An atlas of transcription factor cooperation reveals how motif readers shape regulatory output

Inferred mechanisms of TF-motif dependencies converge on three operating routes: direct sequence recognition, protein-mediated recruitment or exclusion, and regulatory context. Whereas motifs are conventionally associated directly with named TFs, this work uses ARES to test competing mechanisms in specific cellular contexts, expanding mechanism inference from single sequence recognition to a multi-route framework including recruitment, exclusion, and context. Based on multi-omic analysis of 1,552 TF binding datasets across 10 cell types; a large-scale computational inference, with no specific statistics or effect sizes provided in the text.

Predictive motifs of target TF binding were read by their conventionally 'canonical' TFs in only one third of resolved dependencies, and these 'canonical' TFs were expressed much less often than the inferred readers. This directly challenges the conventional assumption that a motif label identifies the reading protein, showing that motif identity and reader identity diverge in most resolved dependencies. Based on statistics over resolved dependencies; the text gives the proportion 'only one third' but no confidence intervals or test details.

Motif similarity was associated with shared regulatory region type but not shared transcriptional outcome, whereas reader identity was associated with both and was the only feature among those examined associated with outcome. This decouples motif sequence similarity from functional output, indicating that reader identity rather than the motif itself is the key feature predicting transcriptional outcome. Based on association analysis of multi-omic data; the text reports no specific correlation coefficients or model performance metrics.

In validation cases, perturbing an inferred reader reduced target TF occupancy in proportion to reader binding before perturbation, and a natural variant disrupting the predictive motif altered target TF binding at every intermediate step of the inferred mechanism. Perturbation experiments and a natural variant validate the inferred mechanisms, linking computational inference to causal perturbation evidence. Validation case studies, further supported by single-cell perturbation, in vitro cooperativity, and evolutionary constraint, though the text provides no sample sizes or effect sizes.

This work provides a framework for reinterpreting motif function in specific cellular contexts, useful for researchers studying TF binding and gene regulation, especially in cell types with available multi-omic data. It supports viewing motifs as addresses and readers as trans factors determining regulatory outcomes, guiding future perturbation experiments and variant interpretation.

Readers may wonder about the specific algorithmic details of ARES mechanism inference, the criteria for the three routes, and the statistical uncertainty of the 'only one third' proportion; additionally, sample sizes and effect sizes of validation cases and the specific designs of single-cell perturbation and in vitro cooperativity are not expanded in the summary, which are open questions for understanding the robustness of the conclusions.