Indian Institute Of Technology, Bombay Engineered Versus Natural - - PowerPoint PPT Presentation

Indian Institute Of Technology, Bombay

Engineered Versus Natural System

11/1/2009 Team IIT Bombay, Jamboree 2009 2

Design Operation Optimization Control

ENGINEERED SYSTEM Engineered system: bottom-up design with known functionality of components Natural system: top down design with unknown inherent property of various motifs

Engineered Systems : Room Heater

11/1/2009 Team IIT Bombay, Jamboree 2009 3

Temperature Controller Process Measuring Temperature Thermostat Set point for a Temperature Decides to switch on/off electric supply to bring temperature to set point Negative feedback

SINGLE INPUT SINGLE OUTPUT (SISO)

Multiple Input Multiple Output: a motif observed in Biological System

11/1/2009 Team IIT Bombay, Jamboree 2009 4

Process 1 Measurement Set point

Controlled Variable

Single output is regulating the multiple upstream processes

Process 2

Tryptophan in E. coli (bacteria)

11/1/2009 Team IIT Bombay, Jamboree 2009 5

• Ref. Venkatesh K V et al, 2004

Osmotic Stress Pathway in Yeast

11/1/2009 6 Team IIT Bombay, Jamboree 2009

• Ref. Parmar et al, 2009

Insulin Signaling Pathway in Mammals

11/1/2009 7 Team IIT Bombay, Jamboree 2009

11/1/2009 Team IIT Bombay, Jamboree 2009 8

• Ref. Freeman, 2000

11/1/2009 Team IIT Bombay, Jamboree 2009 9

Designed and Implemented a synthetic genetic network with multiple feedbacks

Linking protein expression to growth

Modeling and Experiments for characterization

• f the network

Approach

Modeling –

• Detailed molecular

mechanisms based model

• Stochastic modeling
• Control analysis

Experiments

• Protein expression by FACS
• Characterization of

phenotype in the synthetic constructs

Components of Synthetic Constructs

• Use of existing bio-bricks
• Four promoter sites used for the

constructs: pTet, pLac, pMB1 and pLacOP .

• pMB1 and pLacOP : promoters for

plasmid replication.

• To characterize amount of LacI:

LacI-CFP fusion protein.

• To characterize plasmid copy number:

YFP expression.

11/1/2009 10 Team IIT Bombay, Jamboree 2009

Promoter site

On addition of IPTG Plasmid copy number does not change Plasmid copy number increases

Characteristics of promoters used for Plasmid Replication

On addition of IPTG

pMB1 pLacOP

11/1/2009 11 Team IIT Bombay, Jamboree 2009

LacI regulation in pTet and pLac

pTet pLac

LacI

RNA/DNA Polymerase

11/1/2009 12 Team IIT Bombay, Jamboree 2009

lacI lacI

Plasmid Strain 2 (Single Input Single Output – LacI regulation, BBa_K255003) Plasmid Replication

Constructs

11/1/2009 Team IIT Bombay, Jamboree 2009 13

Plasmid Strain 3 (Single Input Single Output – Copy Number, BBa_K255002) pTet YFP Plasmid Replication Plasmid Strain 4 (Multiple Input Multiple Output, BBa_K255001) pLac pTet YFP Plasmid Replication pLac pTet YFP pMB1 pTet Plac pLacOP pLacOP Plasmid Strain 1 (Open Loop, BBa_K255004) pTet LacI +CFP pTet YFP Plasmid Replication pMB1 Promoter

• ve Feedback

Protein LacI +CFP LacI +CFP LacI +CFP STOP

SYNTHETIC CONSTRUCTS

NO CONTROL OPEN LOOP (STRAIN 1) SINGLE INPUT SINGLE OUTPUT SISO_LacI : Regulation of LacI (STRAIN 2) SISO_CN : Regulation of Plasmid Copy Number (STRAIN 3) MULTIPLE INPUT MULTIPLE OUTPUT MIMO: Regulation of Plasmid Copy Number and LacI (STRAIN 4)

11/1/2009 Team IIT Bombay, Jamboree 2009 14

Molecular Map of the Construct

11/1/2009 Team IIT Bombay, Jamboree 2009 15

LacI-IPTG complex Replicated Plasmids

Modeling Methodologies

• Detailed Dynamic Modeling using all known

molecular interactions

• Stochastic Analysis on a simplified model

using Langevin approach

• Frequency response analysis on the linearised

model

11/1/2009 Team IIT Bombay, Jamboree 2009 16

Prediction of Steady State Expression

• f YFP (Plasmid Copy Number)

11/1/2009 Team IIT Bombay, Jamboree 2009 17

11/1/2009 Team IIT Bombay, Jamboree 2009 18

+ +

• Set-point

+

LacI level

Error

Control Analysis to Characterize System Behavior

Block diagram for the Linearised LacI system Block diagram for the LacI system

C1(s) F(Cs)/(s+µ+β1-F’(Cs) C1s) k3C2s/(s+µ+β3) k3Css/(s+µ+β3) C2(s) Controllers Controllers

Frequency Response Analysis

• Higher bandwidth
• Higher Phase margin
• Noise Attenuation

11/1/2009 19 Team IIT Bombay, Jamboree 2009

Experimental Validation

• Experiments with various IPTG concentrations

were conducted.

