The authors of this recently published paper in nature structural & molecular biology (http://www.nature.com/nsmb/journal/v20/n2/full/nsmb.2472.html) provide many arguments against contribution of facilitated diffusion (1D hopping/sliding along the DNA) as a promoter search mechanism for Escherichia coli RNA polymerase. According to them the contribution of 3D diffusion, especially at physiological protein concentrations outweighs the contribution of any form of facilitated diffusion.
Their experimental system involves a curtain of λ dna molecules tethered at both ends in the same orientation. Using quantum dot tagged RNAP they were able to visualize the RNAP molecules at the DNA curtain using TIRFM. Based on the lifetimes of the quantum dot labeled single molecules of RNAP they discriminate various intermediates: (in order of increasing lifetimes) random diffusion in absence of DNA interaction, random interactions with DNA, closed complexes and open complexes. They find that most events where RNAP engages the promoter were preceded by 3D diffusion and 1D diffusion was virtually not seen.
They also come up with a theoretical model to determine the significance of contribution of the various forms of diffusion to promoter search. They find that with greater concentrations of the protein, 3D diffusion overcomes any possible accelerating effects of 1D diffusion and thus come up with the concept of ‘facilitation threshold’, the concentration of (any) DNA-interacting protein below which facilitated diffusion would be faster in target search than 3D diffusion. They surmise that for the levels of RNAP in the cell 3D diffusion would be a faster mechanism for promoter search.
To demonstrate the significance of facilitation threshold experimentally they use the lac repressor and insert tandem lac operator sequences in the λ DNA curtain. Under conditions where non-specific DNA binding and hence facilitated diffusion is favoured they see that the lac repressor at low concentrations engages its operator mainly by 1D diffusion, however when the concentration of the repressor was increased there was an increase in the number of events in which operator binding was preceded with 3D diffusion of repressor rather than 1D diffusion clearly adding weight to the concept.
Finally the authors also discuss how under various in-vivo conditions seen by the RNAP like presence of nucleoid associated proteins and higher chromatin architecture as well as molecular crowding why 3D diffusion would be a more prevalent mechanism for promoter search rather than 1D diffusion.