologic observations on isolated bac-teriocytes.29,30

Our aim in this study was to extend previous research byproviding molecular and biochemical data on R. pachyptilaCAs, because in this symbiotic system involving bothmetazoan and bacterial components, the number andorigin of CA isoforms remained an open issue. In addition,the unique life mode of R. pachyptila prompted us toexamine possible functional and structural adaptations ofCAs in this model organism.

MATERIAL AND METHODSAnimals and Sampling

Juvenile specimens of R. pachyptila were collected at the9° N site (2600-m depth) along the East Pacific Rise during

the HOPE 99 cruise. Animals collected with the telema-nipulated arm of the submersible Nautile were brought tothe surface in a temperature-insulated container. Uponarrival onboard, young specimens (2–5 cm) were checkedfor tactile responsiveness after removal from their tube.Whole animals were immediately frozen and stored inliquid nitrogen for total DNA or RNA extraction. Tissuepieces of branchial plume, vestimentum, body wall, andtrophosome (the symbiotic tissue31) were rapidly dissectedon ice, washed several times with DNase- and RNase-freeRiftia saline,23 and frozen separately in liquid nitrogenuntil used.

Characterization of Carbonic Anhydrase cDNACA-specific primers design

Degenerate forward and reverse primers were designedaccording to conserved regions of deduced amino acidsequences from the GenBank database, belonging to eithercytoplasmic, mitochondrial, membrane-associated, or se-creted isoforms from �-, �-, and �-CAs. Computer-assistedprimer design (Oligo 4.0) was based on the CA amino acidsequence of the symbiotic cnidarian Anthopleura elegan-tissima (GenBank accession number AF140537) to refer-ence the relative position of the primers along the se-quence. The resulting primers are presented in Table I.

RNA extraction and cDNA synthesis

The four main body compartments (vestimentum, plume,trophosome, and body wall) of R. pachyptila were pulver-ized separately in liquid nitrogen. Total RNAs from eachorgan were then extracted and recovered using RNAble�

buffer (Eurobio). Messenger RNAs were purified from totalRNAs using the oligodT resin of the mRNA PurificationKit� (Pharmacia Biotech). We initiated reverse transcrip-tion polymerase chain reaction (RT-PCR) with an an-chored oligodT primer (Table I) using 200 U/�L M-MLVReverse Transcriptase (Promega).

Production of a CA cDNA probe

Partial CA cDNA was amplified from plume and tropho-some cDNA matrix by PCR with a set of degenerateprimers on a Perkin Elmer GenAmp PCR System 2400�.

TABLE I. Sequences of Oligonucleotide Primers Used to Amplify CA cDNA from Plume and Trophosome ofRiftia pachyptila (EMBL Accession No. AJ439711)

Primer set Corresponding number Nucleotide position Nucleotide sequence (5�3 3�)

CA4A For* 1 246–267 GAG CAR TTY CAY TTY CAY TGGCA8 Rev* 2 450–469 CGG ART ANG TCC ART ART CCA For 3 103–123 GTG TAA GCA AGA AGT CAA CCCA Rev 4 825–843 TAC TGA TGA CGG CGC TGRaceCA2b Rev 5 223–244 CTT TGT ATT CGT TGC CCA GTGAnchor Rev 6 Downstream CTC CTC TCC TCT CCT CTTPoly G For 7 Upstream GGG GGG GGG GGG GGG G

*Primer design relative to Anthopleura elegantissima CA cDNA (GenBank Accession No.: AF140537) for the design of the firstdegenerated primers: N, A/C/G/T; R, A/G; Y, C/T. corresponds to complete coding sequence; corresponds to identical 225 bp cDNAfragment obtained from trophosome and plume with degenerated primers.

328 M.-C. DE CIAN ET AL.

Page 3: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

Amplifications were carried out in 25-�L reaction mix-tures containing 10–50 ng cDNA target, 50–100 ng of eachdegenerate primer, 200 �M deoxyribonucleoside 5�-triphosphates (dNTPs), 2.5 mM MgCl2, and 1 unit of TaqDNA polymerase (Promega). PCR products with the ex-pected size were cloned with a TOPO-TA Cloning� kit(Invitrogen). The cloned CA cDNA fragment of 225 bpamplified with CA4 For and CA8 Rev primers (Table I) waslabeled with radioactive dCTP32 and used as a probe forlibrary screening.

Construction of cDNA library

Purified poly(A) RNA from vestimentum and branchialplume was used. Double-stranded cDNAs were fraction-ated through a Sephacryl S-400 column. The fraction withan average size of 500–2000 bp, as estimated on anacrylamide gel, was selected for the preparation of the5�-3�-oriented �ZAP II library (Stratagene, La Jolla, CA).The library was constructed following the manufacturer’sinstructions. The titer of the primary library was 7 � 107

pfu/mL�1.

Trophosome CA cDNA amplification

In order to amplify the CA sequence upstream anddownstream from the 225 kb cDNA fragment obtainedfrom trophosome, we applied the manufacturer’s instruc-tions of 5�/3�-RACE Kit (Roche Diagnostics). Internalprimer (CA4A For) and 3�-Anchor Rev primer sequencesare detailed in Table I. For 5� amplification, multipledeoxycytidine 5�-triphosphate (dCTP) were added to the 5�end of trophosome cDNA using 14 U/�L Terminal Deoxy-nucleotidyl-Transferase (Amersham). C-tailed cDNA wasused in an initial PCR containing 10 pmol of polyG (16 bp)forward primer and 10 pmol of RaceCA2b reverse primer(Table I). We performed a second PCR reaction with thesame primer set using 10 �L of the first PCR reaction as amatrix for the second PCR reaction. The cDNA fragmentamplified was visualized on an agarose gel and cloned asdescribed above.

DNA sequencing

Plasmid DNA from individual colonies containing theCA insert were purified with a FlexiPrep kit� (Amersham)and used in a dye-primer cycle sequencing reaction witheither labeled Texas Red universal T7 primer (5�-GTAATA CGA CTC ACT ATA GGG C-3�) or M13 reverseprimer (5�-GGA AAC AGC TAT GAC CAT G-3�) and theThermo Sequenase™ premixed cycle sequencing kit fromAmersham. PCR products were subsequently run on aVistra automated DNA Sequencer 725 (Amersham).

