By the early 1990s, prokaryotic DNA-binding transcription factors were known to directly contact and recruit RNA polymerase subunits to stimulate transcription. However, the situation in eukaryotes was turning out to be more complicated. Initial studies indicated that transcription factors directly recruit components of the basal transcription machinery to activate transcription. But activated transcription could not be reconstituted in the test tube when transcription factors were combined with the known components of the basal machinery. The existence of intermediary proteins was therefore proposed.

In 1991, Dynlacht and colleagues from the Tjian group showed that fruitflycells contain a complex of proteins, which they termed 'co-activators', that associate with the TATA-binding protein and are important for activation in vitro by specific transcription factors, such as Sp1 and NTF1. Subsequently, co-activators were found to also participate in regulatory signalling pathways — for example, Chrivia et al. showed that the transcription factor CREB (cyclic AMP response-element binding protein) recruits the co-activator CBP (CREB-binding protein) during cyclic-AMP-regulated transcription, with phosphorylation of CREB being a prerequisite for co-activator binding.

Biochemical studies by Kelleher et al. from the Kornberg group had shown, in 1990, that transcriptional interference caused by overexpressing a transcription factor (which was presumed to sequester a basal transcription factor) could be relieved by a partially purified fraction from yeast, and that this fraction did not contain a known basal transcription factor. This observation led the authors to propose the existence of a factor termed 'Mediator'.

In genetic studies reported in 1993, Youngand colleagues isolated genes from a screen for mutants that suppressed the defective phenotype caused by truncations in the carboxy-terminal domain of RNA polymerase II (also known as RNA poly-merase B). These 'suppressors of RNA polymerase B', or SRB genes, were shown by Thompson et al. to be required for efficient basal and activated transcription in vitro. In subsequent work, Koleske and Young found that a preparation of RNA polymerase II purified from yeast contained the SRB proteins and general transcription factors in a holoenzyme that could mediate the response to activators in vitro.

A key breakthrough came in 1994 with the difficult purification of the Mediator complex by Kim et al. from the Kornberg group. They were able to reconstitute activated transcription in vitro with purified components, finally demonstrating the elusive Mediator activity. There were two particularly surprising findings. First, the Mediator was a huge complex of about 20 subunits that associated with RNA polymerase II. Second, the same SRB proteins identified by Young and colleagues were found to be components of the Mediator complex. Thompson and Young subsequently made the important observation that the Mediator is generally required for transcription in yeast, and we now know that it is also conserved in mammalian cells.

The discoveries of co-activators and the Mediator complexes have led to our present appreciation of the enormous intricacy and diversity of eukaryotic transcription, with large multisubunit complexes integrating the regulated function of transcription factors with the core transcription machinery.