Content uploaded by Zsolt Bodor
Author content
All content in this area was uploaded by Zsolt Bodor on Jun 08, 2020
Content may be subject to copyright.
Copyright © 2020 University of Bucharest Rom Biotechnol Lett. 2020; 25(3): 1666-1676
Printed in Romania. All rights reserved doi: 10.25083/rbl/25.3/1666.1676
ISSN print: 1224-5984
ISSN online: 2248-3942
*Corresponding author: ZSOLT BODOR, Sapientia Hungarian University of Transylvania, Department of
Bioengineering, Libertatii square, No. 1, 530104, Miercurea Ciuc, Romania
E-mail: bodorzsolt@uni.sapientia.ro
Received for publication, December, 15, 2018
Accepted, August, 5, 2019
Original paper
Using genome-scale model to predict the metabolic
engineering impact on Escherichia coli metabolism
during succinic acid production optimization
ZSOLT BODOR1,*, ANDREA (IUHÁSZ) FAZAKAS2, KATALIN BODOR3,
ERIKA KOVÁCS1, ILDIKÓ MIKLÓSSY1, BEÁTA ALBERT1
1Sapientia Hungarian University of Transylvania, Department of Bioengineering, Libertatii
square, No. 1, 530104, Miercurea Ciuc, Romania
2“Politehnica” University of Bucharest, Department of Inorganic Substances Technology
and Environment Protection, Polizu street, No 1-7, 011061, Bucharest, Romania
3University of Pécs, Faculty of Natural Sciences, Doctoral School of Chemistry, Ifjúság 6,
7624 Pécs, Hungary
Abstract
In this study, we investigated the production possibility of succinic acid from renewable
resources, such as glucose and glycerol (by-products resulted from different industries), and
the impact of metabolic engineering on cellular metabolism was estimated using systems
biology approaches. The cellular metabolic flux was estimated for aerobic, microaerobic and
anaerobic conditions. The in silico simulation of metabolic processes in Escherichia coli has
been concluded the possibility of increasing the yield of succinic acid from 0.008 to 0.9 (mol
mol-1 glucose) and from 0.0003 to 0.6 (mol mol-1 glycerol) by removing three specific genes
from the genome (pyruvate formate lyase – pflB, lactate dehydrogenase – ldhA and alcohol
dehydrogenase – adhE) (anaerobic conditions). Comparing the microaerobic and anaerobic
conditions was clear that a very small amount of oxygen has a vital contribution on cell survival.
We have determined the deletion impact (p) on growth rate of each of the 1363 genes one by
one under aerobic, microaerobic and anaerobic conditions in wild-type and mutant strains. Wet
experiments were carried out and highest molar yield was obtained for glycerol, 0.51 mol mol-1.
This study informs other studies that environmental conditions have a stronger influence on the
genes with reduced maximal growth than the elimination of different genes from redox
metabolism. The knowledge gained in this study would help in microbial cell factories design
for succinic acid production from renewable resources under various environmental conditions
Keywords
In silico, Escherichia coli, succinic acid, metabolic engineering, glucose, glycerol.
To cite this article: BODOR Z, FAZAKAS AI, BODOR K, KOVÁCS E, MIKLÓSSY I,
ALBERT B. Using genome-scale model to predict the metabolic engineering impact on
Escherichia coli metabolism during succinic acid production optimization. Rom Biotechnol
Lett. 2020; 25(3): 1666-1676. DOI: 10.25083/rbl/25.3/1666.1676
Using genome-scale model to predict the metabolic engineering impact on Escherichia coli metabolism
1667
Introduction
The role of in silico platforms is essential for
the rational design of networks in order to improve the
productivity of a microorganism, to increase the production
of a target compound as well as to analyze the intracellular
processes (LEWIS & al [1]; CHAN & al [2]). Systems
biology has been shown to be successful in metabolic
engineering (BODOR & al [3]; FONG [4]; FONG & al [5];
LEE & al [6], [7], [8]; MIKLÓSSY & al [9]; NA & al [10];
YIM & al [11]) especially in predicting the outcomes
(e.g., production-, growth rates, etc.), while constraint-based
modelling predicts the gene knockouts with reasonable
accuracy (COOPER & DUFFIELD [12]; HUANG & al
[13]; TEPPER & SHLOMI [14]). Different genome-scale
models of metabolism (GSM) have been generated
describing the cell and the processes inside (FEIST &
PALSSON [15]; KIM & al [16]; ORTH & al [17]; SAHA,
CHOWDHURY & MARANAS [18]). Recently, the
generation of cell factories is directly linked to a holistic
approach which involves several disciplines and techniques
like genomics, proteomics, metabolomics, fluxomics,
transcriptomics and phenomics (PEY & al [19]). Rational
analysis, based on knowledge of cellular processes
provides the opportunity for optimization using the highly
curated genome-scale metabolic models. This widely used
systems metabolic engineering enables target-oriented
redesign of metabolism to increase the production of the
target metabolite and to cope with huge complexities in
biology (FAULON & CARBONELL, [20]; LEE & al [6];
LEE & al [21]; LEWIS & al [1]; STEPHANOPOULOS,
[22]). Hence, in silico analysis is mandatory in deciphering
metabolic changes under different environmental condi-
tions and genetic manipulations (FEIST & al [23]; LEWIS
& al [24]; BORDBAR & al [25]; ANDREOZZI & al [26).
