Rapid birth-and-death evolution of the xenobiotic metabolizing NAT gene family in vertebrates with evidence of adaptive selection

Abstract

Background: The arylamine N-acetyltransferases (NATs) are a unique family of enzymes widely distributed in nature that play a crucial role in the detoxification of aromatic amine xenobiotics. Considering the temporal changes in the levels and toxicity of environmentally available chemicals, the metabolic function of NATs is likely to be under adaptive evolution to broaden or change substrate specificity over time, making NATs a promising subject for evolutionary analyses. In this study, we trace the molecular evolutionary history of the NAT gene family during the last ~450 million years of vertebrate evolution and define the likely role of gene duplication, gene conversion and positive selection in the evolutionary dynamics of this family.

Results: A phylogenetic analysis of 77 NAT sequences from 38 vertebrate species retrieved from public genomic databases shows that NATs are phylogenetically unstable genes, characterized by frequent gene duplications and losses even among closely related species, and that concerted evolution only played a minor role in the patterns of sequence divergence. Local signals of positive selection are detected in several lineages, probably reflecting response to changes in xenobiotic exposure. We then put a special emphasis on the study of the last ~85 million years of primate NAT evolution by determining the NAT homologous sequences in 13 additional primate species. Our phylogenetic analysis supports the view that the three human NAT genes emerged from a first duplication event in the common ancestor of Simiiformes, yielding NAT1 and an ancestral NAT gene which in turn, duplicated in the common ancestor of Catarrhini, giving rise to NAT2 and the NATP pseudogene. Our analysis suggests a main role of purifying selection in NAT1 protein evolution, whereas NAT2 was predicted to mostly evolve under positive selection to change its amino acid sequence over time. These findings are consistent with a differential role of the two human isoenzymes and support the involvement of NAT1 in endogenous metabolic pathways.

Conclusions: This study provides unequivocal evidence that the NAT gene family has evolved under a dynamic process of birth-and-death evolution in vertebrates, consistent with previous observations made in fungi.

Figure 1

Phylogenetic tree of the NAT gene family based on 77 vertebrate nucleotide sequences. Shown is the maximum likelihood tree built using PhyML (GTR + I + G) and rooted with the three fish species (Danio rerio, Gasterosteus aculeatus, Oryzias latipes). The bootstrap values of 1,000 replicates are shown as percentages at nodes. Bootstrap values are only shown for nodes with greater than 50% support. The clades of Afrotheria, Laurasiatheria, Lagomorpha, Rodentia and Primates are shown in aqua blue, purple, green, blue and red, respectively.

Figure 2

Gene order and orientation in regions surrounding NAT genes on chicken (Gallus gallus) chromosome 11, turkey (Meleagris gallopavo) chromosome 13 and zebra finch (Taeniopygia guttata) chromosomes 11 and Un. Gene lengths and intergenic distances are not drawn to scale. Grey boxes indicate NAT-like sequences. Black boxes represent chicken-specific genes with no homologous sequences in turkey or zebra finch. Double slashes (//) indicate continuing sequence data extending toward the centromeric (cen) and telomeric (tel) parts of the chromosome. In each species, NAT genes are arrayed linearly along a single chromosome, indicating an expansion of the NAT gene family through tandem repeats, except in zebra finch where a fifth copy is found on an unidentified (Un) chromosome. Chicken NAT4 (chr11:17,483,758-17,484,630 in galGal3 assembly) represents a short sequence of 873 bp, identical to the NAT3 protein-coding sequence, which is embedded within the NAT3 gene (chr11:17,481,972-17,493,917) upstream of the NAT3 coding sequence.

