1 25 OCTOBER 2002 VOL 298 SCIENCE Two types of motifs heavily over-represented in transcriptional networks: 2 Simple expression Negative auto-regulation Positive auto-regulation 3 Simple expression Negative auto-regulation Positive auto-regulation a b c 4 Up to 40% of E. coli TFs negatively regulate their own expression. Rosenfeld, Elowitz, & Alon. J. Mol. Biol. (2002) 323, 785793 Negative Autoregulation (NAR) speeds response time (relative to steady state) Anhydrotetracycline (aTc) inactivates TetR repressor Constitutive TetR TetR represses its own expression 5 To reach the *same steady state* NAR requires a stronger promoter Concentration of aTc Auto-feedback doesnt kick in until reach aTc threshold 6 Simple expression Negative auto-regulation Positive auto-regulation a b c 7 NAR can reduce cell-cell variation, PAR can accentuate it (even = bimodal distribution) Bimodal distribution = Bistability: cells in a population can exist in TWO DIFFERENT states given the SAME ENVIRONMENT 8 Bistability can also give rise to hysteresis: history-dependent response to environmental cues Maeda & Sano. J. Mol. Biol. (2006) 359, 11071124 (inducer) 9 Feed-forward loops are also recurring network motifs 8 possible structures, each can be AND or OR gate = 16 possibilities 10 AND gate: both X AND Y required to activate Z = response delay, but rapid shutoff Activating Signal delay comes from need to make Y Sign-sensitive delay element 11 Mangan, Zaslaver, Alon. J. Mol. Biol. (2003) 334, 197204 vs. FFL was significantly slower turning on in response to cAMP but same kinetics switching off 12 AND gate: both X AND Y required to activate Z = response delay, but rapid shutoff Activating Signal Sign-sensitive delay element: Delay in turning system on buffers against small changes in activating signal OR gate: opposite trends: Rapid on (need EITHER X OR Y) but delay in shut-off resistant to fluctuations in input signal once the system is on Activating Signal Example from review: bacterial flagella production continues despite subtle fluctuations in activator 13 Activating Signal Generates a pulse of output Also speeds up response time (similar to negative auto-regulation of TFs) 14 15 Signaling networks in development tend to be more complicated Double-positive feedback loops: can activate persistent states (memory) even after activating signal is gone. Double-negative feedback loops: can activate persistent states (memory). Often acts as a toggle switch between two states. 16 Combination of network motifs into larger signaling networks can generate a myriad of outputs: pulses oscillations ultra-sensitive fate switches toggle switches and more Future directions: - better methods for motif identification in large datasets - identification of motifs unique to different types of data/biology - better understanding of motif effects prediction of outputs 17