Seeing is believing

Elucidating the molecular mechanisms that underlie the control of initiation of transcription in animals has been central to our research for almost 40 years. The first breakthrough in this field was, at the end of the 1960s, the independent findings by Roeder and Rutter and by our own laboratory of multiple forms of nuclear DNAdependent RNA polymerases (Chambon, 1975). Their extensive purification and the determination of their subunit structure, which were completed in the early 1970s, opened the era of in vitro functional studies aimed at investigating their role in the selectivity of transcription in vivo. However, we quickly established that these purified enzymes were unable to initiate RNA synthesis on either intact linear double-stranded animal viral DNAs or high molecular weight cellular DNAs devoid of singlestranded breaks, which Maria Gross-Bellard prepared in our laboratory (see references in Chambon, 1975). Speculating that chromatin rather than naked DNA was likely to be the natural template in vivo, we decided to isolate chromatin in a state as native as possible. To this end, one of us (P.O.) set up in our lab an electron microscopy (EM) technique that had just been devised by Jacques Dubochet in Edouard Kellerberger's group in Basel for the direct visualization of protein-DNA complexes on charged grids without a fixation, under conditions where stretching and modifications of DNA-protein interactions were clearly minimized (Dubochet et al., 1971). Previous electron microscopic and physical studies published in the early 1970s had indicated that the basic structure of chromatin corresponded to an approximately 100 A thick fiber. However, there was no definitive information concerning the substructure of this fiber, which might correspond to the DNA double helix with attached histones (see Oudet et al., 1975, for references). As X-ray data had indicated that, unlike the four histones H3 (F3), H4 (F2al), H2A (F2a2), and H2B (F2b), the lysine-rich histone H1 (termed F1 at that time) was not essential for the basic chromatin structure, we initially used HI-depleted chromatin preparations, which are more easily dispersed. We still vividly remember our excitement when Pierre showed us in the fall of 1973 his first electron micrographs revealing that the fundamental structure of our chromatin preparations looked like a flexible chain of 120-130 A spherical particles