Extended Data Fig. 7. Structural and computational analysis of RPC4-RPC5 extensions.

Schematic domain organisation of human RPC4 and RPC5 with DNA binding predictions below as predicted by DP-Bind⁷⁰. Sumo-modifications of human RPC4 (K78, K141, K190, K199, K206, K220, K285, K288, K302) are marked on the sequence and were obtained from the PhosphoSitePlus database⁷¹. Dashed lines indicate regions that were not visible in the human Pol III cryo-EM maps. Human RPC5 contains, according to sequence homology prediction⁷², four WH domains that are absent in S. cerevisiae. b, Close-up view onto K141 of RPC4 being buried within the interface of RPC4 DM* (shown as cartoon), RPC5 and RPC10 (shown as transparent surfaces). c, Structural model of newly built RPC5 WH1 and WH2 domains binding the Pol III core subunits RPC2, RPAC1, and RPABC5. Corresponding contact points are labelled. d, Conformational dynamics of the WH1-WH2 domains. Left: Cryo-EM maps of human Pol III, in which RPC5 WH domains adopt conformation 1 (map F, top) and conformation 2 (map H, bottom). Cryo-EM maps were low-pass filtered to 6 Å and are shown as transparent surfaces. RPC4 and RPC5 are shown as thick ribbons. Right: Superimposition of the WH domains in the two conformations. The black arrow illustrates their movement. The WH domains in conformation 2 are shown as transparent cartoons e, Phylogenetic analysis of RPC5 and its WH domains, of TFIIIB components Brf1 and Brf2, as well as of the SNAPc genes. The colour code for RPC5 is the same as in a. Coloured and white squares indicate identified and not-identified homologues, respectively. The stars mark species, in which all the SNAPc building blocks were found that are likely required to form a functional SNAPc complex (SNAPc1, SNAPc3, SNAPc4) For further details of phylogenetic analysis see Methods. f, Model of the human Pol III PIC assembled onto the U6 snDNA promoter. Pol III core is shown as surface and RPC5 WH domains, Brf2, Bdp1, TBP and DNA are represented as cartoons. SNAPc is illustrated as transparent sphere. The model was generated using Chimera and Coot by: i) superimposing human TBP from PDB 5n9g onto yeast TBP in PDB 6f42, ii) extending the upstream DNA in 5n9g manually with ideal B-DNA carrying the human U6 snDNA sequence in Coot, iii) superimposing the elongating human Pol III model onto the model of yeast Pol III with Chain C (RPAC1) being used for alignment. SNAPc was placed onto the modelled downstream DNA according to information retrieved from the literature^48,73.