Correlated Evolution of Androgen Receptor and Aromatase Revisited

Correlated Evolution of Androgen Receptor and Aromatase Revisited Adam M. Reitzel and Ann M. Tarrant* Biology Department, Woods Hole Oceanographic Institution, Woods Hole, MA *Corresponding author: E-mail: . Associate editor: David Irwin Abstract Key words: androgen receptor, aromatase, correlated evolution, cytochrome p450, nuclear receptor. Tiwary and Li (2009) claimed to identify apparent AR and CYP19 genes with similarity searches throughout the Eumetazoa, including insects, a cnidarian, and other invertebrates. These authors reported a strong and significant correlation of protein distances between AR and CYP19, but not among background control proteins, suggesting that AR and CYP19 have evolved at similar rates throughout most of animal evolution. Tiwary and Li (2009) concluded that AR and CYP are evolving in a correlated fashion, which they termed parallel evolution. However, to test hypotheses for correlated evolution, whether it be co- or parallel evolution, practitioners must use orthologous sequences. Confirmation of parallel evolution additionally requires identification of identical yet independent replacements in particular amino acids, which involve sitespecific analyses in a phylogenetic context (Rokas and Carroll 2008). We show that their analysis violates the requirement for orthologous sequences and further that the relationship is not strongly different from other surveyed proteins without a functional interaction. The reported identification of AR and CYP19 across much of the animal kingdom by Tiwary and Li (2009) conflicts with previous literature. Phylogenetic studies of the NR superfamily have shown that AR differentiated from an ancestral steroid receptor early in the vertebrate lineage (Thornton 2003; Bertrand et al. 2004; Bridgham et al. 2008). Similarly, CYP19 most likely evolved in the lineage leading © The Author 2010. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: Mol. Biol. Evol. 27(10):2211–2215. 2010 doi:10.1093/molbev/msq129 Advance Access publication May 21, 2010 2211 Letter Molecular coevolution is the correlated evolution of two or more interacting molecules due to selection imposed by changes in each on the other. Molecular coevolution has been demonstrated in several cases where proteins directly interact, particularly when they form obligate complexes within molecular networks. These interactions have largely been elucidated by testing for correlated changes in amino acid or nucleic acid sequences using phylogenetic methods and/or structural models (Yeang and Haussler 2007; Pazos and Valencia 2008). Molecular coevolution of pairs of proteins may be more indirectly mediated through their interactions with a conserved third molecule, such as a cofactor, ligand, or substrate (McPartland et al. 2007). Particularly, good candidates to investigate this later form of correlated evolution are components of steroid-signaling pathways, which require the actions of enzymes and receptors with specific, high-affinity interactions with hormones. In a recent study, Tiwary and Li (2009) tested for correlated evolution of androgen receptor (AR) and aromatase (CYP19) throughout the animal lineage. AR is a ligandactivated member of the nuclear receptor (NR) superfamily (NR3C4) that specifically binds androgens. Aromatase, a cytochrome p450 (CYP) enzyme, catalyzes the synthesis of estrogens from androgen precursors. Thus, these two proteins share specificity for androgens, and this indirect interaction may link their evolutionary and functional histories. Conserved interactions among proteins or other molecules can provide strong evidence for coevolution across their evolutionary history. Diverse phylogenetic methods have been applied to identify potential coevolutionary relationships. In most cases, these methods minimally require comparisons of orthologous sequences and appropriate controls to separate effects of selection from the overall evolutionary relationships. In vertebrates, androgen receptor (AR) and cytochrome p450 aromatase (CYP19) share an affinity for androgenic steroids, which serve as receptor ligands and enzyme substrates. In a recent study, Tiwary and Li (Tiwary BK, Li W-H. 2009. Parallel evolution between aromatase and androgen receptor in the animal kingdom. Mol Biol Evol. 26:123–129) reported that AR and CYP19 displayed a signature of ancient and conserved interactions throughout all the Eumetazoa (i.e., cnidarians, protostomes, and deuterostomes). Because these findings conflicted with a number of previous studies, we reanalyzed the data set used by Tiwary and Li. First, our analyses demonstrate that the invertebrate genes used in the previous analysis are not orthologous sequences but instead represent a diverse set of nuclear receptors and CYP enzymes with no confirmed or hypothesized relationships with androgens. Second, we show that 1) their analytical approach, which measures correlations in evolutionary distances between proteins, potentially led to spurious significant relationships due simply to conserved domains and 2) control comparisons provide positive evidence for a strong influence of evolutionary history. We discuss how corrections to this method and analysis of key taxa (e.g., duplications in the teleost fish and suiform lineages) can inform investigations of the coevolutionary relationships between AR and aromatase. Reitzel and Tarrant · doi:10.1093/molbev/msq129 MBE to the cephalochordate–vertebrate ancestor (Campbell et al. 2004; Baker 2007). In a recent study, Markov et al. (2009) stated that in contrast to the report by Tiwary and Li, they found no evidence for an aromatase gene outside of the chordate lineage, but they did not explain the incongruent results. We tested the evolutionary relationships of the NRs and CYPs used by Tiwary and Li (2009) (see supplementary file, Supplementary Material online, for all methods). Because their data set lacked designated outgroup sequences, we retrieved a set of sequences to represent the diversity within these superfamilies. We found that none of the invertebrate sequences included in this earlier study are orthologous to either AR (fig. 1A) or CYP19 (fig. 1B). Our analysis strongly supports placement of the invertebrate NRs within diverse NR families, mostly in NR family 2. As previously reported (Holland et al. 2008; Schubert et al. 2008), the Branchiostoma steroid receptor was positioned as ancestral to the NR3 steroidogenic receptors. Similarly, the invertebrate CYPs used by Tiwary and Li (2009) represent diverse CYP families. As expected, the Branchiostoma CYP19 gene formed a strongly supported clade with the vertebrate CYP19s. All the vertebrate sequences were strongly supported as orthologs of AR and CYP19. 2212 Analysis of the evolutionary relationships between AR and CYP19 is complicated because both of these proteins have been retain (...truncated)