Transcript Document

Gene regulation and metabolic flux
reorganization in aerobic/anaerobic
switch of E. coli
Chao WANG
July 19, 2006
E. coli is a prokaryote model organism with relatively complete knowledge
on both transcriptional regulation (TR) and metabolism.
In response to external oxygen level, two global regulators, FNR and ArcA,
activate or repress a large number of enzymes, which in turn switches on/off
certain metabolic pathways.
Based on metabolic flux simulations and the known regulatory network, we
investigate the regulatory mechanisms underlying the presumably efficient
switch.
The target genes regulated by FNR and ArcA are compared with the
metabolic flux pattern generated from the Flux Balance Analysis (FBA)
under aerobic, micro-aerobic and anaerobic conditions, and their
physiological role examined.
We also compare the theoretical study with the microarray gene expression
data to cross-validate the data from different sources, thereby gaining a more
complete view of the regulatory processes involved.
Introduction
E. coli can grow with glucose as the sole organic constituent and metabolically it can
transform glucose into all of the macromolecular components that make up the cell. E.
coli can grow in the anaerobic or aerobic environments.
Fnr, DNA binding activity is found to
be associated with the [4Fe-4S]2+
form, not with the [2Fe-2S]2+ species.
ArcB can sense changes in the
electron transport chain. Also ArcB
responds to metabolites. ArcB
undergoes autophosphorylation, and
the phosphoryl group is transferred to
ArcA
Transcriptional Regulation Network
Metabolic Network from KEGG
The metabolic sub network corresponding to Fnr and/or ArcA. Blue box:
enzymes up-regulated by Fnr. Black diamond: enzymes up-regulated by ArcA.
Yellow triangle: enzymes up-regulated by both Fnr and ArcA. Red ellipse:
compounds.
Expression Data
1
ArcA deletion aerobic
2
ArcA deletion
anaerobic
3
FNR deletion aerobic
4
FNR deletion
anaerobic
5
ArcA/FNR double
deletion aerobic
6
ArcA/FNR double
deletion anaerobic
7
Wild type aerobic
8
Wild type anaerobic
Flux Balance Analysis of iJR904 Model
carbon source (e.g. glucose)
flux fixed
Freely available compounds Na+, K+,
NH4+, SO4-2, H2O, CO2, H+, P, Fe
aerobic/anaerobic
(oxygen)
biomass
S v =0
waste
89 external compounds can sustain model growth. Carbon source is
used both as carbon atoms and energy source.
The Simulation from Anaerobiosis to Aerobiosis
We feed the glucose as the carbon source and set the
glucose uptake rate with a max of 10.
We set the oxygen uptake rate from 0 to 20, gradually
increased by 1, which can simulate the external oxygen
level from anaerobiosis to aerobiosis.
Metabolic Flux Patterns
There are totally 334 flux
carrying reactions with different
frequency. And 260 of these
reactions have none zero flux for
all the 21 conditions.
In anaerobic and aerobic
conditions, 258 reactions
increase or reduce their flux rate
in the same direction, 50
reactions switch on or off their
flux and only 10 reactions
change their flux’s direction.
Fnr/ArcA Target Portion in the Simulation Process
Anaerobiosis
Aerobiosis
TCA Cycle Metabolic Flux Reorganization
Missing information, for example, ‘akg + coa + nad --> co2 + nadh + succoa’,
catalyzed by enzyme with EC number ‘1.2.4.2’ , coded by genes sucAB.
Aerobiosis
Anaerobiosis
Linear Superpostion of Biomass Solutions
With the initial onset of anaerobiosis, ArcA is activated, and if this condition
persists or becomes more severe, Fnr is activated. We assume both Fnr and ArcA
are involved in this adaptation process with respective effect. And this effect will
gradually vary with the change of oxygen level. Given the biomass yields in
anaerobic and aerobic conditions, we can get a linear superposition solution for the
simultaneous optimization of them with the linear proportion.
