ORIGINAL ARTICLE

Nuclear hormone receptor signaling in amphioxus

Michael Schubert & Frédéric Brunet & Mathilde Paris &

Stéphanie Bertrand & Gérard Benoit & Vincent Laudet

Received: 11 July 2008 /Accepted: 5 September 2008 /Published online: 25 September 2008# Springer-Verlag 2008

Abstract The nuclear hormone receptors (NRs) form asuperfamily of transcription factors unified by conservedprotein structure and mode of function. While mostmembers of this superfamily are activated by ligands, suchas thyroid hormones, steroids, vitamin D or retinoic acid,other NRs are called orphan receptors because they have noknown ligand. NR-dependent signaling is crucial forvertebrate development with the majority of receptors beingexpressed in the developing embryo. Due to massive geneduplications during vertebrate diversification, there areusually more NRs in vertebrates than in invertebrates. Inthis study, we examine the evolutionary diversification ofthe NR superfamily and of NR-dependent signaling inchordates (vertebrates, tunicates, and amphioxus). We takeadvantage of the unique features of the genome of theinvertebrate amphioxus, which is characterized by avertebrate-like gene content without having undergone

massive duplications, to assess the NR signaling comple-ment (NRs and NR coregulators) of the ancestral chordate.We find 33 NRs in amphioxus, which are more NRs thanoriginally anticipated. This increase is mainly due to anamphioxus-specific duplication of genes encoding receptorsof the NR1H group. In addition, there are three heterolo-gous NRs in amphioxus that could not be placed within theframework of the NR superfamily. Apart from theseexceptions, there is usually one amphioxus NR or NRsignaling coregulator for each paralogous group of two,three, or four human receptors suggesting that the ancestralchordate had a set of 22 different NRs plus one copy ofeach NR coregulator.

Keywords Branchiostoma floridae .

Cephalochordate . Chordate . Development . Evolution .

Invertebrate-to-vertebrate transition .

Nuclear hormone receptor coregulator

Introduction

Nuclear hormone receptors (NRs) constitute a superfamilyof DNA-binding transcription factors that are involved in awide variety of physiological processes both duringdevelopment and in the adult (reviewed in Gronemeyer etal. 2004). NRs are best known for their functions as ligand-activated transcription factors that mediate the response tohormones, such as thyroid hormones, steroids, vitamin D orretinoic acid (reviewed in Gronemeyer et al. 2004). Inaddition to this role, there is also a non-negligible numberof NRs that either do not bind a ligand at all or for which aligand has not yet been described. These NRs withoutknown ligand are called orphan receptors (reviewed inBenoit et al. 2006; Germain et al. 2006c). Phylogenetic

Dev Genes Evol (2008) 218:651–665DOI 10.1007/s00427-008-0251-y

Michael Schubert and Frédéric Brunet are equal first authors.

Communicated by J.J. Gibson-Brown

Electronic supplementary material The online version of this article(doi:10.1007/s00427-008-0251-y) contains supplementary material,which is available to authorized users.

M. Schubert : F. Brunet :M. Paris :G. Benoit :V. Laudet (*)CNRS UMR5242-INRA 1288-ENS-UCBL,Institut de Génomique Fonctionnelle de Lyon, IFR128BioSciences Lyon-Gerland, Ecole Normale Supérieure de Lyon,46 allée d’Italie,69364 Lyon Cedex 07, Francee-mail: Vincent.Laudet@ens-lyon.fr

S. BertrandCNRS UMR7628, UPMC Univ Paris 06,Observatoire Océanographique,66651 Banyuls-sur-Mer, France

analyses have shown that NRs arose very early in themetazoan lineage long before the divergence of proto-stomes and deuterostomes and diversified through complexseries of gene duplications and gene losses (Escriva et al.1997; Bertrand et al. 2004; reviewed in Escriva et al. 2000).

NR proteins are characterized by a modular structureconsisting of six domains (reviewed in Germain et al.2006c). At the N-terminus, the A and B domains contain atranscriptional activation unit called activation function 1(AF-1). The C region of the NR is the DNA-bindingdomain (DBD), which is required for the interaction of theNR protein with the DNA. The DBD is the most highlyconserved domain of NRs and consists of cysteine-rich zincfinger motifs, α-helices, and a C-terminal extension. The Ddomain acts as a hinge between the C region (the DBD) andthe E region, which corresponds to the ligand-bindingdomain (LBD) and which includes a second activationdomain (AF-2) crucial for interactions of NRs withcoactivators and corepressors and often also a nuclearlocalization signal. Finally, the F region, which is notpresent in all NRs, is not very well characterized and has noclear known function.

Interactions of NRs with coactivators and corepressorsare required for defining the transcriptional activities ofNRs, which can function both as activators and repressorsof gene transcription depending on the target gene and onthe presence or absence of a ligand (reviewed in Bastienand Rochette-Egly 2004; Germain et al. 2006c). Althoughtranscriptional regulation by NRs is very complex anddiverse, in the absence of ligand, a liganded NR (such asthe retinoic acid receptor, RAR, or the thyroid hormonereceptor, TR) is bound to DNA predominantly as hetero-dimer with the NR retinoid X receptor (RXR) and isassociated with a corepressor complex that represses thetranscriptional activity of the NRs (reviewed in Hu andLazar 2000; Bastien and Rochette-Egly 2004; Goodson etal. 2005; Germain et al. 2006c). In some cases, NRs canalso bind to DNA as homodimers (e.g. the hepatocytenuclear factor 4, HNF4) or as monomers (e.g. the nervegrowth factor-induced clone—B, NGFI-B) (reviewed inBenoit et al. 2006; Germain et al. 2006c). Ligand bindinginduces a conformational change in the NR leading to thedissociation of the corepressors and to the binding of acoactivator complex (reviewed in Bastien and Rochette-Egly 2004; Germain et al. 2006c). The activity of thiscoactivator complex leads to unfolding of the condensedchromatin (reviewed in Bastien and Rochette-Egly 2004;Germain et al. 2006c; Lonard and O’Malley 2006). Afterthis relaxation of the chromatin structure, the coactivatorsare replaced by a mediator complex that recruits the basaltranscription machinery to the DNA locus for initiation oftranscription (reviewed in Bastien and Rochette-Egly 2004;Germain et al. 2006c).

Most NRs are expressed during vertebrate development(Bertrand et al. 2007) and have important functions in thedeveloping embryo (reviewed in Laudet and Gronemeyer2002). For example, NRs are involved in survival andproper development of the early embryo (PPAR, HNF4,ERR, NOR1, LRH1, GCNF), in patterning and formationof the central nervous system (TR, RAR, REV-ERB, ROR,TR2/4, TLX, PNR, COUP-TF, EAR2, ER, NURR1, NOR1,SF1), in organogenesis of, among others, the heart, lung,kidney, liver, pancreas, and the immune and reproductivesystems (DAX1, RAR, PPAR, ROR, HNF4, RXR, COUP-TF, GR, PR, AR, SF1), and in the development of thevisual system (RAR, ROR, TLX, PNR) (reviewed in Benoitet al. 2006; Dahlman-Wright et al. 2006; Flamant et al.2006; Germain et al. 2006a; Germain et al. 2006b; Lu et al.2006; Michalik et al. 2006; Moore et al. 2006). Moreover,in vertebrates (most prominently in amphibians and teleostfish) and in invertebrate chordates (such as amphioxus), thethyroid hormone receptor (TR) controls metamorphosis, thepostembryonic transition from larva to adult (Paris et al.2008a). Along the same lines, the functions of another NR,the retinoic acid receptor (RAR), in early development arealso conserved between invertebrate chordates and verte-brates. Thus, it has been shown that in amphioxus andvertebrates, RAR controls expression of Hox genes, whichare crucial mediators of anteroposterior axial patterning indevelopment (Schubert et al. 2006b; reviewed in Marlétazet al. 2006; Campo-Paysaa et al. 2008).

In humans, there are a total of 48 NRs. In contrast, theinvertebrates studied so far generally have fewer genesencoding NR proteins. For example, there are 21 NRs inthe genome of the fruit fly Drosophila melanogaster andonly 17 NRs in the genome of the sea squirt Cionaintestinalis (Yagi et al. 2003; Bertrand et al. 2004). Thisdifference in the total number of NRs between invertebratesand vertebrates can be attributed to massive whole genomeduplications that accompanied the diversification of thevertebrates (Holland et al. 1994; Dehal and Boore 2005;Holland et al. 2008; Putnam et al. 2008; reviewed inSchubert et al. 2006a). Phylogenetic analyses of thecomplete NR superfamily have shown that the vast majorityof NRs from metazoan animals fall into six distinctsubfamilies (numbered NR1 through NR6). In addition,there is a seventh “subfamily” (called NR0), which containsall receptors lacking the typical NR protein structure(Fig. 1) (Nuclear Receptors Nomenclature Committee1999; reviewed in Germain et al. 2006c). Within eachsubfamily, the NRs are further subdivided into severalgroups (Fig. 1) (reviewed in Germain et al. 2006c). Tobetter understand the evolutionary diversification of the NRsuperfamily and of NR-dependent signaling in chordates(i.e. in vertebrates, tunicates, and amphioxus), we decidedto characterize the complement of NRs and of key

652 Dev Genes Evol (2008) 218:651–665

components of NR signaling in the genome of thecephalochordate amphioxus (Branchiostoma floridae). Ofthe two invertebrate chordate groups (tunicates and amphi-oxus), we chose amphioxus as a model for our studies,because it is now widely regarded as the best availablestand-in for the chordate ancestor (Yu et al. 2007; Hollandet al. 2008; Putnam et al. 2008). Moreover, amphioxus ischaracterized by an overall body plan and a genome that arevertebrate-like, but simpler (reviewed in Schubert et al.2006a; Garcia-Fernàndez et al. 2007). For example, due towhole genome duplications in the vertebrate lineage, thereare often two, three, or four vertebrate genes, whereamphioxus only has a single ortholog (Holland et al.2008; Putnam et al. 2008). Moreover, unlike the genomesof representatives of the other invertebrate chordate group,the tunicates (e.g., the sea squirt Ciona intestinalis), theamphioxus genome has not experienced any major reorga-

nization by lineage-specific gene losses or duplications(Dehal et al. 2002; Holland et al. 2008; Putnam et al. 2008).