Sequence and molecular phylogeny analyses

The EMBL accession number for Riftia CA is AJ439711.All sequences for comparisons were retrieved from proteindatabases using the Entrez program (National Center forBiotechnology Information, NCBI). Deduced protein se-quence was checked for CA similarity using the BLASTprogram.32 Multiple sequence alignments were obtainedwith the Clustal X software.33 NCBI protein accessionnumbers are mouseCAI (P13634), humanCAI (P00915),

mouseCAII (P00920), chickCAII (P07630), ratCAIII(014141), mouseCAIV (Q64444), ratCAIV (P48284), rabbit-CAIV (P48283), humanCAIV (P22748), mouseCAV(Q9Q7A0), humanCAV (Q9Y2D0), ratCAV (P43165), hu-manCAVI (P23280), mouseCAVI (P18761), humanCAVII(P43166), mouseCAVII (AAG16230), D. melanogasterCA(NP_523561), zebrafishCA (T08463), A. elegantissimaCA(AF140537), C. jejuniCA (CAB72706), and B. melitensisCA(AAL51404). A molecular phylogenetic tree was con-structed with the use of a Neighbor-Joining (NJ) methodcomputed with PHYLIP software.34 Phylogenetic dis-tances were computed from expected under the Dayhoff’sPAM matrix fixed in ProtDist step. Bootstrap proportionswere obtained from 1000 resampling sequence alignmentsfor NJ. We performed hydrophobic cluster analysis (HCAplots) to predict the secondary structure of primary CAamino acid sequences, using the DrawHCA software.35

Protein Extraction

Protein extracts were prepared from 1 g of frozen tissueof each body part. The tissues were homogenized on iceusing a Tissuemizer (Ultra-Turrax, Janke-Kuntel, IkaLabortechnik, Staufen) in three 30-s bursts in 4 mLextraction buffer composed of 50 mM Tris-HCl, pH 7.5; 200mM sucrose; 150 mM KCl containing protease inhibitors,including 2 mM ethylenediaminetetraacetic acid (EDTA);1 mM Pefablock (Boerhinger Mannheim); 5 �g/mL chymo-statin; 10 �g/mL leupeptin; and 10 �g/mL aprotinin.Samples were kept on ice during the whole procedure andcentrifuged at 10,000 g for 20 min at 4°C. The supernatant(S1) was transferred to a new tube and stored at 4°C. Thepellet was washed twice and then solubilized for 1 h in 2mL of membrane buffer containing Triton X-100 1% (3 vol),Tris buffer used above (7 vol), and saponin 0.2% (1 vol).36

Solubilization of the proteins was achieved by breaking upthe membrane pellet with sonication. S1 and the solubi-lized pellet (P1) were then ultracentrifuged at 100,000 gfor 90 min at 4°C, resulting in supernatants S2 (from S1,cytosolic proteins) and S3 (from P1, membrane-associatedproteins). Protein concentrations were determined usingthe Bradford method,37 and samples were stored in homog-enization buffer at �80°C. Proteins from the differentfractions were resolved in one dimension using 12.5%polyacrylamide gels with sodium dodecyl sulfate38 in aHoeffer electrophoresis system according to the manufac-turer’s instructions. Two-dimensional polyacrylamide gelelectrophoresis (PAGE) was performed on a Multiphor IISystem (Pharmacia Biotech) according to the manufactur-er’s instructions, with the following modifications: Theproteins were precipitated with 10% trichloroacetic acid(TCA)-acetone at �40°C for 2 h, centrifuged, and resus-pended in 2 M thiourea, 7 M urea, 4% CHAPS {3-[(3-cholamidopropyl)dimethyammonio]-1-propane-sulfonate},and 0.6% ampholyte 4-6 and 5-7 mixture (Biorad). Aliquotsof 200 �g of proteins were used to rehydrate ReadyStrips™IPG strips (Biorad) overnight.

Biochemical AnalysisWestern blots

We transferred proteins to a nitrocellulose membrane(Bio-Rad) using a wet transfer unit for 2 h at 4°C and 80 V.

CARBONIC ANHYDRASE IN RIFTIA PACHYPTILA 329

Page 4: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

All blocking and antiserum incubations were performed inTBST (20 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 0.05% Tween20). The membrane was blocked 2 h in 5% skim milk-trisbuffer solution-Tween 20 (TBST) at room temperature,and then split up in two identical parts. The first part wasincubated overnight at 4°C with rabbit antichick CA-IIantiserum (generous gift from Dr. P. Linser, WhitneyLaboratory, University of Florida, St. Augustine) diluted1:1000, and the second part with rabbit antirat lung CA IVantiserum (generous gift from Dr. W. S. Sly, Departmentof Biochemistry and Molecular Biology, St. Louis Univer-sity School of Medicine, St Louis, MO) diluted 1:1000.Bovine CA present in the molecular weight marker mix[low molecular weight (LMW) electrophoresis calibrationkit, Amersham Pharmacia Biotech. SA, France] was usedas a positive standard for CA. Preimmune rabbit serum(Sigma) and omission of the primary antibody were usedas negative controls. The membrane was then rinsed 3times in TBST and incubated for 2 h at room temperaturewith a 1:3000 dilution of pig antirabbit horseradish peroxi-dase (HRP)-conjugated Immunoglobulin G (IgG) antibody(DAKO A/S, Denmark). The antigen–antibody complexwas revealed by a chemiluminescence system (ECL, Amer-sham Pharmacia Biotech, SA, France) according to themanufacturer’s instructions, and by exposing the nitrocel-lulose membrane to autoradiographic film X-OMAT AR(Eastman Kodak Co., NY).

In-gel digestion procedure

For internal peptide sequencing, proteins of interestfrom the trophosome and the branchial plume were visual-ized in gel with Coomassie R250 staining. The bandscorresponding to anti-CA-IV and anti-CA-II immunoreac-tive proteins were excised from the gel with a razor bladeand washed with 100 �L of 25 mM NH4HCO3, andagitated for 8 min with a vortex mixer before removing theNH4HCO3 solution.

Gel pieces were then submitted to the following steps,the solution covering the gel being removed between eachstep: dehydration with acetonitrile; vacuum drying; alkyla-tion–reduction of the proteins with 10 mM DTT in 25 mMNH4HCO3 at 57°C for 1 h, followed by incubation with 55mM iodoacetamide in 25 mM NH4HCO3 in the dark atroom temperature for 1 h; 3 washing steps with NH4HCO3

and acetonitrile; vacuum drying; and then digestion of theproteins with three volumes of trypsin (12.5 ng/�L) in 25mM NH4HCO3 (freshly diluted) for one volume of gel at35°C overnight. The gel pieces were eventually centri-fuged, and 5 �L of a 25% H2O-70% acetonitrile-5% HCOOHsolution was added. The mixture was sonicated 5 min andcentrifuged again. The supernatant containing the elutedproteins was reduced under nitrogen flow to 4 �L. Then, 1�L of H2O/5% HCOOH was added, and 0.5 �L was used forMALDI-TOF analysis.