In silico techniques were successfully used to design
industrially important strains (HANSEN & al [27]; LEE &
al [28]; MAIA & al [29]; YIM & al [11]), to analyse
metabolic processes (CAMPODONICO & al [30]; LISHA
& al [31]; RAMAN & al [32]). The relationship between
genotype and phenotype can be analysed by using the well-
known methods used in genetic engineering and phenotype
analysis (CHOI & al [33]; NIELSEN & al [34]).
In vitro mutations, for example, can be performed
by the λ-Red recombinase method for chromosomal gene
disruption (DATSENKO & al [35]) and changes in meta-
bolite profiles can be identified by using HPLC or GC-MS.
The production of biofuel generates large quantities
of glycerol as waste (WENDISCH & al [36]) what could be
an inexpensive carbon source (such as glucose) for many
microorganisms (GARLAPATI & al [37]; YANG & al
[38]; YIM & al [11]) and can be used as an appealing
substrate for high-value added metabolites production
like succinic acid. Succinic acid has received increasing
attention during the last decades because it has an important
role in the synthesis of high value added chemicals
(WERPY and PETERSEN, [39]; LIANG & al [40];
SALVACHÚA & al [41]). The chemical synthesis through
catalytic hydrogenation of maleic acid is highly dependent
on fossil resources, which gives rise to major concerns
about sustainability. On the other hand, the environmental
pollution even at regional levels should be taken into
consideration (ILIE & al [42]; KERESZTESI & al [43];
SZÉP & al [44-50]). Taken together, biotechnology offers
new sustainable alternatives, however, there are still many
questions to be answered in order to better understand the
interactions inside a cell, especially if different genetic
modifications are carried out or the environmental condi-
tions are modified.
The present work aimed at the analysis of succinic
acid production impact on E. coli metabolism, under
different genetic and environmental conditions. For this
purpose systems biology tools were applied, namely the
constraint-based approach of metabolic modeling (LEWIS
& al [1]). The most important contributions are as follows:
construction of an industrially important strain for succinic
acid production from renewable resources such as glucose
or glycerol, identification of the impact of gene knockouts
and conditions on gene essentiality and finally the flux
variations in wild-type and mutant strains were identified
using flux variability analysis (FVA). To test the produc-
tion potential basic experiments were carried out under
anaerobic conditions.
Materials and Methods
A previously developed metabolic model of
E. coli K12 MG1655 (ORTH & al [17]), available
in SBML format at BioModels online database
(http://www.ebi.ac.uk/ biomodels-main/), was used for
in silico studies. In order to avoid unrealistic behaviors and
to improve the predictions quality the upper and lower
bounds for the reactions CAT (catalase), DHPTDNR
(dihydropteridine reductase), DHPTDNRN (dihydropte-
ridine reductase (NADH)), FHL (Formate-hydrogen lyase),
SPODM (superoxide dismutase), SPODMpp (superoxide
dismutase – periplasm), SUCASPtpp (succinate:aspartate
antiporter (periplasm)), SUCFUMtpp (succinate:fumarate
antiporter (periplasm)), SUCMALtpp (succinate:malate
antiporter (periplasm)), and SUCTARTtpp (succinate:
D-tartrate antiporter (periplasm)) were constrained to
zero. The steady-state flux distribution was calculated
by using MATLAB (The MathWorks Inc., Natick, MA,
USA) and COBRA Toolbox (BECKER & al [51];
SCHELLENBERGER & al [52]) software packages with
TOMLAB CPLEX ((Tomlab Optimization Inc., San
Diego, CA, USA) and GUROBI (Gurobi Optimizer version
6.0, Houston Texas) solvers.
