A conserved retinoid X receptor (RXR) from the molluskBiomphalaria glabrata transactivates transcription in thepresence of retinoids

D Bouton, H Escriva1, R L de Mendonça, C Glineur2, B Bertin, C Noël,M Robinson-Rechavi1, A de Groot3, J Cornette, V Laudet1 and R J PierceINSERM U 547, Institut Pasteur, 1 rue du Professeur Calmette, 59019-Lille, France

1CNRS UMR 49, École Normale Supérieure de Lyon, 46 allée d’Italie, 69364-Lyon, France

2INSERM U 545 Institut Pasteur, 1 rue du Professeur Calmette, 59019-Lille, France

3Laboratoire d’Écologie Microbienne de la Rhizosphère (LEMiR), Département d’Écophysiologie Végétale et de Microbiologie (DEVM), CEA Cadarache, 13108 Saint Paul LezDurance, France

(Requests for offprints should be addressed to R J Pierce; Email: Raymond.Pierce@pasteur-lille.fr)

(D Bouton is now at Department for Biosciences at Novum, Karolinska Institute, 141 57 Huddinge, Sweden)

(R L de Mendonça is now at Université Libre de Bruxelles, Service de Microbiologie, Hôpital Erasme, Route du Lennik, 1070-Bruxelles, Belgium)

(C Noël is now at School of Biology, Institute for Research on Environment and Sustainability, Devonshire Building, University of Newcastle upon Tyne, Newcastle upon TyneNE1 7RU, UK)

(M Robinson-Rechavi is now at Joint Center for Structural Genomics, UCSD and Burnham Institute, 10901 North Torrey Pines Road, La Jolla, CA 92037, USA)

Abstract

Retinoid X receptors (RXR) are members of the nuclear receptor superfamily of ligand-activated transcription factorsthat have been characterized in a wide variety of metazoan phyla. They act as heterodimer partners of other nuclearreceptors, and in vertebrates also activate transcription as homodimers in the presence of a ligand, 9-cis retinoic acid.In order to test the hypothesis that retinoic acid signaling pathways involving RXRs are present in the Lophotrochozoa,we have sought to isolate conserved members of this family from the platyhelminth parasite Schistosoma mansoni andits intermediate host, the mollusk Biomphalaria glabrata. Here we report that an RXR ortholog from B. glabrata(BgRXR) is better conserved, compared with mouse RXR�, both in the DNA-binding domain (89% identity) and in theligand-binding domain (LBD) (81% identity), than are arthropod homologs. In EMSA, BgRXR binds to the direct repeatresponse element DR1 as a homodimer or as a heterodimer with mammalian RAR�, LXR, FXR or PPAR�. Whentransfected alone into mammalian cell lines, BgRXR transactivated transcription of a reporter gene from the Apo-A1promoter in the presence of 9-cis retinoic acid or DHA. Constructs with the Gal4 DNA binding domain fused to thehinge and LBDs of BgRXR were used to show that ligand-dependent activation of transcription by BgRXR required itsintact AF-2 activation domain, and that the LBD can form homodimers. Finally, the binding of 9-cis retinoic acidpreferentially protected the LBD of BgRXR from degradation by trypsin in a proteolysis protection assay. Our resultsshow that BgRXR binds and is activated by retinoids and suggest that retinoid signaling pathways are conserved in theLophotrochozoa. The nucleotide sequence reported in this paper has been submitted to the GenBank/EBI Data Bankwith accession no. AY048663.

Journal of Molecular Endocrinology (2005) 34, 567–582

Introduction

Retinoic acids (RA) play a key role in embryo patterningand organogenesis in vertebrates (reviewed in Ross et al.2000), but there is growing evidence that their role incellular signaling is ancient within the Metazoa.Retinoids and carotenoids were detected in sponges and13-cis RA was found to regulate gene expression(Biesalski et al. 1992). In the freshwater spongeEphydatia muelleri, retinoic acid initiates morphogeneticevents (Imsiecke et al. 1994) and the downregulationof a homeobox-containing gene (Nikko et al. 2001).

Moreover, in the marine sponge Suberites domuncula,incubation with all-trans RA leads to reversible tissueregression in intact individuals (Wiens et al. 2003).However, the only evidence for the influence of RA inprotostomes is its active uptake by the parasiticnematode Brugia malayi and its influence on thedevelopment of the worm within its mammalian host(Wolff & Scott 1995). There are no reports of theeventual role of RA in lophotrochozoan invertebrates.

Retinoid signals are transduced in vertebrates by twofamilies of nuclear receptor, the retinoic acid receptorsRAR�, � and �, which preferentially bind all-trans RA,

567

Journal of Molecular Endocrinology (2005) 34, 567–5820952–5041/05/034–567 © 2005 Society for Endocrinology Printed in Great Britain

DOI: 10.1677/jme.1.01766Online version via http://www.endocrinology-journals.org

and the retinoid X receptors RXR �, � and �, whichbind 9-cis RA. The latter play a central role in a varietyof nuclear signaling pathways (Mangelsdorf & Evans1995) and, in vertebrates, modulate the activities ofother nuclear receptors (including RAR, the vitamin Dreceptor or the peroxisome proliferator-activatedreceptor (PPAR)) by forming heterodimers. These bindspecifically to direct repeats (DR) of the consensusresponse element PuGGTCA spaced by 1–5 nucleotides(DR1 to DR5) (Mangelsdorf et al. 1995). The spacingdetermines the specificity of the binding of the variousheterodimers.

Whereas the RAR family has been found only invertebrates, arthropods possess homologs of RXR,including Drosophila ultraspiracle (USP), but none ofthese receptors has been conclusively shown to bind 9-cisRA, or indeed any ligand. However, an RXR-likenuclear receptor has been characterized in the marinesponge S. domuncula, and although its capacity to bindRA was not demonstrated, treatment of the sponge withall-trans RA strongly upregulated the expression of itsmRNA (Wiens et al. 2003). An RXR in the jellyfishTripedalia cystophora has a well-conserved ligand-bindingdomain (LBD) and has been shown to bind 9-cis RA(Kostrouch et al. 1998), but transactivation of transcrip-tion by T. cystophora RXR in the presence of the ligandwas not demonstrated.

Among the non-moulting triploblastic invertebrates,the Lophotrochozoa (Aguinaldo et al. 1997), RXR familymembers have been characterized only in the platy-helminth parasite of man, Schistosoma mansoni. How-ever, both receptors described (Freebern et al. 1999,De Mendonça et al. 2000b) have poorly conserved LBDsand are unlikely to be able to bind 9-cis RA. Moreover,they also show atypical DNA-binding properties and intheir native forms do not bind to conventional DRelements. Nevertheless, RA signaling remains a possibleelement in the molecular dialog that exists between theparasite and its definitive and intermediate hosts(reviewed in De Mendonça et al. 2000a), and in order totest this hypothesis we have sought to characterize novelRXRs, both from the parasite and from its intermediatehost, the freshwater snail Biomphalaria glabrata.

Here we describe a structurally and functionallyconserved RXR from B. glabrata (BgRXR) that bothbinds to conventional DR elements as a homodimer oras a heterodimer partner with mammalian nuclearreceptors, and transactivates transcription from theseresponse elements in the presence of 9-cis RA whentransfected alone into mammalian cell lines. We presentevidence showing that the transcriptional activity ofBgRXR observed is due to the direct binding ofretinoids by its LBD. Our results argue both for theconservation of nuclear receptor–cofactor interactions inthe mollusk, but also for the importance of RA signalingand its conservation in the Lophotrochozoa.

Materials and methods

Biologic material

B. glabrata snails were maintained in a mixture (1:1, v/v)of deionized water and mineral water at 24 �C and fedwith fish pet food maintained solid with alginate, underan approximately natural diurnal light cycle. Snails werestarved for 6 h before use. The B. glabrata embryonic cellline (Bge) (Hansen 1976) was maintained in SchneiderDrosophila medium (Invitrogen), 22% (v/v); lactalbuminhydrolysate (Invitrogen), 0·45% (w/v); D+ galactose(Sigma), 0·13% (w/v); phenol red (Sigma) 0·01‰ (w/v);fetal bovine serum (Invitrogen), 10% (v/v); anddouble-distilled water 68% (v/v). S. mansoni cercariaewere released from infected snails and harvested on ice.Balb/c mice were obtained from D. Dombrowicz(INSERM U 547, Institut Pasteur de Lille, France) andkilled by cervical dislocation before removal of livers.Total RNA was obtained by the guanidine thiocyanate/caesium chloride method, as described by Chirgwin et al.(1979), from mollusk hepatopancreas, Bge cells, mouseliver or S. mansoni cercariae. Poly A+ RNA was purifiedwith the Oligotex mRNA kit (Qiagen). Genomic DNAwas extracted by standard techniques (Sambrook &Russell 2001) from Bge cells, whole mollusk bodies,S. mansoni cercariae and mouse liver.