MALDI-TOF

Mass measurements were carried out on a BruckerBIFLEX III™ matrix-assisted laser desorption time-of-flight mass spectrometer (MALDI-TOF) equipped with theSCOUT™ High Resolution Optics with X-Y multisample

probe and gridless reflector. This instrument was used at amaximum accelerating potential of 19 kV and operated inreflector mode. Ionization was accomplished with a 337-nmbeam from a nitrogen laser with a repetition rate of 3 Hz.The output signal from the detector was digitized at asampling rate of 2 GHz. A saturated solution of cyano-4-hydroxycinnamic acid in acetone was used as a matrix. Afirst layer of fine matrix crystals was obtained by spread-ing and fast evaporation in 0.5 �L of matrix solution. Onthis fine layer of crystals, a droplet of 0.5 �L aqueousHCOOH (5%) solution was deposited. Afterwards, 0.5 �Lof sample solution was added and a second droplet of 0.2�L matrix saturated solution (in 50% H2O/50% acetoni-trile) was added. The preparation was dried under vacuum.The sample was washed one to three times by applying 1�L of aqueous formic acid (HCOOH, 5%) solution on thetarget and then flushed after a few seconds. The calibra-tion was performed in internal mode with 4 peptides:angiotensin 1046.542 Da; Substance P 1347.736 Da; bomb-esine 1620.807 Da; and trypsine autolysis fragment 842.510Da, 2211.107 Da.

CA purification and microsequencing

Cytosolic protein extracts from the plume- and mem-brane-associated proteins from the trophosome were frac-tionated onto a HR-75 gel filtration column (AmershamPharmacia Biotech) with the use of veronal buffer [25 mMbarbital buffer pH 8.2 (Sigma), 5 mM EDTA, 5 mMdithiothreitol (DTT), 10 mM MgSO4] as the eluant. In theplume, the fraction corresponding to putative cytosolic CAwas further purified with reverse-phase high pressureliquid chromatography (HPLC) before submission to Ed-man reaction and N-terminal microsequencing throughmass spectrometry analysis. In the trophosome, the differ-ent fractions obtained from membrane-associated proteinextract were tested for CA activity separately.

CA activity

CA activity was tested for qualitative purposes and notfor the quantitative measurement of its maximal activity.From each fraction, 50 �L was mixed with 70 �L ofbromothymol blue (BTB)-veronal buffer (25 mM barbitalbuffer, pH 8.2, 5 mM EDTA, 5 mM DTT, 10 mM MgSO4,0.7% BTB) and 90 �L of ice-cold CO2-saturated distilledwater in a tube held at 4°C. The rate of change in pHfollowing addition of the CO2-saturated water was fol-lowed, from pH 8.2 to 6.9. The specific activities of CA foreach assay were calculated as (pHs � pHb)/min�1/mgprotein�1, where pHs is the rate of pH change in thesample, and pHb is the rate of pH change in the samesample boiled for 5 min, equivalent to noncatalyzed hydra-tion time of CO2.14,28,39

RESULTSSequence Analysis

Using the degenerate primers CA4A For and CA8 Rev(Table I), an identical 225-bp fragment was obtained fromtrophosome and plume cDNAs.

330 M.-C. DE CIAN ET AL.

Page 5: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

Trophosome CA

Since no cDNA library relative to the trophosome wasavailable, the complete putative CA sequence was ob-tained by overlapping PCR amplifications from tropho-some cDNA with different primer sets: CA For-CA Rev,and CA4A For-Anchor Rev (Table I). In order to obtain the5�-UTR (untranslated region), we used a rapid amplifica-tion of 5� cDNA ends (5� RACE-PCR) strategy. A singleproduct, 1209 bp in length minus the poly-A tail, with a731-nucleotide open reading frame (ORF), was sequenced.A poly-A tail signal (AATAAA) occured 248 nucleotidesdownstream from the inframe stop codon and 11 nucleo-tides upstream from the poly-A tail. The resulting se-quence (Fig. 1) [European Molecular Biology Laboratory(EMBL) accession number AJ439711] was translated andsubmitted to a BLAST search (NCBI). The sequence fromRiftia was assumed to correspond to a putative �-CAsequence, because all the proteins that matched with oursequence belonged to �-CA gene family. The closest pro-tein sequence in the database was the CA from the seaanemone Anthopleura elegantissima (GenBank accessionnumber AF140537), with 45% of identities (115/255) and59% of similarities (152/255) [Gaps 15/255 (5%) andExpect 6/e�54]. A subset of relevant amino acid se-quences was chosen from GenBank to perform the align-ment with the complete deduced amino acid sequence fromRiftia (Fig. 1), including A. elegantissima CA, mouseCAVII, human CAI, CAV, CAVI, chicken CAII, rat CAIII,CAIV, D. melanogaster CA, and zebrafish CAH. Sequenceswere chosen to represent a large range of vertebrate CAisoforms among the enzymatically active proteins from the�-CA group. In addition, a sequence from B. melitensis�-CA was included to show the divergence between se-quences from organisms across broad phylogenetic dis-tances. From BLAST results and the alignment shown inFigure 1, the possibility that the CA sequenced from Riftiabelonged to bacterial �-CA could be eliminated. he puta-tive CA amino acid sequence from Riftia exhibited strongsimiliarities with the other �-CA sequences. The threehistidine residues (shown by a Z) directly implied inbinding the zinc cofactor were totally conserved in Riftiasequence. The amino acids involved in the associatedhydrogen bond network surrounding the active site (indi-cated with “plus” signs) were also highly conserved. TheN-terminal region was consistent with the N-terminalsequences found in other cytosolic isoforms (CA-I,-II,-III,and -VII) from the �-CA multigenic family. Indeed, theconserved regions first encountered in �-CA sequences,“WxY”(#41), “NGPxxW”(#48), and “SPIDI”(#70) were allpresent in Riftia sequence (Fig. 1). Moreover, the initiatorcodon (AUG) assumed to correspond to the initiation siteshowed a molecular environment conserved among themarine invertebrate nucleotide sequences: an adenosineresidue in position �11 and �3, and a guanosine residuein position �4 (Fig. 2).40–42 Noteworthy, membrane-bound (CA-IV), mitochondrial (CA-V), and secreted (CA-VI) isoforms exhibited isoform-specific signatures in theirN-terminal region (Fig. 1), confirmed by HCA plot analy-sis. None of these patterns could be found in Riftiasequence (data not shown).

Branchial CA

None of the six putative CA sequences obtained from theplume cDNA library screening was full length. All se-quences began in the same region of the open readingframe (ORF), corresponding to amino acid number 17 ofthe trophosome CA (Fig. 2). The plume sequence was 100%identical (in nucleotide and deduced amino acid se-quences) to the trophosome sequence, except for aminoacids 1 to 10. In plume, the amino acid sequence ARA-GAEFGRR replaced the trophosome sequence SFP-LAAGKKQ (position 17-26, Fig. 2).