Flux balance analysis (FBA)
Growth was simulated by maximizing flux through
a defined biomass objective function. Steady-state balances
ZSOLT BODOR et al
1668
and stoichiometry of all metabolites are imposed as linear
constraints as follows:
(1)
- where S is an m · n stoichiometric matrix containing
all the stoichiometric coefficients in the model of m
metabolites and n reactions, v is the flux vector.
- constraints vl≤v≤vu; vl and vu are vectors with n
elements each, which represent the lower and upper bounds
on the fluxes, respectively. Because relationships are linear,
the metabolic objective can be estimated by linear
optimization:
(2)
In this equation, c is a row vector containing
weighting factors for individual fluxes (v), while z is
the objective function. Flux units are in mmol
gDW-1h-1 (millimoles dry cell weight per hour), except
biomass which has units of h-1 (BECKER & al [51];
SCHELLENBERGER & al [52]).
In silico implementation and prediction of
growth on glucose and glycerol under different
environmental conditions
The inputs were restricted to minimal media (M9)
containing only inorganic salts and glucose or glycerol as
carbon sources. Growth on glucose was simulated by
constraining the lower bound of the glucose exchange
to the experimentally determined 10 mmol gDW-1h-1
(VARMA & PALSSON, [53]). For growth on glycerol, the
lower bound of the glucose exchange reaction was set to
zero, while the lower bound of the glycerol exchange
reaction was set to 10 mmol gDW-1h-1. Under aerobic
growth, the oxygen uptake was set to 1000 mmol gDW-1h-1
(unlimited oxygen uptake), for microaerobic we used
5 mmol gDW-1h-1 and 0 to create anaerobic conditions.
The eliminated reactions were: ΔpflB (pyruvate
formate lyase), ΔldhA (lactate dehydrogenase), ΔadhE
(alcohol dehydrogenase). Using glycerol as the sole carbon
source the cell was able to grow aerobically and micro-
aerobically, meanwhile, the pflB gene elimination blocked
the biomass formation under anaerobic conditions.
Different co-substrates were tested to improve cell viability
including glucose and amino acids (data not shown here).
Large-scale in silico evaluation of single gene
deletions was conducted to determine the effect of gene
deletion on cellular growth including the wild-type and
designed mutant strains. Genes were categorized as non-
lethal (unchanged maximal growth), with reduced-lethality
(reducing maximal growth) and lethal/essential (no growth).
Flux variability analysis (FVA)
FVA is a method widely used to identify the
maximum and minimum possible fluxes through each
reaction of the genome-scale metabolic network for a
given maximum objective value (GHOSH, & al [54];
MAHADEVAN & SCHILLING, [55]). Calculations were
carried out by following the descriptions presented by
(BECKER & al [51]). Briefly, the optimal predicted
growth rate was constrained to 100% of the optimal value
under various environmental conditions and the minimum
and maximum values were estimated to decipher the
redundancy of reactions in the network. Two optimization
problems need to be solved during FVA simulations for
each flux vj of interest,
maxνj/minνj
subject to
(6)
where Zbiomass is biomass optimal solution. Considering n as
the number of reactions than FVA requires the 2n LP
problems to be solved as mentioned previously.
Bacterial strain, culture conditions
The strain used in this study was E. coli K12 MG1655
from “Deutsche Sammollung von Mikroorganismen
und Zellkulturen GmbH” (DSMZ 18039). All the work
was done at 37°C, in minimal medium under anaerobic
conditions with glucose or glycerol at a concentration of
5 g/L. Bacterial cells were grown in minimal medium in
5 mL medium to produce a starter culture and the seed
culture was used to inoculate the fermentation medium
(optical density 0.1). Cells were grown with shaking at
150 rpm (Certomat BS-1 Sartorius) for 24 h in serum
bottles (50 mL) with 20 mL M9 medium. Samples were
taken in every two hours for the analysis of cell growth.
The optical density of the cell cultures was measured at
550 nm (OD550) to quantify cell growth, using a Cary 50
Conc UV-Visible spectrophotometer.
Chromosomal gene deletion
For gene deletion strategy we used the λ-Red recom-
bineering methods previously described by DATSENKO &
al [35]. Plasmids (5 Strain Wanner Lambda Red Gene
Disruption Kit) were obtained from the E. coli Genetic
Stock Center (Yale University).