Touchdown RT-PCR

Touchdown RT-PCR was performed by the method ofDon et al. (1991). Briefly, reverse transcription of 5 µgtotal RNA was carried out with 110 pmol randomprimers (Promega) and the Superscript Kit (Invitrogen).cDNA (2 µl) was then amplified with 40 pmol forward(5�-ggI AAR CAY TAY ggI gTI TAC Ag-3�) and reverse(5�-ARC TCI gWI ARI ACI CKR TCR AA-3�)degenerate primers in a total volume of 50 µl with 2·5 Uof TaqGold DNA polymerase (Applied Biosystems,Courtaboeuf, France), the supplied buffer and 2 mMMgCl2. After 10 min of denaturation at 95 �C, four setsof five cycles of 95 �C for 15 s, at annealing temperaturesof respectively 55 �C, 50 �C, 45 �C and 40 �C for 30 sand 72 �C for 90 s were performed, followed by 25cycles of 95 �C for 15 s, 37 �C for 30 s and 72 �C for 90s with an Applied Biosystems 9700 Thermocycler.Nested PCR was performed with forward (5�TgY gARggI TgY AAR ggI TTY TTY AA-3�) and reverse (5�-TKS IKI CgI SWR TRC TCY TC-3�) primers, 2 µl ofthe previous reaction and the same amplificationprogram. Analysis of the product was carried out on 1%agarose gels in TAE buffer stained with ethidiumbromide. Fragments of interest were excised from thegel, purified on silica beads (Geneclean kit, BIO 101,QBiogene, Irvine, CA, USA) and cloned into pCR2·1-TOPO (Invitrogen). Top 10F’electrocompetent cellswere transformed with an Equibio (Boughton, Kent,

D BOUTON and others · A conserved RXR from Biomphalaria glabrata568

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582

UK) electroporator under 2500 V and 25 µF in a2-mm-wide cuvette. Plasmids from positive clones wereprepared by alkaline lysis (Birnboim & Doly 1979).Sequencing was performed on an ABI 377 automatedsequencer (Applied Biosystems), by the methods andreagents of the supplier. Full-length cDNA was obtainedby performing 5�and 3�RACE with cDNA preparedfrom Bge cells with the SMART RACE kit (Clontech) asa template (a gift from Colette Dissous, INSERM U 547,Institut Pasteur de Lille, France), and primers derivedfrom the sequence of the initial fragment as follows:5�-gAA AgA gAA Agg ggA CggTgA ggg gAg AgC-3�and 5�-gAT CCA gTg ACC AAC ATA TgC CAg gCAgCT g-3� for 3�-RACE and 5�-ATg gAg Cgg AgT gCAggT AAT CgA AgA Agg Ag-3� and 5�-Tgg CCA CCAgCT CAg TCA ggA CAC ggT-3� for 5�-RACE.

Sequence analysis and phylogenetic tree

The protein sequence of BgRXR was manually alignedto homologs with Seaview (Galtier et al. 1996). All othersequences than BgRXR were taken from the non-redundant database of nuclear receptors NUREBASE(Duarte et al. 2002), families NRB010202 andNRB010201. These data can be accessed at www.ens-lyon.fr/LBMC/laudet/nurebase/nurebase.html. Thetree was rooted with T. cystophora RXR as an outgroup.Only complete sites (no gap, no X) were used forphylogenetic analysis. Trees were constructed byMaximum Likelihood, as implemented in PhyML(Guindon & Gascuel 2003), with the JTT substitutionmodel; rate heterogeneity between sites was corrected bya gamma law with eight parameters (alpha estimated byPhyML). The reliability of branches was evaluatedby bootstrap with 1000 replicates (Felsenstein 1985), andby likelihood tests as implemented in Tree-Puzzle(Strimmer & Von Hessler 1996).

Northern and Southern blotting

Electrophoresis of poly A+ RNA from the mollusk wascarried out alongside RNA size markers from NewEngland Biolabs (Beverley, MA, USA) in a 1%agarose/3% formaldehyde gel in MOPS buffer (Lehrachet al. 1977). The gel was then blotted onto a Hybond N+

membrane (Amersham). The full-length cDNA probewas �32P-labeled by the Random Primers DNALabelling System (Invitrogen). A probe corresponding tothe complete coding sequence of B. glabrata actin(Lardans et al. 1997) was labeled in the same way andused as a control. Hybridization was carried out asdescribed elsewhere (Sambrook & Russell 2001), andblots were exposed overnight to X-Omat AR film(Kodak). Genomic DNA (25 µg) was digested withEcoRV and BamHI and separated on a 1·2% agarosegel in TAE buffer. After transfer to a Hybond N+

membrane, hybridization was carried out by a standardmethod (Sambrook & Russell 2001) with a 474 bp probe(nucleotides 589–1062 of the BgRXR cDNA) radio-labeled as above. After stringent washes, the blot wasexposed overnight to X-Omat AR film.

Antibodies

Two ovalbumin-coupled peptides (supplied by Synt:em,Nimes, France) covering the residues 177–191 and371–386 of BgRXR were used to immunize two NewZealand rabbits (IFFA-Credo, Larbresle, France) (Vaitu-kaitis et al. 1971). Sera were collected 2 months after thefirst injection and tested for the presence of specificantibodies by ELISA against uncoupled peptidesadsorbed onto Maxisorp immunoplates (Nunc, Roskilde,Denmark). The purified immunoglobulin (Ig) G fractionof these sera, as well as commercial antibodies (SantaCruz Biotechnologies, Santa Cruz, CA, USA) directedagainst the LBD of human RXR� (residues 198–468),was also tested by Western blotting.

Western blotting and immunohistochemistry

Mollusks were ground in lysis buffer containing NH4Cl150 mM, EDTA 100 mM, NaHCO3 100 mM andPMSF 1 mM with an Ultraturax (IKA, Labortechnik,Stauten, Germany). Bge cells were resuspended in thesame buffer, and both were sonicated three times for10 s (maximum power; Microson XL, Misonix,Farmingdale, NY, USA). The homogenates werecentrifuged 20 800 g for 30 min, and the protein contentof the supernatant was measured with the BCA assay kit(Pierce, Rockford, IL, USA). Each extract (1·4 mg) wasseparated on a 10% SDS/polyacryamide gel and blottedonto a nitrocellulose membrane (Schleicher & Schuell,Dassel, Germany) (Towbin et al. 1979). Blots weredeveloped with either the purified IgG fraction ofantipeptide antisera or commercial antibodies at adilution of 1/4000 in PBS/skim milk 1% (w/v) andperoxidase-coupled antirabbit IgG antiserum (Sigma) ata 1/20 000 dilution in PBS/skim milk 1% (w/v).Detection was carried out by chemiluminescence withthe West-Pico detection kit (Pierce). Immunofluores-cence detection of recombinant BgRXR expressed intransfected cells was performed as described elsewhere(De Mendonça et al. 2000b).

Electrophoretic mobility shift assay (EMSA)

BgRXR cDNA was cloned into XhoI/KpnI digestedpTL1 (a modified version of pSG5; Stratagene, La Jolla,CA, USA), for transient transfection assays and in vitrotranslation. Vectors expressing potential heterodimerpartners were as follows: human vitamin D receptorand human RAR�, cloned in pSG5, were gifts from

A conserved RXR from Biomphalaria glabrata · D BOUTON and others 569

www.endocrinology-journals.org Journal of Molecular Endocrinology (2005) 34, 567–582

J M Vanacker (CNRS UMR 49, Ecole NormaleSupérieure de Lyon, France); human LXR, FXR andPPAR�, cloned in pcDNA3, were gifts from B. Staels(INSERM U 545, Institut Pasteur de Lille, France);mouse RXR�, cloned in pTL-1, was used as previouslydescribed (De Mendonça et al. 2000b). Recombinantproteins were obtained in vitro with the rabbitreticulocyte lysate TNT kit (Promega). The EMSA wasperformed with 40�103 c.p.m. of a �-32P-ATP-end-labeled, double-stranded oligonucleotide probe and 2 µlin vitro synthesized proteins. Binding reactions wereperformed as previously described (Vanacker et al. 1993).Reaction products were run on a 5% native polyacryla-mide gel in Tris/borate/EDTA. Gels were dried andexposed overnight to X-Omat AR film.

