NEW YORK (GenomeWeb News) – Bimodal gene expression — shifts between expression states that are "on" and "off" without intermediate states — can occur in positive feedback loops even in the absence of cooperative transcription factor binding, according to a new paper in Science.
A pair of Massachusetts Institute of Technology researchers developed a synthetic gene expression system in yeast and used it to explore bimodal gene expression. Their results show that — contrary to previous theories — bimodal gene expression doesn't necessarily require multiple transcription factors working together in concert. Instead, the pair reported, it's possible to get bimodal gene expression in the presence of rapidly degraded transcription factors and noisy promoters.
The researchers exploited the tet-transcriptional activator system in yeast to create an open loop system as well as closed loop systems akin to positive feedback loops in which the promoter contained either one or seven tet operator binding sites.
Because the transcriptional activator only binds the tet operator in the absence of the chemical doxycycline in this "tetOff" system, the researchers were able to control activator binding by treating cells with varying doxycycline concentrations. They then monitored expression levels using a fluorescent reporter gene.
In the open system, they found that gene expression increased slowly but surely in a "graded steady state response" with decreasing doxycycline concentrations.
On the other hand, in the closed system, the researchers saw a bimodal expression response when there were seven tet operator binding sites in the promoter and also when there was just one binding site, suggesting cooperative transcription factor binding is not required for bimodal gene expression.
That contradicts the old idea that the binding of one transcription factor makes it easier for other transcription factors to bind the promoter, eventually triggering a switch between expression states, senior author Narendra Maheshri, a chemical engineering researcher at the Massachusetts Institute of Technology, told GenomeWeb Daily News.
"This paper says that that's not how it works," Maheshri said. "You don't need multiple binding sites."
Based on their results, the researchers argued that bimodal gene expression patterns are a consequence of variability or "noise" in the transcriptional system.
If bursts of transcription are large or occur relatively infrequently, Maheshri explained, there tends to be greater transcriptional noise. Multiple binding sites in the promoter seem to increase this noise as well as bursts of gene expression.
In this system bimodal expression also relied on transcriptional activator instability — in this case the tet activator. That ties in to past research in yeast suggesting transcription factors tend to be less stable than other proteins, the researchers noted.
Indeed Maheshri and co-author Tsz-Leung To demonstrated that the tet activator is degraded when transported into the nucleus, with expression patterns shifting from bimodal to graded when they artificially stabilized the activator by tweaking its ubiquitination patterns.
"The hallmarks of noise-induced bimodality in gene expression — positive-feedback loops and unstable proteins — are characteristic of many [transcription factors] and promoters and likely widespread in biological systems," the researchers wrote.
They say the research has implications both for understanding and modeling gene expression networks and for designing such systems. Maheshri noted that they are currently looking at a gene expression network involved in drug response in yeast.
In a perspectives article appearing in the same issue of Science, National Cancer Institute pathology researcher David Levens and graduate student Ashutosh Gupta discussed how the new study could impact researchers' understanding of regulatory circuits and, more broadly, switching events in cells.
There has traditionally been a lot of interest in biostability, the existence of stable expression states, in synthetic biology and biology — particularly biological systems pertaining to development, Levens told GWDN. But researchers often assume that biostability-related switches are permanent, he added, and the extent to which cells revert to their pre-switch states is unclear.
He argued that the new research highlights the importance of considering not only the way noise influences transcription but also how this relates to the stochastic nature of cells overall.
Although the current study uses a synthetic system in yeast to assess gene expression, Levens believes that the phenomenon that To and Maheshri detected will translate to other types of cells and systems. "I think this will be generalizable," he said, calling the new paper "very important."
To accurately assess and understand these transcriptional networks in other cells, though, he explained, researchers will need to find ways to better monitor expression in single cells rather than looking at the average expression across populations of cells.
While such work remains to be done, Levens said that he's convinced there are numerous situations in which noise influences transcription. When it doesn't, he added, researchers should consider the possibility that cellular mechanisms are suppressing this effect.