Analytical procedure
Metabolites were analysed with GC-MS (6890N/5975
Agilent) based on solid-phase micro extraction (SPME)
with on-fiber silylation. Silylation was carried out using
N, O bis trifluoroacetamide (BSTFA) following the
procedures described by (LUAN & al [56]). For chromato-
grams and spectra analyses we used the MassLab software
(ThermoQuest, Manchester, UK). Compounds were identified
by comparing the mass spectra obtained with commercially
available MS libraries (Wiley, NIST and LIBTX).
Results and Discussions
Under aerobic conditions, E. coli converted the
substrates quantitatively to biomass and CO2, without any
by-product formation (highest biomass production). During
microaerobic and anaerobic conditions in minimal
medium, pflB is highly active, catalyses the CoA-
dependent cleavage of pyruvate to form acetyl-CoA and
formic acid. Acetyl-CoA is then converted to acetic acid
and ethanol (Fig. 1).
Using genome-scale model to predict the metabolic engineering impact on Escherichia coli metabolism
1669
Figure 1. Routes for overproduction of succinic acid in the mutant strain. Crosses represent the reaction knockouts;
the following pathways were eliminated (represented by a black x): pyruvate formate lyase- ΔpflB (formate), lactate dehydrogenase-
ΔldhA (lactate), alcohol dehydrogenase-ΔadhE (ethanol); Abbreviations: pts-D-glucose transport via PEP:Pyr PTS (periplasm);
pgi – glucose-6-phosphate isomerase; pfk – phosphofructokinase; fba – fructose-bisphosphate aldolase; tpi – triose-phosphate
isomerase; gapd glyceraldehyde-3-phosphate dehydrogenase; glpK – glycerol kinase; gldA – glycerol dehydrogenase;
glpD – Sn-glycerol-3-phosphate dehydrogenase; dhaKLM – dihydroxyacetone kinase; pgk – phosphoglycerate kinase; pgm –
phosphoglycerate mutase; eno – enolase; ppc – phosphoenolpyruvate carboxylase; pyk – pyruvate kinase; ldh-D–lactate dehydrogenase;
pfl – pyruvate formate lyase; pdh – pyruvate dehydrogenase; pta – phosphotransacetylase; ackA – acetate kinase; acald – acetaldehyde
dehydrogenase (acetylating); adhE2 – alcohol dehydrogenase (ethanol); cs – citrate synthase; acontA – aconitase (half-reaction A, Citrate
hydro-lyase); acontB – aconitase (half-reaction B, Isocitrate hydro-lyase); icd – isocitrate dehydrogenase (NADP); akgdh –
2-Oxoglutarate dehydrogenase; sucoas – succinyl-CoA synthetase (ADP-forming); sucd – succinate dehydrogenase (irreversible);
fum – fumarase; mdh – malate dehydrogenase; cat – catalase; sucCD – succinyl-CoA synthetase.
Deletion of pflB had little impact on cell growth in
M9 mineral salts medium using glucose as the sole carbon source because the carbon flow is diverted to
lactic acid (Table 1).
Table 1. Inactivating alternative NADH oxidizing pathways in M9 mineral salts medium (anaerobic conditions)
Wild-type
ΔpflB
ΔpflB, ΔldhA
ΔpflB, ΔldhA, ΔadhE
µ
(h-1)
0.24
0.19
0.19
0.11
YSucca
(mol mol-1)
0.008
0.006
0.006
0.910
YForb
(mol mol-1)
1.730
0
0
0
YLacc
(mol mol-1)
0
1.720
0
0
YEtOHd
(mol mol-1)
0.803
0
1.720
0
YAce
(mol mol-1)
0.820
0
0
0.430
YAlaf
(mol mol-1)
0
0
0
0.490
YCO2
(mol mol-1)
0
0.040
1.760
0
Yield (Y); aSuccinic acid; bFormic acid; cLactic acid; dEthanol; eAcetic acid; fAlanine
ZSOLT BODOR et al
1670
Lactic acid production was blocked by blocking
the lactate dehydrogenase enzyme (ldhA), however, the
production yield of succinic acid did not change. The yield
of succinic acid increased to 0.9 mol mol-1 glucose with
adhE deletion when both of the alternative NADH
oxidation pathways were inactivated, but growth was
reduced by more than 40% compared to the double mutant
(ΔpflB, ΔldhA).
The main difference between microaerobic and
anaerobic conditions is as follows: acetic acid production
(higher during microaerobic) and succinic acid (higher
during anaerobic conditions). On the other hand, alanine
was present only under anaerobic conditions and in
mutant strains.
Glycerol as carbon source
Glycerol is generated in huge quantities as a
by-product during biofuel production, hence the value-
added utilization in biotechnology is a promising future.