Transactivation assays

Transactivation experiments with intact BgRXR ormurine RXR� cloned into pTL1 were done with theApo-A1-tk-luc reporter gene (de Urquiza et al. 2000) (agift from T Perlmann, Ludwig Institute, Stockholm,Sweden). In order to test the activity of the LBDindependently of the DNA-binding properties of thereceptors, constructs allowing fusion of the LBD to theGal4 protein were used. An EcoRI fragment encodingamino acids 1–143 of the Saccharomyces cerevisiae Gal4protein was cloned into pTL1 (pTL1-Gal4). Theoligonucleotide primers 5�-ggA AgC TTA AgC gAgAAg CTg TCC Agg Agg AA-3� and 5�-AAg Cgg CCgCgg ATT ggT TgC TgA CTg CAT CCT-3� or 5�-AAgCgg CCg CCT AAA ATg TgT gTC AAT ggg CTggTC A-3� were used respectively to amplify the regionsencoding BgRXR hinge and ligand domains (aminoacids 183–436) and the hinge and ligand domaintruncated at the C-terminal end (amino acids 183–424;�AF2). Resulting PCR products were cloned into pCR2·1-TOPO and sequenced. Inserts containing no errorswere excised with HindIII and NotI and cloned into thecorresponding sites in pTL1-Gal4 in frame with theGal4 DBD. The expression plasmid pGal4 was fromStratagene, and the pGal4-RXR containing theDNA-binding domain of the yeast Gal4 protein fused tothe LBD of murine RXR� was kindly provided by P.Lefebvre (INSERM U459, Lille, France). The plasmidpGal5-TK-pGL3 was obtained by inserting five copiesof the Gal4 DNA binding site in front of the thymidinekinase (TK) promoter of the TK-pGL3 plasmid(Promega). The construct expressing the BgRXR hingeand ligand domains fused to the VP16 activationdomain was constructed with the oligonucleotideprimers, 5�-AAg gAT CCT gAA gCg AgA AgC TgTCCA ggA ggA A-3� and 5�-ggA TTT TAT TgC TgCAgC TAT TgT ggC Agg-3� for PCR. The resultingamplification products were cloned into pCR 2·1-TOPO and sequenced. Inserts were excised with

BamHI and PstI and cloned into the corresponding sitesin the pVP16 vector (Clontech) in frame with the VP16activation domain.

The CV-1 cell line (ATCC CCL-70), the 293Thuman embryonic kidney cell line (ATCC CRL-11268)and the Cos-1 cell line (ATCC CRL-1650) weremaintained in DMEM (Invitrogen) supplemented with10% fetal bovine serum. Bovine aortic endothelial cells(BAE cells) were obtained by scraping the aorta excisedfrom a freshly slaughtered cow. The different clones ofcells were cultured in DMEM containing fibroblastgrowth factor (1 ng/ml) with 10% newborn calf serum.The cultures used in the present study had undergonesix passages. Cells were transfected with 1–1·2 µg totalDNA per assay and polyethyleneimine (4 µl Ex Gen 500,Euromedex, Mundolsheim, France) under the con-ditions recommended by the supplier. The pTL1plasmid was used as a carrier when necessary. Ligands(ethanol solution of 9-cis retinoic acid 10 mM; ethanolsolution of cis-4,7,10,13,16,19-docosahexanoic acid(DHA) (de Urquiza et al. 2000) (Sigma) (100 mM); aDMSO solution of the RXR specific agonist LG268(10 mM; a gift from B. Staels) or a 10 mM solution ofWY-14643 (4-chloro-6-(2,3-xylidino)-2-pyrimidinylthioacetic acid) (Alexis Biochemicals, Lausen, Switzerland)in DMSO were incubated with the cells overnight. Cellswere lysed in the commercial reporter lysis buffer(Promega) and assayed for luciferase (assay kit, Promega)in a Wallac Victor 1420 multilabel counter (Perkin-Elmer, Wellesley, MA, USA).

Limited proteolytic digestion

For limited proteolytic digestions (Vivat et al. 1997),in vitro synthesized 35S-labeled BgRXR (TNT kit,Promega) was used. Briefly, after incubating on ice for30 min with ligands, receptor proteins were digested at25 �C for 10 min with 50 mg/ml trypsin. The proteo-lytic fragments were separated on a 15% SDS-PAGEpolyacrylamide gel and visualized by autoradiography.

Results

Isolation of BgRXR from the mollusk B. glabrata

Touchdown PCR followed by nested touchdown PCRon cDNA from Bge cells with degenerate primersflanking the DNA-binding domain (DBD) and LBD ofRXR allowed us to obtain a strong single band of833 bp, which contained an open reading frame (ORF),including the nuclear receptor signature DNA andLBDs, sharing a high level of peptide sequence identity(89% and 81% respectively) with those of mouse RXR�.5� and 3� RACE allowed the assembly of a contig of1721 bp which contained the entire ORF of 1308 bpcorresponding to a 436 residue protein. Primers were

D BOUTON and others · A conserved RXR from Biomphalaria glabrata570

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582

then designed, flanking this ORF, and RT-PCR wascarried out to obtain a unique clone encompassing thefull-length coding sequence of BgRXR. Although thisclone carried the complete ORF, it did not represent theentire mRNA sequence, since Northern blotting of snailpoly A+ RNA revealed a unique band of 5 kb (Fig. 1A,lane 1), while a blot probed with the B. glabrata actincDNA gave a band at the expected size of about 3 kbp(lane 2). This indicates that about 3280 bases of UTRare missing from the BgRXR cDNA. Although theresulting sequence terminated with a poly A tail, theabsence of a consensus polyadenylation signal (AAGAAA instead of AATAAA) suggested that the 3� UTR,as well as the 5�UTR, was incomplete. No varianttranscripts were identified during RACE-PCR experi-ments, suggesting that the BgRXR gene does not producealternatively spliced or truncated forms of the mRNAencoding protein isoforms, although this phenomenonhas been described for RXRs in other organisms (Guoet al. 1998, Kostrouch et al. 1998).

The snail origin of the sequence was furtherconfirmed by Southern hybridization using EcoRV andBamHI-digested genomic DNA from Bge cells and, as

controls, DNA from S. mansoni cercariae and from mice.No signal was obtained with either mouse or schistosomeDNA, whereas a single band of about 4·5 kbp wasobtained with Bge cell DNA (not shown). Neitherrestriction enzyme used cuts within the BgRXR cDNAsequence.

The BgRXR peptide sequence is highly conservedcompared with mammalian RXRs

The sequence we obtained encodes a protein of 436amino acids with an estimated size of 48 kDa. Thisprotein shares the modular structure characteristic of thenuclear receptor superfamily with a divergent Nterminal A/B domain, a highly conserved DBD and aless well conserved LBD. As usual in other RXRs, thereis no F domain following the LBD (Chung et al. 1998).

The DBD of BgRXR is extremely well conserved(Fig. 2A) compared with Drosophila USP and vertebrateRXR DBDs, whereas the LBD shows greater identity tovertebrate RXRs than to Drosophila USP (Fig. 2B). Withinthe BgRXR DBD, the P-box, involved in determiningDNA binding specificity, is 100% identical with those ofother RXRs and USPs, while the D-box, which con-tributes to the homodimerization interface, contains anamino acid substitution compared with vertebratehomologs (D156 substituted for N in mouse RXR�)(Fig. 2A). Furthermore, in another helix known to partici-pate in this interface (Lee et al. 1993) (amino acids 170–180), Q175 and A179 are replaced by M and S respectively(Fig. 2A). These differences could explain specificdimerization properties of BgRXR as discussed below.

The LBD displays a higher level of identity tovertebrate RXRs than to arthropod USPs (75- 81·5%and 43·8–68·7% respectively (Fig. 2B), except for thelocust Locusta migratoria USP: 80·1% (Hayward et al.1999). The overall sequence conservation of the BgRXRLBD is consistent with the maintenance of the canonicalnuclear receptor fold (Wurtz et al. 1996) with 11�-helices (H1, H3–H12) and two short � strands (s1 ands2). Several subdomains, such as the helix-turn zipper inhelix H4 and the regulatory zipper (helices H7-H8), arewell conserved. The identity or I-box, which is involvedin determining the specificity of heterodimerization,(Zhang et al. 1994, Perlmann et al. 1996) and otherresidues involved in interactions with heterodimerpartners (Bourget et al. 2000) are perfectly conserved(Fig. 2B). The � turn, a subdomain involved in bindingpocket formation, mainly in association with helices H3,H5, H7 and H11 (Egea et al. 2000), is completelyconserved, as are the amino acids involved in the ligandentering and anchoring in the ligand-binding pocket(Fig. 2B). The helix H11 is also 100% identical to thoseof vertebrate RXRs. The consensus signature of theLBD ((F,W,Y) (A,S,I)(K,R,E,G)XXX(F,L)XX(L,V,I)

Figure 1 Expression of BgRXR mRNA and protein.(A) Northern blot of poly A+ RNA from B. glabrata hybridizedwith probes for BgRXR (lane 1, 20 µg RNA) and B. glabrataactin (lane 2, 5 µg RNA). Sizes of the fragments are given inbases and were calculated relative to an RNA ladder (Biolabs).(B) Proteins blotted onto a nitrocellulose membrane weredetected with commercial antiserum directed against the LBDof human RXR� (Santa Cruz Biotechnologies). Identical resultswere obtained with sera directed against BgRXR peptides.