Taking advantage of these unique features of theamphioxus genome, we have identified the complete setof NRs in amphioxus. In addition, we have characterizedthe most important coregulators of the NR signalingmachinery in amphioxus. We found that amphioxus has atotal of 33 NRs, which is significantly more than expected(Holland et al. 2008). This high number of NRs inamphioxus is mainly due to a lineage-specific duplicationof NR1H receptors, which could be related to an elabora-tion in amphioxus of protection mechanisms against xeno-biotics and toxins. Apart from this amphioxus-specificduplication, there is generally one amphioxus receptor fortwo, three, or four human NRs and NR coregulators. Forexample, we identified one ortholog of the estrogenreceptor (ER) and one ortholog of the steroid receptor

NR2F (COUP-TF) *

NR2E (TLX, PNR, NR2E)

100

73

NR2D

NR2C (TR2/4) 84

49

100

NR2A (HNF4)

71

100

NR2B (RXR)

52

99

NR6A (GCNF)

NR5B (NR5B)

100

99

NR5A (NR5A)

46

97

94

93

NR3B (ERR)

NR3A (ER)

NR3C (SR)

58

47

100

85

59

96

NR4A (NR4A)

NR1E

NR1C (PPAR) 100

NR1F (ROR)

NR1D (REV-ERB)

100

54

66

18

NR1B (RAR)

NR1I / NR1J / NR1K

NR1H (NR1H-1 to -10)

98

100

NR1A (TR)

76

96

100

85

100

51

100

100

0.2

NR2

NR6

NR5

NR3

NR4

NR1

NR0NR0A / NR0B (NR0B)

95

Fig. 1 Diagrammatic phyloge-ny of the nuclear hormone re-ceptor (NR) superfamily. Theoverall topology and bootstrapsupport values are derived froma NR superfamily tree includingNRs from humans, sea squirts,fruit flies, and amphioxus. NR0has been added for illustrationpurposes. In red are NR groupsthat include amphioxus sequen-ces. The names of the respectiveamphioxus NRs are indicated inbrackets. In the NR1H group,there are ten amphioxus NRscalled NR1H-1, NR1H-2,NR1H-3, NR1H-4, NR1H-5,NR1H-6, NR1H-7, NR1H-8,NR1H-9, and NR1H-10. * indi-cates that a COUP-TF receptorhas previously been described inamphioxus that could not beidentified in the genome

Dev Genes Evol (2008) 218:651–665 653

(SR) in amphioxus, while there are two ERs and four SRsin humans. Thus, both ER and SR have already beenpresent in the last common ancestor of all chordates andhave secondarily been lost in sea squirts and othertunicates. In sum, the data on NRs and NR coregulatorsin the amphioxus genome suggest that the ancestralchordate had a set of 22 different NRs and a single copyof each NR coregulator.

Materials and methods

Data mining

The NR data set was based on sequence alignments ofprevious studies (Bertrand et al. 2004). The alignment filealready included all NRs from humans (Homo sapiens), seasquirts (Ciona intestinalis), and fruit flies (Drosophilamelanogaster). In contrast, the NR coregulator data setwas based on the Ensembl families of the NR coregulatorsof interest (NCOA, CBP and P300, PGC1, NCOR/SMRT,MED1/TRAP220), which are available at the Ensemblgenome browser (www.ensembl.org/index.html). We usedEnsembl version v48 for this analysis. All NR andcoregulator sequences were BLASTed against the amphi-oxus genome (version v1.0), which is available at the JGIwebpage (genome.jgi-psf.org/Brafl1/Brafl1.home.html).The best hits were retrieved and aligned using MUSCLE(Edgar 2004) followed by manual refinement. We createdboth nucleotide and amino acid alignments. All alignmentfiles are available from the authors. The nucleotide align-ments were used to identify and eliminate allelic poly-morphs of the different NR and NR coregulator sequencesretrieved from the amphioxus genome. Amphioxus allelicpolymorphs are characterized by nucleotide sequences thatdiffer almost exclusively in third codon positions, whichallows their identification and elimination (Holland et al.2008; Putnam et al. 2008).

Phylogenetic analyses

The amino acid alignments of amphioxus NR and NRcoregulator sequences were used for phylogenetic treereconstruction experiments. Phylogenetic analyses on thecomprehensive NR data set including all NRs fromhumans, sea squirts, fruit flies and amphioxus were carriedout using both the neighbor joining (NJ) and maximumlikelihood (ML) method. Likewise, the different NRcoregulator alignments were analyzed by phylogenetic treereconstruction using both the NJ and ML method. The NJanalyses, carried out with the program Phylo_Win (Galtieret al. 1996), were based on a Poisson correction andpairwise gap removal. The ML trees were calculated using

PhyML (Guindon and Gascuel 2003) with default param-eters (a JTT model with a gamma distribution parameterof 2). Bootstrap support values were calculated in 100bootstrap replicates. The NJ and ML analyses yieldedidentical results and, hence, only the ML trees arediscussed.

Results

In this study, we have annotated from the amphioxusgenome all NR genes and the genes of some of the maincoregulators of NR signaling. Gene annotation was basedon phylogenetic analyses of the identified sequences in thecontext of the NR superfamily or of the coregulator proteinfamilies (Supplementary Figs. S1–S6). The sequence IDs inthe amphioxus genome of the amphioxus NRs and of theNR coregulators are given in the supplementary material(Supplementary Tables S1, S2).

Amphioxus NRs

In total, there are 33 NR genes in amphioxus, 32 of whichwe were able to identify directly in the amphioxus genome(Fig. 1, Table 1). Although amphioxus COUP-TF hadpreviously been cloned from amphioxus (Langlois et al.2000), we were not able to identify this receptor in thegenome probably because it is located in an unassembledregion of the amphioxus genome (Putnam et al. 2008). Inthe following, we will discuss the different NR subfamiliesaccording to the overall structure and nomenclature of thesuperfamily (Fig. 1).

In the NR1 subfamily, there are a total of 15 amphioxusNRs (Table 1). As expected, there is a single amphioxusortholog in most NR1 groups. For example, there is a singleamphioxus thyroid hormone receptor (TR), retinoic acidreceptor (RAR), REV-ERB receptor, retinoid-related or-phan receptor (ROR) and peroxisome proliferator-activatedreceptor (PPAR; Fig. 2a). The amphioxus TR and RARhave previously been cloned and functionally characterized(Escriva et al. 2002; Paris et al. 2008a). In contrast, wewere not able to find any amphioxus representatives of theNR1I/J/K group, which includes, for example, the verte-brate vitamin D receptors (VDRs), pregnane X receptors(PXRs) and constitutive androstene receptors (CARs). Inthe NR1H group (Fig. 2b), which contains the vertebrateliver X receptors (LXRs) and farnesoid X receptors (FXRs),the amphioxus receptors have experienced a lineage-specific duplication resulting in ten NR1H-like receptors(Table 1). Since two of these independent amphioxusduplicates are very short, containing only the DBD(NR1H-9 and NR1H-10), they could not be included inour phylogenetic analyses of the whole NR superfamily

654 Dev Genes Evol (2008) 218:651–665

(Fig. 2b). We have, thus, verified that these two amphioxusreceptors (NR1H-9 and NR1H-10) are members of theNR1H group by constructing a phylogenetic tree basedonly on the DBD regions of the NR1A, NR1H, and NR1I/J/K sequences (data not shown). Interestingly, in theamphioxus genome NR1H-1, NR1H-2, NR1H-3, andNR1H-4 are located on the same scaffold (on scaffold153), as are NR1H-5 and NR1H-7 (on scaffold 291) and

NR1H-8 and NR1H-9 (on scaffold 91). This linkage patternstrongly suggests that the diversity of amphioxus NR1Hreceptors is the result of a series of lineage-specific tandemduplication events. Within the NR1 subfamily, the branchsupport for the association of the amphioxus receptors withtheir respective NR group is very strong except foramphioxus REV-ERB, where bootstrap support is only at73%. This bootstrap value increases to 82% in a tree based