Phylogenetic Tree

To further characterize putative Riftia CA-deducedamino acid sequences, we performed an analysis of theevolution of the CA isoforms on 22 sequences usingPHYLIP software. The unrooted CA molecular phyloge-netic tree (Fig. 3) clearly demonstrated the dichotomybetween bacterial CA (�-CA), represented here by Campy-locater jejuni and Brucella melitensis, and animal CA(�-CA). Within the multigenic �-CA group, the isoformswith different subcellular localizations were clustered inthis analysis in five distinct groups: the membrane-boundCA (CA-IV), the secreted CA (CA-VI), the mitochondrialCA (CA-V), and the cytosolic CA (CA-I, -II, -III), withCA-VII apart. These isoform sequences all belonged tovertebrates. R. pachyptila CA clearly emerged with thetwo other invertebrate CA proteins, from sea anemoneAnthopleura and

Drosophila (Fig. 3), and appeared to be closer to Drosoph-ila CA than to Anthopleura CA. However, the bootstrapvalues internal to each cluster remained relatively low inthe invertebrate cluster (56% and 70%), whereas theterminal lineage between CA-VII, CA-V, and CA-I exhib-ited strong bootstrap values (100%, 99%, and 99%). Theinvertebrate cluster was clearly separated from membrane-bound and secreted CA isoforms, and emerged with verte-brate cytosolic CA, before the divergence of CA-V, CA-VII,and CA-I, -II, -III. The complete CA sequence from theinvertebrate R. pachyptila presented here thus appearedto correspond to a putative cytosolic CA. According to thededuced amino acid sequence alignment (Fig. 1), thisputative CA exhibited all features required for CA activity.

Biochemical Characteristics of Putative CAProteinsCytosolic isoform

Western blot analyses were performed on cytosolic andmembrane-associated protein extracts from the tropho-some and the plume of Riftia (Fig. 4). Both one- andtwo-dimensional gels were tested for immunoblot reactiv-ity analysis. The polyclonal rabbit antibody raised againstchick cytosolic 30 kDa CA cross-reacted with a 27 kDacytosolic protein in the plume and a 28 kDa cytosolicprotein in the trophosome [Fig. 4(A and B)]. Two-dimensional electrophoresis on the same protein samplesalso discriminated the 27 kDa and 28 kDa through theirrespective pI’s. Indeed, the 27-kDa band evident in theplume appeared to be composed of 3 spots with pI’s of 6.1,6.5, and 6.75 [Fig. 4(A)]. The spot corresponding to the

CARBONIC ANHYDRASE IN RIFTIA PACHYPTILA 331

Page 6: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

Fig. 1. Alignment of Riftia pachyptila predicted amino acid sequence (EMBL accession number AJ439711) with CA protein sequences from human,mouse, rat, chicken, and zebrafish isoforms identified in vertebrates, and nonvertebrate CA sequences from bacteria, drosophila, and sea anemone. Thealignment was generated by Clustal-X software with multiple parameters given by default with use of the PAM series as protein weight matrix, and editedwith GeneDoc software. Conserved regions among the different isoforms are shown in black, and protein patterns with similarity superior to 80% areshown in dark gray. Histidine residues involved in zinc binding in the catalytic site are indicated by a Z, and ‡ signs represent amino acids participating inthe hydrogen bond network forming the active site.60,61

332 M.-C. DE CIAN ET AL.

Page 7: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

27-kDa protein with a pI of 6.5 appeared to be the mostreactive isoform. The 28-kDa trophosome CA did notexhibit distinct pI isoforms but exhibited a broad spotaround pI 6.75 [Fig 4(B)]. The lack of resolution and focusfor this spot is likely due to protein-associated pigmentcontaminations, according to silver nitrate–stained bidi-mensional gels (data not shown). These greenish pigmentsare specific to the trophosome, and no similar contamina-tion could be observed in the protein extracts from theplume. As an indication for the specificity of the antibodyused, the rabbit antichick CA antibody did not reveal anyother protein present in the extract, either from cytosolicor membrane-bound fractions.

The immunoreactive proteins from the plume and thesymbiotic tissue analyzed by MALDI-TOF mass spectrom-etry (MS) after a trypsin digestion step, and the internalfragments generated, appeared to be identical in bothproteins (Fig. 5). Table II summarizes comparisons be-tween experimental masses obtained by MS and theoreti-cal masses generated by computer-assisted trypsin diges-tion of deduced amino acid sequence from completetrophosome CA cDNA. Identical peptide signatures werefound in both plume and trophosome CA, with, for ex-ample, peptides 26-39 QSPIDIDPASVSKK, 82-91 AASFH-FHWSK, or 105-119 AYAAEAHIVHYNAAH as an illustra-tion of the occurrence of the peptides comprising thecharacteristic histidine residues involved in CA-activesites. We could thus assume that (1) the nucleotide se-quence of trophosome CA (Fig. 1) corresponded to thecytosolic proteins immunoblotted; (2) the CA sequencesfrom both tissues were at least 90% identical except for theN-terminal region, which remained undetermined; and (3)they both belonged to the �-CA multigenic family. Lookingat the N-terminal region, the experimental mass of 790.481corresponding to the peptide sequence (K)SFPLAAGK(K)(theoretical mass 790.45; Table II) was found in thetrophosome CA, but not in the plume CA, where no exactcorrespondence between experimental and theoreticalmasses could be established (Table II). Up to now, N-terminal microsequencing of the CA protein from theplume has failed for unknown reasons.

Membrane-bound isoform

We examined the possible existence of a membrane-associated CA isoform, in addition to a cytosolic isoform,using membrane-bound protein fractions of branchialplume and trophosome of Riftia. The protein extracts wereblotted with antirat membrane-bound CA-IV antiserumafter SDS-PAGE [Fig. 6(A)] and bidimensional electro-phoresis [Fig. 6(B)]. Three bands were revealed in themembrane-bound fraction of the trophosome, at 50, 42,and 39 kDa, with pI’s of 5.4, 5.1, and 5.1, respectively [Fig.6(B)]. No protein of the same molecular weight wasdetected in the membrane-bound fraction of the plume.The antirat membrane-bound CA-IV antiserum also cross-reacted with cytosolic CA isoforms (bovine CA at 30 kDa inthe ladder, BrS2 and TrS2). The three membrane-boundimmunoreactive proteins of 50, 42, and 39 kDa wereisolated on a Coomassie-stained gel and subjected to MSmicrosequencing. Unfortunately, the presence of multipleproteins in the sample, or the low quantity of material,prevented us from clearly identifying the proteins throughthis method. Considering that R. pachyptila samples arerare and difficult to obtain, we examined specific CAactivity as an alternative. The membrane-bound extractwas fractionated through gel filtration on a SephadexHR-75 column in order to separate the proteins of 50, 42,and 39 kDa from the 28 kDa band, corresponding tocytosolic CA. All fractions were assayed on SDS-PAGE(data not shown): Fractions 2, 3, and 4 did not contain the28 kDa band, which was eluted in fractions 8, 9, and 10.Two peaks of specific CA activity were evidenced with amaximal activity of 1.25 � 0.37 pH/min�1/mg protein�1

in fraction 4, and 4.86 � 1.04 pH/min�1/mg protein�1 infraction 9 (Fig. 7).

DISCUSSION AND CONCLUSIONS

Published �-CA sequences concern mainly vertebrates,especially mammals, with several thousand entries inGenBank compared to only a few sequences outside thevertebrates, including the arthropod D. melanogaster andthe cnidarian A. elegantissima. Previously, the presence of

Fig. 2. Nucleotide and predicted amino acid sequences of putative CA from the plume and the symbiotic tissue of Riftia pachyptila. The sequence is a1209-nucleotide cDNA with a 731-nucleotide ORF. The deduced amino acid sequence is 243 amino acids in length. AUG initiator codon is boxed. Thenucleotide residues (�11, �3, and �4) involved in consensus translational initiation sequence40,42 are boxed and shaded in gray.

