1. Programming Biological Cells Ron Weiss, George Homsy, Radhika Nagpal Tom Knight, Gerald Sussman, Nick Papadakis MIT Artificial Intelligence Laboratory

2. ? Goal: program biological cells ? Characteristics  small ( E.coli : 1x2  m, 10 9 /ml)  self replicating  energy efficient ? Potential applications  “smart” drugs / medicine  agriculture  embedded systems Motivation

3. Approach ? Building biological digital circuits  compute, connect gates, store values ? High-level programming issues Outline: logic circuit microbial circuit compiler genome high-level program

4. Compute: Biological Inverter ? signal = protein concentration level ? computation = protein production + decay [A][Z][A][Z] = 0 = 1 [A][A][Z][Z] [A][Z][A][Z] = 0 Z A

5. active gene Inverter Behavior  Simulation model based on phage biochemistry [A][A] [Z][Z] [ ] time (x100 sec)

6. Connect: Ring Oscillator ? Connected gates show oscillation, phase shift time (x100 sec) [A][A] [C][C] [B][B]

7. B _S_S _R_R Memory: RS Latch time (x100 sec) _[R]_[R] [B][B] _[S]_[S] [A][A] = A

8. Microbial Circuit Design ? Assigning proteins is hard. ? BioSPICE: Simulate a colony of cells logic circuit microbial circuit compiler genome protein DB BioSPICE

9. ? Prototype protein level simulator  intracellular circuits, intercellular communication Simulation snapshot cell protein concentration

10. High Level Programming ? Requires a new paradigm  colonies are amorphous  cells multiply & die often  expose mechanisms cells can perform reliably ? Microbial programming language  example: pattern generation using aggregated behavior

11. Conclusions + Future Work ? Biological digital gates are plausible ? Now:  Implement digital gates in E. coli ? Also:  Analyze robustness/sensitivity of gates  Construct a protein kinetics database  Study protein  protein interactions for faster logic circuits