It's easy to forget that somewhere between the now high-tech fields of genomics and transcriptomics once lay the fundamental problem of how DNA is transcribed. Perhaps most important was the question of how RNA polymerase is guided to the right promoters and prevented from binding unproductively to random sequences. This question was answered almost 40 years ago for bacteria with the finding that sigma factors lead RNA polymerase to its target sequences, thereby kick-starting a flurry of biochemical studies of transcription.

In 1969, knowledge about bacterial transcription was based largely on in vitro assays using purified RNA polymerase and a (typically bacteriophage) template DNA. Such experiments had shown that the polymerase was specific about the sequences it transcribed in vivo, but was less choosy in vitro — in fact, its specificity in vitro varied depending on the template used and how the polymerase was purified.

The solution to this confusing state of affairs depended on having a reliable means of separating an active RNA polymerase into its components and ascribing a function to each of them. Richard Burgess and Andrew Travers at Harvard, and John Dunn and Ekkehard Bautz at Rutgers, did just that: they purified Escherichia coli RNA polymerase on a phosphocellulose column and asked which combination of fractions, and then which factor in particular, conferred transcriptional activity to the core enzyme on T4 phage DNA.

This was the landmark paper that identified that the selective fraction was a protein — a result that was achieved largely thanks to the ability of a new biochemical technology, SDS–PAGE, to separate the protein components by size. Yet how this protein, which they called sigma, worked could only be speculated about: did it simply increase the affinity of the enzyme for any DNA, or was it more specific? And how and where did it associate with the enzyme?

In a second paper, Travers and Burgess showed that sigma increased the number of RNA chains that were initiated by the polymerase. In particular, as Dunn and Bautz pointed out in a third study, sigma functions just before the first RNA phosphodiester bond is formed. In experiments that were to alter how we view regulatory protein–protein interactions, Travers and Burgess then showed that sigma is released by the polymerase after initiation and then reassociates with another polymerase enzyme in a cyclical fashion. These conclusions were reinforced by work in a different system by Joseph Krakow and colleagues, who reported that a protein component equivalent to sigma is released from the polymerase of the bacterium Azotobacter vinelandii.

These studies — along with work from the groups of Michael Chamberlin and Wolfram Zillig, who reported further evidence for sigma initiation factors at about the same time — spurred decades of study into promoter recognition and transcription initiation. Although a different mechanism operates in eukaryotes (see Milestone 12), the cyclical model of transcription has been amply validated and expanded in recent years, both as a general strategy for how auxiliary factors operate in transcription and through the identification of a family of prokaryotic sigma factors that determine the different promoter specificities of core polymerase enzymes.