developing designable genetic control systems for applied ......developing designable genetic...

Developing designable genetic control systems for applied synthetic biology

James M. Carothers

December 9, 2013

BIO Pacific Rim Summit on Industrial Biotechnology & Bioenergy

San Diego, CA

Chemical Engineering and BioengineeringCenter for Synthetic Biology

University of Washington, Seattle, WA

Control functions balance the supplies and demands for cellular resources

After Carothers, Goler, Keasling. Curr. Opin. Biotechnol. 2009.

DNA

Enzymes

Reaction1

Reaction2

ReactionN

Product

Enzyme1

E1

E1

E2

EN

EN

EN

EN

s1

Enzyme2

EnzymeN

s2

sN P

Control functions balance the supplies and demands for cellular resources

After Carothers, Goler, Keasling. Curr. Opin. Biotechnol. 2009.

DNA

Enzymes

Reaction1

Reaction2

ReactionN

Product

Enzyme1

E1

E1

E2

EN

EN

EN

EN

s1

Enzyme2

EnzymeN

s2

sN P

Targeted device outputs

Genetic devicedesign goals

Engineercomponents

mechanisticmodels

Systems-levelmodels

0...1

Biochemicalmodels for identifying

component specifications

Functionaldesign

Physical implementation

Kinetic foldingmodels

Biophysical models for designing

transcripts

Assembly & verification

mechanisticmodels

Biochemical models for predicting

device outputs

StaticRNA devices

DynamicRNA devices

Devices generating

targeted outputs

Systems-levelfunctions

Refine Systems-level models

Approach: Develop RNA-based systems to solve regulatory control problems and enable high levels of chemical production from engineered cells

Goal: Designable RNA-based genetic control systems that dramatically increase the sizes and complexities of systems that can be engineered

Computational modeling

In vitro selection/evolution

Genetic engineering

E. coli platform for p-aminostyrene (p-AS) production 14 enzymatic transformations

Goal: Engineer RNA-based genetic systems to solve pathway control problems and enable production of p-AS for advanced polymer composites

DHA

DHQ

Shikimate

Glucose

DAHP

PEPE4P

Chorismate

p-AF

PpsATktA

AroG*

aroB

aroD

aroE

EPSC

S3P

p-ADC

p-AP

p-APA

aroL

aroA

aroC

papA

papB

papC

deaminase

p-ACA

deaminase

decarboxylase

Opiate analgesics

p-aminostyrene(p-AS)

NH2

COO-

NH3

+

NH2

COO-

OHOH

HO

Shikimate p-AF

Dipeptidylpeptidase IV

inhibitors

Pristinamycinantibiotics

After Carothers. Syst. Synth. Biol. 2013.Sariaslani. Ann. Rev. Microbiol. 2007.

E. coli platform for p-aminostyrene productionLast 6 of 14 steps shown

Goal: Engineer RNA-based genetic systems to solve pathway control problems and enable production of p-AS for advanced polymer composites

Functional RNAs regulate gene expression and control cellular activities

S. mansoni hammerhead ribozyme B. subtilis purine riboswitch aptamer domain

Phosphodiester bond cleavage

5'

3' Purine ligand

Dynamic, ligand-responsive

5' 3'

5'

3'

Scissilephosphate

Functional RNAs can be evolved in the laboratory with in vitro selection

Aptamers GTP aptamer

Ribozymeshammerhead rbz

A G U

C

G

C

U

AG

C

U G

AA A

C

G

C

C

UU

AG

A

GAA

GA

U

C

G

G

C

GC

UU A

UA

GG

C G

G C

G C

G UG

A A

AAptamer

Ribozyme

Bridge(evolved)

5'3'

AptazymesAptamer-controlled ribozyme

Scissilephosphate

kd range: mM to nM k

cat (+/-) up to 1000

5' 3'

5'

3'

Scissilephosphate

kcat

: up to ~10 min-1

GTP ligand

3'

5'

Carothers & Szostak. The Aptamer Handbook. 2006.Carothers, Goler et al. & Keasling. Nucl. Acids Res.2010.

Formalized framework for designing static and dynamic genetic devices with quantitatively-predictable outputs

Targeted device outputs

Genetic devicedesign goals

Engineercomponents

mechanisticmodels

Systems-levelmodels

0...1

Biochemicalmodels for identifying

component specifications

Functionaldesign

Physical implementation

Kinetic foldingmodels

Biophysical models for designing

transcripts

Assembly & verification

mechanisticmodels

Biochemical models for predicting

device outputs

StaticRNA devices

DynamicRNA devices

Devices generating

targeted outputs

Systems-levelfunctions

Refine Systems-level models

Ribozyme (rREDs)

Aptazyme (rREDs)

Predicted and observed functions for 25 static rREDs and dynamic aREDs

r = 0.94, P = 2 x 10-7

For devices w/ kfold

> 0

rel. rms-d = 0.19

For devices w/ kfold

< 0

rel. rms-d = 0.60

Excellent correspondence between predictions and observations validates underlying models and overall approach

Carothers, Goler, Juminaga & Keasling. Science, 2011.