Table 1 List of nuclear hormone receptors (NRs) in humans (Homo sapiens), sea squirts (Ciona intestinalis), fruit flies (Drosophilamelanogaster) and amphioxus (Branchiostoma floridae)

NR group NR name in humans NR name in sea squirts NR name in fruit flies NR name in amphioxus

NR0A KNI, KNRL, EGNR0B DAX1, SHP NR0BNR1A TRα, TRβ TR TRNR1B RARα, RARβ, RARγ RAR RARNR1C PPARα, PPARβ, PPARγ PPAR PPARNR1D REV-ERBα, REV-ERBβ REV-ERB E75 REV-ERBNR1E E78NR1F RORα, RORβ, RORγ ROR HR46 RORNR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-1NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-2NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-3NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-4NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-5NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-6NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-7NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-8NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-9a

NR1H LXRα, LXRβ, FXRα LXR, FXR ECR NR1H-10a

NR1I VDR, PXR, CARNR1J HR96NR1K NR1K-1, NR1K-2NR2A HNF4α, HNF4γ HNF4 HNF4 HNF4NR2B RXRα, RXRβ, RXRγ RXR USP RXRNR2C TR2, TR4 TR2/4 TR2/4NR2D HR78NR2E TLX TLL, DSF TLXNR2E PNR HR51 PNRNR2E NR2ENR2E FAX1NR2F COUP-TFα, COUP-TFβ, EAR2 COUP-TF SVP COUP-TFb

NR3A ERα, ERβ ERNR3B ERRα, ERRβ, ERRγ ERR ERR ERRNR3C GR, MR, AR, PR SRNR4A NGFI-B, NURR1, NOR1 NR4A HR38 NR4ANR5A SF1, LRH1 NR5A FTZ-F1 NR5ANR5B HR39 NR5BNR6A GCNF GCNF HR4 GCNFNRa NRaNRb NRbNRc NRca

Total 48 NRs in humans 17 NRs in sea squirts 21 NRs in fruit flies 33 NRs in amphioxus

a DNA-binding domain only.b COUP-TF has been cloned from amphioxus (Branchiostoma floridae), but could not be identified in the genome sequence.

Dev Genes Evol (2008) 218:651–665 655

NR1H

NR1I/J/K

HR38 fruit flyNR4A amphioxus

NR4A sea squirtNOR1 human

NURR1 human NGFI-B human

16

38

76

23

E78 fruit flyPPAR amphioxus

PPAR sea squirtPPARγ human

PPARβ human PPARα human 87

100 85

100

HR46 fruit flyROR amphioxus

ROR sea squirtRORβ human

RORγ human RORα human 49

97 63

34

REV-ERB sea squirtE75 fruit fly

REV-ERB amphioxus REV-ERBβ human

REV-ERBα human 100 73 57

100

54

66

18

RAR sea squirtRAR amphioxus

RARγ human RARβ human RARα human 81 98

63 100

85

100

51

100

100

NR1K-1 sea squirtNR1K-2 sea squirt

CAR human

HR96 fruit fly 71

VDR human

PXR human 100

17

82

ECR fruit flyFXR sea squirt

FXRα human 97 LXR sea squirtLXRβ human

LXRα human 99

41

98

98

100

TR amphioxus TR sea squirt

TRβ human TRα human 100

37

76

96

0.2NR4

NR1A

NR1H

NR1I/J/K

NR1B

NR1D

NR1F

NR1C

NR1E

NR1A

10097

5233

8697

10088

68

ECR fiddler crabECR mealworm

ECR hornwormECR fruit fly

NR1H-3 amphioxusNR1H-1 amphioxus

NR1H-2 amphioxusNR1H-4 amphioxus

NR1H-6 amphioxusNR1H-7 amphioxus

NR1H-5 amphioxusFXR sea squirt

FXRa humanNR1H-8 amphioxus

LXR sea squirtLXRβ human

LXRα human9894

47

50

100

40

NR1K-2 sea squirtNR1K-1 sea squirt

HR96 fruit flyVDR human

CAR humanPXR human98

85

61

50

TR amphioxusTR sea squirt

TRβ humanTRα human100

85

98

100

98 0.2

a

b

Fig. 2 Nuclear hormone receptor (NR) phylogeny of the NR1 andNR4 subfamilies. a The NR1 and NR4 branches of the overall NRsuperfamily tree based on NRs from humans, sea squirts, fruit flies,and amphioxus. The grey shading highlights the NR1 groups used forcalculating the phylogeny shown in (b). Human NRs are shown inblack, sea squirt NRs in green, fruit fly NRs in blue, and amphioxus

NRs in red. Bootstrap support values are given for each branch. bPhylogenetic analysis of the NR1A, NR1H, and NR1I/J/K groupshighlighting the lineage-specific duplication of NR1H receptors inamphioxus. Human NRs are shown in black, sea squirt NRs in green,fruit fly NRs in blue, and amphioxus NRs in red. Bootstrap supportvalues are given for each branch

656 Dev Genes Evol (2008) 218:651–665

only on sequences of the NR1C, NR1D, NR1E, and NR1Fgroups (data not shown). We are, hence, confident that theamphioxus genome contains a true REV-ERB ortholog.

We found a total of seven amphioxus receptors of theNR2 subfamily (Table 1). In the amphioxus genome, thereis a single gene encoding retinoid X receptor (RXR),hepatocyte nuclear factor 4 (HNF4), testicular receptor 2/4(TR2/4) and tailless (TLX) (Fig. 3). Moreover, there is asingle amphioxus photoreceptor-specific nuclear receptor(PNR) and an additional amphioxus receptor that isassociated with the NR2E group (Fig. 3). It is very likelythat this second amphioxus NR2E receptor is the result of alineage-specific duplication, although our phylogeny doesnot provide conclusive evidence for this hypothesis (Fig. 3).The amphioxus NR2 receptors strongly branch within their

respective NR groups, the weakest being amphioxus TLXwith a branch support of 92%. In addition to RXR and TR2/4,a chicken ovalbumin upstream promoter—transcriptionfactor (COUP-TF) ortholog has previously been describedin amphioxus (Langlois et al. 2000; Escriva et al. 2002).Thus, although we were not able to find amphioxus COUP-TF in the amphioxus genome sequence, this receptor doesexist in amphioxus.

In the NR3 subfamily, there is a single amphioxusreceptor for each group (Table 1). Thus, there is oneamphioxus steroid receptor (SR), one estrogen receptor(ER) and one estrogen receptor-related receptor (ERR), allof which have previously been described (Horard et al.2004; Bardet et al. 2005; Paris et al. 2008b). Although thebranch support for the association of amphioxus SR with

EAR2 human

SVP fruit fly

COUP-TF amphioxus *COUP-TF sea squirt

80

COUP-TFβ human COUP-TFα human

78

50

99

FAX1 fruit flyNR2E amphioxus

HR51 fruit flyPNR amphioxus

PNR human

81 99

TLL fruit flyDSF fruit fly

TLX amphioxus TLX human

68 50

57

92

82

100

73

HR78 fruit flyTR2/4 amphioxus

TR2/4 sea squirtTR4 human

TR2 human 92 49

84

49

100

HNF4 fruit flyHNF4 amphioxus

HNF4 sea squirt

71

HNF4γ human

HNF4α human

48

83

71

100

USP fruit flyRXR sea squirt

RXR amphioxus RXRβ human RXRγ human

RXRα human 47 67

56

86

52

99

HR4 fruit fly

GCNF amphioxus GCNF sea squirt

GCNF human 60 100

HR39 fruit flyNR5B amphioxus

100

99

NR5A sea squirt FTZ-F1 fruit fly

NR5A amphioxus LRH1 human

SF1 human

26

77

48

46

97

94

93

ERR amphioxus ERR sea squirt

ERR fruit fly

ERRa human 31

ERRγ human ERRβ human

55

61

42

ER amphioxus ERβ human

ERα human 99

SR amphioxus GR human

MR human PR human

AR human

53

81

100

58

47

100

85

59

96

0.2

NR2

NR6

NR5

NR3

Fig. 3 Nuclear hormone recep-tor (NR) phylogeny of the NR2,NR3, NR5, and NR6 subfami-lies. The tree shows the NR2,NR3, NR5, and NR6 branchesof the overall NR superfamilytree calculated with NRs fromhumans, sea squirts, fruit flies,and amphioxus. Human NRs areshown in black, sea squirt NRsin green, fruit fly NRs in blue,and amphioxus NRs in red.Bootstrap support values aregiven for each branch. Asteriskindicates that a COUP-TFreceptor has previously beendescribed in amphioxus thatcould not be identified in thegenome