A conserved RXR from Biomphalaria glabrata · D BOUTON and others 571

www.endocrinology-journals.org Journal of Molecular Endocrinology (2005) 34, 567–582

XXX(D,S)(Q,K)XX (L,V)(L,I,F)) (Brzozowski et al. 1997,Xu et al. 1998, Kraichely et al. 1999), extending from theC terminal of H3 to the middle of H5, is completelyconserved in BgRXR. All these elements suggest that,unexpectedly, BgRXR may have a similar ligand-binding pocket to that of mouse RXR�. This is alsoindicated by the perfect conservation of residues thathave been shown to interact with 9-cis RA in mouseRXR� (Egea et al. 2000) (Fig. 2B). Moreover, apart fromone substitution, helix H12, and within it the activatingdomain AF2–AD, are completely conserved comparedwith vertebrate RXRs. The consensus core motif ofthe activating domain, designated ϕϕXEϕϕ where ϕ

represents a hydrophobic amino acid, is present in themollusk receptor. The AF2–AD is known to interact in aligand- or phosphorylation-dependent manner with thetranscriptional machinery through transactivation inter-mediary factors (TIF) (Durand et al. 1994, Voegel et al.1996, vom Baur et al. 1996, Hammer et al. 1999), and itsconservation could indicate that BgRXR may interactwith TIFs that are closely related to those identified invertebrates. The only major difference in the LBDstructure of BgRXR compared with mouse RXRconcerns the loop connecting helix H1 to helix H3,which is eight residues shorter in the former (Fig. 2B).However, this region is often found to be extremely

Figure 2 Sequence alignments. (A) The DBD and (B) the LBD of BgRXR aligned withmouse RXR� (mRXR�; accession no. P28700), T. cystophora RXR (TcRXR;AAC80008) and D. melanogaster USP (DmUSP; AAL39538). Identical residues areshown in boldface. Subdomains (P-box, D-box, I-box) are delimited by arrows abovethe sequences. In (B) residues known to interact with 9-cis RA (Egea et al. 2000) areindicated by dark-gray dots above the BgRXR sequence. Helices (H1 and H3–H12)and �-strands (s1 and s2) are as defined in Billas et al. (2001), and are indicatedrespectively by boxes outlined in black and by arrows above the alignment.

D BOUTON and others · A conserved RXR from Biomphalaria glabrata572

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582

variable in length and in sequence, consistent with itsnature as a flexible and loosely structured region (Billaset al. 2001).

Phylogenetic analysis

Because of the similarity between RXR LBDs, asimple distance phylogenetic reconstruction, with rateheterogeneity between sites not taken into account inthis reconstruction, groups BgRXR with vertebrateRXRs (data not shown). With a more realistic modelunder maximum likelihood, BgRXR is placed in amanner consistent with known relationships betweenmetazoans (Fig. 3). In addition to the bootstrap test,likelihood phylogenetic tests significantly support theposition of BgRXR as being neither an arthropod nor a

vertebrate RXR (Expected likelihood weight test(Strimmer & Rambaut 2002) P�0·011). The relation-ships we recover are consistent with previous analyses(Laudet et al. 1992, Gronemeyer & Laudet 1995, Laudet1997), notably including rapid evolution of the USPs.This fast evolution is again consistent with the highersequence and functional resemblance between verte-brate RXRs and BgRXR, rather than between BgRXRand USPs.

The protein BgRXR

We raised antibodies against synthetic peptides derivedfrom the sequence of BgRXR in order to detect theexpression of the BgRXR protein by Western blotting.

Figure 3 Phylogenetic tree of the RXR subfamily. The tree was constructed withMaximum Likelihood, rooted with the RXR from T. cystophora, a cnidarian. The newBiomphalaria sequence is in boldface. Branch lengths are proportional to evolutionarychange; the measure bar represents 0·1 substitutions per site. Figures at nodescorrespond to bootstrap support expressed as the percentage of 1000 replicates.

A conserved RXR from Biomphalaria glabrata · D BOUTON and others 573

www.endocrinology-journals.org Journal of Molecular Endocrinology (2005) 34, 567–582

In addition, in view of the marked sequence identity ofthe LBD with mouse RXR�, we also tested whethercommercial antibodies specific for the C terminalportion of this NR could recognize BgRXR. After in vitrotranscription and translation, the 58 kDa proteinproduced was recognized by both commercial andBgRXR-specific antibodies (not shown) on a Westernblot (Fig. 1B). This size does not correspond to thetheoretical molecular mass of the unmodified BgRXRprotein (48 kDa) and may be due to incorrect folding ofthe molecule synthesized in vitro. Western blotting(Fig. 1B) using either our antibodies (not shown) orcommercial antibodies with the soluble protein fractionof mollusk homogenate as antigen led to the detection ofa single band of 52 kDa, closer to the theoreticalmolecular mass. Moreover, we always obtained a uniqueband indicating that the presence of several Kozak

consensus sequences (Kozak 1991) in the ORF do notinterfere with the translation mechanism.

BgRXR binds to direct repeat elements as a homo-or heterodimer

In order to determine whether BgRXR could bind toconsensus RXR response elements, electrophoreticmobility shift assays (EMSA) were carried out withlabeled probes containing the DR response elementincubated with recombinant protein transcribed andtranslated in vitro. A weak band was obtained withBgRXR alone on the DR1 response element (Fig. 4),whereas mouse RXR� used as a control bound stronglyto DR1 as expected (Mader et al. 1993). Variousmammalian nuclear receptors were tested as hetero-dimer partners since, so far, no other B. glabrata nuclear

Figure 4 BgRXR binds to DR1 as a homo- or heterodimer. EMSA on DR1 of in vitrotranscribed and translated BgRXR (lane 2) or mouse RXR� (mRXR; lane 3) alone or inthe presence of human RAR� (hRAR�, lanes 9 and 14), human peroxisomeproliferator-activated receptor (hPPAR�, lanes 10 and 15), human vitamin D receptor(hVDR, lanes 11 and 16), human liver X receptor (hLXR, lanes 12 and 17) or humanFXR (hFXR, lanes 13 and 18). The EMSAs using the mammalian receptors on theirown are shown in lanes 4–8. The expression vector pSG5 was used on its own as anegative control (lane 1). The autoradiogram of lanes 1 and 2 was exposed at –80 °Covernight and for lanes 3–18 at room temperature for 8 h.

D BOUTON and others · A conserved RXR from Biomphalaria glabrata574

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582

receptors have been characterized. BgRXR hetero-dimerizes weakly with human RAR� and LXR andmore strongly with human FXR and PPAR�(Fig. 4). Both direct binding and competition experi-ments showed that binding was strongest with bothRAR� and PPAR� as partners on the DR1 element,was weaker on DR2 and DR5, and was much weakeron DR3, DR4 and a palindromic response element(data not shown). Unexpectedly, BgRXR was unableto form heterodimers on the DR1 element with oneother common partner of vertebrate RXRs, VDR(Fig. 4), in contrast to Drosophila USP, which does formheterodimers with this receptor.

BgRXR transactivates reporter gene transcription inthe presence of 9-cis retinoic acid

In order to determine whether BgRXR was capable oftransactivating a reporter gene in mammalian cell lines,we constructed a pTL-1-based plasmid expressingBgRXR under the control of the SV40 promoter. Severaldifferent cell lines were used in this study in order toeliminate line-specific effects in these heterologous ex-pression systems. Immunolocalization showed that theBgRXR protein was expressed in the nucleus of trans-fected CV-1 cells (not shown). Surprisingly, preliminaryexperiments showed that BgRXR was unable to promotetransactivation of a reporter gene as a heterodimer part-ner of PPAR�, either with the ligand of the latter,WY-14643, or with the RXR ligand 9-cis RA, despite thestrong binding of PPAR� to BgRXR shown in EMSA. Incontrast, BgRXR on its own consistently transactivatedtranscription in the presence of 9-cis RA or LG268, anRXR-specific agonist, despite the fact that it failed toform homodimers on DR elements in EMSA. We there-fore concentrated our efforts on determining whether thistransactivation was specific and could be unambiguouslyattributed to functional interaction between the verte-brate RXR ligands and the BgRXR LBD.