The growth of wild-type E. coli K12 MG1655 was slowly
on glycerol even under aerobic conditions. The reason is
the difference between substrates molecular weights, but
after calculations, it was clear that mass yield of cell-mass
on glucose was 0.43 gDW g-1 glucose while in case of
glycerol 0.53 under aerobic conditions. With an uptake
rate of 10 mmol glycerol the maximal growth rate was
0.55 under aerobic conditions, 0.33 under microaerobic
and 0.08 (h-1) under anaerobic conditions.
To block formic acid production the pflB reaction
must be deleted as stated out for glucose, but our results
show that the ΔpflB mutant fails to grow anaerobically.
We decided to test different co-substrates including glucose
and amino acids to improve cell viability. Three best
candidates were detected, glutamic acid, glutamine and
glucose. To modify the redox pathway and to estimate
the effect on cell metabolism all 3 genes (ΔpflB, ΔldhA,
ΔadhE) were eliminated as mentioned before. Using
glucose as co-substrate the triple mutant succinic acid
yield increased to 0.6 (mol mol-1 glycerol) with 1 mmol
gDW-1 h-1 glucose uptake rates, but the growth rate was
significantly reduced (only 0.001 h-1). Changing the environ-
mental conditions- (limited oxygen availability/elimination
of O2) reduced biomass formation nearly by 58%-89%
(wild-type – mutants) on glucose and 47%-99% on glycerol.
Figure 2 shows the deletion impact of the total genes
(1363) on wild-type and mutant strains under different
environmental conditions and substrate availability, and
the distribution of relative growth rates for all the gene
deletions in the model is presented. Using this method
a good agreement with experimental deletion studies was
reported earlier by (BECKER & al [51].
With glucose as the sole source of carbon 200 of the
1363 model genes (Fig. 2. IA) were found to be lethal
(critical genes), and 48 gene deletions resulting in reduced
maximal growth rates under aerobic conditions in minimal
medium. We were interested to investigate if the triple
mutation has any effect. We found that the number of
critical genes did not change and minimal changes were
found in the list of genes with reduced maximal growth
rates (Fig. 2. ID). Similar simulations were carried out for
microaerobic and anaerobic conditions. Microaerobic
predictions were in strong correlation with the aerobically
obtained ones, 200 critical genes for wild-type and mutant
strains, 45 (Fig. 2. IB) respectively 54 genes (Fig. 2. IE)
with reduced growth rates. Eliminating the oxygen from the
system has increased the number of essential genes (Fig. 2.
IC) to 204 and the growth rate was reduced significantly
(~40%) in case of ten genes. The list of essential genes of
the triple mutant strain (Fig. 2. IF) remained the same as in
the case of wild-type (anaerobic conditions), suggesting
that the modification of the redox metabolism does
not change the number of essential genes. However, a
significant difference was observed between genes with
reduced growth rates (wild-type 25, mutant 47).
A similar tendency was observed after changing
the carbon source to glycerol. Out of the 1363 genes in
the model, 199 genes were considered lethal under aerobic
(Fig. 2. IIA) and microaerobic conditions (Fig. 2. IIB) and
204 under anaerobic (Fig. 2. IIC) conditions. Genes with
reduced maximal growths were as follows: 43 aerobic, 42
microaerobic and 21 for anaerobic conditions. Eliminating
important genes from pyruvate metabolism has minimal
impact on essential genes under aerobic (Fig. 2. IID) and
microaerobic (Fig. 2. IIE) conditions (199) in wild-type.
Interestingly under anaerobic conditions in the mutant
strain with singleGeneDeletion we were unable to detect
genes with reduced maximal growth rates (Fig. 2. IIF).
Changing the co-substrate to amino acids (glutamic
acid or glutamine) under anaerobic conditions no signifi-
cant differences were detected in essential genes compared
to glucose (as co-substrate), and 8 genes were detected
with reduced maximal growth.
The results indicate that E. coli metabolic network
is robust to different changes and may activate other
pathways (on different conditions) to strive alive. As we
expected, minimal changes were observed in lethal genes
after environmental and genetic modifications, however,
genes with reduced maximal growths were mostly affected.
Taking into account the environmental conditions (aerobic,
microaerobic, anaerobic and the carbon sources) as well
as the mutations (ΔpflB, ΔldhA, ΔadhE) we can conclude,
that the profile of genes with reduced maximal growth
rates was much more affected by substrates and the
presence of oxygen than by genetic modifications.
