Protein evolution by molecular tinkering: diversification of the nuclear receptor superfamily from a ligand-dependent ancestor

Abstract

Understanding how protein structures and functions have diversified is a central goal in molecular evolution. Surveys of very divergent proteins from model organisms, however, are often insufficient to determine the features of ancestral proteins and to reveal the evolutionary events that yielded extant diversity. Here we combine genomic, biochemical, functional, structural, and phylogenetic analyses to reconstruct the early evolution of nuclear receptors (NRs), a diverse superfamily of transcriptional regulators that play key roles in animal development, physiology, and reproduction. By inferring the structure and functions of the ancestral NR, we show--contrary to current belief--that NRs evolved from a ligand-activated ancestral receptor that existed near the base of the Metazoa, with fatty acids as possible ancestral ligands. Evolutionary tinkering with this ancestral structure generated the extraordinary diversity of modern receptors: sensitivity to different ligands evolved because of subtle modifications of the internal cavity, and ligand-independent activation evolved repeatedly because of various mutations that stabilized the active conformation in the absence of ligand. Our findings illustrate how a mechanistic dissection of protein evolution in a phylogenetic context can reveal the deep homology that links apparently "novel" molecular functions to a common ancestral form.

Figure 1. Phylogeny of the NR superfamily.

A reduced representation of the NR phylogeny inferred by maximum likelihood and Bayesian method is shown. Colored circles indicate the presence of each clade in major metazoan taxa: sponges (light blue), placozoans (purple), Cnidaria (green), protostomes (orange), and deuterostomes (dark blue). Clades are labeled with their common protein names; NR nomenclature codes are in parentheses (see also Table S6), with the number of sequences analyzed in each group after the slash. Branch labels show support measured as approximate likelihood ratios (the ratio of the likelihood of the best tree with that node to the best tree without it), Bayesian posterior probabilities, and chi-square confidence estimates (the probability of a likelihood ratio at least as great as the observed ratio if the node is not resolved on the true tree). INRs, clade of invertebrate-only nuclear receptors with no standard nomenclature. Scale bar, probability of substitutions per site. A key to abbreviations of protein names is shown below the phylogeny. For unreduced phylogenies and a list of species, genes, accessions, and receptor abbreviations, see Figures S2–S3 and Tables S1 and S6.

Figure 2. Duplications and losses in NR evolution.

(A) Reconciliation of NR gene family phylogeny with the species tree shown in panel B. The phylogeny is rooted to minimize the total number of gene duplications and losses; each branch is labeled with the number of additional duplications and losses required for the tree to be rooted on that branch. Gene duplications are shown as green circles, gene losses as Xs, and speciation events as yellow shapes (circles, split of demosponges from eumetazoans; squares, split of placozoans from eumetazoans (s.s.); diamonds, split of cnidarians from bilaterians). Colors of gene names indicate the taxon from which they are derived, using the color scheme in panel B. (B) Gene duplication history implied by the reconciled tree in panel A. Green bars, duplications; red bars, losses. Duplications are labeled with the named NR lineages they generated. The NR1/NR4/INR ortholog lost in the placozoans (marked *) was generated in the duplication marked **. The large bar comprises numerous duplications that cannot be temporally ordered.

Figure 3. Ligand binding and transcriptional activation by A. queenslandica nuclear receptors.

(A) Transcriptional effects of AqNR1 and AqNR2 LBDs in a luciferase reporter assay. Transfected cells were treated with stripped or complete serum (white and black bars, respectively). Activation is scaled relative to vector-only control; error bars, SEM for 3 replicates. (B) Fatty acid (FA) binding profile by AqNR1 and AqNR2, as determined by electrospray ionization/mass spectrometry. Binding of each FA species is shown as a percentage of the total FAs bound by each receptor. The most abundantly bound FA for each receptor is in bold. FA binding by rat HNF4α is shown for comparison. Measurements are shown relative to a labeled palmitic acid standard and represent the average of two runs. (C) Quantitative fatty acid binding by AqNR1 and mutant AqNR1-R492A as determined in a colorimetric enzymatic assay. Receptors were expressed in E. coli, purified, and treated with no serum (gray), complete serum (black), or stripped serum (white). Errors bars, SEM for 3 replicates. The percentage of receptor molecules occupied by FAs (after subtraction of background binding by MBP/no serum control) is shown for each experimental condition.