We therefore sought to confirm the transactivationobserved with BgRXR alone in the presence of RXRligands by determining the dose-dependency of trans-activation with the Apo-A1-tk-luc reporter gene (deUrquiza et al. 2000) in the 293T and Cos-1 cell lines.Figure 5A shows the results obtained in the 293T cellline (Cos-1 cells gave identical results) and thedose-dependent activation of transcription by 9-cis RAobtained, whether with mouse RXR� or BgRXR. Aclearly dose-dependent activation was obtained with thelatter, although it was significantly lower than thatobserved with mouse RXR�. Activation above thebackground level was significant at a concentration of10�8 M 9-cis RA. We next used the specific RXRligand DHA (de Urquiza et al. 2000) at a concentrationof 10�4 M in the same cell lines (Fig. 5B). Activation bymouse RXR� was much greater, but again a significant

activation of transcription in the presence of BgRXRwas noted.

Despite the absence of ligand-dependent transactiva-tion by control cells transfected with the reporterplasmid alone, the observed transactivation by BgRXRcould still have been due to the formation of apermissive heterodimer with an endogenous RXR, theexpression of which had been induced by that ofBgRXR. In order to test our hypothesis that the liganddomain of BgRXR binds 9-cis RA and that it is thisbinding that allows transactivation of transcription, twochimeric proteins were expressed in mammalian cells.The first comprises the hinge and LBD of BgRXR(amino acids 183–436) fused to the Gal4 DBD(Gal4–BgRXR) and the second comprises the hingedomain and a C-terminally truncated LBD of BgRXR(amino acids 183–424) that is deleted for the AF2-AD,again fused to the Gal4 DBD (Gal4–BgRXR �AF2).The plasmids encoding these proteins were cotransfectedinto Cos-1 or BAE cells together with a reporter plasmidcontaining the Gal4 recognition sequence upstream ofthe luciferase gene. Figure 6 shows results obtained fora representative experiment using BAE cells. Resultsobtained with Cos-1 cells were very similar. Transactiva-tion of transcription above the background level obtainedwith the vector expressing the Gal4 DBD alone occurredonly in the presence of 9-cis RA and with the intactBgRXR LBD. The �AF2 construct failed to transactivatetranscription in the presence of the ligand. Indeed, thelevels of luciferase expression obtained were consistentlybelow those produced by the Gal4 DBD alone. Thisindicates that the �AF2 construct squelches the back-ground transcription level due to the Gal4 DBD. Moreimportantly, the results demonstrate that transactivationobtained in the presence of intact BgRXR is due tothe ligand-dependent interaction between the AF2–ADand the cellular transcriptional machinery. Moreover,although, again, we cannot rule out heterodimer forma-tion, the activation we observed is clearly not due toheterodimerization with an endogenous retinoid receptorin the mammalian cells that itself binds the ligand anddrives transcription. A construct expressing a Gal4DBD fusion protein with the mouse RXR� hinge andLBD (Gal4-mRXR�) also transactivated transcriptionexclusively in a ligand-dependent manner in BAE cells(Fig. 6) and Cos-1 cells. The levels of activation above thebackground were superior (about threefold) to thoseobtained with BgRXR (about twofold), but this is consist-ent with the results obtained with the complete receptorson their response element (Fig. 5) and may indicate that9-cis RA is not an optimal ligand for BgRXR.

Since BgRXR bound to DR1 elements only weakly asa homodimer, we sought to demonstrate homodimerformation after transfection into mammalian cells. Usingthe mammalian two-hybrid system, with the BgRXRhinge and LBD fused to the Gal4 DBD, on the one hand,

A conserved RXR from Biomphalaria glabrata · D BOUTON and others 575

www.endocrinology-journals.org Journal of Molecular Endocrinology (2005) 34, 567–582

and to the VP16 activation domain, on the other hand,we were able to show that homodimerization of the LBDdid indeed occur (Fig. 7). Increasing amounts of theBgRXR-VP16 constructs increased the transactivation ofthe luciferase reporter gene in a dose-dependent mannerup to threefold over the background obtained with theGal4-BgRXR construct alone and in the absence ofligand. In contrast, the heterodimerization between theBgRXR-VP16 construct and a fusion between Gal4DBD and the human RAR� LBD produced a strongersignal up to sixfold background. The stronger interactionwith a heterodimer partner in part supports the EMSAresults and agrees with the generally observed strongerinteraction of mammalian RXR and heterodimerpartners, compared with homodimerization of these re-ceptors (Vivat-Hannah et al. 2003).

BgRXR binds 9-cis RA in vitro

To demonstrate that BgRXR physically binds 9-cis RA,we carried out a proteolysis protection assay. BgRXR

produced in vitro in rabbit reticulocyte lysate waspreincubated with 9-cis RA, DHA or other fatty acids atconcentrations from 10�4 to 10�6 M and thenincubated with trypsin. Only the addition of 9-cis RAstabilized the protein (LBD) against proteolysis at allconcentrations (Fig. 8). The failure of DHA to do so,despite the fact that this ligand did have a moderateeffect on transactivation of transcription by BgRXR maybe linked to the high concentrations necessary for thelatter, and the low affinity of BgRXR for DHA impliedby this. Other fatty acids (but not juvenile hormone III)had a stabilizing effect only at the highest concentration,perhaps indicating that they can enter the ligand-binding pocket, but are not bound with a high affinity.

Discussion

The RXR family nuclear receptor from the freshwatersnail B. glabrata that we describe is highly conserved,both in structure and function, particularly with respect

Figure 5 Dose dependency of transactivation of transcription with 9-cis RA and activity of DHA onBgRXR. (A) Different doses of 9-cis RA (0, 10−8, 10−7, 10−6 M) were incubated with 293T cellstransfected with the reporter plasmid (Apo-A1-tk-luc (de Urquiza et al. 2000); DRLuc) alone or withplasmids expressing BgRXR or mouse RXR� (mRXR�). Results are expressed as fold activationcompared with the expression vector control incubated in the absence of ligand. (B) Activation oftranscription in 293T cells transfected with the same reporter and expression plasmids in the presenceand absence of 10−4 M DHA. Results presented are representative of two separate triplicate assays.

D BOUTON and others · A conserved RXR from Biomphalaria glabrata576

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582

to vertebrate RXRs, and is the first RXR from aninvertebrate that has been shown to interact functionallywith vertebrate RXR ligands. This suggests that retinoidsignaling, which evolved early in the Metazoa but hasnot been demonstrated in the ecdysozoan moulting

invertebrates, is shared between the lophotrochozoanbranch of the protostomes (Aguinaldo et al. 1997) andthe deuterostomes.

The observed sequence conservation which extendsover both the DBD and the LBD is particularly high.

Figure 6 Activation of transcription in BAE cells transfected with Gal4 constructs. Plasmidsexpressing the hinge and ligand-binding domains of BgRXR (Gal4–BgRXR), or mouse RXR�(Gal4–mRXR�), or the hinge and ligand domains of BgRXR deleted for the AF2-AD (Gal4–BgRXR�AF2) as fusion proteins with Gal4 DNA binding domain, or the Gal4 DBD alone were cotransfectedwith the pGal5-TK-pGL3 reporter plasmid into BAE cells incubated in the presence or absence of10−6 M 9-cis RA. Results are presented as luciferase units and are representative of two separatetriplicate assays.

