Fig. 5: Three crosslinking hotspots identified on TAF1 correspond to distinct anchor points for specific TFIID building blocks.

a, Summary of TAF1-centred crosslinking-MS metanalysis derived from three independent studies^10,18,19. Interprotein crosslinks between TAF1 and other TFIID subunits (TAFs are indicated with their corresponding numbers) are displayed along the protein as dotted lines or shaded ranges. Numbers above each crosslinked subunit indicate TAF1 positions or the range involved in the crosslinks, and the values in brackets indicate the number of distinct TAF1crosslinked positions. Only crosslinks reported in at least two independent datasets are shown. The known structural domains of TAF1 are depicted. b, Heatmaps of TAF1 conservation (ConSurf), structural disorder (Metapredict) and pLDDT (AlphaFold structural prediction confidence) scores are shown. Structures with pLDDT > 90 are expected to be modeled to high accuracy. A pLDDT between 70 and 90 means that the structure is modeled well, and structures with a pLDDT between 50 and 70 should be treated with caution. The scales of the different prediction scores are shown on the right. c, Full-length human TAF1 AlphaFold model. The initial structure was extended to better appreciate the different domains indicated along the protein. d–f, Structural models of the three main TAF1 anchor points (labeled A, B and C) in TFIID are shown in the insets. The model in d is the result of AlphaFold prediction of the TAF1–TAF11–TAF13–TBP subcomplex. Distinct TAF1 domains are colored as in a. Partner subunits are shown in shades of gray. TAF1 segments part of each anchor point are indicated. g, Schematic summary of the distinct interaction hotspots along the protein and the length of the intervening linker regions. TFIID lobes A and B are shown as circles. aa, amino acids; HMG: HMG-box domain; ZnK: zinc-knuckle domain.