Genomic clustering tendency of transcription factors reflects phase-separated transcriptional condensates at super-enhancers

Abstract

Many transcription factors (TFs) have been shown to bind to super-enhancers, forming transcriptional condensates to activate transcription in various cellular systems. However, the genomic and epigenomic determinants of phase-separated transcriptional condensate formation remain poorly understood. Questions regarding which TFs tend to associate with transcriptional condensates and what factors influence their association are largely unanswered. Here we systematically analyzed 571 DNA sequence motifs across the human genome and 6650 TF binding profiles across different cell types to identify the molecular features contributing to the formation of transcriptional condensates. We found that the genomic distributions of sequence motifs for different TFs exhibit distinct clustering tendencies. Notably, TF motifs with a high genomic clustering tendency are significantly associated with super-enhancers. TF binding profiles showing a high genomic clustering tendency are further enriched at cell-type-specific super-enhancers. TFs with a high binding clustering tendency also possess high liquid-liquid phase separation abilities. Compared to nonclustered TF binding, densely clustered TF binding sites are more enriched at cell-type-specific super-enhancers with higher chromatin accessibility, elevated chromatin interaction and stronger association with cancer outcomes. Our results indicate that the clustered genomic binding patterns and the phase separation properties of TFs collectively contribute to the formation of transcriptional condensates.