Understanding the Nutritional Control of Metabolic Flux in ... - shackett

hostitchAI and Robotics

Oct 23, 2013 (3 years and 11 months ago)

96 views

The

variation

in

flux

through

any

reaction

can

be

related

to

its

reaction

mechanism,

where

the

flux

through

the

reaction

is

described

as

a

function

of

kinetic

parameters,

concentration

of

metabolites

and

enzyme

abundance
.






Within

this

framework,

variation

in

V

(found

above)

will

be

related

to

variation

in

intracellular

concentration

of

substrates

and

other

competitive

species,

and

Vmax

related

changes

in

the

level

of

an

enyzme

or

its

activity
.


To

describe

this

relationship

exactly,

we

need

precise

measurements

of

the

intracellular

concentration

of

all

relevant

species,

to

know

the

kinetic

nature

of

their

interaction

and

to

know

which

species

are

sufficient

to

describe

the

mechanism
.



Understanding the Nutritional Control of Metabolic Flux in
S. cerevisiae

Sean R Hackett and Josh
Rabinowitz

Summary

Yeast

must

efficiently

balance

anabolism

with

nutrient

availability

in

order

to

grow

optimally

in

diverse

environments
.

Previous

efforts

to

characterize

the

physiology

underlying

this

control

have

focused

on

looking

at

patterns

of

transcriptional

and

metabolite

variability

under

25

nutritional

conditions

differing

either

in

the

nature

of

the

limiting

nutrient,

or

how

stringent

the

limitation

is
.

This

has

revealed

major

patterns

associated

with

nutrient

invariant

and

nutrient

specific

limitation,

but

given

us

little

insight

into

how

the

relative

flux

through

reactions

is

shaped

by

nutrient

availability

and

how

this

flux
-
control

is

accomplished

at

a

mechanistic

level
.

As

an

attempt

to

address

this,

I

will

characterize

the

fluxes

across

conditions

by

measuring

the

rate

that

nutrients

are

absorbed

and

the

rates

that

they

are

being

polymerized

to

create

the

macromolecules

necessary

for

growth
.

From

this

point,

the

flux

through

all

reactions

can

be

predicted

using

flux

balance

analysis
.

To

compliment

this

direct

measurement

of

flux,

we

can

use

proteomics

and

metabolomics

to

evaluate

whether

predicted

fluxes

are

consistent

with

measured

fluxes

and

whether

modifications

of

the

mechanism

are

supported
.

This

will

allow

us

to

determine

how

flux

through

individual

enzymes

is

controlled

across

nutrient

conditions

and

ultimately

how

flux

is

regulated

at

a

systems
-
level
.

Background

In

order

for

microbes

to

have

prospered,

they

have

evolved

regulatory

mechanisms

that

allow

them

to

adjust

their

metabolic

strategy

under

varying

conditions
.

On

their

quest

for

division,

two

of

the

largest

challenges

that

they

face

are

how

to

attain

a

relatively

invariant

quantity

of

macromolecules

from

diverse

nutrient

sources,

where

some

resources

may

be

abundant

and

others

scarce
;

and

how

to

chemically

translate

increased

nutrient

availability

into

faster

production

of

macromolecules
.

Looking

at

nutrient

conditions

that

differ

in

the

nature

of

the

limiting

nutrient

and

how

much

is

available,

the

physiology

underlying

these

central

regulatory

questions

can

be

investigated
.

This

has

previously

been

done

at

the

transcriptional

and

metabolomic

level
.

Across

these

conditions

there

is

both

nutrient

independent

and

nutrient
-
specific

variation

in

flux

associated

with

changes

in

the

rate

of

division
.

This

suggests

that

the

control

of

flux

across

these

conditions

could

be

accomplished

either

through

hierarchical

control

of

flux,

where

variation

in

the

flux

through

a

reaction

is

related

to

the

level

or

activity

of

the

enzyme

or

through

metabolic

regulation

where

flux

is

controlled

by

substrate

occupancy
.


The

space

of

possible

fluxes

through

the

metabolic

network

can

be

bounded

by

determining

the

rates

that

molecules

are

taken
-
up

from

the

media,

and

the

rates

that

the

monomers

that

are

polymerized

to

create

macromolecules

are

created
.

Under

steady
-
state

conditions,

the

concentration

of

metabolites

won’t

change

over

time

so

flux

balance

will

be

satisfied

for

each

metabolite

and

mass

balance

of

nutrients

and

macromolecules

will

exist
.

