Isolation and partial characterisation of the relaxation protein from nuclei of cultured mouse and human cells.

A protein, called relaxation protein because of its ability to remove superhelical turns in closed-circular DNA, has been isolated and partially characterized from the nuclei of LA9 mouse and HeLa cells. The purification was facilitated by an assay method, with PM2 DNA, which used the fluorescence enhancement of the intercalating dye ethidium bromide upon binding to the closed-circular DNA. The amount of dye bound depends upon the degree of the superhelix density of the DNA. The relaxation products were analysed by the buoyant separation method in CsCl containing ethidium bromide and were shown to be completely relaxed. The purification resulted in a single band in a dodecylsulfate gel electrophoresis with an apparent molecular weight of 37000. The pH optimum is 7.0 and the optimal salt concentration is 0.2 M NaCl. The relaxation protein removes negative as well as positive supercoils, the latter generated by the interaction of ethidium bromide with closed-circular DNA. Relaxation of positive supercoils results, after removal of the dye, in the formation of molecules with superhelix densities exceeding that of native PM2 DNA (0.054). The highest negative superhelix density observed was -0.098 +/- 0.001. The corresponding positive superhelix density has been calculated to be + 0.023. A nicking--swivelling--closing mechanism is postulated, but nicked intermediates have so far not been demonstrated. The relaxation protein is not inhibited by known mammalian endonuclease I inhibitors, except for denatured DNA, and does not possess a conventional polynucleotide ligase activity. The relaxation activity was found to be predominantly in the nuclei, with only small amounts present in the cytoplasm and mitochondria. The biological function of transient swivels induced by the relaxation protein is not known. However, transient swivels are considered necessary or useful in the replication of closed-circular DNA or long linear DNA, respectively. Relaxation protein could replace the combined action of an endonuclease and a ligase ahead of the replication fork. Alternatively, transient swivels could be involved in the transcription process.