• Protein expression measured as YFP using

FACS to quantify plasmid copy number.

• Mean and Variance obtained from the

distribution.

11/1/2009 20 Team IIT Bombay, Jamboree 2009

Experimental YFP expression (characterizing Plasmid Copy Number)

11/1/2009 Team IIT Bombay, Jamboree 2009 21

Higher variance in open loop

• 20

20 40 60 80 100 120 140 100 200 500

Normalised YFP count Normalised YFP v/s IPTG

OPEN LOOP SISO_pLAC SISO_CN MIMO

IPTG µM

• Open Loop and SISO_LacI:

No increase in YFP with inducer

• SISO_CN and MIMO:

expression increase with inducer

Characterization of LacI expression

• The detection of LacI-CFP fusion protein was

not possible due to technical problems.

• An indirect measure of LacI was obtained by

measuring β-galactosidase from the lacZ of the host.

• Further the growth rate of the four

transformants were also enumerated.

11/1/2009 Team IIT Bombay, Jamboree 2009 22

Schematic representation of the network in presence of lactose

11/1/2009 Team IIT Bombay, Jamboree 2009 23

lacZ

LacI-IPTG Replicated Plasmids LacI-Lactose

Growth Response from Modeling

11/1/2009 Team IIT Bombay, Jamboree 2009 24

dual feedback system: high β-gal expression with low variance

Stochastic Modeling on Growth Rate

SPECIFIC GROWTH RATE NORMALIZED β-gal EXPRESSION For perturbation of the kinetic parameters around the mean value, we see MIMO has the least variance compared to open loop or a single feedback system

11/1/2009 25 Team IIT Bombay, Jamboree 2009

Experimental Results

11/1/2009 Team IIT Bombay, Jamboree 2009 26

• 0.02

0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 1 2 3 4 5 6 Specific Growth Rate (in hr-1) Lactose (g/L)

Specific Growth Rate v/s Lactose

OPEN LOOP MIMO

• 0.2

0.2 0.4 0.6 0.8 1 1.2 1 2 3 4 5 6

β-gal/βa-gal max

Lactose (g/L)

Normalized β-gal expression v/s Lactose

MIMO OPEN LOOP

Noise in protein expression propagates to growth The variance in specific growth rate is less compared to that

• bserved in protein expression.

Agar Plate Experiments

11/1/2009 Team IIT Bombay, Jamboree 2009 27

1 2 3 4 1 5 CFU (in 106 /ml) Lactose (g/L)

Agar Plate Experiment (without IPTG)

OPEN LOOP

10 20 30 40 1 5 CFU (in 106 /ml) Lactose (g/L)

Agar Plate Experiments (without IPTG) MIMO

Strains were grown on agar plate with different lactose concentrations. Colony Forming Units in the agar plates were counted. Variance in Open Loop is 40 % and MIMO is 10%.

Recapitulating…

• Robustness in protein expression which leads

to low variance in specific growth rate.

• The noise in protein expression is filtered

leading to a decrease in the variance in growth

• rate. This may be due to metabolism and

division process.

• The transformants with the synthetic network

yields distinct phenotypic response.

11/1/2009 Team IIT Bombay, Jamboree 2009 28

Optimality

11/1/2009 Team IIT Bombay, Jamboree 2009 29

Optimal production of enzyme : growth rate for MIMO. MIMO has optimized its burden for optimal Normalized Growth Rate.

MIMO NGR OPEN LOOP NGR

11/1/2009 Team IIT Bombay, Jamboree 2009 30

MULTIPLE FEEDBACK SYSTEM

IMPROVED PERFORMANCE

OPTIMALITY FASTER RESPONSE TIME ROBUSTNESS TO INTRINSIC NOISE PRECISION

Phd mentors:

• Pushkar Malakar,
• Navneet Rai ,
• Vinay Bavdekar,

Acknowledgements

11/1/2009 Team IIT Bombay, Jamboree 2009 31

Mentors:

• Prof. K V Venkatesh
• Prof. Sharad Bhartiya
• Prof. Vishwesh Kulkarni

Sponsors:

IIT Bombay Contributors:

• Mukund Thattai, NCBS, Bangalore
• Dr. Manjula Reddy, CCMB, Hyderabad

Thank you!!

11/1/2009 Team IIT Bombay, Jamboree 2009 32

Q&A

• Bode plot analysis

11/1/2009 Team IIT Bombay, Jamboree 2009 33

11/1/2009 Team IIT Bombay, Jamboree 2009 34

Magnitude and Phase Bode plots

ZERO IPTG HIGH IPTG BACK