A conserved RXR from Biomphalaria glabrata · D BOUTON and others 577

www.endocrinology-journals.org Journal of Molecular Endocrinology (2005) 34, 567–582

On the phylogenetic tree obtained, it is notable thatBgRXR has evolved much more slowly than eitherUSPs or the RXRs from the lophotrochozoan parasite S.mansoni, SmRXR1 and SmRXR2 (De Mendonça et al.2000b), and is thus much less divergent from vertebrateRXRs (Fig. 3). Both vertebrate and Biomphalaria RXRsappear to evolve slowly, and seem to represent themost conserved form of RXR. Given the structuralconservation of the DBD of BgRXR, it is surprising that

it fails to bind more strongly to the DR responseelements as a homodimer in EMSA and differs in thisrespect from vertebrate RXRs. This might be due to theresidue substitution described above (Fig. 2A) in part ofthe DBD known to be important for homodimerformation. However, it is well known that some othermembers of the RXR subfamily, vertebrate orarthropod, are dependent on a heterodimer partner tobind to their response element. For example, Drosophila

Figure 7 BgRXR LBD forms homodimers in mammalian cells. (A) Dose–response of interactionbetween the Gal4–BgRXR and BgRXR–VP16 constructs, expressing the BgRXR LBD fused to theGal4 DBD or the VP16 activation domain respectively, in the mammalian two-hybrid assay (CV-1cells). Amounts of 200, 500 and 800 ng BgRXR–VP16 construct were transfected in the presence of200 ng Gal4 BgRXR and 200 ng reporter plasmid. Amounts of DNA were adjusted to 1·2 µg with thepVP16 plasmid. Results are presented as fold activation compared with the pVP16 plasmidcotransfected with Gal4–BgRXR and are representative of two separate triplicate assays. (B) TheBgRXR LBD heterodimerizes with the human RAR� LBD in the mammalian two-hybrid assay. Aconstruct expressing the human RAR� LBD fused to the Gal4 DBD (Gal4-hRAR�, 200 ng) wascotransfected with the BgRXR-VP16 construct (800 ng) and reporter plasmid (200 ng) in CV-1 cells.Results are presented as fold activation compared with the pVP16 plasmid cotransfected withGal4–hRAR�.

D BOUTON and others · A conserved RXR from Biomphalaria glabrata578

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582

USP can only bind as a heterodimer with a variety ofpartners (Yao et al. 1992). Similarly, Danio rerio RXR�can bind to DR1 as a heterodimer with D. rerio RAR�and produce a shifted band in EMSA but is unable to doso alone (Jones et al. 1995). It is also noteworthy thatBgRXR is not able to bind as a heterodimer with one ofthe vertebrate nuclear receptors, VDR, that interactswith Drosophila USP (Yao et al. 1992). The very stronginteraction noted with PPAR� probably does notindicate the presence of a PPAR in B. glabrata, however,as this subfamily (as is the case for RAR) has beendetected only in vertebrates (Laudet 1997), even if thepresence of subfamily I nuclear receptors in lophotroco-zoans cannot be excluded (Bertrand et al. 2004). Theheterodimeric complexes containing BgRXR were ableto bind to either the DR1 or the DR5 response elements.Vertebrate RXR–RAR heterodimers bind to eachelement with an opposite polarity and this in turndetermines different responses to ligands and coactiva-tors. This property is associated with a switch in activityfrom repression to activation of retinoid responsive genes(Kurokawa et al. 1994, 1995). The BgRXR-PPAR�heterodimer should bind only to the DR1 element(Kliewer et al. 1992) and it is possible that the amino-acidsubstitution in the D box of BgRXR may induce achange in binding specificity due to a modification of thedimerization interface.

In mammalian RXR the major dimerization interfaceis present on the LBD and implicates residues fromhelices H7 to H10 and loops L8–9 and L9–10 (Bourguetet al. 2000). Among these, the I-box in H10 is critical forthe determination of identity in heterodimeric inter-actions (Perlmann et al. 1996), whereas the mutation of asingle Tyr residue in H9 to Ala favors the formation ofhomo- over heterodimers (Vivat-Hannah et al. 2003).Both the I-box and the H9 Tyr residue are perfectlyconserved in BgRXR, suggesting that, like its mam-malian counterparts, the mollusk receptor functionsprimarily as a heterodimer partner. However, while theresults obtained by the mammalian two-hybrid assayconfirm that the heterodimeric interactions of theBgRXR LBD with partners are more stable, possibly

explaining why they are stronger in EMSA, they alsosupport the assertion that BgRXR can function as ahomodimer.

The transactivation of transcription observed in threedifferent cell lines by BgRXR on its own in the presenceof 9-cis RA or of specific RXR agonists is the first suchdemonstration for an invertebrate RXR. However, analternative explanation for the observed transactivationwith BgRXR alone is the presence of an endogenousnuclear receptor in the mammalian cells that formed aheterodimer with BgRXR, resulting in a 9-cis RAactivated complex. This explanation applies, providedthat these heterodimers are permissive; that is, that theycould be activated by ligands of both receptorsregardless of the order of binding (Kliewer et al. 1992,Willy et al. 1995, Willy & Mangelsdorf 1997). However,this hypothetic partner is not an RXR or an RAR sinceno transactivation was detected in cells transfected withthe reporter plasmid alone in the presence of ligand.Moreover, the fact that DHA activates BgRXR is ofparticular interest since it has been proposed as a naturalligand of mammalian RXR (de Urquiza et al. 2000).Moreover, it is specific for RXR and will not activateRAR, unlike 9-cis RA, which does have some effect onthe latter and this indicates that the observedtransactivation is indeed due to the activation ofBgRXR. This conclusion is further supported by theresults obtained with chimeric Gal4-BgRXR constructstransfected into mammalian cells. In this system,transactivation of a reporter gene was dependent on thepresence of the ligand and the AF2-AD of the receptor.This shows that transactivation occurred via theligand-dependent interaction of the BgRXR AF2domain with the host cell transcriptional machinery andwas not due to the formation of an active heterodimerwith an endogenous receptor. In the latter casetransactivation should have occurred with the �AF2construct, whereas in fact, in this situation, the reportergene expression was greatly reduced compared with thebackground level obtained with the Gal4 DBD alone.Moreover, we have shown that the BgRXR LBD canform homodimers in transfected mammalian cells with

Figure 8 9-cis RA protects the BgRXR LBD against trypsin digestion. 35S-labeled BgRXR produced inrabbit reticulocyte lysate was incubated with each ligand at 10−4, 10−5 and 10−6 M. Ligands usedwere 9-cis retinoic acid (9-cis RA), docosahexanoic acid (DHA), phytanic acid (Phyt.), juvenilehormone III (JHIII), methoprene acid (Metop.) and oleic acid (oleic). The solvent (ethanol, Eth) wasused as a negative control. The first lane (U) represents the input.

A conserved RXR from Biomphalaria glabrata · D BOUTON and others 579

www.endocrinology-journals.org Journal of Molecular Endocrinology (2005) 34, 567–582

the two-hybrid assay, supporting the view that thetransactivation of transcription in cells transfectedwith the full-length BgRXR is due to this receptoralone. Finally, the observation that 9-cis RA stabilizesBgRXR toward trypsin proteolysis further supports thehypothesis that it binds this ligand.

The transactivation of transcription by BgRXR in thepresence of both 9-cis RA and DHA was clearlysignificant, but much less marked than that obtainedwith mouse RXR�. This could be due in part to the useof heterologous mammalian cell lines and the conse-quent suboptimal interaction between the molluskreceptor and the endogenous transcriptional coactiva-tors. However, it may also suggest that neither of thecompounds tested represents the natural BgRXR ligand.In mammals, the status of 9-cis RA as a natural RXRligand is controversial since it is undetectable in normaltissues in vivo, even using highly sensitive techniques(Ulven et al. 2001). DHA has since been suggested as anatural ligand in mammals (de Urquiza et al. 2000)although it is active at much higher concentrations than9-cis RA. Its presence or absence in B. glabrata, as is thecase for other ‘rexinoids’, is at present unknown andrequires urgent study.

The only other invertebrate RXR that has beenshown to interact with 9-cis RA is the T. cystophorareceptor, which was shown to bind the ligand, but forwhich no ligand-dependent transactivation of transcrip-tion was demonstrated (Kostrouch et al. 1998). The factthat BgRXR was able to transactivate transcription in aligand-dependent manner may be due to the highdegree of conservation, not only of the LBD itself,including residues known to interact with the ligand inmouse RXR� (Egea et al. 2000), but also of the interfacesresponsible for cofactor binding, allowing its activity in aheterologous cell system. Our results favor theconservation of RA signaling in triploblastic inverte-brates, but this may have been lost in the Ecdysozoa. Inthe arthropods the LBD sequences can be divided intotwo groups (Hayward et al. 1999), with RXR/USPfamily members from crabs, ticks and hemimetabolousinsects showing far greater similarity to vertebrate RXRLBDs than do those from Diptera and Lepidoptera (Fig. 3).However, the RXR/USPs from the former group do notactivate transcription in the presence of either 9-cis RAor a synthetic retinoid (Guo et al. 1998). It is clear thatvertebrate-like RXRs are absent from both Drosophila(Adams et al. 2000) and the nematode Caenorhabditiselegans (Sluder et al. 1999), and the latter even lacks aUSP homolog, although one is present in the parasiticnematode Dirofilaria immitis (Sluder & Maina 2001, Sheaet al. 2004).

