Programming Bacteria for Optimization of Genetic Circuits

Principles – Math Problems

• Computation of solutions to Math Problems such as NP complete problems– Bacterial computers

• We can encode these math problems in biological terms and solve prototype versions of them

• We have a problem scaling to enormous sizes because of the number of bacteria in a culture or the number of DNA molecule in a reaction

– Silicon computers• As long as the problem is not too large, they can outperform

bacterial computers at this task

Maybe bacteria cannot beat Bill Gates at his own game…

Principles – Biological Problems• Computation of solutions to Biological problems such as

Optimization of Genetic Circuits for Synthetic Metabolic Pathways– Silicon computers

• Programs have been developed for the determination of the best genetic circuit elements for use in controlling pathways

• Incomplete inputs and models lead to inaccurate predictions• Computers can only model the biological system

– Bacteria • Could be programmed to compute solutions to these problems• Bacteria are not models of the system, they are the system

But perhaps bacteria can beat Bill Gates at their own game.

Biological Problem

• Say we have a synthetic metabolic pathway– Examples? How would we pick one? We could pick one that

enables selection• Assume that we don’t know how to optimize the output

of the pathway in terms of the following variables– Promoters– RBS– Degradation tags– Order and orientation of genes

• How do we built a system that would allow us to explore combinations of the above variables?

Mathematic Expression of ProblemO = output of metabolic pathway in terms of the concentration of the product

P = promoter elementsR = RBS elementsD = degradation tagsG = order and orientation of genes

O = fcn (P,R,D,G)Fitness = fcn (O)

• We need to explore this 4 dimensional sequence space for each of the genes in the pathway

• We need to examine the relationship between the optimized function for each of the genes

• We need to connect the output of the pathway to fitness of clones

Genetic Circuit and Metabolic Pathway

Gene Expression A Gene Expression B Gene Expression C Gene Expression D

Precursor X

Intermediate A

Intermediate B

Intermediate C

Product D

Enzyme C

Enzyme B

Enzyme A

Enzyme D

Note: Since we are developing a method here, we can pick a pathway that suits our purpose

Gene Expression Cassette

A

LVA

Gene Expression A =

A

= one of the elements of the promoter set

= one of the elements of the C dog set

= fixed as coding sequence A, B, C, or D

= one of the elements of the degradation set, eg. LVA, GGA, PEST, Ubi-Lys

LVA

Element Insertion

• Use GGA to insert elements• Elements carry BbsI sites for initial insertion• But we want to be able to reinsert elements

later, after selection of other elements• So, elements carry BsaI sites for reinsertion• Alternate between BsaI and BbsI for multiple

rounds of insertion

GGA - BbsI Element Insertion

A

BsaIBsaI BbsIBbsI

BbsI BbsI

BbsI, Ligase

ABsaI BsaI

LVAA

To be inserted

Same idea for

To be replaced

final product

GGA - BsaI Element Insertion

A

LVAA

BbsIBbsI BsaIBsaI

BsaI BsaI

BsaI, Ligase

ABbsI BbsI

Same idea for

To be inserted

To be replaced

final product

Genetic Circuit

A LVA

B GGA

C LVA

D GGA

Protocol Step 1

• Use GGA in vitro to place one promoter element from the promoter set into each of the four Gene Expression Cassettes

• Transform E. coli• This establishes the Starting Population promoter

allele frequencies• Culture for one or more generations under selection

for optimal production of product D• Do minipreps and measure Selected Population

allele frequencies

Genetic Circuit

A LVA

B GGA

C LVA

D GGA

Protocol Step 2

• Use GGA in vitro to place one C dog element from the promoter set into each of the four Gene Expression Cassettes

• Transform E. coli• This establishes the Starting Population C dog allele

frequencies• Culture for one or more generations under selection

for optimal production of product D• Do minipreps and measure Selected Population C

dog allele frequencies

Protocol Step 3

• Use GGA in vitro to place one Degradation Tag element from the promoter set into each of the four Gene Expression Cassettes

• Transform E. coli• This establishes the Starting Population Degradation

Tag allele frequencies• Culture for one or more generations under selection

for optimal production of product D• Do minipreps and measure Selected Population

Degradation Tag allele frequenciesImportant note: Maybe using degradation tags is redundant with the transcriptional controls

Protocol Step 4

• Express Hin and reshuffle the orientation and order of the Gene Expression cassettes– Allow complex effects of readthrough transcription– Eg. 384 combinations for 4 genes??

• Transform E. coli• This establishes the Starting Population Order/Orientation

allele frequencies• Culture for one or more generations under selection for

optimal production of product D• Do minipreps and measure Selected Population

Order/Orientation allele frequencies

Protocol Additional Steps

• Go back and repeat Step 1, if desired• Repeat Step 2, or Step 3• Explore the sequence space in whatever way

you want, informed by mathematical modeling

w

x

y

z

w

x

y

z

w = 1

w

x

y

z

z = 2

w

x

y

z

Fitness

• We need to connect the optimization of the metabolic pathway to bacterial cell fitness:

Fitness = fcn (amount of product D)

• Easier Idea– Product D is tied to cell generation time

• Harder Idea– Product D will do the following

• Increase Fitness by protecting the cell that makes it (Protection)• Decrease fitness of surrounding cells (Attack?)

Fitness Easier Idea

• Product D will cause derepression of a gene product that shortens generation time

Product D

Repressor 1

Fitness Gene

Fitness Harder Idea• Product D will cause Hin and Blue luminescence expression• Blue luminescence will interact with optogenetic system to express Death Gene (Attack)• Hin will enable expression of a repressor that will turn off the Death Gene expression

(Protection)

Product D

Repressor 1

Hin Blue

SacB Death Gene

Bacteriorhodopsin

Signal TransductionSee Jeff Tabor work“Multichromatic Control of Gene Expression” JMB

Flip

Repressor 2

Important note: this is a placeholder genetic circuit that could certainly be improved upon

Why separate steps for element insertion?

• We cannot explore all the combinations at once

• For 16 promoters, 8 C dogs, 4 degradation tags, and 4 genes in all orders/orientations, there are over 1014 combinations

Is this just screening?

• Perhaps the answer is Yes, but maybe that is Ok, since the goal is to optimize a pathway, not to compute the answer to a math problem

• Perhaps the answer is No, and the bacteria are computing– The bacteria are evaluating the inputs and

applying a Fitness function– The bacteria are rearranging gene

order/orientation

CRIM

• Lambda bacteriophage system commercially available for insertion of plasmid DNA into genome– Uses insertion and excision and attachment

sequences• For a pathway that was too large for plasmids,

we could park circuits into the genome