<i>ATG</i> gene duplication in vertebrates: evolutionary divergence and its functional implications

Macroautophagy (hereafter referred to as autophagy) requires the coordinated action of approximately 20 <i>ATG</i> (autophagy related) genes. Duplication of <i>ATG</i> genes has had a major impact on the evolution of the autophagy pathway among major lineages. One duplication hotspot is in vertebrates. However, the exact duplication timing, post-duplication evolutionary divergence patterns, and its relation to functional differences among paralogs have not been investigated in detail. Here, we demonstrate that most <i>ATG</i> genes were likely duplicated by whole-genome duplication events near the root of vertebrates. We compared the sequence and gene expression divergence between paralogs and categorized the evolutionary fates (i.e., how ancestral function is divided between paralogs). Within the paralog pairs that evolved most asymmetrically, namely <i>BECN</i>, <i>WIPI</i> (<i>WIPI1</i> and <i>WIPI2</i>), and <i>ATG16</i>, one paralog likely retained the ancestral function, allowing the other to evolve under less constraint. While no obvious asymmetry was observed between <i>ATG9A</i> and <i>ATG9B</i> in non-mammalian vertebrates, <i>ATG9B</i> experienced marked sequence divergence and expression level reduction in mammals, suggesting a shift in balance. Expression patterns among the <i>ULK-1</i> (<i>ULK1</i> and <i>ULK2</i>), <i>GABARAP</i> (<i>GABARAP</i> and <i>GABARAPL1</i>), and <i>LC3</i> (<i>LC3A</i> and <i>LC3B</i>) pairs were more consistent with hypofunctionalization/dosage sharing, such that ancestral function depends on both paralogs. We also demonstrate that both <i>ULK1</i> and <i>ULK2</i> can support autophagy, whereas only <i>BECN1</i>, but not <i>BECN2</i>, has autophagic function and discuss the relationship between autophagic function and evolutionary divergence. The present detailed analysis of <i>ATG</i> gene duplication in vertebrates provides a critical timeline for interpreting functional differentiation between homologs.<b>Abbreviations</b>: ATG: autophagy related; BLAST: Basic Local Alignment Search Tool; DKO: double knockout; GFP: green fluorescent protein; GLMM: generalized linear mixed model; KO: knockout; LC3: MAP1LC3; MEF: mouse embryonic fibroblast; ns: non-significant; PAML: Phylogenetic Analysis by Maximum Likelihood; RPKM: reads per kilobase per million mapped reads; SVA: surrogate variable analysis; TMM: trimmed mean of M values; TMR: tetramethylrhodamine; WT: wild type.