Dev Genes Evol (2008) 218:651–665 657

the vertebrate SRs (i.e. with the glucocorticoid receptor,GR, the mineralocorticoid receptor, MR, the androgenreceptor, AR, and the progesterone receptor, PR) and ofamphioxus ER with vertebrate ERs is not very strong(bootstrap of 47% and 58%, respectively), the overallassociation of these two amphioxus receptors with the NR3subfamily is very well supported (bootstrap of 96%)(Fig. 3). In a phylogeny of the NR3 subfamily using theNR2B group (i.e. the RXRs) as outgroup, the bootstrapsupport for the branching of amphioxus SR with vertebrateSRs increases to 86%, while the grouping of amphioxus ERwith its vertebrate orthologs is still only weakly supported(bootstrap of 57%) (data not shown). Similar conclusionshave been reached in a recent detailed phylogeneticanalysis of the NR3 subfamily (Paris et al. 2008b). Tofurther characterize the three amphioxus NR3 receptors, wecompared the amino acid sites required for DNA and ligandbinding in human NR3 receptors with those of amphioxus(Table 2). Unfortunately, the amphioxus SR sequence weidentified in the genome is incomplete and lacks the C-terminus. We were, hence, unable to analyze three sites inthe C-terminal region of the LBD of amphioxus SR(Table 2). Within the DBD, the P-box of amphioxus ERis identical to that of human ERs. In contrast, the LBDsequences of amphioxus ER and the human ERs are lessconserved with one of the three amino acids involved inligand binding in human ERs being different in amphioxusER (Table 2). Our comparison also indicates that theamphioxus SR is more ER-like than SR-like, both as faras DNA and ligand binding is concerned (Table 2). Thus,the P-box and the D-box in the DBD of the amphioxus SRare more like human ERs and of the seven key amino acidpositions in the LBD, four are ER-like (T141, E147, L181,and I218) and none is SR-like (two amino acids, R188 andF199, are conserved with both human ERs and SRs andone, A138, is divergent in amphioxus SR) (Table 2).

In humans, the NR4 subfamily comprises three recep-tors: the nerve growth factor-induced clone—B (NGFI-B),the NUR-related factor 1 (NURR1) and the neuron-derivedorphan receptor 1 (NOR1). As expected, there is only asingle NR4 subfamily member in amphioxus, which iscalled NR4A (Table 1). This amphioxus receptor groupsvery strongly within this NR subfamily (100% bootstrapsupport) (Fig. 2a).

In our phylogenetic analysis, some NR5 subfamily mem-bers are weakly associated with the NR6 subfamily (Fig. 3).Thus, one of the two amphioxus NR5 subfamily members(NR5A) robustly groups (with a bootstrap support of97%) with the other members of the NR5A group, whichincludes the human steroidogenic factor 1 (SF1) and liverreceptor homolog 1 (LRH1), while the second amphioxusNR5 receptor (NR5B), together with the fruit fly hormonereceptor-like in 39 (HR39), is weakly associated with the T

able

2Aminoacidsim

portantforDNA

(DBD)andlig

and(LBD)bind

ingin

NR3subfam

ilyreceptors

Nuclear

receptor

DBD

LBD

P-box

D-box

Hum

anERα

CEGCKA

PATNQ

G(344

)T(247

)E(353

)L(387

)R

(394

)F(404

)I(424

)H

(524

)L(525

)M

(528

)Hum

anERβ

CEGCKA

PATNQ

M(296

)T(299

)E(305

)L(339

)R

(346

)F(356

)I(376

)H

(475

)L(476

)M

(479

)Amph

ioxu

sER

CEGCKA

PGTNQ

E(481

)S(484

)E(490

)L(524

)C(531

)L(542

)M

(570

)H

(659

)L(660

)V

(663

)Hum

anAR

CGSCKV

ASRND

S(702

)N

(705

)Q

(711)

S(745

)R

(752

)F(764

)Q

(783

)F(876

)T(877

)L(780

)Hum

anGR

CGSCKV

AGRND

T(561

)N

(564

)Q

(570

)M

(604

)R

(611)

F(623

)Q

(642

)Y

(735

)C(736

)T(739

)Hum

anMR

CGSCKV

AGRND

S(767

)N

(770

)Q

(776

)S(810

)R

(817

)F(829

)L(848

)F(941

)C

(942

)T(945

)Hum

anPR

CGSCKV

AGRND

S(716

)N

(719

)Q

(725

)S(759

)R

(766

)F(778

)L(797

)Y

(890

)C(891

)T(894

)Amph

ioxu

sSR

CEGCKS

PANNN

A(138

)T(141

)E(147

)L(181

)R(188

)F(199

)I(218

)–

––

Hum

anERRα

CEACKA

PASNE

A(323

)C(326

)E(332

)L(366

)R(373

)F(383

)L(402

)H

(495

)F(496

)C(499

)Hum

anERRβ

CEACKA

PASNE

T(241

)C(244

)E(250

)L(284

)R(291

)Y

(301

)L(320

)H

(408

)F(409

)C(412

)Hum

anERRγ

CEACKA

PASNE

T(243

)C(246

)E(252

)L(286

)R(293

)Y

(303

)L(322

)H

(410

)F(411)

I(414

)Amph

ioxu

sERR

CEACKA

PATNE

T(289

)C(292

)E(298

)L(332

)R(339

)F(349

)L(358

)H

(456

)F(457

)I(460

)

Aminoacidskn

ownto

interactwith

thelig

andin

human

ERα(Tanenbaum

etal.1

998),E

Rβ(Pikeetal.1

999),A

R(Pereira

deJésus-Tranetal.2

006),G

R(Bledsoe

etal.2

002),M

R(Bledsoe

etal.20

05)andPR(W

illiamsandSigler19

98)areshow

nin

bold.Aminoacid

positio

nsaregivenin

parentheses.

658 Dev Genes Evol (2008) 218:651–665

NR6 subfamily (Fig. 3). A phylogenetic analysis based onthe NR5 and NR6 subfamily sequences and using the NR2Bgroup (i.e. the RXRs) as outgroup recovers the integrity ofthe NR5 and NR6 subfamilies, with strong support for theNR6 subfamily (97% bootstrap support) and weak supportfor the NR5 subfamily (46% bootstrap support) (data notshown). It is interesting to note that there are no NR5Breceptors in humans and sea squirts (Table 1) suggesting thatmembers of the NR5B group might have secondarily beenlost in the lineages leading to extant sea squirts andvertebrates. In the NR6 subfamily that includes the verte-brate germ cell nuclear factor (GCNF), there is only a singleamphioxus receptor (Table 1), which branches very robustlywithin this subfamily (Fig. 3).

The NR superfamily also includes receptors that are notcharacterized by the typical NR protein structure. Thesereceptors are grouped together in the NR0 “subfamily”(Nuclear Receptors Nomenclature Committee 1999;reviewed in Germain et al. 2006c). In the amphioxusgenome, we found only one member of this NR0“subfamily”, a member of the NR0B group that is ortho-logous to human dosage-sensitive sex reversal–adrenalhypoplasia congenita critical region of the X chromosomeprotein 1 (DAX1) and small heterodimer partner (SHP)(Table 1). Since the amphioxus NR0B protein lacks aDBD, it could not be included in the overall NR super-family phylogeny, but an alignment of human DAX1 andSHP with the single amphioxus NR0 subfamily memberclearly identifies this amphioxus receptor as a member ofthe NR0B group (data not shown). Amphioxus NR0Bshares almost identical levels of sequence identity with thetwo human NR0B receptors DAX1 (38.4%) and SHP(37.9%).

Apart from the amphioxus NRs that we were able tocategorize into existing NR subfamilies, we also foundthree divergent amphioxus NRs (called NRa, NRb, andNRc) that we failed to place within the existing frameworkof the NR superfamily (Fig. 1, Table 1). Thus, while NRa ischaracterized by the typical organization of a NR receptor,NRb has a conserved DBD, but very divergent C-terminalLBD region, and NRc is very short consisting only of aputative DBD region. Since we were unable to reliablyplace the three divergent amphioxus NRs in the context of a

phylogenetic tree of the NR superfamily, we carried outBLAST analyses of the protein sequences to obtain an ideaabout the sequence similarities of amphioxus NRa, NRb,and NRc with representatives of the different NR sub-families. The BLAST results indicate that NRa can looselybe associated with the NR2 or NR3 subfamilies, NRb withthe NR5 or NR6 subfamilies and NRc with the NR1subfamily (data not shown). Interestingly, an analysis of thethree divergent amphioxus receptors in the context of thesea urchin NR complement (Howard-Ashby et al. 2006;Sea Urchin Genome Sequencing Consortium 2006)revealed that amphioxus NRa robustly branches with thesea urchin receptor Sp-nr2C (data not shown). This resultsuggests that with the availability of more genomic data(both from amphioxus as well as from other species), thesethree amphioxus receptors can hopefully be further charac-terized and categorized in the near future.