Figure 4. Conserved architecture and function of the AqNR1 ligand pocket.

(A) The ligand cavity in the predicted structure of AqNR1 (green, with palmitic acid) and in the crystallographic structure of rat HNF4α (blue, with lauric acid) are shown. Cavities (volumes of 835 and 450 ų , respectively) are shown as surfaces, and the ligands as sticks. Labeled side chains at homologous positions that contact the ligand were subject to experimental characterization. Dashed lines show hydrogen bonds. (B) Effect of mutagenesis of ligand-contacting residues on transactivation in a luciferase reporter assay by AqNR1 and rat HNF4α in stripped (white bars) and complete serum (black bars). Bar labels show fold activation relative to vector-only control with stripped serum; error bars, SEM for 3 replicates.

Figure 5. Ancient origin of ligand-activation in the NR superfamily.

(A) Reconstruction of structural and functional characters on the NR phylogeny. Ligand-regulated transcriptional activators are shown in green, with ligands in parentheses. Red, activators with no known ligand. Underlined, receptors with transcriptional activity in the absence of ligand or other modifications. Black, repressors that do not activate transcription. The ancestral NR (AncNR, green circle) is shown, with the most parsimonious reconstructions at that node for the characters in the table. Hash marks on branches show gains of ligand-independent activity with or without loss of ligand binding (filled and empty red boxes, respectively). States of protein structural characters are shown in the table. “3 layers”: 10 to 12 helices arranged in conserved three-layer sandwich as in panel B. “Ligand pocket open or filled”: O, open internal cavity bounded by helices 3, 5, 7, 10, and 12 in crystal structure (cavity volume in ų); O*, open pocket in homology model; O**, open pocket inferred from experimental data on ligand-binding or ligand-regulation; F, filled cavity, <50 ų. “H12 coactivator sequence”: presence (+) or absence (−) of canonical co-activator interface φφ*κφφ motif in helix 12 (φ, hydrophobic; *, any residue; κ, charged). “L6-11 salt bridge”: presence (+) or absence (−) of salt bridge from the L6-7 loop to helix H10-11. “Rigid HX and H10-H12 h-bonds”: presence (+) or absence (−) of short additional helix N-terminal to H12 and hydrogen bonds between H10-11 and H12. “Bulky residues in pocket”: presence (1,2,3) or absence (−) of Phe or Trp residues in space occupied by ligand in other receptors; 1, 2, 3 indicate bulky residues at different sets of sequence sites (see Figure 6). Blank cells, no data available. Rodent LRH-1 has been coded as a ligand-independent activator based on its crystal structure, but functional assays have not established its ligand-independence. (B) Shared structural basis for ligand-dependent activation in NR LBDs. The peptide backbones and ligands are superimposed for five distantly related NRs: human HNF4α, mouse RXRα, human ERα, mouse RARβ, and human PPARα. Gray spheres show ligands. Helices (shown as cylinders) are numbered, and the coactivator peptide (Co-Ac) is a magenta ribbon. The beta-sheet between H5 and H6 is pink. The region between H1 and H3, which is structurally variable, has been removed to show the ligands more clearly. For details and PDB identifiers, see Tables S4, S5, and S7.

Figure 6. Independent structural mechanisms for ligand-independent activation in nuclear receptors.

(A) Receptors with filled ligand cavities. Bulky side chains (shown as sticks) that fill the cavity are shown for rat Nurr1 (cyan), human ERRα (magenta), and oyster ER (blue), with α-carbons as balls. Green ribbon, backbone of the loop with side chains that fill the pocket in D. melanogaster Ftz-F1. The van der Waals space occupied by estradiol in human ERα is in white. (B) In human CAR (bronze), a network of hydrogen bonds (blue) stabilizes the activation-function helix H12 and a novel adjacent helix (orange). These interactions are absent from other NRs, such as the close paralog VDR (white), shown with its ligand vitamin D (blue spheres). (C) In the LRH-1 protein of rodents (left, blue), a salt bridge and hydrogen bond between side chains on helices 7 and 10 replaces similar interactions in human LRH-1 (right, green) between each helix and the ligand (white). The ligand-mediated interactions and the amino acids involved in human LRH-1 are also present in the closest NR paralog SF-1 . For PDB identifiers, see Table S7.