Synthetic probes of natural oscillations

One of the motivations for systems biology is the gathering realization that biological systems are not simply composed of on/off switches.  Instead of thinking of signal transduction as a simple relay race — A passes the information to B, who passes it to C — we need to understand the information processing in multiple layers of feedback and feed-forward loops.  The dynamics of the components of the pathway are our best window onto the behaviors of these loops.  But the complexity of natural systems is such that interpreting protein dynamics is often not easy.  The Lahav lab has been chewing away at one such problem for a while: the dynamics of the transcription factor p53, one of the body’s most important defenses against cancer. After DNA damage, the p53 network is responsible for making the decision of whether the cell should arrest the cell cycle and attempt to repair the damage, carry on, or die.  Most cancer cells have lost the ability to choose correctly when faced with this situation.

When DNA is damaged by gamma-irradiation, p53 undergoes a series of pulses with interesting characteristics: their height does not depend on the dose of irradiation, and their width is constant.  Individual cells may show variable numbers of pulses, but the average height of the 5th pulse is no different from the average height of the first pulse.  This series of pulses is sometimes talked of as oscillatory behavior, but it’s not really clear whether that’s the right way to think about it; an oscillation, to me, is something you kick off once and it keeps on going, whereas there’s evidence that in the case of p53 the pulses may be repeatedly initiated by DNA damage that hasn’t yet been repaired.  Given how complex the p53 network is — I show a small section of it in the left panel of the figure, but the list of proteins that interact with p53 is about 100 long — how are we going to be able to understand what controls the pulses?

The Lahav lab’s response to this follows Thoreau’s mantra — simplify, simplify.  In a new paper (Toettcher et al. 2010 A synthetic-natural hybrid oscillator in human cells  Proc. Natl. Acad. Sci. doi: 10.1073/pnas.1005615107) they decided to take out all the feedbacks except the key negative feedback loop between p53 and Mdm2, then add back synthetic feedbacks and determine how the behavior changes.  The simplest semi-synthetic system they built is shown on the right.  Zinc activates a zinc-inducible metallothionein promoter (P-MTF1), which drives the expression of a CFP-labeled p53.  Induction of Mdm2 expression by p53 remains normal, and Mdm2 induces the destruction of p53 via the ubiquitin pathway.  There are no other important feedback loops (that we know of) active in the system.  How does this very simple circuit behave?

Well, it’s interesting.  You get oscillations, and they have the right timing.  But, unlike the oscillations driven by the full circuit, the size of the pulses gradually declines with time; in other words, the oscillations are damped.  And, again unlike the natural system, the synthetic system is dose-responsive; higher zinc concentrations lead to higher initial pulses.  Toettcher et al. put together a mathematical model for the zinc-driven oscillations, and used it to ask in silico what would happen if you added back different kinds of feedback on p53.  They found that extra feedbacks would only weakly change the timing of the oscillations (which is good, because the timing is already correct in the minimal circuit), but would have a strong effect on the damping of the oscillations; positive feedback should reduce the damping, and negative feedback should increase it. [Now, bear in mind that I didn’t show you a positive feedback on p53 in the diagram of the circuit above; we don’t (yet) know of a positive feedback that contributes to the dynamics.  But perhaps the modeling is telling us that we should look for one.]

To test these predictions, they created artificial positive and negative feedbacks on p53.  For the positive feedback, they put the MTF1 transcription factor — which binds zinc and activates the P-MTF1 promoter — under a p53-dependent promoter.  When the synthetic p53 circuit is activated by the addition of zinc, p53 goes up and MTF1 production is increased; then the MTF1 induces the production of more p53.  For the negative feedback they used the same basic strategy, this time using a variant of MTF1 that includes a dominant repressor domain.  As predicted, the positive feedback considerably slows the damping of the oscillations, while the negative feedback accelerates damping; and no effect is seen on timing.

What do you have to do to change the timing?  Using the model, Toettcher et al. determined that changing the affinity between Mdm2 and p53 should do the job.  This is not easy to do at the protein level, but fortunately a small molecule, Nutlin3A, is available that interferes with Mdm2-p53 binding.  Using it, the authors are able to decrease the frequency of oscillations by about 20%, close to the model prediction.

The authors argue that this mixed synthetic-natural approach has the potential to help understand which features of a natural network are crucial, and for what.  Strictly synthetic networks are relatively easy to understand but may not teach you much about the behavior of the real system; and the natural pathway may be too complex to understand all at once.  But the compromise approach of combining synthetic elements with natural elements may, as in this case, help you identify informative ways of perturbing the system.  The p53 community now has a way to change the timing, dose responsiveness and damping of p53 oscillations; the next question will be whether these new perturbation methods can provide insight into what information is encoded in the dynamics of the system.  To be continued.

Toettcher JE, Mock C, Batchelor E, Loewer A, & Lahav G (2010). A synthetic-natural hybrid oscillator in human cells. Proceedings of the National Academy of Sciences of the United States of America PMID: 20837528