Amphioxus coregulators

Coregulators are very important components of NRsignaling. Thus, a search of NR coregulators in theamphioxus genome allows important insights into theconservation (but also into possible differences) of the NRsignaling machinery in amphioxus and vertebrates. Wehave focused on three classes of NR signaling coregulators:coactivators, corepressors, and components of the mediatorcomplex (Table 3). We identified amphioxus orthologs ofthe classical vertebrate nuclear receptor coactivator(NCOA), of CREB-binding protein (CBP) and histoneacetyltransferase p300 (P300), both of which are membersof the same coregulator family, as well as of peroxisomeproliferator-activated receptor gamma coactivator 1(PGC1). For each of these coactivator families, we onlyfound a single amphioxus ortholog (Table 3): amphioxusNCOA (Supplementary Fig. S2), amphioxus CBP/P300(Supplementary Fig. S3) and amphioxus PGC1 (Supple-mentary Fig. S4). For corepressors, we searched theamphioxus genome for orthologs of the classical vertebratenuclear receptor corepressor/silencing mediator for retinoidand thyroid hormone receptors (NCOR/SMRT) and identi-fied only a single amphioxus NCOR (Table 3; Supplemen-tary Fig. S5). Finally, as component of the mediator

Table 3 List of nuclear hormone receptor coregulators (NRCs) in humans (Homo sapiens) and amphioxus (Branchiostoma floridae)

Coregulator class Coregulator name in humans Coregulator name in amphioxus

Coactivators NCOA1, NCOA2, NCOA3 NCOACBP, P300 CBP/P300PRGC1/PGC1α, PRGC2/PGC1β, PPRC1/PGC1-related PGC1

Corepressors NCOR1, NCOR2/SMRT NCORMediator complex MED1/TRAP220 TRAP220

Dev Genes Evol (2008) 218:651–665 659

complex, we looked for orthologs of the vertebratemediator complex subunit 1/thyroid hormone receptor-associated protein complex 220 kDa component (MED1/TRAP220). Again, we identified only a single ortholog inthe amphioxus genome, which we called amphioxusTRAP220 (Table 3; Supplementary Fig. S6).

Discussion

In this study, we identified all NRs and some keycomponents of the NR signaling machinery in the inverte-brate chordate amphioxus, the best available stand-in forthe last common ancestor of all chordates. In comparison to48 NRs in humans, 17 in sea squirts and 21 in fruit flies,there are 33 NRs in amphioxus (32 NRs were identified inthe amphioxus genome and COUP-TF has previously beencloned from amphioxus), which is more than initiallyexpected. For NR signaling coregulators, we found a singleamphioxus ortholog for each coregulator family analyzed.By discussing this amphioxus NR and NR coregulator dataset we will now infer the NR signaling complement of theancestral chordate.

Amphioxus as a simple model to study NR signalingevolution in chordates

There are a large number of NR groups, with a singleamphioxus NR and two, three, or four human receptors. Forthese NR groups, amphioxus is a very good system to studythe evolution of NR function in chordates, which hasalready been shown for NRs belonging to three differentsubfamilies (NR1, NR2, and NR3). In the NR1 subfamily,the thyroid hormone receptor, TR, and the retinoic acidreceptor, RAR, have already been studied in amphioxus. Asin some vertebrates, most prominently in amphibians, TRcontrols metamorphosis in amphioxus (Paris et al. 2008a).Whether, like in vertebrates, the amphioxus TR also hasother roles in late development, for example in the anteriorcentral nervous system or in the intestine (reviewed inFlamant et al. 2006), remains to be established. AmphioxusRAR is conspicuously expressed in hindbrain and spinalcord of developing amphioxus as well as in the middle thirdof the embryo (Escriva et al. 2002), which has beenhypothesized to closely resemble the expression pattern ofRAR in the last common ancestor of all chordates (Escrivaet al. 2006). Moreover, it has been shown that RAR iscrucial for anteroposterior regionalization of the developingamphioxus embryo, where it acts upstream of Hox genes inthe patterning of the amphioxus endoderm, central nervoussystem and general ectoderm (reviewed in Marlétaz et al.2006; Campo-Paysaa et al. 2008). This role of RAR inanteroposterior patterning of the embryo seems to be largely

conserved between amphioxus and vertebrates (reviewed inMarlétaz et al. 2006; Campo-Paysaa et al. 2008).

In the NR2 subfamily, the single amphioxus orthologs ofthe human retinoid X receptor (RXR), testicular receptor2/4 (TR2/4) and COUP-TF have already been studied(Langlois et al. 2000; Escriva et al. 2002). In amphioxus,RXR is very broadly expressed in the developing embryo(Escriva et al. 2002), which is not surprising, because it hasimportant functions as heterodimeric binding partner ofother NRs in amphioxus and vertebrates (Escriva et al.2002) and ubiquitous availability of this NR is, thus, acrucial component of NR-dependent signaling (reviewed inGermain et al. 2006b). It will be very interesting to furtherstudy the functions of this single RXR in amphioxus, bothas a proper transcription factor and as heterodimericbinding partner of other NRs. In vertebrates, as inamphioxus, RXRs have broad or even ubiquitous expres-sion patterns. Moreover, loss-of-function experiments inmice have shown that RXR function is required fordevelopment of the vertebrate heart and eyes (reviewed inGermain et al. 2006b). In contrast to RXR, amphioxusTR2/4 expression is detectable only in the anterioramphioxus central nervous system and in the anterior andposterior thirds of the endoderm (Escriva et al. 2002). Invertebrates, TR2/4 is expressed in the developing kidney,gut endoderm and in neuronal precursors (reviewed inBenoit et al. 2006) suggesting that at least some of thefunctions of TR2/4 in endoderm and central nervous systemmight be conserved between amphioxus and vertebrates.

Although we were not able to identify a COUP-TFortholog in the amphioxus genome, this receptor haspreviously been cloned from amphioxus (Langlois et al.2000) suggesting that the COUP-TF locus is located in aregion of the amphioxus genome that contains gaps and,hence, could not be properly assembled (Putnam et al.2008). Developmental expression of amphioxus COUP-TFhas been assessed. In developing amphioxus, COUP-TFexpression is limited to hindbrain and anterior spinal cordof late larvae (Langlois et al. 2000). In contrast to thisrelatively restricted expression in amphioxus, vertebrateCOUP-TFs are expressed in various different tissues duringdevelopment and are required for central nervous systempatterning as well as for the development of the heart and ofthe circulatory system (reviewed in Benoit et al. 2006).Thus, it is possible that at least some of the functions ofCOUP-TF in the developing central nervous system mighthave evolved in the last common ancestor of amphioxusand vertebrates (Langlois et al. 2000).

The NR3 subfamily includes the estrogen receptors (ERs),steroid receptors (SRs) and the estrogen receptor-relatedreceptors (ERRs). All three amphioxus representatives of thissubfamily (amphioxus ER, SR, and ERR) have already beenstudied (Horard et al. 2004; Bardet et al. 2005; Paris et al.

660 Dev Genes Evol (2008) 218:651–665

2008b). For example, it has been shown that the amphioxusER is unable to bind estradiol, the classical ligand ofvertebrate ERs, as well as a variety of other known naturaland synthetic ligands of vertebrate ERs (Paris et al. 2008b).In vertebrates, ERs are crucial for the development of thereproductive organs and of the brain (reviewed in Dahlman-Wright et al. 2006). In contrast, ERRs in vertebrates areimportant for early development, which has been shown inboth mice and zebrafish (Bardet et al. 2005; reviewed inBenoit et al. 2006). In zebrafish, for example, ERRs controlthe morphogenetic movements during gastrulation (Bardet etal. 2005). Interestingly, in both zebrafish and amphioxus,ERRs are expressed in a segmented manner in hindbrain andmuscular somites suggesting that ERRs might have alreadybeen required for the development of these two tissue layersin the last common ancestor of all chordates (Bardet et al.2005; Bertrand et al. 2007).

For the NR coregulator families we studied, there isalways a single amphioxus ortholog for each paralogousgroup of two, three, or four human coregulator familymembers. Hence, we identified the unique representativesof key components of the NR signaling machinery in theamphioxus genome including classical NR coactivators(nuclear receptor coactivator, NCOA, CREB-binding pro-tein, CBP, and histone acetyltransferase p300, P300, as wellas peroxisome proliferator-activated receptor gamma coac-tivator 1, PGC1), classical NR corepressors (nuclearreceptor corepressor/silencing mediator for retinoid andthyroid hormone receptors, NCOR/SMRT) and an impor-tant component of the NR mediator complex (mediatorcomplex subunit 1/thyroid hormone receptor-associatedprotein complex 220-kDa component, MED1/TRAP220)(reviewed in Hu and Lazar 2000; Bastien and Rochette-Egly 2004; Goodson et al. 2005; Germain et al. 2006c;Lonard and O’Malley 2006). For each of these coregulatorfamilies analyzed, there are multiple representatives invertebrates (reviewed in Hu and Lazar 2000; Goodson et al.2005; Germain et al. 2006c; Lonard and O’Malley 2006).As mentioned above, in amphioxus we only isolated asingle ortholog in each of these coregulator families: thereis only one coactivator of the NCOA, CBP/P300, andPGC1 families (amphioxus NCOA, CBP/P300 and PGC1,respectively), one corepressor of the NCOR/SMRT family(amphioxus NCOR) and one MED1/TRAP220 componentof the mediator complex (amphioxus TRAP220). Thiscoregulator complement in amphioxus represents a verte-brate-like set with the important difference that theamphioxus NR coactivators, corepressors, and mediatorcomplex components have not been duplicated. It is, hence,very likely that the ancestral chordate had a NR coregulatorcomplement that very closely resembled that of amphioxus,and that in the vertebrate lineage this basic coregulatorset has been expanded by the massive whole genome

duplications that characterized the evolution of vertebrates(Holland et al. 1994; Holland et al. 2008; Putnam et al.2008; reviewed in Schubert et al. 2006a). These data alsoshow that the amphioxus NR signaling machinery isvertebrate-like, but simpler. This relative simplicity inamphioxus highlights the usefulness of amphioxus as amodel to study the evolution and elaboration of NRsignaling in chordates. For example, experimental targetingof a single one of these amphioxus factors will provideinformation that, in vertebrates, can only be obtained bytargeting two, three, or even four paralogous genes.