CARBONIC ANHYDRASE IN RIFTIA PACHYPTILA 333

Page 8: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

different CAs in the symbiotic hydrothermal vent annelidR. pachyptila has been suggested by investigation of thecytosolic isoform mRNA expression pattern and the corre-sponding protein localization in the worm,28 and thephysiologic response of isolated bacteriocyte cell suspen-sions to different selective CA inhibitors regarding intracel-lular pH and major ion concentrations.29 In this study, twoCA cDNAs have been cloned.

Alignment of amino acid sequences obtained fromthese cDNAs indicate that R. pachyptila CA revealed

patterns common to cytosolic CA and clearly distinctfrom either membrane-bound, mitochondrial, or se-creted enzyme isoforms, and no N-terminal peptidesignal for specific subcellular addressing message wasdetected in the cDNA sequence. The respective positionsof upstream Kozak pattern, start and stop codons, anddownstream polyadenylation signal were all congruentto previously published sequences, indicating that thesequence presented here is likely to be correct. Thebiochemical protein features, molecular weight, and

Fig. 3. Unrooted neighbor-joining (NJ) tree obtained with multiple alignment of 21 CA amino acidsequences, and computed with PHYLIP software. Phylogenetic distances were computed from those expectedunder the Dayhoff’s PAM matrix fixed in ProtDist step. Accession numbers of the protein sequences used as amatrix are as follows (NCBI Entrez Proteins): mouseCAI (P13634), humCAI (P00915), mouseCAII (P00920),chickCAII (P07630), ratCAIII (014141), mouseCAIV (Q64444), ratCAIV (P48284), rabitCAIV (P48283),humCAIV (P22748), mouseCAV (Q9Q7A0), humCAV (Q9Y2D0), ratCAV (P43165), humCAVI (P23280),mouseCAVI (P18761), humCAVII (P43166), mouseCAVII (AAG16230), DrosoCA (NP_523561), zebrafishCA(T08463), A. elegantissimaCA (AF140537), C. jejuniCA (CAB72706), and B.melitensisCA (AAL51404). Theemergent clusters are shaded in gray. The sequences of �-CA from bacteria were used as the outgroup. Thesubcellular localization of the isoforms identified in the �-CA multigenic family of vertebrates is indicated inbold.

334 M.-C. DE CIAN ET AL.

Page 9: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

isoelectric points also appeared to be similar to those ofcytosolic �-CAs.2,7 These molecular and biochemicaldata suggest that the cloned gene codes are for acytosolic CA, and are consistent with the in situ hybrid-ization of CA mRNA performed both on plume andtrophosome tissues, and the corresponding protein immu-nolocalization, which appeared restricted to the cytosolof plume epidermal cells, and bacteriocyte and perito-neal cell cytosol in the trophosome.29

Interestingly, the strong cross-reactivity between a poly-clonal antivertebrate CA and an annelid protein suggestsa relatively high level of conservation of the protein amongthe �-CA gene family.

The nucleotide sequences from the plume and the tropho-some are 90% identical, which strongly suggests that theyoriginate from a single gene. The difference between these

two nucleotide sequences is restricted to their N-terminalregion. We propose two alternative hypotheses to accountfor this difference.

The first hypothesis is that there is a posttranscriptionalevent such as a differential splicing of the mRNA thatwould be tissue-specific.43 Indeed, the first 30 nucleotidessequenced from the plume cDNA library, corresponding to“ARAGAEFGRR,” were the same in six positive clones andwere all found in the correct ORF (Fig. 2). These aminoacids were replaced by “SFPLAAGKKQ” in the tropho-some sequence (position 17-26, Fig. 2). In addition, thedifference in molecular weight of 1 kDa between tropho-some (28 kDa) and plume CA (27 kDa) could correspond tothe lack of a maximum of 15 amino acid residues in plumeCA. This missing fragment could include the first tyrosineresidue in position �6, which is involved in the hydrogen

Fig. 5. MALDI-TOF MS-analyzed profiles of cytosolic branchial plume CA (27 kDa, lower) and trophosomeCA (28 kDa, upper) following an in-gel trypsin digestion step. The perfect correspondence of the resultinginternal fragments between plume and trophosome CA is evident, but these analyses cannot discriminate thetwo proteins in their NH2-teminal region with remaining nonidentified peaks in the first part of the spectrum. Thecomparison between the experimental masses of the fragments obtained and the theoretical massesgenerated from the deduced amino acid sequence is given in Table II.

Fig. 4. Western blots on total protein extracts from plume (A) and trophosome (B) of Riftia (around 50 �gloaded) probed with polyclonal rabbit antichick CA-II. (A) Cross-reaction of the rabbit antichick CA-II antibodywith a protein of 27 kDa in the cytosolic fraction (BrS2) of the plume after SDS-PAGE, and three spots with pI’sof 6.1, 6.5, and 6.75 after bidimensional electrophoresis. (B) Same antibody as in (A), blotted with cytosolicfraction of the trophosome (TrS2). A unique band of 28 kDa, with a corresponding pI of 6.75, was detected. Nocross-reaction was observed with preimmune serum (data not shown). Abbreviations: Br, branchial plume; S2,cytosolic protein fraction; S3, membrane-associated protein fraction; Tr, trophosome.

CARBONIC ANHYDRASE IN RIFTIA PACHYPTILA 335

Page 10: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

bond network around the zinc-binding catalytic site of theenzyme (Fig. 1), and could result in modifications of thecatalytic properties of the enzyme.

The second hypothesis involves different posttransla-tional modifications from a unique mRNA in the plume,and in the symbiotic tissue. This would imply a flaw in thecDNA library issued from the plume, and the correct CAsequence would be that of the trophosome. The observeddifference in molecular weight between the two proteinscould thus be attributed to a glycosylation or phosphoryla-tion event. The comparison between the pI’s of the proteinscould also account for posttranslational modifications,because trophosome CA (28 kDa) exhibited a unique pI of6.75, whereas plume CA (27 kDa) exhibited three differentpI’s of 6.1, 6.5, and 6.75.

Our results also suggest that R. pachyptila expresses aputative CA-IV (membrane-bound) isoform in addition to acytosolic isoform. Further studies including microsequenc-ing and PCR amplification are necessary to obtain acomplete characterization of these R. pachyptila mem-brane-bound proteins, but our data provide convincingevidence for a CA-IV–like enzyme. Indeed, the rabbitantirat lung CA-IV antiserum used for Western blotanalysis demonstrated the existence of three proteins of50, 42, and 39 kDa, the latter two being the major ones, in

the membrane-associated protein fraction of the tropho-some. This antibody revealed no proteins in the membrane-associated protein fraction of the plume. These proteinsexhibited a pI congruent with those found in the literaturefor the vertebrate CA-IV isoform44,45 and a substantial CAactivity (Fig. 7). The molecular mass of CA-IV isoformsstill remains equivocal, probably linked to various glycosyl-ation states: 46 and 30 kDa in the mouse,46 48 and 33 kDain the dogfish,47 or 52 and 36 kDa in the cow,48 forexample. The same could be true for Riftia: The observedpattern resulted from the incubation of trophosome cellmembranes for 2 h at room temperature in presence ofTriton X-100 and saponin, which may generate “pro-cessed forms” of 39 and 42 kDa from the native form of50 kDa. It would be interesting to follow the kinetics ofdeglycosylation coupled to mass spectrometry to resolvethis point.

