Engineered Post-Translational Logic (PTL) Samantha C. Sutton , Sara E. Neves , Lauren W. Leung , and Drew Endy Division of Biological Engineering and Abstract Department of Biology, Massachusetts Institute of Technology Modeling as a Design Tool Current synthetic biological circuits make use of protein-DNA and RNA-RNA interactions to control gene expression in bacteria. Systems that rely on the regulation of gene expression are relatively slow and unsuitable for many applications. Here, we describe our work to engineer synthetic biological systems in yeast using posttranslational modifications of proteins to define system state and control cell function; such systems should have faster performance time and enable a wider range of applications. We have specifically chosen to focus on building phosphorylation-driven protein circuits. We modeled a specific instance of a post-translational circuit using methods such as Lyapunov exponents, and showed that the circuit should behave as desired within a large parameter space. We developed a set of peptide tags that can be used to drive the phosphorylation of a chosen substrate by a desired mitogen-activated protein kinase (MAPK). Each phosphorylation event alters a substrate output activity, such as translocation, degradation, or other binding event. These tags were developed using the Phospholocator – a construct whose phosphorylation-mediated translocation is controlled by MAPK activity. Specifically, MAPK phosphorylation of the Phospholocator nuclear localization sequence (NLS) controls recognition of the NLS by cellular import machinery. The Phospholocator serves three purposes: to determine the docking sites of MAPKs of interest, to measure the in vivo activity of such MAP Kinases, and to serve as a first set of post-translational logic parts. Currently, we have built a version of the Phospholocator that is targeted by Cdc28; our next step is to build Fus3-, p38-, and Hog1activated instances. Building and Testing a PTL Device Example PTL Device: Flip-Flop In1 Necessary Components of a PO4-MAPK Part MAPK binds here Out1 Kinase MAPK adds phosphate group here P Docking site In2 State 1 P P 0 0 1 1 P Kinase In2 Kinase Out1 NLS Substrate Protein State 2 Out2 hold hold 0 1 1 0 not allowed 0 1 0 1 PO4 site Out2 Kinase In1 NLS P Kinase P P Kinase When unphosphorylated: Import machinery binds the NLS and brings the device into the nucleus. When phosphorylated: Import machinery cannot bind the NLS, and thus the device remains in the cytosol. The Phospholocator Differences between PDL and PTL Flip-Flops PDL Flip-Flop PTL Flip-Flop Cdc 28 Docking site Swi5 NLS PO4 site Swi5 NLS YFP + -gal Kinase Why Post-Translational Logic? Kinase Our Goal: Engineering Biology Physics • Unlimited species concentrations • Use hill coefficient to describe cooperativity • Can make Pseudo-steady state assumption about protein-DNA binding Electrical Engineering P k1 • Protein-DNA (Transcriptional) logic (PDL) – Engineered around gene expression – Easier to engineer – Slow response time (hours) – Uses one subset of cellular functions + P k1 P k3 + k2 # of fixed points An Example of a PTL Device Activity of P Kinase Kinase Activity of Choice of Modification and Enzyme + 3 k1, k3 fixed Activation Nuclear localization Degradation Binding Powerful tool. Directly control kinase activity PTL Flip-Flop is Robust to Concentration Fluctuations 5 Inactive Cons May involve engineering 3o structure Good visualization, local Must be compatible with expertise, previous translocation machinery, examples of modular screen sensitivity engineering Good assays, well studied system, good screen No examples of modular engineering, confounding fluctuations in expression Well-studied, local expertise Less interesting function • Enzyme of choice: MAP Kinase • Signaling pathways • Well-studied •Yeast has two well-known MAPKs: Fus3, and Hog1 • Examples of modular MAPK docking sites Phospholocator-PO4 • We used Matlab to vary k1, k2, k3, k4 over biologically relevant values, and then used fsolve to locate the fixed points. Shown above is the number of fixed points obtained for different values of k2 and k4 (k1 = k3 = 10-4 (nM s)-1). •Three fixed points can indicate a functional flip-flop, while one cannot. • We computed the Jacobian of the system evaluated at each fixed point, and determined the corresponding eigenvalues. •Two fixed points are asymptotically stable because they have all negative eigenvalues. •The remaining fixed point is an unstable fixed point because it has one positive and three negative eigenvalues, indicating it has 3D stable and 1D unstable manifolds. 