Gene duplication and loss accompanied the diversificationof chordate lineages

In the amphioxus genome, we found 33 NRs, which is morethan we initially expected. This elevated number of NRs ismainly due to a duplication in amphioxus of the receptors ofthe NR1H group. In this NR group, instead of single liver Xreceptor (LXR) and farnesoid X receptor (FXR) orthologs, asis the case in sea squirts (Dehal et al. 2002; Bertrand et al.2004), there are ten amphioxus members. These NR1Hduplicates most likely arose from an amphioxus-specificduplication. In vertebrates, NR1H receptors control lipid andcholesterol metabolism and, together with NR1I/J/K recep-tors (the vitamin D receptor, VDR, the pregnane X receptor,PXR, and the constitutive androstene receptor, CAR), areinvolved in the metabolism of toxins and xenobiotics(reviewed in Moore et al. 2006). The duplication of membersof the NR1H group in amphioxus suggests that the signalingnetworks controlled by these receptors might have also beenelaborated in amphioxus. It is, hence, conceivable that thiselaboration lead to the evolution of novel mechanisms fordegradation of and protection from various kinds of toxinsand xenobiotics derived from marine microorganisms andthe environment surrounding the sedentary, filter-feedingamphioxus adult. In this context, it is also very interesting tonote that there is no NR1I/J/K member in amphioxus, whilerepresentatives of this group exist in humans (VDR, PXR,and CAR), sea squirts (NR1K-1 and NR1K-2), and evenfruit flies (the hormone receptor-like in 96, HR96). Thus, theamphioxus ortholog of this group is either located in a regionof the genome that contains gaps and could not be assembledor has been lost in the amphioxus lineage. In the case of alineage-specific loss, since members of the NR1I/J/K groupare very closely related to the NR1H receptors that haveextensively been duplicated in amphioxus, it is possible thatin amphioxus one or several of the NR1H duplicates haveassumed the functions of the lost NR1I/J/K receptor.

Another example of an amphioxus-specific duplicationcan be found in the NR2E group. In vertebrates, the twomembers of this NR group, tailless (TLX) and thephotoreceptor-specific nuclear receptor (PNR), have impor-

Dev Genes Evol (2008) 218:651–665 661

tant roles in brain and retina development (reviewed inBenoit et al. 2006). In amphioxus, in addition to TLX andPNR, there is an additional NR2E group member. From thephylogenetic analysis, it can be assumed that this receptor,called NR2E, is the result of a lineage-specific duplicationevent in amphioxus. Since both TLX and PNR functionduring development of the visual system in vertebrates, it istempting to suggest that the additional NR2E receptor inamphioxus functions alongside TLX and PNR in therelatively complex visual system of amphioxus with itsfrontal eye, pineal eye, and multiple photoreceptors(Glardon et al. 1998; reviewed in Lacalli 2001). Finally, itis very interesting to note that there is no NR2E receptor inthe sea squirt genome (Dehal et al. 2002; Bertrand et al.2004) suggesting that both TLX and PNR have been lost inthe tunicate lineage, which could be related to the tendencyin tunicates towards modification and reduction of thevisual system (reviewed in Lacalli 2001). Functionalanalyses of the three amphioxus NR2E receptors willprovide very interesting insights into the evolution of novel(neofunctionalization) and shared (subfunctionalization)roles for gene copies after lineage-specific duplication.

In the recent past, members of the NR3 subfamily (ERs,SRs, and ERRs) have been used repeatedly as molecularmodels to understand the evolution of ligand-bindingcapacities within the NR superfamily (Escriva et al. 2000;Baker 2003; Thornton et al. 2003; Bridgham et al. 2006). Inhumans, there are two ERs, four SRs (the glucocorticoidreceptor, GR, the mineralocorticoid receptor, MR, theandrogen receptor, AR, the progesterone receptor, PR),and three ERRs. Interestingly, of the three NR3 groups,only ERRs have been identified in the genomes of seasquirts and fruit flies (Bertrand et al. 2004). In amphioxus,there seems to be a single ortholog in each of the three NR3groups, although the branch support for the grouping of theamphioxus ER and SR with their respective group is notvery good. In contrast, the amphioxus ERR groups veryrobustly with the ERRs from humans, sea squirts, and fruitflies. Sequence comparisons have shown that both the LBDand the DBD of the amphioxus ER and SR are more similarto vertebrate ERs than to vertebrate SRs (Holland et al.2008; Paris et al. 2008b). In addition, the amphioxus ER isunable to bind estradiol, the classical ligand of vertebrateERs, as well as a variety of other known natural andsynthetic ligands of vertebrate ERs (Paris et al. 2008b). Inits lack of ability to bind estradiol, the amphioxus ER is,hence, more similar to ERs of other invertebrates, such asmollusks (Thornton et al. 2003; Paris et al. 2008b). Sincethe amphioxus SR has not yet been functionally described,it remains to be established, whether the amphioxus SR iscapable to bind steroids (for example estrogens) and toactivate transcription in a ligand-dependent manner. Invertebrates, such as mice, ERs and SRs are crucial for the

development of the reproductive organs (reviewed inDahlman-Wright et al. 2006; Lu et al. 2006). Moreover,ERs are also required, for example, for neuronal migrationin the embryonic mouse brain and SRs, such as GR, areimportant for lung and liver formation in mice (reviewed inDahlman-Wright et al. 2006; Lu et al. 2006). To betterunderstand the evolution of ER and SR function inchordates, it will be necessary to carefully assess thefunctions of ER and SR in developing and adult amphiox-us, in particular because ERs and SRs are absent from thesea squirt genome (Dehal et al. 2002; Bertrand et al. 2004)suggesting that these NR3 receptors have secondarily beenlost in the tunicate lineage.

Within the NR5 subfamily, there are two humanreceptors, two from amphioxus and fruit flies and one fromsea squirts (Bertrand et al. 2004). The two human members,steroidogenic factor 1 (SF1) and the liver receptor homolog1 (LRH1), both belong to the NR5A group, which alsoincludes the single sea squirt NR5, fruit fly fushi tarazu-transcription factor 1 (FTZ-F1) and amphioxus NR5A.Interestingly, the second NR5 subfamily member inamphioxus clusters very strongly with the fruit fly hormonereceptor-like in 39 (HR39) suggesting that this amphioxusreceptor is a member of the NR5B group. We, hence, calledthis amphioxus receptor NR5B. In fruit flies, HR39 isrequired for proper female reproductive tract developmentand function (Allen and Spradling 2008) and, during post-embryonic development, probably also acts early in theecdysone cascade that triggers metamorphosis (Bonneton etal. 2008; reviewed in King-Jones and Thummel 2005). It isintriguing to speculate that the amphioxus ortholog of fruitfly HR39, NR5B, might also have roles in reproduction andmetamorphosis of amphioxus, which would establish amolecular link of these two processes between arthropodsand chordates. In contrast to amphioxus, there are no NR5Breceptors in humans and sea squirts. It is, thus, likely that aNR5B receptor was present in the last common ancestor ofall chordates and that the members of this receptor grouphave secondarily been lost in the lineage leading to extanttunicates and vertebrates (Bertrand et al. 2004).

In the NR0 “subfamily”, there are two representatives inhumans, one in amphioxus and none in sea squirts. WhileNR0 receptors have apparently been lost in sea squirts, boththe two human NR0 receptors and the single amphioxus NR0are members of the NR0B group (Dehal et al. 2002; Bertrandet al. 2004). The two human NR0 receptors, dosage-sensitivesex reversal–adrenal hypoplasia congenita critical region ofthe X chromosome protein 1 (DAX1) and small heterodimerpartner (SHP), lack the C region (the DBD) and are, thus,unable to bind to classical NR DNA response elements(reviewed in Benoit et al. 2006). Based on sequencealignments and on the overall length of the sequence, it isvery likely that the amphioxus NR0B is also unable to bind

662 Dev Genes Evol (2008) 218:651–665

DNA, although DNA-binding experiments with the amphi-oxus NR0B will be necessary to validate this hypothesis.

In addition to the amphioxus NRs that group withinexisting NR subfamilies, we also identified three divergentNRs in the amphioxus genome (called NRa, NRb, andNRc). These three receptors did not branch reliably withany NR subfamily. It is possible that these three receptorsinitially originated from an amphioxus-specific duplication.After duplication, the coding sequences of the duplicatesdiverged possibly resulting in neofunctionalization orsubfunctionalization. Alternatively, these three NRs couldalso be remnants of ancient NR subfamilies, whoserepresentatives have secondarily been lost in most species,for which the NR repertoire has already been assessed. Thefact that one of the three divergent amphioxus NRs (NRa)robustly groups with a sea urchin receptor (Sp-nr2C) lendssupport to the latter hypothesis and suggests that with moregenomic data and whole genome sequences, it might becomefeasible to classify these three divergent amphioxus NRs.