Dynamic FBA of diauxic growth
Simulations were carried out by combining FBA
with an iterative approach based on a quasi-steady-
state assumption to identify diauxic growth. Briefly,
the substrate concentration (Sc) is determined from the
previous step (Sco) if it is the first time step from the initial
substrate concentration. The amount of substrate available
per unit of biomass per unit of time is estimated using
the equation:
(3)
where Sa is the amount of substrate available and X
(gDW/L) represents cell density. Then the substrate uptake
(Su) and growth rate (µ) is calculated using FBA. The final
concentrations for the next step are calculated from the
differential equations:
(4)
(5)
Using genome-scale model to predict the metabolic engineering impact on Escherichia coli metabolism
1671
Considering the predictions, diauxic growth was
not possible under anaerobic conditions, however under
aerobic and microaerobic conditions the metabolites
(formic acid, acetic acid, ethanol or succinic acid) initially
secreted were subsequently metabolized after glucose or
glycerol exhaustion.
Figure 2. Single gene deletion results of 1363 genes (wild-type) and 1360 (mutant), distribution of relative growth rates;
wild-type on glucose (I.) and glycerol (II.) under aerobic (A), microaerobic (B) and anaerobic (C) conditions: mutant strains
(ΔpflB, ΔldhA, ΔadhE) under aerobic (D), microaerobic (E) and anaerobic (F) conditions.
Flux variability analysis within the metabolic
network for wild-type and mutant strains
FVA was carried out by setting growth rates as the
objective function and the maximum and minimum
possible fluxes through each reaction was identified.
Briefly, the optimized objective value obtained with FBA
was used as a constraint during calculations of feasible
ranges of reaction fluxes (minimization and maximization
of each flux value). Substrate consumption rates were
set to 10 mmol gDW-1h-1, while the oxygen uptake rate
ZSOLT BODOR et al
1672
was modified to create anaerobic conditions. To analyse
the impact of environmental conditions and the genetic
modifications on flux ranges the changes were calculated
for each condition. In Fig. 3 are presented reactions with a
normalized span >0.01 and sorted by magnitude. FVA is
presented for wild-type and mutant strains on glucose and
glycerol under anaerobic conditions.
Figure 3. Range of flux values for wild-type and mutant strains analysed by flux variability analysis, A. for glucose and
B. for glycerol. The span was normalized to the maximal flux span identified. Reactions with a normalized span >0.01 are
shown for strains under anaerobic conditions.
In order to test the impact of knockouts FVA was
implemented to calculate for each reaction in the model the
flux variations under certain conditions. In this analysis two
situations were compared: i) wild-type and mutant on
glucose; ii) wild-type and mutant on glycerol. Fig. 3. shows
the normalized values of the reactions span. Some reactions
had an infinite span in both cases, (totally 6 reaction: ADK1
(adenylate kinase), ADK3 (adentylate kinase (GTP)),
NDPK1 (nucleoside-diphosphate kinase (ATP:GDP)),
ALATA_L (L-alanine transaminase), VALTA (valine
transaminase), VPAMTr (Valine-pyruvate aminotrans-
ferase)), meaning that these reactions are not constrained.
Many other reactions had a normalized span of 1.5 totally
70 and this number did not change during substrate
availability. The remaining reactions are characterized by a
normalized span <0.02. Important reactions with relevant
biological meaning are as follows: TALA (transaldolase),
FBA (fructose-bisphosphate aldolase), FBA3 (Sedohep-
tulose 1,7-bisphosphate D-glyceraldehyde-3-phosphate-
lyase), PFK (phosphofructokinase), PFK_3 (phospho-
fructokinase (s7p)), F6PA (fructose 6-phosphate aldolase),
DHAPT (dihydroxyacetone phosphotransferase), PYK
(pyruvate kinase) for WT on glucose, and CITt3pp (citrate
transport out via proton antiport (periplasm)), CITt7pp
(citrate transport via succinate antiport (periplasm)),
SUCCt3pp (succinate transport out via proton antiport
(periplasm)), FRD2 (fumarate reductase), FRD3 (fumarate
reductase – 2-Demethylmenaquinol 8), NADH17pp
(NADH dehydrogenase (menaquinone-8 & 3 protons)
(periplasm)), NADH18pp (NADH dehydrogenase
(demethylmenaquinone-8 & 3 protons) (periplasm)),
FBA, TALA, FBA3, PFK, PFK_3, EX_h(e) (H+ exchange),
Htex (proton transport via diffusion (extracellular to
periplasm), EX_co2(e) (CO2 exchange), CO2tex (CO2
transport via diffusion (extracellular to periplasm)),
CO2tpp (CO2 transporter via diffusion (periplasm)),
THD2pp (NAD(P) transhydrogenase (periplasm)), FADRx
(FAD reductase), GLUDy (glutamate dehydrogenase),
PDH (pyruvate dehydrogenase), EX_ala-L(e) (L-Alanine
exchange), EX_nh4(e) (ammonia exchange), ALAtex
(L-alanine transport via diffusion (extracellular to periplasm)),
NH4tex (ammonia transport via diffusion (extracellular
to periplasm)), NH4tpp (ammonia reversible transport
(periplasm)), EX_ac(e) (acetate exchange), ACtex (acetate
transport via diffusion (extracellular to periplasm)) for
mutant, respectively. On the other hand, on glycerol the
normalized flux span with at least 0.01 was predicted only
for mutant strains and the reactions are: CITt3pp (citrate
transport out via proton antiport (periplasm)), CITt7pp
(Citrate transport via succinate antiport (periplasm)),
FRD2, SUCCt3pp (succinate transport out via proton
antiport (periplasm)), FRD3, NADH17pp, NADH18pp,
EX_ala-L(e), ALAtex. The result of the FVA analysis
shows which fluxes are influenced by genetic engineering
in a certain way; there are changes in reactions with
large and small flux span in both substrates.