Oxygen
uptake
rate
Optimal solution
Linear superposition
solution
0
0.216255
0.216255
1.000000
1
0.251540
0.257339
0.977464
2
0.286824
0.298144
0.962033
3
0.322109
0.338949
0.950317
4
0.357394
0.379754
0.941119
5
0.392678
0.420557
0.933710
6
0.427963
0.461363
0.927606
7
0.463248
0.500188
0.926147
8
0.498532
0.537513
0.927479
9
0.533817
0.574843
0.928631
10
0.569102
0.612168
0.929649
11
0.604386
0.649498
0.930544
12
0.639671
0.686825
0.931345
13
0.674955
0.724153
0.932062
14
0.710240
0.761480
0.932710
15
0.745525
0.790232
0.943425
16
0.780809
0.816981
0.955725
17
0.816094
0.843727
0.967249
18
0.851379
0.870480
0.978057
19
0.886663
0.897230
0.988223
20
0.921948
0.921948
1.000000
Ylsp / Yopt
The interruption of TCA cycle in
anaerobic condition is due to the
silence of reaction ‘akg + coa +
nad --> co2 + nadh + succoa’,
which should be catalyzed by the
enzyme with EC assignment
[1.2.4.2].
The corresponding gene to the enzyme [1.2.4.2] is sucA. From Chapter 2 we
know this gene is repressed by regulator Fnr in anaerobic condition. We check the
gene expression data, in the eight experimental conditions that are described in
Chapter 4, to obtain the expression ratio that is [-4.6860 -4.5878 -5.1599 6.6164 -4.9745 -4.7465 -4.8279 -7.1273] respectively. In wild type condition
its expression ratio is the lowest. Such information assume that the gene sucA
plays a very key role in the anaerobic/aerobic switch.
Once the TCA cycle is interrupted
the flux rate through related
reactions will reduce significantly,
which can be illustrated in 7.2.
Particularly we investigate the
reaction ‘atp + coa + succ <==> adp
+ pi + succoa’, which is the reaction
following the interrupted reaction.
The enzyme catalyzing this reaction has the EC assignment [6.2.1.5]. The
corresponding genes are sucC and sucD. Both them are repressed by Fnr in
anaerobic condition. After deleting the aerobic/anaerobic response factors the
expression ratios of these two genes increase significantly more than 50%.
These three reactions, 'fum +
mql8 --> mqn8 + succ' (FRD2),
'2dmmql8 + fum --> 2dmmq8
+ succ' (FRD3) and 'fad + succ
--> fadh2 + fum' (SUCD1i), are
all catalyzed by the enzyme
with assignment [1.3.99.1]
whose corresponding genes are
sdhABCD and frdABCD. It is
very interesting that sdhABCD
are repressed by Fnr whereas
frdABCD are activated by Fnr.
Obviously enzymes coded by sdhABCD play the role in reaction SUCD1i whose
flux is from Succinate to Fumarate, which is reverse to the other two reactions.
Such direction should be present in TCA cycle. In anaerobic condition the flux rate
of FRD2 is not zero, which indicate that this reaction should be catalyzed by the
enzymes coded by gene frdABCD.
Many other reactions’ flux rate
change can also be
characterized based on our
transcriptional regulation and
gene expression information.
But some reactions can not be
interpreted clearly. For
example, the reaction 'cit
<==> icit' shows some unclear
mechanism.
This reaction is catalyzed by the enzymes with assignment [4.2.1.3], which can
be coded by genes acnB and acnA. These two genes are both repressed by
ArcA. Maybe ArcA only plays its role when the oxygen uptake rate is less than
14.
In aerobic conditions ArcA has no influence on these anaerobic
respiratory pathway genes, which can be validated by gene expression
date.
In aerobic conditions, oxygen is the main electron acceptor. While in
aerobic conditions, electron can be accepted by some metabolites or
nitrogen. The reaction ‘(2) h[c] + no3[c] + q8h2[c] --> (2) h[e] +
h2o[c] + no2[c] + q8[c]’ is very important in anaerobic conditions,
which can discharge the electrons by transforming NO3 into NO2.
This reaction is catalyzed by the enzymes with assignment [1.7.99.4],
which can be coded by genes narGHJI. Gene narGHJI are anaerobic
respiratory pathway genes (Cotter PA and Gunsalus RP, 1992), which
are activated by Fnr in anaerobic conditions. But in our iJR904 model,
we obtain zero flux rate for this reaction. So from this aspect we
should update and consummate our model.