Chat with Dr. Tom Ellis

4/30/2015 Update: Watch the complete Q&A here.

On Thursday April 30 at 11 AM Eastern Standard time (4 PM GMT), we will be hosting Dr. Tom Ellis, a professor in synthetic biology at Imperial College London in a Google hangout webinar to discuss challenges in synthetic biology and most importantly, those addressed by a recent Nature Methods publication from his lab on an in vivo method to assay the burdens posed by synthetic constructs and genes in E. coli.

A note from Dr. Ellis:

Synthetic biology has come on leaps and bounds in the past decade, but it is still a challenge to predict how a synthetic construct will behave when in a living cell. One of the main reasons it’s so hard to predict how a construct will behave is that most approaches only consider how the different parts of each construct interact with one another and don’t take into account how they interact with the cell’s own machinery. Even if a construct is designed to be wholly orthogonal with no specific interactions with the cell’s own genes, it still uses up precious cell resources just for maintenance of its synthetic DNA and expression of its genes. Countless researchers witness this every day when they see smaller colonies than usual for many of their engineered cells; the addition of synthetic DNA is visibly placing a burden on the cells and this is seen time-and-time again as slower growth rates.

To put more precision on this important phenomenon we developed a new method to measure the burden different designs place on E. coli cells and used this to assess a variety of typical synthetic biology and iGEM constructs as well as a library of different designs of the same synthetic construct, an enzyme expression construct typical of that used in industrial biotechnology. This work gave us interesting new data on the cost different parts impose on the gene expression capacity of a growing bacteria and pointed out some important phenomena too.

Our method exploits the fact that gene expression resources are universally shared around a bacterial cell, so that depletion of the resources due to one gene being suddenly being turned on quickly causes a shortage of resources for all other genes. We simply placed a constitutive synthetic GFP gene in the genome and watched the rate of GFP production from this over a couple of hours while synthetic constructs added on plasmids were turned on and off using inducers. In an analogy to electrical engineering, it was like we were watching the brightness of our bedroom light dim while someone turns the washing machine, the TV and the iron on downstairs. We can infer from the dimming what the resources are that are used by the different systems.

It’s a very accessible method, taking only a few hours with a plate-reader. But what’s great is how much is revealed by making measurements of the cost to capacity of different designs. We learned that the decrease in gene expression capacity due to increased burden happens quickly and precedes (and actually predicts!) future decreased growth rates. We also learned that longer synthetic constructs and higher plasmid copy numbers aren’t really much of an issue for the cell, what really matters is the how many ribosomes are being used. So when we looked at a library of different designs of the same construct we saw some really interesting results. Two versions of a construct producing similar amounts of the same enzyme end up doing so with very different burden. Counter-intuitively, it is the design with the higher plasmid copy number (more DNA per cell) but weaker RBS (less translation per mRNA) that has less burden than the lower-copy, strong RBS design. By bringing in an elegant model and Python simulation of the process of translation we were able to work out that this was because strong RBS designs overload mRNAs and this is an inefficient sequestering of the cell’s precious ribosomes. It seems that the main effect of burden we see is all to do with how many of the cell’s free ribosomes a construct takes up. The free ribosome pool is really critical to optimal expression and growth.

There are many implications from the work; how to optimise DNA design, the effects of codon choice, how growth decreases predicts inactivating mutations, what parts can be used that are less cost for the cell, and lots lots more – way more than we could fit into a short paper. It’s exciting because burden is something a lot of people in synthetic biology are aware of and everyone in the field is looking for answers that will help the predictability of their designs and better long-term performance of constructs at scale-up. There are a lot of avenues this work could go down next and we really hope others begin to use our method and with it uncover more important information too.

As always, we want your inputs and participation! The webinar will be hosted here, and send us your comments and questions at synbiocommunity@plos.org, tweet at us at @PLOSSynbio (with #EllisChat), in the comment box below, or on the Hangout-on-Air event page (LINK, on the top right corner of the video labeled Q&A)!