Time-course fermentation experiments were carried
out for wild-type and mutant strains (ΔpflB, ΔldhA, ΔadhE)
to follow the changes of succinic acid. Results presented
here (Fig. 4) have been obtained during anaerobic
fermentations. Identification of metabolites was performed
in a GC-MS experiment, data analysis was carried out
with MassLab via comparison with mass spectra obtained
from different libraries.
A
B
Using genome-scale model to predict the metabolic engineering impact on Escherichia coli metabolism
1673
Figure 4. GC-MS analysis of determination of succinic acid from supernatants; I. glucose and II. glycerol
(wild-type (A), mutant (B), (arrows indicate succinic acid peaks).
The compounds present in the fermentation mixtures
were identified by GC-MS. The heights of the peak indicate
the relative concentrations of the metabolites present in the
fermentation mixtures. For example, in the fermentation
mixtures the succinic acid was identified as trimethyl-
silylated succinic acid at 10’71” together with acetic acid,
phosphoric acid, lactic acid, propanoic acid, butanoic acid,
glucose and glycerol. The peak of succinic acid in mutant
strains was significantly higher compared to wild-type
strains (in both cases) and the molecular ion peak was at
m/z 262 (succinic acid+2 trimethylsilyl) and the fragment
ion peaks at m/z 247 (succinic acid+2 trimethylsilyl-
methyl) and m/z 147 (2 trimethylsilyl+H+). Succinic acid
peak area for wild-type and mutant strains on glucose
was 0.65% and 12.14%, and on glycerol 3.77% and
29.69%, respectively.
As a proof of concept experiment, we demonstrate
that using systems biology approach the production
potential of succinic acid in E. coli under anaerobic
conditions is a sustainable approach. Significant changes
were observed between wild-type and mutant strains on
both substrates. To our surprise, the presence of succinic
acid in the fermented broth was significantly higher in
case of glycerol compared to glucose. Succinic acid
concentration after 6 days was as follows: wild-type
3.43±0.15 mM and 13.5±0.08 mM with a molar yield
of 0.51 mol mol-1. The production in mutant strain was
4x higher compared to wild-type. As stated out earlier
(JEONG & al [57]) during adaptive laboratory evolution
the production potential can be significantly improved in
order to reach a stable and industrially important strain with
increased succinic acid yield and productivity on both
substrates. This is an initial strategy for strain development
and further analysis will be carried out using knowledge,
genome-scale metabolic models and constraint/based
modeling strategies to identify new gene targets to further
improve the production potential.
Conclusions
Sustainable development is environment-connected
and needs to be harmonized to sustain a continuous growth,
and hence, the application of (bio-based) green techno-
logies is crucial. A potential approach for conversion of
biomass and renewable resources to fuels and chemicals
is to combine biochemical and chemical processes.
Genetically engineered strains were designed and
constructed using systems biology and metabolic engine-
ering methods, which are capable of metabolizing glucose
and glycerol (resulted from different industries as by-
products) and producing succinic acid a key platform
chemical. The present work presents in silico and
experimental methods for strain simulation and genetic
modification to design strains by deleting different genes
and testing the substrate and environmental effects on
cell metabolism. The λ-Red recombineering technology
was successfully used for chromosomal modifications
in E. coli.
