Design of Digital Logic by Genetic Regulatory Networks Ron Weiss - - PDF document

1

Design of Digital Logic by Genetic Regulatory Networks

Ron Weiss Department of Electrical Engineering Princeton University

Computing Beyond Silicon Summer School, Caltech, Summer 2002

• E. coli

Diffusing signal

Programming Cell Communities

proteins

Program cells to perform various tasks using:

• Intra-cellular circuits

– Digital & analog components

• Inter-cellular communication

– Control outgoing signals, process incoming signals

2

Programmed Cell Applications

analyte source Analyte source detection Reporter rings Pattern formation

• Biomedical

– combinatorial gene regulation with few inputs; tissue engineering

• Environmental sensing and effecting

– recognize and respond to complex environmental conditions

• Engineered crops

– toggle switches control expression of growth hormones, pesticides

• Cellular-scale fabrication

– cellular robots that manufacture complex scaffolds

Outline

• In-vivo logic circuits
• Intercellular communications
• Signal processing and analog circuits
• Programming cell aggregates

3

A Genetic Circuit Building Block

Digital Inverter Threshold Detector Amplifier Delay

Logic Circuits based on Inverters

• Proteins are the wires/signals
• Promoter + decay implement the gates
• NAND gate is a universal logic element:

– any (finite) digital circuit can be built!

X Y R1 Z

R1 R1 X Y Z

=

gene gene gene

NAND NOT

4

Why Digital?

• We know how to program with it

– Signal restoration + modularity = robust complex circuits

• Cells do it

– Phage ? cI repressor: Lysis or Lysogeny? [Ptashne, A Genetic Switch, 1992] – Circuit simulation of phage ? [McAdams & Shapiro, Science, 1995]

• Also working on combining analog &

digital circuitry

BioCircuit Computer-Aided Design

SPICE BioSPICE

steady state dynamics intercellular

• BioSPICE: a prototype biocircuit CAD tool

– simulates protein and chemical concentrations – intracellular circuits, intercellular communication – single cells, small cell aggregates

5

Genetic Circuit Elements

input mRNA

ribosome

promoter

• utput

mRNA

ribosome

• perator

translation transcription

RNAp RBS RBS

Modeling a Biochemical Inverter

input

• utput

repressor promoter

6

A BioSPICE Inverter Simulation

input

• utput

repressor promoter

“Proof of Concept” Circuits

• Work in BioSPICEsimulations [Weiss, Homsy, Nagpal, 1998]
• They work in vivo

– Flip-flop [Gardner & Collins, 2000] – Ring oscillator [Elowitz & Leibler, 2000]

• However, cells are very complex environments

– Current modeling techniques poorly predict behavior

time (x100 sec)

[A] [C] [B]

B _ S _ R A _ [R] [B] _ [S] [A]

time (x100 sec)

time (x100 sec)

RS-Latch (“flip-flop”) Ring oscillator

7

The IMPLIES Gate

• Inducers that inactivate repressors:

– IPTG (Isopropylthio-ß-galactoside) Lac repressor – aTc(Anhydrotetracycline) Tet repressor

• Use as a logical Implies gate: (NOT R) OR I
• perator

promoter gene RNAP active repressor

• perator

promoter gene RNAP inactive repressor inducer no transcription transcription Repressor Inducer Output 1 1 1 1 1 1 1

Repressor Inducer Output

The Toggle Switch

[Gardner & Collins, 2000]

pIKE = lac/tet pTAK = lac/cIts

8

Actual Behavior of Toggle Switch

[Gardner & Collins, 2000]

promoter protein coding sequence

The Ring Oscillator

[Elowitz, Leibler 2000]

9

Example of Oscillation Evaluation of the Ring Oscillator

Reliable long-term oscillation doesn’t work yet: Will matching gates help? Need to better understand noise Need better models for circuit design

[Elowitz & Leibler, 2000]

10

A Ring Oscillator with Mismatched Inverters

A = original cI/?P(R) B = repressor binding 3X weaker C = transcription 2X stronger

Device Physics in Steady State

Transfer curve:

gain (flat,steep,flat) adequate noise margins

[input]

“gain”

1

[output]

• Curve can be achieved with certain dna-binding proteins
• Inverters with these properties can be used to build complex circuits

“Ideal” inverter

11

Measuring a Transfer Curve

• Construct a circuit that allows:

– Control and observation of input protein levels – Simultaneous observation of resulting output levels

“drive” gene

• utput gene

R

YFP CFP

inverter

• Also, need to normalize CFP vs YFP

Flow Cytometry (FACS)

12

Drive Input Levels by Varying Inducer

100 1000

IPTG YFP lacI

[high] (Off)

P(lac)

P(lacIq)

lacI

P(lacIq)

YFP

P(lac)

IPTG

IPTG (uM)

promoter protein coding sequence

1.00 10.00 100.00 1,000.00 0.1 1.0 10.0 100.0 1,000.0 10,000.0 IPTG (uM) FL1

pINV-112-R1 pINV-102

Also use for CFP/YFP calibration

Controlling Input Levels

13

Cell Population Behavior

Red = pPROLAR Rest = pINV-102 with IPTG (0.1 to 1000 uM)

CFP: a Weak Fluorescent Protein

Induction of CFP expression

IPTG Fluorescence

14

Measuring a Transfer Curve

for lacI/p(lac)

aTc YFP lacI CFP tetR

[high]

(Off)