Figure 3

Evidence for lineage-specific positive selection in the vertebrate NAT phylogeny. The branches with ω ratios > 1, as estimated by the free-ratio model (branch-specific test), are shown with black thick lines and are labeled from a to r. The estimated ω ratio and numbers of nonsynonymous and synonymous changes for each branch are as follows: a (1.26; 56.5/16.3), b (1.76; 22.8/4.7), c (1.07; 6.2/2.1), d (∞; 8.1/0.0), e (231.27; 45.7/0.1), f (1.07; 43.8/15.0), g (1.07; 3.3/1.1), h (1.11; 6.7/2.2), i (2.85; 10.7/1.4), j (2.54; 14.0/2.0), k (∞; 4.9/0.0), l (2.43; 13.4/2.0), m (1.41; 9.2/2.4), n (∞; 8.2/0.0), o (6.31; 12.0/0.7), p (∞; 5.1/0.0), q (1.11; 3.1/1.0), r (1.56; 4.5/1.1). In addition, branch-site tests were carried out to test for positive selection in 13 pre-specified lineages (highlighted in red): the ‘B’ letter refers to the branch-site tests performed at the branch level and the ‘C’ letter refers to those performed at the clade level (multiple branch-site analysis, see Methods). We focused on the major taxonomic groups of the vertebrate phylogeny: Primates (#13), Rodentia (#12), Lagomorpha (#11), Glires (Rodentia + Lagomorpha, #10), Euarchontoglires (Primates + Glires, #9), Laurasiatheria (#8), Afrotheria (#7), Eutheria (#6), Mammalia (#5), Sauropsida (#2), and the fish species (#1). As the group of avian NAT sequences was not monophyletic, we separately tested the subgroup of NAT1 sequences and the subgroup of NAT2/NAT3/NAT4/NAT5 sequences in birds (#3 and #4, respectively). * Statistically significant branch-site tests at the conventional P-value threshold of 0.05 (not corrected for multiple testing). ** Statistically significant branch-site tests at the Bonferroni-corrected threshold of 0.0038 (13 tests).

Figure 4

Phylogenetic tree of the NAT gene family in primates. Shown is the maximum likelihood tree built using PhyML (GTR + I + G) with the multiple alignment of 59 NAT nucleotide sequences from 19 distinct primate species. The tree was rooted with the non simian species (Otolemur garnetti, Microcebus murinus, Tarsius syrichta). The bootstrap values of 1,000 replicates are shown as percentages at nodes. Bootstrap values are only shown for nodes with greater than 50% support. The clades of Strepsirrhini, Platyrrhini, Cercopithecidae, Hylobatidae and Hominidae are shown in turquoise, green, blue, pink and red, respectively.

Figure 5

Evidence for lineage-specific positive selection in the primate NAT phylogeny. Branches with evidence of positive selection (ω > 1), as estimated by the free-ratio model (branch-specific test), are shown with black thick lines and are labeled from a to e. The estimated ω ratio and numbers of nonsynonymous and synonymous changes for each branch are as follows: a (∞; 8.1/0.0), b (∞; 5.1/0.0), c (∞; 7.0/0.0), d (1.07; 3.9/1.3), e (1.63; 4.5/1.0). In addition, branch-site tests were carried out to test for positive selection in 9 pre-specified lineages (highlighted in red): the ‘B’ letter refers to the branch-site tests performed at the branch level and the ‘C’ letter refers to those performed at the clade level (multiple branch-site analysis, see Methods). *Statistically significant branch-site tests at the conventional P-value threshold of 0.05 (not corrected for multiple testing). **Statistically significant branch-site tests at the Bonferroni-corrected threshold of 0.0056 (9 tests).

Figure 6

Variable selective pressures (ω) along NAT protein sequence. The posterior mean of ω was estimated for each of the 290 amino acid residues by the codeml program (a) in the vertebrate dataset (n = 77 coding sequences), (b) in the mammalian dataset (n = 55 coding sequences), (c) in the primate dataset (n = 43 coding sequences), (d) in the set of 19 simian NAT2 coding sequences. A graphical representation of the three-domain structure of human NAT1 and NAT2 proteins is shown at the bottom. Residues 1–83 in the N-terminus form domain I, mainly consisting of α-helices; residues 84–192 form domain II, mainly consisting of β-strands, followed by inter-domain comprising residues 193–229, and residues 230–290 consisting of both α-helix and β-strands form domain III. The amino and carboxyl termini are labeled as NT and CT, respectively. The 17-residue insertion and carboxyl-terminal tail of human NATs, which are lacking in the structures of prokaryotic NATs, are highlighted with thick lines. Residues of the catalytic triad are highlighted with stars, residues interacting with CoA are highlighted with black triangles, and residues involved in substrate binding are highlighted with white triangles.