The acquisition of totipotency is a fundamental process in early mammalian development, characterized by zygotic genome activation (ZGA) and the transient expression of 2-cell-stage (2C) specific genes and retrotransposons, such as MERVL. While the Double Homeobox (DUX) family proteins (mouse Dux and human DUX4) are recognized as master transcription factors for this transition, the underlying biophysical mechanisms by which they orchestrate global chromatin remodeling and coordinate distal gene activation have remained elusive.
This study reveals that DUX family proteins undergo liquid-liquid phase separation (LLPS) to drive the transition to a totipotent-like state. The authors identified that the intrinsically disordered regions (IDRs) within the conserved Homeobox domains (HD), specifically mediated by key arginine residues (e.g., R70/R72 in mouse Dux; R71/R73 in human DUX4), are essential for forming dynamic nuclear condensates. These DUX condensates act as specialized transcriptional hubs, recruiting co-activators like CBP/p300 and structural proteins such as CTCF. Notably, although the phase-separation-deficient DUX mutant retains its DNA-binding capacity and the ability to interact with CBP/p300, it completely loses the ability to activate MERVL.
Mechanistically, DUX phase separation facilitates the de novo assembly of totipotency-specific super-enhancers (SEs). By leveraging Hi-C analysis, we demonstrate that DUX condensates trigger a profound reorganization of the 3D genome, including the shifting of TAD boundaries and the formation of novel enhancer-promoter loops. This spatial reorganization allows DUX to activate a vast network of 2C-specific genes, even those lacking direct DUX binding sites, through long-range chromatin interactions.
Functionally, Dux mutants deficient in phase separation fail to activate the 2C transcriptome and lose the capacity to contribute to extraembryonic lineages in chimeric assays. Furthermore, the phase separation of human DUX4 is shown to be a critical driver of myotoxicity in Facioscapulohumeral Muscular Dystrophy (FSHD), suggesting that targeting LLPS could offer a novel therapeutic strategy. Collectively, our findings establish a phase-separation-mediated model for 3D genome reprogramming to obtain totipotency.
DOI: 10.1093/procel/pwag014