Engineered Gene Circuits Jeff Hasty

Engineered Gene Circuits

Jeff Hasty

OR

cI857

gfp

RBS

RBST1T2

ampR

ColE1

T1T2

How do we predict cellular behavior from the genome? Sequence data gives us the components, now how do we understand the full system?

How can we control or monitor cellular behavior? Diseases, pathogenic invasions involve alterations of natural dynamics - can we reestablish normal function?

Gene Regulation

Gene regulatory networks• Proteins affect rates of production of other

proteins (or themselves)

• This allows formations of networks of interacting genes/proteins– Sets of genes whose expression levels are

interdependent

A

B

C

D E

““Using gene and protein network wiring Using gene and protein network wiring diagrams to try to deduce cellular behavior diagrams to try to deduce cellular behavior is akin to using a VCR circuit diagram to try is akin to using a VCR circuit diagram to try to deduce how to program it.”to deduce how to program it.”

Mathematical models are needed to Mathematical models are needed to translate gene-protein wiring diagrams translate gene-protein wiring diagrams into “manuals” explaining cellular into “manuals” explaining cellular processes.processes.

But how do we construct reliable and useful But how do we construct reliable and useful mesoscopic models?mesoscopic models?

John Tyson’s Analogy

Engineered Gene Circuits

Faithful modeling of large-scale networks is difficult…

Alternative: Design and build simpler networks

Decouple complexity

Use model to design experiments

Systematic comparison of model and experiment

“Forward Engineering” of useful circuits

Design networks to perform tasks

Couple to host - control or monitor cellular function

Engineered Toggle Switch

Gardner, Cantor & Collins, Nature 403:339 (2001)

Gene A onGene B off Reporter

G FPRepressor A

Gene A offGene B on Reporter

Repressor B

“On”

“Off”

Model - design criteria:

Construction/experiments:

The RepressilatorG FPA

B C

Gene A

Gene B Gene C

Elowitz and Leibler, Nature 403:335 (2001)

A Detailed Example: Single-Gene Autoregulatory Module

Well-characterized: Kinetic parms known

Tunable control: CI857 denatures with temp

Build network with off-the-shelf molecular biology

Theoretical predictions: Bistablity and hysteresis

(Hasty et al PNAS 97:2075, 2000)

Biochemical Reactions

Rate Eqs For cI Monomers and GFP Reporter

Model predictions as the temperature is varied?

Model Prediction: Multistability

Experimental Protocol

Bistability Results

Prediction

Observation

Model the Fluctuations

- OK when fluctuations dominated by production and degradation

- Distributions numerically check with Monte Carlo “gold standard”

- Still working on systematic demonstration of validity

Model Versus Experiment

Coefficient of Variation

Genetic Relaxation Oscillator

O 2R O 2R

Prom oter P R M

cI

O 3R

Prom oter P R M

Plasm id 2P lasm id 1

RcsA

O 3R

Hasty et al, Chaos (2001)

Relaxation Oscillator Analysis

Design network so that y is a slow variable:

Drive Oscillator With Cell Division Cycle

Identify known oscillating gene product and its target promoter

SWI4 forms a complex and activates the HO promoter

Resonant Dynamics

Drive Period (Minutes)

Regions of synchronization

Summary

• Use of biochemical kinetics to describe gene regulation (in bacteria)

• Models can be used to develop “tailor-made” circuits

• Gene circuits lead naturally to problems relevant to nonlinear dynamics, statistical physics and engineering

• Noise from small molecular numbers is a dominant source

• Genetic “states” accessed through fluctuations (noise-induced transitions between attractors)

OR

cI857

gfp

RBS

RBST1T2

ampR

ColE1

T1T2

Milos Dolnik (Brandeis)

David McMillen (Boston University)

Vivi Rottschafer (Leiden)

Farren Isaacs (BU)

Charles Cantor (BU-UCSD)

Jim Collins (BU)

Funding: NSF, DARPA and the Fetzer Institute

Collaborators:

The Human Genome Project

• Why is this not true?

• Network dynamics not yet understood