mRNA expression in mammalian cells

Newest from PLoS Biology:

Raj A, Peskin CS, Tranchina D, Vargas DY, Tyagi S
Stochastic mRNA Synthesis in Mammalian Cells.
PLoS Biol (2006) 4(10): e309

The authors genomically incorporated a gene with

32 tandem copies of a 43-base-pair probe-binding sequence at the 3′ end of a coding sequence for a fluorescent protein

into CHO (chinese hamster overy) cells and probed fixed cells with fluorescent oligos (in other words they used FISH). The high signal (32 oligos/transcript) allowed the group to see individual mRNAs. The incorporated gene was under an inducible promoter.

What did they find? RNA is transcribed in bursts. When you look at two genes, the transcription burst is uncorrelated except if the two genes are right next to eachother. Interpretation?

The fact the mRNA is produced in bursts points to new means by which the cell may control transcription. There are three apparent means by which a cell would be able to upregulate a gene's transcription: it could (i) increase the rate of gene activation, (ii) increase the rate of transcription when the gene is in the active state, or (iii) decrease the rate of gene inactivation (the opposite behaviors, of course, apply should a cell decide to downregulate a gene's transcription). These mechanisms, while all resulting in the same average increase in transcription, differ markedly in the nature of the cell-to-cell variations induced. Our data indicate that in our system, either case (ii) or (iii) applies, whereas case (i) does not; in other words, the average burst size is being modulated rather than their frequency. The observation that altering the level of transcriptional activator does not reduce the rate of gene activation supports this hypothesis. Furthermore, the fact that altering the level of transcriptional activator does not reduce the rate of gene activation again argues for the intrinsic nature of the variations observed: if the primary source of cell-to-cell variation is the infrequent events of gene activation and those events are independent of the level of transcriptional activator, then the variations are likely due to some intrinsic fluctuations gene activation that do not depend on transcriptional activators. If gene activation does indeed correspond to chromatin remodeling, this points to the possibility that the nucleation of chromatin decondensation at a gene locus may be an inherently random event that does not require the presence of transcription factors but, once initiated, requires those factors to sustain the decondensed state.

Translation (excuse the pun): You can regulate how often the burst occur (i.e. turning on a gene), the rate of transcription during a burst, or the length of a burst (i.e. turning off a gene). Their data argues that the later two effects play important roles. Furthermore the correlation between the activation of two nearby genes suggests that local chromatin remodeling may play a significant role in regulating groups of genes in one chromosomal neighborhood.