Conclusions

In this study, we identified the NR and NR coregulatorcomplement of the invertebrate chordate amphioxus, whichis located at the base of the chordates and as such is the bestavailable stand-in for the last common ancestor of allchordates. Our data suggest that the chordate ancestor had22 NRs, more specifically one NR0 (NR0B), eight NR1s(TR, RAR, PPAR, REV-ERB, ROR, LXR, FXR, NR1I),six NR2s (HNF4, RXR, TR2/4, TLX, PNR, COUP-TF),three NR3s (ER, ERR, SR), one NR4 (NR4A), two NR5s(NR5A, NR5B), and one NR6 (GCNF). In addition, theancestral chordate had a single copy of the NR coactivatorsNCOA, CBP/P300, and PGC1, of the NR corepressorNCOR, and of the NR mediator complex componentTRAP220. Thus, the last common ancestor of chordateswas already characterized by a very elaborate set of NRsand NR coregulators that have independently been modi-fied and expanded both in amphioxus and vertebrates. Insum, this work highlights the utility of the amphioxusgenome for understanding the evolution of chordates andalso shows that the amphioxus genome has by no meansbeen fixed in an ancestral state: this genome has undergonevarious lineage-specific gene duplications and losses afterthe split from the lineage leading to extant tunicates andvertebrates some 550 million years ago.

Acknowledgments The authors would like to thank Michael E.Baker for critical reading of the manuscript. This work was supportedby grants from the ANR and the CNRS to M.S., the MENRT to M.S.and V.L. and by CRESCENDO, a European Union Integrated Projectof FP6, and by CASCADE, a Network of Excellence of FP6. S.B. isfunded by an ARC postdoctoral fellowship.

References

Allen AK, Spradling AC (2008) The Sf1-related nuclear hormonereceptor Hr39 regulates Drosophila female reproductive tractdevelopment and function. Development 135:311–321

Baker ME (2003) Evolution of adrenal and sex steroid action invertebrates: a ligand-based mechanism for complexity. Bioessays25:396–400

Bardet PL, Schubert M, Horard B, Holland LZ, Laudet V, HollandND, Vanacker JM (2005) Expression of estrogen-receptor relatedreceptors in amphioxus and zebrafish: implications for theevolution of posterior brain segmentation at the invertebrate-to-vertebrate transition. Evol Dev 7:223–233

Bastien J, Rochette-Egly C (2004) Nuclear retinoid receptors and thetranscription of retinoid-target genes. Gene 328:1–16

Benoit G, Cooney A, Giguere V, Ingraham H, Lazar M, Muscat G,Perlmann T, Renaud JP, Schwabe J, Sladek F, Tsai MJ, Laudet V(2006) International Union of Pharmacology. LXVI. Orphannuclear receptors. Pharmacol Rev 58:798–836

Bertrand S, Brunet FG, Escriva H, Parmentier G, Laudet V, Robinson-Rechavi M (2004) Evolutionary genomics of nuclear receptors:from twenty-five ancestral genes to derived endocrine systems.Mol Biol Evol 21:1923–1937

Bertrand S, Thisse B, Tavares R, Sachs L, Chaumot A, Bardet PL,Escriva H, Duffraisse M, Marchand O, Safi R, Thisse C, LaudetV (2007) Unexpected novel relational links uncovered byextensive developmental profiling of nuclear receptor expression.PLoS Genet 3:e188

Bledsoe RK, Montana VG, Stanley TB, Delves CJ, Apolito CJ,McKee DD, Consler TG, Parks DJ, Stewart EL, Willson TM,Lambert MH, Moore JT, Pearce KH, Xu HE (2002) Crystalstructure of the glucocorticoid receptor ligand binding domainreveals a novel mode of receptor dimerization and coactivatorrecognition. Cell 110:93–105

Bledsoe RK, Madauss KP, Holt JA, Apolito CJ, Lambert MH, PearceKH, Stanley TB, Stewart EL, Trump RP, Willson TM, WilliamsSP (2005) A ligand-mediated hydrogen bond network requiredfor the activation of the mineralocorticoid receptor. J Biol Chem280:31283–31293

Bonneton F, Chaumot A, Laudet V (2008) Annotation of Triboliumnuclear receptors reveals an increase in evolutionary rate of anetwork controlling the ecdysone cascade. Insect Biochem MolBiol 38:416–429

Bridgham JT, Carroll SM, Thornton JW (2006) Evolution ofhormone-receptor complexity by molecular exploitation. Science312:97–101

Campo-Paysaa F, Marlétaz F, Laudet V, Schubert M (2008) Retinoicacid signaling in development: tissue-specific functions andevolutionary origins. Genesis, in press

Dahlman-Wright K, Cavailles V, Fuqua SA, Jordan VC, KatzenellenbogenJA, Korach KS, Maggi A, Muramatsu M, Parker MG, GustafssonJÅ (2006) International Union of Pharmacology. LXIV. Estrogenreceptors. Pharmacol Rev 58:773–781

Dehal P, Satou Y, Campbell RK, Chapman J, Degnan B, De TomasoA, Davidson B, Di Gregorio A, Gelpke M, Goodstein DM,Harafuji N, Hastings KE, Ho I, Hotta K, Huang W, KawashimaT, Lemaire P, Martinez D, Meinertzhagen IA, Necula S, NonakaM, Putnam N, Rash S, Saiga H, Satake M, Terry A, Yamada L,Wang HG, Awazu S, Azumi K, Boore J, Branno M, Chin-Bow S,DeSantis R, Doyle S, Francino P, Keys DN, Haga S, Hayashi H,Hino K, Imai KS, Inaba K, Kano S, Kobayashi K, Kobayashi M,Lee BI, Makabe KW, Manohar C, Matassi G, Medina M,Mochizuki Y, Mount S, Morishita T, Miura S, Nakayama A,Nishizaka S, Nomoto H, Ohta F, Oishi K, Rigoutsos I, Sano M,Sasaki A, Sasakura Y, Shoguchi E, Shin-i T, Spagnuolo A,

Dev Genes Evol (2008) 218:651–665 663

Stainier D, Suzuki MM, Tassy O, Takatori N, Tokuoka M, YagiK, Yoshizaki F, Wada S, Zhang C, Hyatt PD, Larimer F, DetterC, Doggett N, Glavina T, Hawkins T, Richardson P, Lucas S,Kohara Y, Levine M, Satoh N, Rokhsar DS (2002) The draftgenome of Ciona intestinalis: insights into chordate andvertebrate origins. Science 298:2157–2167

Dehal P, Boore JL (2005) Two rounds of whole genome duplication inthe ancestral vertebrate. PLoS Biol 3:e314

Edgar RC (2004) MUSCLE: multiple sequence alignment with highaccuracy and high throughput. Nucleic Acids Res 32:1792–1797

Escriva H, Safi R, Hänni C, Langlois MC, Saumitou-Laprade P,Stehelin D, Capron A, Pierce R, Laudet V (1997) Ligand bindingwas acquired during evolution of nuclear receptors. Proc NatlAcad Sci USA 94:6803–6808

Escriva H, Delaunay F, Laudet V (2000) Ligand binding and nuclearreceptor evolution. Bioessays 22:717–727

Escriva H, Holland ND, Gronemeyer H, Laudet V, Holland LZ (2002)The retinoic acid signaling pathway regulates anterior/posteriorpatterning in the nerve cord and pharynx of amphioxus, achordate lacking neural crest. Development 129:2905–2916

Escriva H, Bertrand S, Germain P, Robinson-Rechavi M, UmbhauerM, Cartry J, Duffraisse M, Holland LZ, Gronemeyer H, Laudet V(2006) Neofunctionalization in vertebrates: the example ofretinoic acid receptors. PLoS Genet 2:e102

Flamant F, Baxter JD, Forrest D, Refetoff S, Samuels H, Scanlan TS,Vennström B, Samarut J (2006) International Union of Pharma-cology. LIX. The pharmacology and classification of the nuclearreceptor superfamily: thyroid hormone receptors. Pharmacol Rev58:705–711

Galtier N, Gouy M, Gautier C (1996) SEAVIEW and PHYLO_WIN:two graphic tools for sequence alignment and molecularphylogeny. Comput Appl Biosci 12:543–548

Garcia-Fernàndez J, D’Aniello S, Escriva H (2007) Organizingchordates with an organizer. Bioessays 29:619–624

Germain P, Chambon P, Eichele G, Evans GM, Lazar MA, Leid M, deLera AR, Lotan R, Mangelsdorf DJ, Gronemeyer H (2006a)International Union of Pharmacology. LX. Retinoic acid recep-tors. Pharmacol Rev 58:712–725

Germain P, Chambon P, Eichele G, Evans GM, Lazar MA, Leid M, deLera AR, Lotan R, Mangelsdorf DJ, Gronemeyer H (2006b)International Union of Pharmacology. LXIII. Retinoid X recep-tors. Pharmacol Rev 58:760–772