With

these

boundary

fluxes

constrained,

pathway

degeneracies

can

be

resolved

by

incorporating

either

kinetic

information

from

isotopic

tracer

methods

or

through

an

estimate

of

flux

carried

related

to

predicted

absolute

flux
.

For

each

condition,

this

information

can

be

integrated

using

flux

balance

analysis,

in

order

to

find

an

optimal

set

of

fluxes

that

conforms

to

boundary

constraints

and

experimental

data














Brauer

2008

Boer 2010

Transcriptomics

Metabolomics

Quantifying fluxes

Explaining flux variation

While

the

information

necessary

to

quantify

flux

in

a

metabolism
-
wide

manner

is

still

accumulating,

analysis

of

a

limited

number

of

reactions

is

currently

possible
.



For

these

reactions,

the

flux

will

be

proportional

to

the

product

of

the

concentration

of

a

macromolecule

and

the

rate

which

volume

is

removed

from

the

reaction

vessel
.



Looking

at

ribonucleotide

synthesis,

the

intracellular

concentration

of

RNA

in

faster

dividing

cultures

implies

a

quadratic

relationship

between

dilution

rate

and

flux

into

ribonucleotides
.

Metabolism
-
wide strategy

A simplified case

P50 GM071508

Acknowledgements

To

determine

whether

modifications

to

reaction

mechanisms

should

be

sought

or

whether

predictions

of

flux

from

the

abundance

of

established

metabolites

and

enzymes

are

consistent

with

the

carried

flux,

the

best

linear

relationship

between

v

and

v
pred

can

be

sought
.

This

can

be

found

by

minimizing

the

coefficient

of

variation

of

v/
v
pred

over

a

set

of

parameters

Θ
.





For

a

set

of

reactions

where

the

flux

is

proportional

to

the

rate

of

RNA

synthesis,

this

approach

reveals

that

for

some

reactions

the

established

mechanisms

are

pretty

good,

while

for

others

the

predicted

flux

is

inconsistent

with

the

observed

flux
.


Ascertaining

the

concentration

of

species

is

an

experimental

challenge
.

What

we

will

actual

use

is

the

relative

concentration

of

enzymes

(using

proteomics)

and

the

absolute

concentration

of

metabolites

(we

currently

only

have

relative

for

both)

and

we

will

have

to

make

assumptions

about

the

subcellular

provenance

of

species,

initially

assuming

that

concentrations

are

uniform

regardless

of

compartment
.


For

most

reactions,

published

mechanisms

are

available,

as

well

as

binding

affinities

for

a

fair

number

of

metabolites,

but

these

affinities

are

inconsistent

and

we

may

be

missing

some

molecules

(or

covalent

modifications)

that

are

necessary

to

adequately

describe

flux

as

a

function

of

species

and

parameters
.

To

address

these

concerns

we

need

to

compare

predicted

to

true

flux

across

a

set

of

models

characterized

by

the

set

of

species,

the

nature

of

their

interaction

and

the

kinetic

parameters

involved

in

a

reaction
.

Recap

Scaling

the

simplified

case

up

to

a

metabolism
-
wide

strategy

will

be

conceptually

similar
.

Optimization

will

still

be

carried

out

on

a

per
-
reaction

basis,

but

the

desire

to

evaluate

different

classes

of

reaction

mechanisms

will

require

the

use

of

B
ayesian

statistics

and

more

powerful

optimization

tools,

namely

genetic

algorithms
.

The

likelihood

of

each

model,

characterized

by

a

set

of

parameters,

can

be

evaluated

as

the

product

of

the

densities

of

v
i

evaluated

at

a

mean

v
i_pred
.

This

likelihood

is

then

combined

with

the

prior

probability

of

the

values

of

parameters

to

create

a

bayes

factor
.

These

bayes

factors

can

then

be

viewed

as

the

fitness

of

a

model

and

this

fitness

can

be

optimized

through

the

process

of

parameter

mutation

and

selection
.






There

are

large

differences

in

patterns

of

flux

across

nutrient

conditions
.

These

are

primarily

related

to

the

growth

rate,

but

there

likely

limiting
-
nutrient

specific

differences

as

well
.


These

patterns

are

linked

at

the

level

of

reaction

mechanisms

to

variation

in

metabolites

and

enzymes


Understanding

the

source

of

variation

driving

flux

through

individual

reactions

will

allow

us

to

study

global

regulation

and

flux

control
.


Josh

Rabinowitz


Tomer

Shlomi


John

Storey


Keyur

Desai


David

Botstein


Pat

Gibney


Sandy

Silverman


Jonathan

Goya


David

Perlman