Inability to switch from ARID1A-BAF to ARID1B-BAF impairs exit from pluripotency and commitment towards neural crest formation in ARID1B-related neurodevelopmental disorders

Abstract

Subunit switches in the BAF chromatin remodeler are essential during development. ARID1B and its paralog ARID1A encode for mutually exclusive BAF subunits. De novo ARID1B haploinsufficient mutations cause neurodevelopmental disorders, including Coffin-Siris syndrome, which is characterized by neurological and craniofacial features. Here, we leveraged ARID1B^+/- Coffin-Siris patient-derived iPSCs and modeled cranial neural crest cell (CNCC) formation. We discovered that ARID1B is active only during the first stage of this process, coinciding with neuroectoderm specification, where it is part of a lineage-specific BAF configuration (ARID1B-BAF). ARID1B-BAF regulates exit from pluripotency and lineage commitment by attenuating thousands of enhancers and genes of the NANOG and SOX2 networks. In iPSCs, these enhancers are maintained active by ARID1A-containing BAF. At the onset of differentiation, cells transition from ARID1A- to ARID1B-BAF, eliciting attenuation of the NANOG/SOX2 networks and triggering pluripotency exit. Coffin-Siris patient cells fail to perform the ARID1A/ARID1B switch, and maintain ARID1A-BAF at the pluripotency enhancers throughout all stages of CNCC formation. This leads to persistent NANOG/SOX2 activity which impairs CNCC formation. Despite showing the typical neural crest signature (TFAP2A/SOX9-positive), ARID1B-haploinsufficient CNCCs are also aberrantly NANOG-positive. These findings suggest a connection between ARID1B mutations, neuroectoderm specification and a pathogenic mechanism for Coffin-Siris syndrome.