P(LtetO-1) λP(R) P(lac)

measure TC

tetR

λP(R)

P(Ltet-O1)

aTc

YFP

P(lac)

lacI CFP

Transfer Curve Data Points

01 10

1 ng/ml aTc

200 400 600 800 1,000 1,200 1,400 1 10 100 1,000 10,000 Fluorescence (FL1) Events

undefined 10 ng/ml aTc 100 ng/ml aTc

200 400 600 800 1,000 1,200 1,400 1 10 100 1,000 10,000 Fluorescence (FL1) Events 200 400 600 800 1,000 1,200 1,400 1 10 100 1,000 10,000 Fluorescence (FL1) Events

15

1 10 100 1000 1 10 100 1000 Input (Normalized CFP) Output (YFP)

lacI/p(lac) Transfer Curve

aTc YFP lacI CFP tetR

[high]

(Off)

P(LtetO-1) λP(R) P(lac)

gain = 4.72 gain = 4.72

Evaluating the Transfer Curve

• Noise margins:

200 400 600 800 1,000 1,200 1,400 1 10 100 1,000 Fluorescence Events

30 ng/ml aTc 3 ng/ml aTc

1 10 100 1,000 0.1 1.0 10.0 100.0 aTc (ng/ml) Fluorescence

• Gain / Signal restoration:

high gain high gain

* note: graphing vs. aTc (i.e. transfer curve of 2 gates)

16

10

1

10

2

10

3

10 10

1

10

2

10 10

1

10

2

10

3

IPTG (mM) aTc (ng/ml) Median FLR

Transfer Curve of Implies

YFP lacI aTc IPTG tetR

[high]

The Cellular Gate Library

Add the cI/λP(R) Inverter

OR1 OR2

structural gene

λP(R-O12)

• cI is a highly efficient repressor

cooperative binding

IPTG YFP cI CFP lacI

[high]

(Off)

λ

P(R)

P(lac)

• Use lacI/p(lac) as driver

high gain

cI bound to DNA

17

Initial Transfer Curve for cI/λP(R)

1.00 10.00 100.00 1,000.00 0.1 1.0 10.0 100.0 1,000.0 IPTG (uM) Output (YFP)

lacI

P(lacIq)

P(lac)

IPTG

YFP λ

P(R)

cI CFP

Recall Inverter Components

input mRNA

ribosome

promoter

• utput

mRNA

ribosome

• perator

translation transcription

RNAp RBS RBS

18

Functional Composition of an Inverter

“clean” signal digital inversion scale input invert signal ψΑ φΑ

translation

φΑ ρ Α

1 1

ρ Α ψΖ

1

+ + =

ψΖ ψΑ

“gain”

1

+ + =

cooperative binding transcription inversion

ψΑ = input mRNA φΑ = input protein ρ Α = bound operators ψΖ = output mRNA

Genetic Process Engineering I:

Reducing Ribosome Binding Site Efficiency

R B S

translation start Orig: ATTAAAGAGGAGAAATTAAGCATG strong RBS-1: TCACACAGGAAACCGGTTCGATG RBS-2: TCACACAGGAAAGGCCTCGATG RBS-3: TCACACAGGACGGCCGGATG weak

ψΑ φΑ

translation stage

ψΖ ψΑ

Inversion

19

1.00 10.00 100.00 1,000.00 0.1 1.0 10.0 100.0 1,000.0 IPTG (uM) Output (YFP)

pINV-107/pINV-112-R1 pINV-107/pINV-112-R2 pINV-107/pINV-112-R3

Experimental Results for cI/λP(R) Inverter with Modified RBS

Genetic Process Engineering II:

Mutating the λP(R) operator

BioSPICESimulation

• rig:

TACCTCTGGCGGTGATA mut4: TACATCTGGCGGTGATA mut5: TACATATGGCGGTGATA mut6 TACAGATGGCGGTGATA

OR1

φΑ ρ Α

cooperative binding

ψΖ ψΑ

20

Experimental Results for Mutating λP(R)

1.00 10.00 100.00 1,000.00 0.1 1.0 10.0 100.0 1,000.0 IPTG (uM) Output (YFP)

pINV-107-mut4/pINV-112-R3 pINV-107-mut5/pINV-112-R3 pINV-107-mut6/pINV-112-R3

Genetic Process Engineering

• Genetic modifications required to make circuit work
• Need to understand “device physics” of gates

– enables construction of complex circuits

1.00 10.00 100.00 1,000.00 0.1 1.0 10.0 100.0 1,000.0 IPTG (uM) Output (YFP)

modify RBS mutate

• perator

RBS #1: modify RBS #2: mutate operator

21

Self-perfecting Genetic Circuits

[Arnold, Yokobayashi, Weiss]

• ptical micrograph of

the µFACS device

• Use directed evolution to optimize circuits
• Screening criteria based on transfer curve
• Initial results are promising

Lab-on-a-chip: µFACS [Quake]

Molecular Evolution of the Circuit

22

Prediction of Circuit Behavior

1 10 100 1,000 0.1 1.0 10.0 100.0 1 10 100 1,000 0.1 1.0 10.0 100.0

=

?

Output signal Input signal

1 10 100 1,000 0.1 1.0 10.0 100.0

Can the behavior of a complex circuit be predicted using only the behavior of its parts?

Prediction of Circuit Behavior

preliminary results

YFP aTc # cells

tetR

P(bla)

P(tet)

aTc

cI P(lac) lacI CFP YFP λP(R)