Design of FadR-based dynamic sensor regulator system to improve fatty acid ethyl ester (FAEE) biodiesel production in E. coli

Approach: Interrogate system designs by performing design variable global sensitivity analysis with biochemical parameter uncertainty

Renewable diesel

Original system: Steen et al. & Keasling. Nature. 2010.

Design of FadR-based dynamic sensor regulator system to improve fatty acid ethyl ester (FAEE) biodiesel production in E. coli

Zhang, Carothers & Keasling, Nature Biotech. 2012.

Broad range of dynamic regulatory topologies gave improved production compared to pathway gene expression from constitutive promoters

Modeled parameter ranges where dynamic sensor regulator system increases FAEE production

Experimentally measured 3 fold greater yields with FadR-dynamic sensor regulator system (28% theoretical maximum from glucose)

E. coli platform for p-aminostyrene productionLast 6 of 14 steps shown

Goal: Engineer RNA-based genetic systems to solve pathway control problems and enable production of p-AS for advanced polymer composites

Approach: Generate 728 unique pathway models to map space of dynamic control topology inputs to production outputs

E. coli platform for p-aminostyrene productionPotential genetic control mechanisms

PapC

PapB

PapC

chorismatep-amino-

phenylalanine(p-AF)

Deam.

txn

trans.

LAAO

p-amino-cinnamic acid

Transport out of cell

p-amino-styrene(p-AS)

CO2

p-AF is excreted from cell

p-ACA is a toxicintermediate

LAAO Overexpressiondepletes

metabolites

Pump Overexpression

reducesgrowthtrans. trans.

txn txn

mRNA mRNA mRNA

Finalproduct

From shikimate/chorismate

modules

L-amino Acid Oxidase CDSPap Operon CDS

pump

pump

pump

deg.deg.

deg.

trans.trans.

ControlProblem

ControlNode

Enzyme

Efflux Pump CDS

blue text

NH2

NH3

NH2

COO-

+

COO-COO-

NH2

NH2

OH

COO-

COO-O

9-fold production increases (20% theoretical maximum) readily achievable with dynamic multi-device, RNA-based genetic control systems

E. coli platform for p-aminostyrene productionPredicted production outputs for 728 core regulatory topologies

Co

re T

op

olo

gy

Nu

mb

er

1

26

aRED aRED aRED only + + sRNA TF

0.1

1

0

1

000

Co

re T

op

olo

gy

Nu

mb

er

1

3

5

7

9

1

1

13

1

5

17

1

9

21

23

2

5

1e-10 1e-5 1 1e5p-ACA Output Ratio (αrel)

P L E

Rank 1: Core Topology 4

αrel: 3.23 (2.68)βrel:1.34 (0.20)titer:12.1(15.7)

P L E

Rank 3: Core Topology 20

αrel: 2.37 (1.55)βrel:1.2 (0.13)titer: 8.39 (11.2)

P L E

Rank 5: Core Topology 22

αrel: 1.04 (0.19)βrel:1.04 (0.03)titer: 3.84 (5.34)

P L E

Rank 15: Core Topology 23

αrel: 0.037 (0.054)βrel:1.44 (0.24)titer: 0.098 (0.14)

P L E

Rank 25: Core Topology 6

αrel: 9.1e-7 (1.4e-6)βrel:1.3 (0.15)titer: 2.6e-6 (5.4e-6)

P L E

Rank 2: Core Topology 5

αrel: 2.83 (1.62)βrel:1.27 (0.15)titer: 11.9 (15.4)

P L E

Rank 4: Core Topology 13

αrel: 2.21 (1.06)βrel:1.2 (0.10)titer: 9.05 (12.1)

P L E

Rank 10: Core Topology 2

αrel: 0.18 (0.20)βrel:1.2 (0.13)titer: 0.61 (0.86)

P L E

Rank 20: Core Topology 26

αrel: 3.2e-6 (4.8e-6)βrel:1.27 (0.145)titer: 1.6e-5 (2.35e-5)

P L E

Rank 26: Core Topology 21

αrel: 3.6e-7 (5.3e-7)βrel:1.05 (0.03)titer: 1.4e-6 (2.1e-6)

p-ACA Output Ratio (αrel

)

= dynamic / static production

Ou

tpu

t P

rod

uct

ion

Rat

io

1 1e51e-51e-10

9-fold production increases (20% theoretical maximum) readily achievable with dynamic multi-device, RNA-based genetic control systems

E. coli platform for p-aminostyrene productionPredicted production outputs for 728 core regulatory topologies

Co

re T

op

olo

gy

Nu

mb

er

1

26

aRED aRED aRED only + + sRNA TF

0.1

1

0

1

000

Co

re T

op

olo

gy

Nu

mb

er

1

3

5

7

9

1

1

13

1

5

17

1

9

21

23

2

5

1e-10 1e-5 1 1e5p-ACA Output Ratio (αrel)

