Synthetic serendipity

Synthetic biology is a good candidate for the area of biology that will make the biggest difference to humanity in this century.  Yes, genomics and neurobiology are contenders; stem cell biology, too; and of course, I would argue systems biology is up there.   But when you think about harnessing the power of biology to build things we want — and you remember that this is a technology that has already terraformed our planet — it’s hard not to get excited.

The enthusiasm around synthetic biology can be most clearly seen in the International Genetically Engineered Machines competition (iGEM).  iGEM began as a month-long course at MIT, and grew into a national, then international competition.  This year about 180 teams are working over the summer to produce useful or just cool devices using biology.  Recent teams have worked on making brightly colored bacteria (with the goal of expanding the range of available biosensors), a system to sense and remove mercury, bacteria that encapsulate and deliver drugs, a vaccine for Helicobacter pylorii, and many, many more cool projects.  (Harvard’s team this year is working on developing tools to let you personalize your garden: the goal is to allow you to introduce any flavor you want into chosen plants, and to remove any allergen that bothers you.)  The teams are mostly made up of undergraduates and even high school students, and their enthusiasm and excitement is awe-inspiring. Key to the iGEM philosophy is that biological “parts” — called BioBricks — can be defined, characterized and re-used in multiple devices.  Without well-defined parts, each device would have to be designed and built from scratch.  With standardized parts, there is at least a chance that what is learned in one setting can be applied to other settings.  As well as competing to make a device that performs a defined function, iGEM teams also aim to add new, widely useful, parts to the BioBrick Registry.

Exciting though the potential of the approach is, it isn’t always easy to persuade biology to do what you want it to.  But this is not necessarily bad.  A corollary to Richard Feynman‘s famous mantra “what I cannot create, I do not understand” — is this: what I set out to create, I may discover new information about.  A recent article in PLoS One (Marguet et al. 2010. Oscillations by minimal bacterial suicide circuits reveal hidden facets of host-circuit physiology. PLoS One. 5 e11909. PMID: 20689598) provides an example.

What Marguet et al. set out to do was to create a circuit that would cause oscillations in the density of bacterial populations.  And they did.  Unfortunately — or not, depending on your point of view — it didn’t work the way that they had planned.

The original idea was to link up a quorum-sensing module, the luxI/luxR module from Vibrio fischeri, to a toxin gene.  LuxI is responsible for producing a small-molecule signal, acyl-homoserine lactone (AHL); as cell density goes up, the concentration of AHL goes up, and eventually there is enough to trigger LuxR to become active and act as a transcription factor.  In the circuit Marguet et al. built, the LuxR-responsive gene is the E protein, which inhibits bacterial cell wall synthesis and causes cell lysis.  So, the idea was that as the population of bacteria grew, AHL would build up gradually.  At a certain population density it would trigger LuxR, the E protein would be produced, most of the bacteria would die, cell density would go down so low that the AHL levels were no longer high enough to induce the E protein, and the whole cycle of growth, triggering of toxin, and death would start again.  They called this a synthetic suicide circuit.  And it worked: E. coli cells carrying this set of genes on a plasmid showed unmistakable oscillations in cell density.

The problem was that when they came to look more closely at the circuit, they found that the oscillations weren’t dependent on the quorum sensing mechanism.  Indeed, the version of LuxR they were using turned out to be accidentally truncated by a frame-shift mutation, and didn’t have a DNA binding domain at all.  The oscillations appeared to be caused by a mechanism that had little or nothing to do with the intended behavior of the circuit. This must have been rather an upsetting discovery; and it must have taken some determination to keep on with the project and figure out what was really going on, instead of quietly throwing the whole thing away and pretending it never happened.

After a bit of detective work, Marguet et al. decided that an unknown mechanism was at work that caused the plasmid on which the circuit was carried to be amplified in a density-dependent way.  Since each plasmid carries a copy of the gene encoding the E protein, increased numbers of plasmids mean more copies of the E gene.  And if the promoter controlling the E protein is just a little leaky, the increase in gene copy number would in turn lead to high levels of E protein, and thus cell death.

A plausible mechanism for the density-dependent increase in copy number draws on three sets of published observations: (1) the plasmids frequently used for genetic engineering are impaired in a key mechanism for controlling plasmid replication, inhibition of replication initiation by the RNA I/RNA II system; (2) amino acid starvation can lead to an accumulation of uncharged tRNAs; and (3) uncharged tRNAs can degrade RNA I, further impairing the control of plasmid regulation. In this model, what switches off the expression of E protein is not the reduction in cell density due to cell death, but the release of nutrients from dead cells; now the remaining cells are no longer starved for amino acids. [They could call it a synthetic cannibal circuit.] A mathematical model of the proposed cycle gives good agreement with what they see.  The model also predicts that chloramphenicol should further exacerbate the problem of unconstrained plasmid replication by blocking a second set of control mechanisms that prevent the buildup of uncharged tRNAs, and indeed addition of chloramphenicol seems to increase the frequency of density oscillations.

The idea that cell density can, for some plasmids, cause an increase in per-cell plasmid burden may or may not be interesting to you depending on the role that plasmids play in your life.  The larger point Marguet et al. draw from their journey through serendipity is this: when treating biology as an engineering technology, it’s tempting to assume that you can predict not only the functions of circuit components, but also the interactions among components and between the components and the cell that hosts them.  In this synthetic suicide circuit, plasmid replication was a “background process” that they thought they could ignore.   That turned out not to be true.  Context is always going to be important in engineering biology, and however carefully a biological module has been characterized in one system, it may have surprises for you when it’s moved into another system.  But that may not be bad.  As Marguet et al. put it: “Every circuit whose real-world behavior varies dramatically from our best predictions represents an opportunity to better understand the components, interactions, and control mechanisms of both the system and the host.”  And thus, perhaps, find ways to build even better circuits next time.

Marguet P, Tanouchi Y, Spitz E, Smith C, & You L (2010). Oscillations by minimal bacterial suicide circuits reveal hidden facets of host-circuit physiology. PloS one, 5 (7) PMID: 20689598