The probable presence in R. pachyptila of CA-IV–likeenzymes is of particular interest because, until now,published CA sequences from invertebrates all code forcytosolic enzymes. The present results are consistent withprevious physiologic studies suggesting that additionalmembrane-bound CAs exist in invertebrates.5 Their exis-tence has been suggested in the gills of Callinectes sapidusbased on physiologic and pharmacologic approaches,49 and

TABLE II. Comparison between Experimental Fragments Generated by MALDI-TOF MS Common to BothPlume and Trophosome CA Sequences, and Theoretical Fragments Generated by Computer-Assisted

Trypsin Digestion of the Deduced Amino Acid Sequence from Trophosome CA cDNA Sequence

MS generatedfragments

Theoreticalfragments Corresponding sequence

Position onCATr

Br Tr Br Tr Br Tr

17–24790.481 790.45 (K)SFPLAAGK(K)

863.45 (R)AGAEFGRR(S)

1028.442 1028.51 (K)YASFQDAVK(A) 120–1281050.4231195.629 1195.73 (K)IIDLLPSVPTK(G) 152–1621217.480 1217.59 (K)AASFHFHWSK(T) 82–911239.4681249.4591255.4321273.496 1273.60 (K)TSAEGSEHTVAGK(A) 92–1041356.564 1356.70 (K)QSPIDIDPASVSK(K) 26–381484.656 1484.40 (K)QSPIDIDPASVSKK(S) 26–391628.670 1628.82 (K)AYAAEAHIVHYNAAK(Y) 105–1191650.6921654.7391666.6721692.6461721.7431735.739 1735.91 (R)KITGCNFRPTLGLCGR(Q) 222–2371749.7731874.9722218.9982233.0262257.121 2257.18 (K)ADDGLAVLATFIQPGATNAGVQK(I) 129–1514191.410 4189.06 (K)STSALVASYNPAASNTLTNTGLSFQVSVDGT-

LSGGPLGNEYK(A)

*All the fragments presented are common to both sequences except for the first line corresponding to the smaller fragmentsthat could be determined; “Br” on the left and “Tr” on the right indicate the data from branchial plume CA and trophosome CA,respectively.

336 M.-C. DE CIAN ET AL.

Page 11: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

histologic localization of CA has been reported in cellmembranes from echinoderms,50 bivalves,51 and insects.52

The occurrence of a membrane-associated CA in Riftiamakes sense in light of other information. As stated before,the symbionts of Riftia are sulfur-oxidizing bacteria thatcan only fix molecular CO2 (not HCO3

�) into organiccompounds, and their metabolism produces large amountsof protons. In addition to the regulator role of the cytosolicCA, a CA isoform associated with the external membraneof the bacteriocytes could take advantage of the proton

efflux to facilitate the conversion of blood-borne HCO3� into

CO2 and thus regulate the movement of CO2 through themembranes. Further investigations now focus on thecharacterization of this potential CA-IV–like protein.

Molecular phylogenetic studies on the �-CA multigenicfamily suggest that ancestral cytoplasmic CAs have evolvedin vertebrates through duplication events over the courseof 600 million years.2 However, a controversy on theresemblance of the common ancestor to either CA-II–like53

or CA-VII–like2 sequences remains. CA sequences frominvertebrates could be highly informative for clarifying themolecular phylogeny of these proteins. In particular, wewere wondering if R. pachyptila cytosolic CA was homolo-gous to one of the vertebrate �-CAs, maybe as a conse-quence of a functional convergence.54,55 Indeed, the threenonvertebrate CAs appeared to form a distinct cluster thatstands as the sister group of cytosolic and mitochondrialvertebrate CAs, apart from the secreted (CA-VI) andmembrane-bound (CA-IV) isoforms. It would now be ofinterest to compile these data with additional invertebrateCA sequences, including the putative CA-IV–like sequencefrom R. pachyptila, in order to complete and refine themolecular phylogeny analysis of �-CAs, potentially yield-ing new insights into the incredible diversity of CA.

Our research provides the first evidence of the existenceof two different CA isoforms in a nonmammalian organismthrough molecular and biochemical characterization. In-deed, deep-sea and hydrothermal vent organisms havepreviously been shown to possess altered enzyme forms asa result of adaptation to their extreme environment.20,56–59

Further research is under way to determine whether thisunusual feature is related to the unique mode of life of R.pachyptila.

Fig. 7. Specific carbonic anhydrase activity as a function of elutedfractions from trophosome membrane-bound protein homogenate (TrS3),consecutive to gel filtration step on a sephadex HR-75 column with FPLCsystem (Pharmacia). Specific CA activity was expressed as (pHs �pHb)/min�1 /mg protein�1 (mean � SD, n 3). Two peaks of activitywere evidenced. The first was in fraction 4, the fractions 2, 3, and 4 beingconsidered free of remaining cytosolic CA in the membrane-bound proteinextract as assayed on SDS-PAGE (data not shown). The second was infraction 9, with cytosolic CA being eluted in fraction 8, 9, and 10.

Fig. 6. Cytosolic and membrane-bound protein extracts from the plume and the trophosome blotted withrabbit antirat lung CA-IV antiserum after SDS-PAGE (A) and bidimensional electrophoresis (B). Only themembrane protein fraction of the symbiotic tissue cross-reacted with the CA-IV antibody. Proteins of differentmolecular masses were evidenced with this antibody, weighing 50, 42, and 39 kDa, with pI’s of 5.4, 5.1, and5.1, respectively. The small blot portion boxed in (B) correspond to 1-s exposure of the membrane to X-OMATfilm, allowing a better visualization of the two main spots. The anti-CA-IV antibody also cross-reacted with thecytosolic isoform present in both BrS2 and TrS2, and still remaining in the membrane-bound protein fraction(BrS3, TrS3). No cross-reaction was observed with preimmune serum (data not shown). Abbreviations: Br,branchial plume; S2, cytosolic protein fraction; S3, membrane-associated protein fraction; Tr, trophosome.

