Neofunctionalization in vertebrates: the example of retinoic acid receptors

Abstract

Understanding the role of gene duplications in establishing vertebrate innovations is one of the main challenges of Evo-Devo (evolution of development) studies. Data on evolutionary changes in gene expression (i.e., evolution of transcription factor-cis-regulatory elements relationships) tell only part of the story; protein function, best studied by biochemical and functional assays, can also change. In this study, we have investigated how gene duplication has affected both the expression and the ligand-binding specificity of retinoic acid receptors (RARs), which play a major role in chordate embryonic development. Mammals have three paralogous RAR genes--RAR alpha, beta, and gamma--which resulted from genome duplications at the origin of vertebrates. By using pharmacological ligands selective for specific paralogues, we have studied the ligand-binding capacities of RARs from diverse chordates species. We have found that RAR beta-like binding selectivity is a synapomorphy of all chordate RARs, including a reconstructed synthetic RAR representing the receptor present in the ancestor of chordates. Moreover, comparison of expression patterns of the cephalochordate amphioxus and the vertebrates suggests that, of all the RARs, RAR beta expression has remained most similar to that of the ancestral RAR. On the basis of these results together, we suggest that while RAR beta kept the ancestral RAR role, RAR alpha and RAR gamma diverged both in ligand-binding capacity and in expression patterns. We thus suggest that neofunctionalization occurred at both the expression and the functional levels to shape RAR roles during development in vertebrates.

Figure 1. Phylogenetic View of Deuterostomes and RARs

(A) Current view of deuterostome phylogeny with amphioxus representing the basal chordate [5]. RARs used in the present study are indicated at their respective taxonomic positions—for mouse, Xenopus, zebrafish, lamprey, amphioxus, and tunicates. The position of the synthetic ancestral sequence is indicated by a red circle. The two proposed periods of whole genome duplications in vertebrates are indicated as Phase I and Phase II, occurring respectively before and after the divergence of lampreys. (B) Phylogenetic tree showing the placement of the RARs used in this study. Branch length is proportional to evolutionary change (bar = 0.1 substitutions per site); numbers at nodes are bootstrap support, in percent of 1,000 replicates. Branches supported by bootstrap lower than 70% have been polytomised. The tree was rooted by the amphioxus sequence, in agreement with [5]. Species abbreviations and their groups are indicated as follows. Amphioxus: Amphi, Branchiostoma floridae. Tunicates: Pm, Polyandrocarpa misakiensis; Ci, Ciona intestinalis. Lampreys: Lamp, Petromyzon marinus. Teleost fish: Takifugu, Takifugu rubripes; Tetraodon, Tetraodon nigroviridis; and Danio, Danio rerio. Amphibians: Xenopus, Xenopus laevis; Ambystoma, Ambystoma mexicanum; and Notophthalmus, Notophthalmus viridescens. Birds: Gallus, Gallus gallus; and Coturnix, Coturnix coturnix. Mammals: Homo, Homo sapiens; Mus, Mus musculus; and Rattus, Rattus norvegicus.

Figure 2. Protein Sequence Alignment of Selected Gnathostome RARs

RARs are represented from lamprey (LampRAR, Petromyzon marinus), amphioxus (AmphiRAR, Branchiostoma floridae), tunicate (RAR_POLM1, Polyandrocarpa misakiensis), and the synthetic predicted ancestral RAR (Ancestor). The position of the 12 helices is indicated above the alignment (H1–H12). Residues implicated in direct contacts with the ligand are numbered from 1 to 25 below the alignment. The three divergent residues within the LBP between vertebrate RARs are within vertical rectangles in helices 3, 5, and 11. Gnathostome and Polyandrocarpa sequences are named with the nomenclature code used in the nuclear receptor database NUREBASE (http://www.ens-lyon.fr/LBMC/laudet/nurebase/nurebase.html) [39].

Figure 3. Transcriptional Activity and Binding Selectivity of Vertebrate RARs

Transcriptional activity is shown in (A–C), (G–I), (M), and (N), and corresponding binding selectivity in (D–F), (J–L), (O), and (P). Identities of the vertebrate RARs for each activity-selectivity pair are indicated above each bar graph. In each case, a chimera comprising the RAR LBD fused to the GAL4 DNA-binding domain (GAL-RAR(LBD)) has been used. The analysis of transcriptional activity in (A–C), (G–I), (M), and (N) shows transient transactivation assays in Cos1 cells with the indicated GAL-RAR(LBD) expression vector and the cognate (17m)5x-G-luc reporter plasmid, in the presence of increasing concentrations (10⁻¹⁰ to 10⁻⁶ M) of ATRA (red bars), BMS753 (yellow bars), BMS641 (light green bars), and BMS961 (dark green bars) respectively. The black bars indicate transactivation in the absence of hormone. Partial proteolysis maps of different in vitro-translated RARs are shown in (D–F), (J–L), (O), and (P). For each proteolysis gel lane 1 represents the undigested protein, lane 2 shows digestion of the receptor in the absence of ligand, lanes 3 and 4 show digestion of the receptor in the presence of ATRA (10⁻⁴ to 10⁻⁵ M), lanes 5 and 6 show digestion in the presence of BMS753 (10⁻⁴ to 10⁻⁵ M), lanes 7 and 8 show digestion in the presence of BMS641 (10⁻⁴ to 10⁻⁵ M), and lanes 9 and 10 show digestion in the presence of BMS961 (10⁻⁴ to 10⁻⁵ M). Protected bands in the presence of BMS641 are indicated by an asterisk, and slightly protected bands are indicated by arrowheads.

Figure 4. Transcriptional Activity and Binding Selectivity of Chordate RARs

Transcriptional activity is shown in (A–D) and (I–L), and corresponding binding selectivity in (E–H) and (M–P). Identities of the chordate RARs for each activity-selectivity pair are indicated above each bar graph. Transcriptional activity is shown in (A–D) for LampRAR, AmphiRAR, PmRAR, and AncRAR, and that of AmphiRAR mutants is shown in (I–L). Partial proteolysis maps of the different in vitro-translated RARs are shown in (E–H) and (M–P). Chimeric GAL-RAR(LBD) transactivation methods, colour code of the transactivation figures, and contents of each proteolysis gel are as in Figure 3. Protected bands in the presence of BMS641 are indicated by an asterisk, and slightly protected bands are indicated by arrowheads.

Figure 5. Schematic Representation of the Expression Territories of RARs

Staining of embryos indicates expression of mRARα (A), mRARβ (B), and mRARγ (C) in mouse embryos at E9; of xRARα (G), xRARβ (H), and xRARγ (I) in stage 30 Xenopus embryos, and of AmphiRAR (M) in 20 h old amphioxus larvae. Schematic representations are shown of the expression territories of mRARs (D–F), xRARs (J–L), and AmphiRAR (N) in mouse, Xenopus, and amphioxus embryos, respectively. Regions with high levels of expression are red and those with lower levels of expression are pink. Arrowheads indicate regions in mouse and Xenopus embryos where the RAR expression cannot be correlated with AmphiRAR expression and can be described as “new expression territories.”

Figure 6. Representation of the Transactivation and Binding Capacities of the RARs Used in the Present Study

The three synthetic retinoids are shown as α, BMS753; β, BMS641; and γ, BMS961. The phylogenetic relationships between the RARs have been schematized by a phylogenetic tree (the tunicate and amphioxus RARs have been polytomised, LampRAR is also polytomised with the vertebrate RARs). The putative position in the tree of the ancestral sequence is indicated by a dashed branch in red.