The presence of a highly conserved, functional,vertebrate-type RXR in a mollusk poses the problem ofthe evolution of ligand binding in the RXR family, andin the nuclear receptor superfamily in general. Recently,

an estrogen receptor (ER) �-related receptor was isolatedfrom another mollusk, Aplysia californica (Thornton et al.2003), but it failed to bind estradiol and wasconstitutively active in transactivation assays. However,the structure of a hypothetic ancestral ER was predictedand the corresponding molecule was synthesized. Thisdid bind estradiol, albeit feebly, suggesting that the bonafide ancestral ER may have bound a steroid ligand, andthat this capacity was lost in Aplysia ER. There is noneed to reconstruct an ancestral RXR since BgRXR isboth highly conserved and transactivates transcriptionunder the control of ligand. Taken together with thediploblast T. cystophora RXR which binds 9-cis RA, andthe RXR family receptor in the sponge S. domuncula, theexpression of which is upregulated on treatment withRA, our results support the view that ligand binding andits function in the control of transcription was acquiredduring the evolution of the nuclear receptor superfamily(Escriva et al. 1997).

Acknowledgements

This work was supported by the Institut National de laSanté et de la Recherche Médicale (U 547), the CentreNational de la Recherche Scientifique (UMR 49), theInstitut Pasteur de Lille, the École Normale Supérieurede Lyon and the microbiology program of the Ministèrede l’Education Nationale de la Recherche et de laTechnologie. D B benefited from grants by the InstitutPasteur de Lille, the Région Nord-Pas de Calais and theFondation pour la Recherche Médicale. R M wassupported by the Fondation des Treilles and by theInstitut Pasteur de Lille. H E was supported by theEuropean Molecular Biology Organization. B B wassupported by a grant from the Ministère de l’EducationNationale de la Recherche et de la Technologie.

References

Adams MD, Celniker SE, Holt RA, Evans CA, Gocayne JD,Amanatides PG, Scherer SE, Li PW, Hoskins RA, Galle RF et al.2000 The genome sequence of Drosophila melanogaster. Science 2872185–2195.

Aguinaldo AM, Turbeville JM, Linford LS, Rivera MC, Garey JR,Raff RA & Lake JA 1997 Evidence for a clade of nematodes,arthropods and other moulting animals. Nature 387 489–493.

Bertrand S, Brunet FG, Escriva H, Parmentier G, Laudet V &Robinson-Rechavi M 2004 Evolutionary genomics of nuclearreceptors: from twenty-five ancestral genes to derived endocrinesystems. Molecular Biology and Evolution 21 1923–1937.

Biesalski HK, Doepner G, Tzimas G, Gamulin V, Schroder HC,Batel R, Nau H & Muller WE 1992 Modulation of mybgene expression in sponges by retinoic acid. Oncogene 71765–1774.

Billas IML, Moulinier L, Rochel N & Moras D 2001 Crystalstructure of the ligand-binding domain of the ultraspiracle protein

D BOUTON and others · A conserved RXR from Biomphalaria glabrata580

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582

USP, the ortholog of retinoid-X-receptors in insects. Journal ofBiological Chemistry 276 7465–7474.

Birnboim HC & Doly J 1979 A rapid alkaline extraction procedurefor screening recombinant plasmid DNA. Nucleic Acids Research 71513–1523.

Bourget W, Vivat J, Wurtz JM, Chambon P, Gronemeyer H &Moras D 2000 Crystal structure of the heterodimeric complexof RAR and RXR ligand-binding domains. Molecular Cell 5289–298.

Brzozowski AM, Pike AC, Dauter Z, Hubbard RE, Bonn T,Engstrom O, Ohman L, Greene GL, Gustafsson JA & CarlquistM 1997 Molecular basis of agonism and antagonism in theoestrogen receptor. Nature 389 753–758.

Chirgwin JM, Przybyla AE, MacDonald RJ & Rutter WJ 1979Isolation of biologically active ribonucleic acid from sourcesenriched in ribonuclease. Biochemistry 18 5294–5299.

Chung AC, Durica DS, Clifton SW, Roe BA & Hopkins PM 1998Cloning of crustacean ecdysteroid receptor and retinoid-Xreceptor gene homologs and elevation of retinoid-X receptormRNA by retinoic acid. Molecular and Cellular Endocrinology 139209–227.

De Mendonça RL, Escriva H, Bouton D, Laudet V & Pierce RJ2000a Hormones and nuclear receptors in schistosomedevelopment. Parasitology Today 16 233–240.

De Mendonça RL, Escriva H, Bouton D, Zelus D, Vanacker JM,Bonnelye E, Cornette J, Pierce RJ & Laudet V 2000b Structuraland functional divergence of a nuclear receptor of the RXRfamily from the trematode parasite Schistosoma mansoni. EuropeanJournal of Biochemistry 267 3208–3219.

de Urquiza AM, Liu S, Sjöberg M, Zetterström RH, Griffiths W,Sjövall J & Perlmann T 2000 Docosahexanoic acid, a ligandfor the retinoid X receptor in mouse brain. Science 2902140–2144.

Don RH, Cox PT, Wainwright BJ, Baker K & Mattick JS 1991‘Touchdown’ PCR to circumvent spurious priming during geneamplification. Nucleic Acids Research 19 4008.

Duarte J, Perriere G, Laudet V & Robinson-Rechavi M 2002NUREBASE: database of nuclear hormone receptors. Nucleic AcidsResearch 30 364–368.

Durand B, Saunders M, Gaudon C, Roy B, Losson R & ChambonP 1994 Activation function 2 (AF-2) of retinoic acid receptor and9-cis retinoic acid receptor: presence of a conserved autonomousconstitutive activating domain and influence of the nature of theresponse element on AF-2 activity. EMBO Journal 13 5370–5382.

Egea PF, Mitschler A, Rochel N, Ruff M, Chambon P & Moras D2000 Crystal structure of the human RXR� ligand-bindingdomain bound to its natural ligand: 9-cis retinoic acid. EMBOJournal 19 2592–2601.

Escriva H, Safi R, Langlois MC, Saumitou-Laprade P, Stehelin D,Capron A, Pierce RJ & Laudet V 1997 Ligand binding wasacquired during evolution of nuclear receptors. PNAS 946803–6808.

Felsenstein J 1985 Confidence limits on phylogenies: confidencelimits using the bootstrap. Evolution 39 783–791.

Freebern WJ, Osman A, Niles EG, Christen L & LoVerde PT 1999Identification of a cDNA encoding a retinoid X receptorhomologue from Schistosoma mansoni. Evidence for a role infemale-specific gene expression. Journal of Biological Chemistry 2744577–4585.

Galtier N, Gouy M & Gautier C 1996 SEAVIEW andPHYLO_WIN: two graphic tools for sequence alignment andmolecular phylogeny. Computer Applications in the Biosciences 12543–548.

Gronemeyer H & Laudet V 1995 Transcription factors 3: nuclearreceptors. Protein Profile 2 1173–1308.

Guindon S & Gascuel O 2003 A simple, fast, and accuratealgorithm to estimate large phylogenies by maximum likelihood.Systematic Biology 52 696–704.

Guo X, Xu Q, Harmon MA, Jin X, Laudet V, Mangelsdorf DJ &Palmer MJ 1998 Isolation of two functional retinoid X receptorsubtypes from the ixodid tick, Amblyomma americanum (L.). Molecularand Cellular Endocrinology 139 45–60.

Hammer GD, Krylova I, Zhang Y, Darimont BD, Simpson K,Weigel NL & Ingraham HA 1999 Phosphorylation of the nuclearreceptor SF-1 modulates cofactor recruitment: integration ofhormone signaling in reproduction and stress. Molecular Cell 3521–526.

Hansen EL 1976 A cell line from embryos of Biomphalaria glabrata(Pulmonata), establishment and characteristics. In Invertebrate TissueCulture: Research Applications, pp 75–97. Ed K Maramorosch. NewYork: Academic Press.

Hayward DC, Bastiani MJ, Trueman JW, Truman JW, RiddifordLM & Ball EE 1999 The sequence of Locusta RXR, homologousto Drosophila ultraspiracle, and its evolutionary implications.Development Genes and Evolution 209 564–571.

Imsiecke G, Borojevic R & Müller WEG 1994 Retinoic acid acts asa morphogen in freshwater sponges. Invertebrate Reproduction andDevelopment 26 89–98.

Jones BB, Ohno CK, Allenby G, Boffa MB, Levin AA, Grippo JF& Petkovich M 1995 New retinoid X receptor subtypes in zebrafish (Danio rerio) differentially modulate transcription and donot bind 9-cis retinoic acid. Molecular and Cellular Biology 155226–5234.

Kliewer SA, Umesono K, Noonan DJ, Heyman RA & Evans RM1992 Convergence of 9-cis retinoic acid and peroxisomeproliferator signaling pathways through heterodimer formation oftheir receptors. Nature 358 771–774.

