Sequestering of p53 into DNA–protein filaments revealed by electron microscopy Original Research Article

Using electron microscopy, we analyzed the interaction of bacterially expressed full-length p53, p53(1–393), and its C-terminal fragment, p53(320–393), with long (∼3000 bp) dsDNA in linear and supercoiled (∣ΔLk∣≈4–6) forms containing or lacking the p53 recognition sequence (p53CON). The main structural feature of the complexes formed by either protein was a DNA–protein filament, in which two DNA duplexes are linked (synapsed) via bound protein tetramers. The efficiency of the synapse, reflected in its length and the fraction of molecules exhibiting DNA–protein filaments, was significantly modulated by the molecular form of the protein and the topological state of the DNA. With linear DNA, the synapse yield promoted by the C-terminus fragment was very low, but the full-length protein was effective in linking noncontiguous duplexes, leading to the formation of intramolecular loops constrained at their bases by short regions of synapsed DNA duplexes. When the linear DNA contained p53CON, regions of preferential sequence, i.e., encompassing p53CON and probably p53CON-like sequences, were predominantly synapsed, indicating a sequence specificity of the p53 core domain. With scDNA, the synapse yield was significantly higher compared to the linear counterparts and was weakly dependent on the sign of superhelicity and presence or absence of p53CON. However, the full-length protein was more effective in promoting DNA synapses compared to the C-terminal fragment. The overall structure of the DNA–protein filaments was apparently similar for either protein form, although the apparent width differed slightly (∼7–9 nm and ∼10–12 nm for p53(320–393) and p53(1–393), respectively). No distortion of the DNA helices involved in the synapse was found. We conclude that the structural similarity of DNA–protein filaments observed for both proteins is attributable mainly to the C-terminus, and that the yield is dictated by the specific and possibly nonspecific interactions of the core domain in combination with DNA topology. Possible implications for the sequestering of p53 in DNA–protein filaments are discussed.