15000 10000 Verification of Phosphorylation 1 k2 Active [ ]o divergent (high Lyapunov exponent) 5000 Nocodazole arrest (G2/M): cytosolic • Cdc28-Cln2 is active during late S and G2/M phase in yeast. In cells arrested with nocodazole, Cdc28 should be active, and phosphorylate the Phospholocator. The Phospholocator should then be cytosolic. • Cdc28-Cln2 is inactive during G1 phase in yeast. In cells arrested with pheromone, Cdc28 should be inactive, and unable to phosphorylate the Phospholocator. The Phospholocator should then be nuclear. Nocodazole Pheromone • Modification of choice: phosphorylation •Best studied phospho-mediated functions Pros P k4 k4 Function Pheromone arrest (G1/S): nuclear k4 PTL Flip-Flop is Robust to Parameter Fluctuations • Post-translational logic (PTL) – Engineered around protein modifications – Difficult to engineer – Fast response time (seconds) – Explores new set of applications PTL Inverter . P • A and B are active until doubly phosphorylated by the other. • Non-processive phosphorylation gives rise to the requisite ultrasensitive behavior of pink and green proteins • Conservation of species means that we are dealing with a 4-D system. Types of Intracellular Circuits Activity Out P k3 + k2 + Synthetic Biology Activity In Cell-Cycle Dependent Localization in Yeast Flip-Flop Model + Biology • Capped species concentrations • Must generate cooperativity in new ways • Cannot make pseudo-steady state assumption anywhere. Uses of the Phospholocator • To build sets of post-translational logic (PTL) parts. • To determine the docking site and p-motifs of MAP Kinases of interest. • To detect the activity of MAP Kinases. Phospholocator + – – + – – •We ran a SDS-PAGE gel of crude yeast lysate from cells arrested with nocodazole or pheromone. •The Phospholocator was detected using anti-GFP antibody (gift from Bob Sauer). •Phosphorylated construct runs slower than non-phosphorylated construct. Conclusions •We have shown that a PTL flip-flop will theoretically behave as expected over a wide range of parameter values. •We have specified a system of PTL based on MAPKs and translocation •We have designed a testing scaffold for identifying and characterizing docking and phosphorylation motifs, and are working on a first set of motifs. •We have built a working instance of a PTL device: the Phospholocator Future Directions • Build a Fus3 activated instance of the Phospholocator • Build a simple inverter • Develop a transcription-based localization assay for directed evolution of motifs Active 0 Inactive -5000 -5000 0 Ao 5000 [ ]o 10000 15000 convergent (low Lyapunov -20 exponent) • Our three stable points define a 2D plane in 4D space. We transformed coordinates so the plane was perpendicular to two axes, and thus we could work in two dimensional space. • Varying initial concentrations of the two kinases, we measured the ratio of the change in initial concentration to the change in equilibrium concentration. This is known as the method of Lyapunov exponents. Larger ratios indicate a separatrix, which is the boundary of a domain of attraction. • We can use this map to determine: 1.The range of concentrations over which our flip-flop will hold state. 2.The amount of stimulus needed to switch states, or “flip.” Acknowledgements @Cambridge: Pam Silver, Mike Yaffe, Doug Lauffenburger, Gerry Sussman, the Endy lab, the Bob Sauer lab, the Chris Kaiser Lab, the Steve Bell Lab . . @Berkeley: Alejandro Colman-Lerner, Jeremy Thorner, Kirsten Benjamin, Richard Yu, Roger Brent, Gustavo Pesce Funding: Howard Hughes Medical Institute, National Institute of Health, Merck & Co., Inc. “Our Goal” Images from:http://chemcases.com/cisplat/ cisplat01.htm ; http://www.nature.com/nsu/030421/ 030421-14.html http://www.northern.wvnet.edu/~tdanford/ icons/CELL.JPG; Ricarose Roque Transcriptional Modeling example from Gardner et al, Nature. 2000 Jan 20;403(6767):339-42.