Kostrouch Z, Kostrouchova M, Love W, Jannini E, Piatigorsky J &Rall JE 1998 Retinoic acid X receptor in the diploblast, Tripedaliacystophora. PNAS 95 13442–13447.

Kozak M 1991 An analysis of vertebrate mRNA sequences:intimations of translational control. Journal of Cell Biology 115887–903.

Kraichely DM, Collins JJ 3rd, DeLisle RK & MacDonald PN 1999The autonomous transactivation domain in helix H3 ofthe vitamin D receptor is required for transactivation andcoactivator interaction. Journal of Biological Chemistry 27414352–14358.

Kurokawa R, DiRenzo J, Boehm M, Sugarman J, Gloss B,Rosenfeld MG, Heyman RA & Glass CK 1994 Regulation ofretinoid signaling by receptor polarity and allosteric control ofligand binding. Nature 371 528–531.

Kurokawa R, Soderstrom M, Horlein A, Halachmi S, Brown M,Rosenfeld MG & Glass CK 1995 Polarity-specific activities ofretinoic acid receptors determined by a co-repressor. Nature 377451–454.

Lardans V, Ringaut V, Duclermortier P, Cadoret JP & Dissous C1997 Nucleotide and deduced amino acid sequences ofBiomphalaria glabrata actin. DNA Sequence 7 353–356.

Laudet V 1997 Evolution of the nuclear receptor superfamily: earlydiversification from an ancestral orphan receptor. Journal ofMolecular Endocrinology 19 207–226.

Laudet V, Hanni C, Coll J, Catzeflis F & Stehelin D 1992 Evolutionof the nuclear receptor gene superfamily. EMBO Journal 111003–1013.

Lee MS, Kliewer SA, Provencal J, Wright PE & Evans RM 1993Structure of the retinoid X receptor alpha DNA binding domain:a helix required for homodimeric DNA binding. Science 2601117–1121.

Lehrach H, Diamond D, Wozney JM & Boedtker H 1977 RNAmolecular weight determinations by gel electrophoresis underdenaturing conditions: a critical reexamination. Biochemistry 164743–4747.

Mader S, Chen JY, Chen Z, White J, Chambon P & Gronemeyer H1993 The patterns of binding of RAR, RXR and TR homo- andheterodimers to direct repeats are dictated by the binding

A conserved RXR from Biomphalaria glabrata · D BOUTON and others 581

www.endocrinology-journals.org Journal of Molecular Endocrinology (2005) 34, 567–582

specificities of the DNA binding domains. EMBO Journal 125029–5041.

Mangelsdorf DJ & Evans RM 1995 The RXR heterodimers andorphan receptors. Cell 83 841–850.

Mangelsdorf DJ, Thummel CS, Beato M, Herrlich P, Schatz G,Umesono K, Blumberg B, Mark M, Chambon P & Evans RM1995 The nuclear receptor superfamily: the second decade. Cell83 835–839.

Nikko E, Van de Vyver G & Richelle-Maurer E 2001 Retinoic aciddown-regulates the expression of the EmH-3 homeobox-containinggene in the freshwater sponge Ephydatia muelleri. Mechanisms inAgeing and Development 122 779–794.

Perlmann T, Umesono K, Rangarajan PN, Forman BM & EvansRM 1996 Two distinct dimerization interfaces differentiallymodulate target gene specificity of nuclear hormone receptors.Molecular Endocrinology 10 958–966.

Ross SA, McCaffery PJ, Drager UC & De Luca LM 2000 Retinoidsin embryonal development. Physiological Reviews 80 1021–1054.

Sambrook J & Russell DW 2001 Molecular Cloning: A LaboratoryManual, 3rd edn. Cold Spring Harbor, NY: Cold Spring HarborLaboratory Press.

Shea C, Hough D, Xiao J, Tzertzinis G & Maina CV 2004 Anrxr/usp homolog from the parasitic nematode, Dirofilaria immitis.Gene 324 171–182.

Sluder AE, Mathews SW, Hough D, Yin VP & Maina CV 1999The nuclear receptor superfamily has undergone extensiveproliferation and diversification in nematodes. Genome Research 9103–120.

Sluder AE & Maina CV 2001 Nuclear receptors in nematodes:themes and variations. Trends in Genetics 17 206–213.

Strimmer K & Von Hessler A 1996 Quartet puzzling: a quartetmaximum likelihood method for reconstructing tree topologies.Molecular Biology and Evolution 13 954–959.

Strimmer K & Rambaut A 2002 Inferring confidence sets of possiblymisspecified gene trees. Proceedings of the Royal Society of London. BBiological Sciences 269 137–142.

Thornton JW, Need E & Crews D 2003 Resurrecting the ancestralsteroid receptor: ancient origin of estrogen signaling. Science 3011714–1717.

Towbin H, Staehelin T & Gordon J 1979 Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets:procedure and some applications. PNAS 76 4350–4354.

Ulven SM, Gundersen TE, Sakhi AK, Glover JC & Blomhoff R2001 Quantitative axial profiles of retinoic acid in the embryonicmouse spinal cord: 9-cis retinoic acid only detected afterall-trans-retinoic acid levels are super-elevated experimentally.Developmental Dynamics 222 341–353.

Vaitukaitis J, Robbins JB, Nieschlag E & Ross GT 1971 A methodfor producing specific antisera with small doses of immunogen.Journal of Clinical Endocrinology and Metabolism 33 988–991.

Vanacker JM, Laudet V, Adelmant G, Stehelin D & Rommelaere J1993 Interconnection between thyroid hormone signaling

pathways and parvovirus cytotoxic functions. Journal of Virology 677668–7672.

Vivat V, Zechel C, Wurtz JM, Bourguet W, Kagechika H,Umemiya H, Shudo K, Moras D, Gronemeyer H & Chambon P1997 A mutation mimicking ligand-induced conformationalchange yields a constitutive RXR that senses allosteric effects inheterodimers. EMBO Journal 16 5697–5709.

Vivat-Hannah V, Bourguet W, Gottardis M & Gronemeyer H 2003Separation of retinoid X receptor homo- and heterodimerizationfunctions. Molecular and Cellular Biology 23 7678–7688.

Voegel JJ, Heine MJ, Zechel C, Chambon P & Gronemeyer H 1996TIF2, a 160 kDa transcriptional mediator for theligand-dependent activation function AF-2 of nuclear receptors.EMBO Journal 15 3667–3675.

vom Baur E, Zechel C, Heery D, Heine MJ, Garnier JM, Vivat V,Le Douarin B, Gronemeyer H, Chambon P & Losson R 1996Differential ligand-dependent interactions between the AF-2activating domain of nuclear receptors and the putativetranscriptional intermediary factors mSUG1 and TIF1. EMBOJournal 15 110–124.

Wiens M, Batel R, Korzhev M & Müller WEG 2003 Retinoid Xreceptor and retinoic acid response in the marine sponge Suberitesdomuncula. Journal of Experimental Biology 206 3261–3271.

Willy PJ, Umesono K, Ong ES, Evans RM, Heyman RA &Mangelsdorf DJ 1995 LXR, a nuclear receptor that defines adistinct retinoid response pathway. Genes and Development 91033–1045.

Willy PJ & Mangelsdorf DJ 1997 Unique requirements forretinoid-dependent transcriptional activation by the orphanreceptor LXR. Genes and Development 11 289–298.

Wolff KM & Scott AL 1995 Brugia malayi: retinoic acid uptake andlocalization. Experimental Parasitology 80 282–290.

Wurtz JM, Bourget W, Renaud JP, Vivat V, Chambon P, Moras D& Gronemeyer H 1996 A canonical structure for theligand-binding domain of nuclear receptors. Nature Structural Biology3 87–94.

Xu Y, Berelowitz M & Bruno JF 1998 Characterization of thepromoter region of the human somatostatin receptor subtype 2gene and localization of sequences required forestrogen-responsiveness. Molecular and Cellular Endocrinology 13971–77.

Yao TP, Segraves WA, Oro AE, McKeown M & Evans RM 1992Drosophila ultraspiracle modulates ecdysone receptor function viaheterodimer formation. Cell 71 63–72.

Zhang XK, Salbert G, Lee MO & Pfahl M 1994 Mutations thatalter ligand-induced switches and dimerization activities in theretinoid X receptor. Molecular and Cellular Biology 14 4311–4323.

Received 15 December 2004Accepted 5 January 2005Made available online as an Accepted Preprint 12 January 2005

D BOUTON and others · A conserved RXR from Biomphalaria glabrata582

www.endocrinology-journals.orgJournal of Molecular Endocrinology (2005) 34, 567–582