CARBONIC ANHYDRASE IN RIFTIA PACHYPTILA 337

Page 12: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

ACKNOWLEDGMENTS

Our thanks to Sandrine Boulben from Dr. R. Belle’slaboratory, Marie Knockaert from Dr. L. Meijer’s labora-tory (UMR 7127), and Sabine Genicot from Dr. B. Kloar-eg’s laboratory (UMR 1931) for their useful assistance andadvice concerning immunoblot assays and protein purifica-tion (CNRS-UPMC Roscoff, France). We also are gratefulto Dr. Paul Linser of the Whitney Laboratory at theUniversity of Florida for generously providing the rabbitantichick CA II antibody used in this study, and Dr. W. S.Sly of the Deparment of Biochemistry and MolecularBiology at St. Louis University School of Medicine, for therabbit antirat lung CA IV antibody. We also thank OdileMulner-Lorillon, Arnaud Lacoste, and Patrick J. Walsh forediting and improving the manuscript. Finally, we acknowl-edge the contributions of the crews of the R.V. Atalanteand the D.S.V. Nautile, which were used in the collection ofour samples.

REFERENCES

1. Maren TH. Carbonic anhydrase: Chemistry, physiology and inhibi-tion. Phyiol Rev 1967;47:595–781.

2. Hewett-Emmett D, Tashian RE. Functional diversity, conserva-tion, and convergence in the evolution of the a-, b-, and g-carbonicanhydrase gene family. Mol Phylogen Evol 1996;5:50–77.

3. Roberts SB, Lane TW, Morel FMM. Carbonic anhydrase in themarine diatom Thalassiosira weissflogii (Bacillariophyceae). JPhycol 1997;33:845–850.

4. Sly W, Peiyi YH. Human carbonic anhydrases and carbonicanhydrase deficiencies. Annu Rev Biochem 1995;64:675–401.

5. Henry RP. Multiple roles of carbonic anhydrase in cellular trans-port and metabolism. Annu Rev Physiol 1996;58:523–538.

6. Lucas JM, Knapp LW. Biochemical characterization of purifiedcarbonic anhydrase from the octocoral Leptogorgia virgulata. MarBiol 1996;126:471–477.

7. Chegwidden WR, Carter ND. Introduction to the carbonic anhy-drases. Carbonic Anhydrases New Horizons 2000;90:13–28.

8. Brinkman R, Margaria R, Meldrum NU, Roughton FJW. The CO2catalyst present in blood. J Physiol (London) 1932;75:3–4. [Meet-ing of March 12, 1932. Addendum by Meldrum and Roughton]

9. Meldrum NU, Roughton FJW. Carbonic anhydrase: Its prepara-tion and properties. J Physiol (London) 1933;80:113–142.

10. Peterson RE, Tu C, Linser PJ. Isolation and characterization of acarbonic anhydrase homologue from the zebrafish (Danio rerio). JMol Evol 1997;44:432–439.

11. Peters T, Papadopoulos F, Kubis HP, Gros G. Properties of acarbonic anhydrase inhibitor protein in flounder serum. J Exp Biol2000;203:3003–3009.

12. Myiamoto H, Miyashita T, Okushima M, Nakano S, Morita T,Matsushiro A. A carbonic anhydrase from the nacreous layer inoyster pearls. Proc Nat Acad Sci U S A 1996;93:9657–9660.

13. Henry RP, Cameron JN. The distribution and partial characteriza-tion of carbonic anhydrase in selected aquatic and terrestrialdecapod crustaceans. J Exp Zool 1982;221:309–321.

14. Weis VM, Smith GJ, Muscatine L. A “CO2 supply” mechanism inzooxanthellate cnidarians: Role of carbonic anhydrase. Mar Biol1989;100:195–202.

15. Shchelkunov SN, Blinov VM, Totmenium AV, Marennikova SS,Kolykhalov AA, Frolov I, Chizhikov VE, Gutorov VV, GashnikovPV, Belanov EF. Nucleotide sequence analysis of variole virusHind III M, L, genome fragment. Virus Res 1993;27:25–35.

16. Fisher M, Gokhman I, Pick V, Zamir A. A salt resistant plasmamembrane carbonic anhydrase is induced by salt in Dunaliellasalina. J Biol Chem 1996;271:17718–17723.

17. Newman T, de Brujn FJ, Green P, Keegstra K, Kende H, McIntoshL, Ohlrogge J, Raikhel N, Somerville S, Thomashow M, Retzel E,Somerville C. Genes galore: A summary of methods for accessingresults from large-scale partial sequencing of anonymous Arabi-dopsis cDNA clones. Plant Physiol 1994;106:1241–1255.

18. Chirica LC, Elleby B, Jonsson BH, Lindskog S. The completesequence, expression in Escherichia coli, purification and some

properties of carbonic anhydrase from Neisseria gonorrhoeae. EurJ Biochem 1997;244:755–760.

19. Cavanaugh CM, Gardiner SL, Jones ML, Jannasch HW, Water-bury JB. Prokaryotic cells in the hydrothermal vent tube-wormRiftia pachyptila Jones: Possible chemoautotrophic symbionts.Science 1981;213:340–342.

20. Felbeck H. Chemoautotrophic potential of the hydrothermal venttubeworm, Riftia pachyptila Jones (Vestimentifera). Science 1981;213:336–338.

21. Childress JJ, Fischer CR, Favuzzi JA, Kochevar R, Sanders NK,Alayse AM. Sulfide-driven autotrophic balance in the bacterialsymbiont-containing hydrothermal vent tubeworm, Riftia pachyp-tila Jones. Biol Bull 1991;180:135–153.

22. Goffredi SK, Childress JJ, Desaulniers NT, Lee RW, Lallier FH,Hammond D. Inorganic carbon acquisition by the hydrothermalvent tubeworm Riftia pachyptila depends upon high externalP-CO2 and upon proton-equivalent ion transport by the worm. JExp Biol 1997;200:883–896.

23. Childress JJ, Lee RW, Sanders NK, Felbeck H, Oros DR, Toul-mond A, Desbruyeres D, Kennicut II MC, Brooks J. Inorganiccarbon uptake in hydrothermal vent tubeworms facilitated byhigh environmental PCO2. Nature 1993;362:147–149.

24. Toulmond A, Lallier FH, De Frescheville J, Desbruyeres D,Childress JJ, Lee R, Sanders NK. Unusual carbon dioxide-combining properties of body fluids in the hydrothermal venttubeworm Riftia pachyptila. Deep-Sea Res 1994;41:1447–1456.

25. Shively JM, van Keulen G, Meijer WG. Something from almostnothing: Carbon dioxide fixation in chemoautotrophs. Ann RevMicrobiol 1998;52:191–230.

26. Childress JJ, Fischer CR. The biology of hydrothermal ventanimals: Physiology, biochemistry and autotrophic symbioses.Oceanogr Mar Biol Annu Rev 1992;30:337–341.

27. Goffredi SK, Childress JJ, Lallier FH, Desaulniers NT. How to bethe perfect host: CO2 and HS� accumulation and H� eliminationin the hydrothermal vent tube-worm Riftia pachyptila. Cah BiolMar 1998;39:297–300.

28. Kochevar RE, Govind NS, Childress JJ. Identification and charac-terization of two carbonic anhydrases from the hydrothermal venttubeworm Riftia pachyptila Jones. Molec Mar Biol Biotechnol1993;2:10–19.