P L E

Rank 1: Core Topology 4

αrel: 3.23 (2.68)βrel:1.34 (0.20)titer:12.1(15.7)

P L E

Rank 3: Core Topology 20

αrel: 2.37 (1.55)βrel:1.2 (0.13)titer: 8.39 (11.2)

P L E

Rank 5: Core Topology 22

αrel: 1.04 (0.19)βrel:1.04 (0.03)titer: 3.84 (5.34)

P L E

Rank 15: Core Topology 23

αrel: 0.037 (0.054)βrel:1.44 (0.24)titer: 0.098 (0.14)

P L E

Rank 25: Core Topology 6

αrel: 9.1e-7 (1.4e-6)βrel:1.3 (0.15)titer: 2.6e-6 (5.4e-6)

P L E

Rank 2: Core Topology 5

αrel: 2.83 (1.62)βrel:1.27 (0.15)titer: 11.9 (15.4)

P L E

Rank 4: Core Topology 13

αrel: 2.21 (1.06)βrel:1.2 (0.10)titer: 9.05 (12.1)

P L E

Rank 10: Core Topology 2

αrel: 0.18 (0.20)βrel:1.2 (0.13)titer: 0.61 (0.86)

P L E

Rank 20: Core Topology 26

αrel: 3.2e-6 (4.8e-6)βrel:1.27 (0.145)titer: 1.6e-5 (2.35e-5)

P L E

Rank 26: Core Topology 21

αrel: 3.6e-7 (5.3e-7)βrel:1.05 (0.03)titer: 1.4e-6 (2.1e-6)

p-ACA Output Ratio (αrel

)

= dynamic / static production

Ou

tpu

t P

rod

uct

ion

Rat

io

1 1e51e-51e-10

9-fold production increases (20% theoretical maximum) readily achievable with dynamic multi-device, RNA-based genetic control systems

E. coli platform for p-aminostyrene productionPredicted production outputs for 728 core regulatory topologies

Co

re T

op

olo

gy

Nu

mb

er

1

26

aRED aRED aRED only + + sRNA TF

0.1

1

0

1

000

Co

re T

op

olo

gy

Nu

mb

er

1

3

5

7

9

1

1

13

1

5

17

1

9

21

23

2

5

1e-10 1e-5 1 1e5p-ACA Output Ratio (αrel)

P L E

Rank 1: Core Topology 4

αrel: 3.23 (2.68)βrel:1.34 (0.20)titer:12.1(15.7)

P L E

Rank 3: Core Topology 20

αrel: 2.37 (1.55)βrel:1.2 (0.13)titer: 8.39 (11.2)

P L E

Rank 5: Core Topology 22

αrel: 1.04 (0.19)βrel:1.04 (0.03)titer: 3.84 (5.34)

P L E

Rank 15: Core Topology 23

αrel: 0.037 (0.054)βrel:1.44 (0.24)titer: 0.098 (0.14)

P L E

Rank 25: Core Topology 6

αrel: 9.1e-7 (1.4e-6)βrel:1.3 (0.15)titer: 2.6e-6 (5.4e-6)

P L E

Rank 2: Core Topology 5

αrel: 2.83 (1.62)βrel:1.27 (0.15)titer: 11.9 (15.4)

P L E

Rank 4: Core Topology 13

αrel: 2.21 (1.06)βrel:1.2 (0.10)titer: 9.05 (12.1)

P L E

Rank 10: Core Topology 2

αrel: 0.18 (0.20)βrel:1.2 (0.13)titer: 0.61 (0.86)

P L E

Rank 20: Core Topology 26

αrel: 3.2e-6 (4.8e-6)βrel:1.27 (0.145)titer: 1.6e-5 (2.35e-5)

P L E

Rank 26: Core Topology 21

αrel: 3.6e-7 (5.3e-7)βrel:1.05 (0.03)titer: 1.4e-6 (2.1e-6)

p-ACA Output Ratio (αrel

)

= dynamic / static production

Ou

tpu

t P

rod

uct

ion

Rat

io

1 1e51e-51e-10

1. RNA devices with quantitatively-programmable genetic outputs can be assembled from component parts generated and characterized in vitro, in vivo, and in silico.

2. Simulation analysis can inform the design of Dynamic Sensor-Regulator Systems that increase production of FAEEs.

3. Multi-device, RNA-based genetic systems can be designed to solve control problems and potentially enable high levels of production.

Summary:

Acknowledgements

University of California Berkeley and DOE Joint BioEnergy Institute

Email: jcaroth@uw.edu http://carothersresearch.com

Jay Keasling Jonathan GolerAlex Juminaga

Tim ThimmaiahFuzhong Zhang

Guillaume CambrayRichard Schwartz

Jason Stevens – BioE PhD Student Rodrigo Correa – PostdocDavid Sparkman-Yager – Research Assistant

Willy Voje – ChemE PhD studentMichael Newton - ChemE PhD studentMisha Yatsevitch – ChemE Undergrad

University of Washington

CENTER for SYNTHETIC BIOLOGY