Whole-genome duplication shaped cell-type evolution in the vertebrate brain

Main

It has long been proposed 6 that vertebrates underwent WGDs sometime in their ancestry. This idea was later refined into a ‘2R hypothesis’, with two WGDs identified in the early evolution of jawed vertebrates. Recent studies have shown that the first WGD predated the separation of cyclostome and gnathostome lineages, and subsequent WGDs occurred independently in each lineage 1 , 7 (Fig.  1a ). WGD is not the only way genes duplicate and must be distinguished from extensive small-scale duplications (SSDs) 8 . Most duplicated genes are lost after duplication, but retained genes may undergo complementary loss of function (subfunctionalization) and/or evolve new functions (neofunctionalization) 9 . Retained genes are also frequently co-opted into evolving gene regulatory networks 10 , 11 , and this process is proposed to drive new uses in the development and specification of tissues, organs and cell types 11 , 12 . An evolutionary definition of cell types has been proposed that is based on common descent regardless of form and function 13 . New cell types can evolve through duplication and divergence (the sister cell-type model), which is an inherently hierarchical concept 13 , 14 . Consequently, in many cases, individual cell types are species-specific or clade-specific 15 . These issues highlight the importance of investigating cell-type evolution at different hierarchical levels and across different regions of the body.

Fig. 1: Vertebrate brain atlases and core TF programs that define major cell-type families. The alternative text for this image may have been generated using AI.

Full size image

a , Phylogenetic tree showing the approximate timing (million years ago) for vertebrate-shared auto-tetraploidization (1R V ), a jawed-vertebrate-specific allo-tetraploidization (2R JV ) and a cyclostome-specific hexaploidization (2R CV ) based on recent studies 1 , 7 , 45 . b , Uniform manifold approximation and projection (UMAP) visualization of the integrated neuronal (left) and non-neuronal (right) atlases for the four indicated species. Each dot represents a single nucleus or cell. To ensure balanced representation across datasets, only 20,000 randomly sampled cells or nuclei are shown per species for both neuronal and non-neuronal integrated atlases. c , Dot plot showing conserved TFs that define major cell-type families in vertebrates (a complete list is provided in Supplementary Table 3 ). The dot size represents the percentage of cells in each cell-type family expressing that gene. The colour gradient for each dot is a scaled average expression for each gene in the species (species colours are as in b ). For TF gene families with multiple copies in lamprey, only the copy with the highest expression is displayed. d , Dot plot showing the expression of conserved key TF families of vertebrate astrocyte clusters ( x  axis) in adult amphioxus brain. The dot size represents the percentage of cells in each cell type expressing that gene. Colour represents the scaled average expression for each gene. A complete dot plot for amphioxus expression of key TFs for all vertebrate cell-type families is shown in Extended Data Fig. 3 . AST, astrocytes; CP-EC, choroid plexus epithelial cells; DeExc, diencephalon glutamatergic neurons; DeInh, diencephalon GABAergic neurons; Epen, ependymal cells; Fibro, fibroblasts; MeExc, mesencephalon glutamatergic neurons; MeInh, mesencephalon GABAergic neurons; Micro, microglia; Oligo, oligodendrocytes; OPC, oligodendrocyte precursor cells; ReExc, rhombencephalon glutamatergic neurons; ReInh, rhombencephalon GABAergic neurons; TeExc, telencephalon glutamatergic neurons; TeInh, telencephalon GABAergic neurons; Vasc, vascular cells. Additional information on cluster names is provided in Supplementary Table 2 .

Vertebrates possess well-developed brains that enable rapid and coordinated responses to environmental stimuli and facilitated adaptation to diverse ecological niches. Compared to their closest invertebrate relatives—tunicates and amphioxus—vertebrate brains are highly regionalized and complex. Previous studies 4 , 16 , 17 , 18 have demonstrated similarities and differences in neural cell types within and between vertebrate species. However, ancestral repertoires of neural cell types and their core transcription factor (TF) programs, and the origin of cell types in early vertebrates, remain poorly understood. Potential roles for WGD and SSD paralogues in brain cell-type evolution remain obscure. In this study, we analyse four vertebrate (human ( Homo sapiens ), mouse ( Mus musculus ), lizard ( Pogona vitticeps ) and lamprey ( Petromyzon marinus )) and one amphioxus ( Branchiostoma floridae ) whole brain single-cell transcriptomes to infer ancestral repertoires of neural cell-type families. We then systemically analyse WGD paralogues (ohnologues) and SSD paralogues in cell-type evolution. Our findings indicate that 2R WGDs capacitated cell-type innovation during both early vertebr…