29. De Cian M-C, Andersen AC, Bailly X, Lallier FH. Expression andlocalization of carbonic anhydrase and ATPases in the symbiotictubeworm Riftia pachyptila. J Exp Biol 2003;206:399–409.

30. De Cian M-C, Andersen AC, Toullec J-Y, Biegala I, Caprais J-C,Shillito B, Lallier FH. Isolated bateriocyte cell suspensions fromthe symbiotic hydrothermal vent tubeworm Riftia pachyptila, apotent tool for cellular physiology in a chemoautotrophic symbio-sis. Mar Biol 2003;142:141–151.

31. Gardiner SL, Jones ML. Vestimentifera. In: Gardiner SL, editor.Microscopic anatomy of invertebrates: Onychophora, Chilopodaand lesser Protostomata. Vol.12. New York: Wiley-Liss; 1993. p371–460.

32. Altschul SF, Fish W, Miller W, Myers EW, Lipman DJ. Basic localalignment search tool. J Mol Biol 1990;215:403–410.

33. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, HigginsDG. The CLUSTAL_X windows interface: Flexible strategies formultiple sequence alignment aided by quality analysis tools.Nucleic Acids Res 1997;25:4876–4882.

34. Felsenstein J. PHYLIP: Phylogeny Inference Package (version3.2). Cladistics 1989;5:164–166.

35. Callebaut I, Labesse G, Durand P, Poupon A, Canard L, ChomilierJ, Henrissat B, Mornon JP. Deciphering protein sequence informa-tion through hydrophobic cluster analysis (HCA): Current statusand perspectives. Cell Mol Life Sci 1997;53:621–645.

36. Brion LP, Cammer W, Satlin LM, Suarez C, Zavilowitz BJ,Schuster VL. Expression of carbonic anhydrase IV in carbonicanhydrase II-deficient mice. Amer J Physiol Renal Physiol 1997;42:F234–F245.

37. Bradford MM. A rapid and sensitive method for the quantificationof microgram quantities of proteins utilizing the principle ofprotein-dye binding. Anal Biochem 1976;72:248–254.

38. Laemmli UK. Cleavage of structural proteins during assembly ofthe head of bacteriophage T4. Nature 1970;227:680–685.

39. Drescher DG. Purification of blood carbonic anhydrases andspecific detection of carbonic anhydrae isoenzymes on polyacryl-amide gels with 5-dimethylaminonaphthalene-1-suflonamide(DNSA). Anal Biochem 1978;90:349–358.

40. Kozak M. Point mutations define a sequence flanking the AUG

338 M.-C. DE CIAN ET AL.

Page 13: Characterization of carbonic anhydrases from Riftia pachyptila, a symbiotic invertebrate from deep-sea hydrothermal vents

initiator codon that modulates translation by eukaryotic ribosomebinding site. Cell 1986;44:283–292.

41. Kozak M. At least six nucleotide preceding the AUG initiatorcodon enhance translation by eukaryotic ribosomes. J Mol Biol1987;196:947–950.

42. Mankad RV, Gimelbrant AA, Mcclintock TS. Consensus transla-tional initiation sites of marine invertebrate phyla. Biol Bull1998;195:251–254.

43. Hames RS, Wattam SL, Yamano H, Bacchieri R, Fry AM. APC/C-mediated destruction of the centrosomal kinase Nek2A occurs inearly mitosis and depends upon a cyclin A-type D-box. EMBO J2001;20:7117–7127.

44. Wistrand PJ, Knuuttila KG. Renal membrane-bound carbonicanhydrase: Purification and properties. Kidney Int 1989;35:851–859.

45. Zhu XL, Sly WS. Carbonic anhydrase IV from human lung:Purification, characterization, and comparison with membranecarbonic anhydrase from human kidney. J Biol Chem 1990;265:8795–8801.

46. Brion LP, Suarez C, Zhang H, Cammer W. Up-regulation ofcarbonic anhydrase isozyme IV in CNS myelin of mice geneticallydeficient in carbonic anhydrase II. J Neurochem 1994;63:360–366.

47. Wilson JM, Randall DJ, Vogl AW, Harris J, Sly W, Iwama GK.Branchial carbonic anhydrase is present in the dogfish, Squalusacanthias. Fish Physiol Biochem 2000;22:329–336.

48. Waheed A, Zhu XL, Sly WS. Membrane-associated carbonicanhydrase from rat lung. J Biol Chem 1992;267:3308–3311.

49. Henry RP. Membrane-associated carbonic anhydrase in gills ofthe blue crab, Callinectes sapidus. Am J Physiol 1987;252:R966–R971.

50. Chen CP, Lawrence JM. Localization of carbonic anhydrase in theplumula of the tooth of Lytechinus variegatus (Echinodermata:Echinoidea). Acta Zool 1986;67:27–32.

51. Boer HH, Witteveen J. Ultrastructural localization of carbonicanhydrase in tissues involved in shell formation and ionic regula-tion in the Pond snail Lymnaea stagnalis. Cell Tissue Res1980;209:383–390.

52. Just F, Walz B. Localization of carbonic anhydrase in the salivaryglands of the cockroach, Periplaneta americana. Histochemistry1994;102:271–277.

53. Hewett-Emmett D, Hopkins PJ, Tashian RE, Czelusniak J. Ori-gins and molecular evolution of the carbonic anhydrase isozymes.Ann NY Acad Sci 1984;429:338–358.

54. Kisker C, Schindelin H, Alber BE, Ferry JG, Mees DC. A left-handbeta helix revealed by the crystal structure of a carbonic anhy-drase from the archaeon Methanosarcina thermophila. EMBO J1996;15:2323–2330.

55. Lindskog S. Structure and mechanism of carbonic anhydrase.Pharmacol Ther 1997;74:1–20.

56. Hentschel U, Felbeck H. Nitrate respiration in the hydrothermalvent tubeworm Riftia pachyptila. Nature 1993;366:338–340.

57. Lutz RA, Shank TM, Fornari DJ, Haymon RM, Lilley MD, VonDamm KL, Desbruyeres D. Rapid growth at deep-sea vents.Nature 1994;371:663–664.

58. Young CM, Vazquez E, Metaxas A, Tyler PA. Embryology ofvestimentiferan tube worms from deep-sea methane/sulphideseeps. Nature 1996;381:514–516.

59. Zal F, Leize E, Lallier FH, Toulmond A, Van Dorsselaer A,Childress JJ. S-sulfohemoglobin and disulfide-exchange: Themechanisms of sulfide-binding by Riftia pachyptila hemoglobins.Proc Nat Acad Sci U S A 1998;95:8997–9002.

60. Fukuzawa H, Ishida S. cDNA cloning and gene expression ofcarbonic anhydrase in Chlamydomonas reinhardtii. Can J Bot1991;69:1088–1096.

61. Sheridan RP, Allen LC. The active site electrostatics potential ofhuman carbonic anhydrase. J Am Chem Soc 1981;103:1544–1550.

CARBONIC ANHYDRASE IN RIFTIA PACHYPTILA 339