Our Random World
In confirmation of my suspicions, transcription initiation seems to be more randomized and probabilistic than people had previously thought.
One theory that had been proposed for a while was the idea of “transcriptional oscillation”, where transcription factors would hop onto a gene, transcribe for a little while (20-30 minutes or so), hop off, then start the cycle over again.
There was interesting data showing all this, which was nicely summarized by Métivier et al.1 in the following graph:
The close details of the plot aren’t that important; the main point is that a lot of the machinery for transcriptional initiation seemed to oscillate with a cycle on the order of 30 mins. The plot took a heck of a lot of work to generate; each data point on a curve is one chromatin immunoprecipitation (ChIP), and in one of the group’s papers, they took a time point every five minutes. It must have been a nightmare of a project to make work, considering everything that could go wrong.
Now, the data is interesting, but their interpretation of it goes wrong. Their model supposes that this sort of mechanism is very regulated, that the cycles occur because factors hop on, do their thing, and get knocked off in a very timed, controlled manner.
Deterministic models like this make me sad. Sure, videos like “The Inner Life of a Cell” make biology look pretty, showing a smoothly operating machine in which motors step along nicely, molecules steadily march along tracks in the right direction, and everything swooshes together into assemblies that make things happen. But molecules just don’t work like this! Take a look at the elongation of RNA polymerase in the plots below from Stephen Block’s lab:
This is probably the smoothest anything ever gets in the cell. RNA transcription is fast, the elongation complex is super-stable (with a half-life of weeks), and not much is going to stop the polymerase. And yet, there’s still so much noise!
At long last, a new paper in Science2 shows that the deterministic model (surprise!) might not be accurate! These authors use FRAP (fluorescence recovery after photobleaching) in order to show that the occupancy of promoters (at least in yeast) is very transient (on the order of 30 seconds).
They also show that this is how transcription initiation happens by using a clever experiment. If a few promoters were stable (in the sea of noisy binding) and responsible for the majority of transcription, then with ten copies of a gene, one particular copy would be transcribed only in about 1/10th of the cell. What the authors show is that any one particular copy gets transcribed in basically all of the cells, which means that the noisy bouncing on-and-off of transcription factors is how genes work, and that it isn’t something that’s obscuring a more stable, deterministic mechanism underneath.
On the other hand, they too find the slow-cycle oscillations when they do ChIP and RT-PCR analysis of the promoter, so clearly, there must be some sort of longer time-length phenomenon. Since this reflects an average phenomenon of all the noise, it must mean that some sort of global energetics is changing, something that would be at a slower time scale than just the binding of some other protein (which would seem to have the same noisiness and time scale as a transcription factor); the authors hypothesize that something like chromatin remodeling is happening, and I’m inclined to agree with them.
There’s more on nuclear dynamics within cells in the next blog post, on a paper that goes after the theory of chromosome territories!
—-
1. Métivier, R., Reid, G., Gannon, F. (2006). Transcription in four dimensions: nuclear receptor-directed initiation of gene expression. EMBO reports, 7(2), 161-167. DOI: 10.1038/sj.embor.7400626
2. Karpova, T.S., Kim, M.J., Spriet, C., Nalley, K., Stasevich, T.J., Kherrouche, Z., Heliot, L., McNally, J.G. (2008). Concurrent Fast and Slow Cycling of a Transcriptional Activator at an Endogenous Promoter. Science, 319(5862), 466-469. DOI: 10.1126/science.1150559