Germain P, Staels B, Dacquet C, Spedding M, Laudet V (2006c)Overview of nomenclature of nuclear receptors. Pharmacol Rev58:685–704

Glardon S, Holland LZ, Gehring WJ, Holland ND (1998) Isolationand developmental expression of the amphioxus Pax-6 gene(AmphiPax-6): insights into eye and photoreceptor evolution.Development 125:2701–2710

Goodson M, Jonas BA, Privalsky MA (2005) Corepressors: customtailoring and alterations while you wait. Nucl Recept Signal 3:e003

Gronemeyer H, Gustafsson JÅ, Laudet V (2004) Principles formodulation of the nuclear receptor superfamily. Nat Rev DrugDiscov 3:950–964

Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithmto estimate large phylogenies by maximum likelihood. Syst Biol52:696–704

Holland LZ, Albalat R, Azumi K, Benito-Gutiérrez E, Blow MJ,Bronner-Fraser M, Brunet F, Butts T, Candiani S, Dishaw LJ,Ferrier DEK, Garcia-Fernàndez J, Gibson-Brown JJ, Gissi C,Godzik A, Hallböök F, Hirose D, Hosomichi K, Ikuta T, Inoko H,Kasahara M, Kasamatsu J, Kawashima T, Kimura A, KobayashiM, Kozmik Z, Kubokawa K, Laudet V, Litman GW, McHardyAC, Meulemans D, Nonaka M, Olinski RP, Pancer Z, Pennacchio

LA, Pestarino M, Rast JP, Rigoutsos I, Robinson-Rechavi M,Roch G, Saiga H, Sasakura Y, Satake M, Satou Y, Schubert M,Sherwood N, Shiina T, Takatori N, Tello J, Vopalensky P, WadaS, Xu A, Ye Y, Yoshida K, Yoshizaki F, Yu JK, Zhang Q, ZmasekCM, de Jong PJ, Osoegawa K, Putnam NH, Rokhsar DS, SatohN, Holland PWH (2008) The amphioxus genome illuminatesvertebrate origins and cephalochordate biology. Genome Res18:1100–1111

Holland PWH, Garcia-Fernàndez J, Williams NA, Sidow A (1994)Gene duplications and the origins of vertebrate development.Development Suppl 1994:125–133

Horard B, Castet A, Bardet PL, Laudet V, Cavailles V, Vanacker JM(2004) Dimerization is required for transactivation by estrogen-receptor-related (ERR) orphan receptors: evidence from amphi-oxus ERR. J Mol Endocrinol 33:493–509

Howard-Ashby M, Materna SC, Brown CT, Chen L, Cameron RA,Davidson EH (2006) Gene families encoding transcription factorsexpressed in early development of Strongylocentrotus purpur-atus. Dev Biol 300:90–107

Hu X, Lazar MA (2000) Transcriptional repression by nuclearhormone receptors. Trends Endocrinol Metab 11:6–10

King-Jones K, Thummel CS (2005) Nuclear receptors—a perspectivefrom Drosophila. Nat Rev Genet 6:311–323

Lacalli TC (2001) New perspectives on the evolution of protochordatesensory and locomotory systems, and the origin of brains andheads. Philos Trans R Soc Lond B 356:1565–1572

Langlois MC, Vanacker JM, Holland ND, Escriva H, Queva C, LaudetV, Holland LZ (2000) Amphicoup-TF, a nuclear orphan receptorof the lancelet Branchiostoma floridae, is implicated in retinoicacid signalling pathways. Dev Genes Evol 210:471–482

Laudet V, Gronemeyer H (2002) The nuclear receptor facts book.Academic Press, San Diego

Lonard DM, O’Malley BW (2006) The expanding cosmos of nuclearreceptor coactivators. Cell 125:411–414

Lu NZ, Wardell SE, Burnstein KL, Defranco D, Fuller PJ, Giguere V,Hochberg RB, McKay L, Renoir JM, Weigel NL, Wilson EM,McDonnell DP, Cidlowski JA (2006) International Union ofPharmacology. LXV. The pharmacology and classification of thenuclear receptor superfamily: glucocorticoid, mineralocorticoid,progesterone, and androgen receptors. Pharmacol Rev 58:782–797

Marlétaz F, Holland LZ, Laudet V, Schubert M (2006) Retinoic acidsignaling and the evolution of chordates. Int J Biol Sci 2:38–47

Michalik L, Auwerx J, Berger JP, Chatterjee VK, Glass CK, GonzalezFJ, Grimaldi PA, Kadowaki T, Lazar MA, O’Rahilly S, PalmerCN, Plutzky J, Reddy JK, Spiegelman BM, Staels B, Wahli W(2006) International Union of Pharmacology. LXI. Peroxisomeproliferator-activated receptors. Pharmacol Rev 58:726–741

Moore DD, Kato S, Xie W, Mangelsdorf DJ, Schmidt DR, Xiao R,Kliewer SA (2006) International Union of Pharmacology. LXII.The NR1H and NR1I receptors: constitutive androstane receptor,pregnene X receptor, farnesoid X receptor α, farnesoid Xreceptor β, liver X receptor α, liver X receptor β, and vitaminD receptor. Pharmacol Rev 58:742–759

Nuclear Receptors Nomenclature Committee (1999) A unifiednomenclature system for the nuclear receptor superfamily. Cell97:161–163

Paris M, Escriva H, Schubert M, Brunet F, Brtko J, Ciesielski F,Roecklin D, Vivat-Hannah V, Jamin EL, Cravedi JP, Scanlan TS,Renaud JP, Holland ND, Laudet V (2008a) Amphioxus postem-bryonic development reveals the homology of chordate meta-morphosis. Curr Biol 18:825–830

Paris M, Pettersson K, Schubert M, Bertrand S, Pongratz I, Escriva H,Laudet V (2008b) An amphioxus orthologue of the estrogenreceptor that does not bind estradiol: insights into estrogenreceptor evolution. BMC Evol Biol 8:219

664 Dev Genes Evol (2008) 218:651–665

Pereira de Jésus-Tran K, Côté PL, Cantin L, Blanchet J, Labrie F,Breton R (2006) Comparison of crystal structures of humanandrogen receptor ligand-binding domain complexed with vari-ous agonists reveals molecular determinants responsible forbinding affinity. Protein Sci 15:987–999

Pike AC, Brzozowski AM, Hubbard RE, Bonn T, Thorsell AG,Engström O, Ljunggren J, Gustafsson JÅ, Carlquist M (1999)Structure of the ligand-binding domain of oestrogen receptor betain the presence of a partial agonist and a full antagonist. EMBO J18:4608–4618

Putnam NH, Butts T, Ferrier DEK, Furlong RF, Hellsten U,Kawashima T, Robinson-Rechavi M, Shoguchi E, Terry A, YuJK, Benito-Gutiérrez È, Dubchak I, Garcia-Fernàndez J, Gibson-Brown JJ, Grigoriev IV, Horton AC, de Jong PJ, Jurka J,Kapitonov V, Kohara Y, Kuroki Y, Lindquist E, Lucas S,Osoegawa K, Pennacchio LA, Salamov AA, Satou Y, Sauka-Spengler T, Schmutz J, Shin- IT, Toyoda A, Bronner-Fraser M,Fujiyama A, Holland LZ, Holland PWH, Satoh N, Rokhsar DS(2008) The amphioxus genome and the evolution of the chordatekaryotype. Nature 453:1064–1072

Schubert M, Escriva H, Xavier-Neto J, Laudet V (2006a) Amphioxusand tunicates as evolutionary model systems. Trends Ecol Evol21:269–277

Schubert M, Holland ND, Laudet V, Holland LZ (2006b) A retinoicacid-Hox hierarchy controls both anterior/posterior patterning andneuronal specification in the developing central nervous systemof the cephalochordate amphioxus. Dev Biol 296:190–202

Sea Urchin Genome Sequencing Consortium (2006) The genome ofthe sea urchin Strongylocentrotus purpuratus. Science 314:941–952

Tanenbaum DM, Wang Y, Williams SP, Sigler PB (1998) Crystallo-graphic comparison of the estrogen and progesterone receptor’sligand binding domains. Proc Natl Acad Sci USA 95:5998–6003

Thornton JW, Need E, Crews D (2003) Resurrecting the ancestralsteroid receptor: ancient origin of estrogen signaling. Science301:1714–1717

Williams SP, Sigler PB (1998) Atomic structure of progesteronecomplexed with its receptor. Nature 393:392–396

Yagi K, Satou Y, Mazet F, Shimeld SM, Degnan B, Rokhsar D,Levine M, Kohara Y, Satoh N (2003) A genomewide survey ofdevelopmentally relevant genes in Ciona intestinalis. III. Genesfor Fox, ETS, nuclear receptors and NFκB. Dev Genes Evol213:235–244

Yu JK, Satou Y, Holland ND, Shin-I T, Kohara Y, Satoh N, Bronner-Fraser M, Holland LZ (2007) Axial patterning in cephalochor-dates and the evolution of the organizer. Nature 445:613–617

Dev Genes Evol (2008) 218:651–665 665