The highest succinic acid production was under anae-
robic conditions, using glucose the yield was 0.9 mol mol-1
ZSOLT BODOR et al
1674
and in the case of glycerol with glucose as co-substrate
0.6 mol mol-1. Blocking the NADH reoxidation results a
complex redirection of flux from formic acid, lactic acid
and ethanol to succinic acid. Using minimal medium and
the elimination of tree genes was sufficient for succinic
acid production from glucose or glycerol under micro-
aerobic and anaerobic conditions.
We found that the number of essential genes is less
influenced by factors, such as substrate availability, genetic
engineering or environmental conditions. Meanwhile,
important changes were predicted on genes which reduce
maximal growth. These genes are much more influenced
by environmental conditions or by substrate availability
than by genetic modifications of redox metabolism.
Diauxic growth was predicted to aerobic and microaerobic
conditions when all the glucose or glycerol has been
consumed the cell began to metabolise the metabolites
secreted earlier. Using FVA simulations the flux span
changes were identified for both substrates, as a conclusion,
we can confirm that genetic engineering impact on cell
metabolism can be analysed by redundancy approach.
At this stage at the proof-of concept level, we were
interested to test in vivo the production potential, and
after six days a 0.51 mol mol-1 molar yield was achieved
on glycerol, however with long-term adaptation the
production potential can be significantly improved.
Bio-based production of different chemicals from
renewable feedstocks could be a sustainable solution to
fulfill the demands of organic chemicals.
Conflicts of Interest: none
Acknowledgements
This work was supported by the „Sectoral Operational
Programme Human Resources Development 2007-2013
of the Romanian Ministry of Labour, Family and Social
Protection through the Financial Agreement POSDRU/
6/1.5/S/19.”, “BIOBILD-Synthesis of some C4, C5
carboxylic acid building block chemicals from renewable
biomass resources” PN-II-PCCA-2011-3.2-1367.
Our work was supported by the Sapientia University
2017/2018 Research Program, grant nr. 227/2/17.05.2017
and by Collegium Talentum. The GC-MS analysis was
carried out at Center for Organic Chemistry “Costin D.
Nenitescu”.
References
1. LEWIS, N.E., JAMSHIDI, N., THIELE, I., PALSSON,
B.O. Metabolic systems biology: a constraint-based
approach. Encyclopedia of Complexity and Systems
Science, 5535 (2009).
2. CHAN, H.W., MOHAMAD, S.M., DERIS, S.,
ILLIAS, M.R. A Review of computational approaches
for in silico metabolic engineering for microbial fuel
production. Current Bioinformatics, 8(2), 253-258
(2013).
3. BODOR, ZS., FAZAKAS (IUHASZ), A., KOVÁCS,
E., LÁNYI, S., ALBERT, B. Systems biology and
metabolic engineering for obtaining E. coli mutants
capable to produce succinate from renewable
resources. Romanian Biotechnologycal Letters, 19(4),
9625-9636 (2014).
4. FONG, S.S. Computational approaches to metabolic
engineering utilizing systems biology and synthetic
biology. Computational and Structural Biotechnology
Journal, 11(18), 28-34 (2014).
5. FONG, S.S., BURGARD, A.P., HERRING, C.D.,
KNIGHT, E.M., BLATTNER, F.R., MARANAS,
C.D., PALSSON, B.O. In silico design and adaptive
evolution of Escherichia coli for production of lactic
acid. Biotechnology and Bioengineering, 91(5), 643-
648 (2005).
6. LEE, D.H., PALSSON, B.O. Adaptive evolution of
escherichia coli K-12 MG1655 during growth on a
nonnative carbon source, L-l,2-propanediol. Applied
and Environmental Microbiology, 76(13), 4158-4168
(2010).
7. LEE, S.Y., KIM, H.U. Systems strategies for develo-
ping industrial microbial strains. Nature Biotechnology,
33(10), 1061-72 (2015).
8. LEE, J.W., YI, J., KIM, T.Y., CHOI, S., AHN, J.H.,
SONG, H., LEE, S.Y. Homo-succinic acid production
by metabolically engineered Mannheimia succinicipro-
ducens. Metabolic Engineering, 38, 409-417 (2016).
9. MIKLÓSSY, I., BODOR, ZS., SINKLER, R.,
ORBÁN, K.C., LÁNYI, S., ALBERT, B. In silico and
in vivo stability analysis of a heterologous biosynthetic
pathway for 1,4-butanediol production in metabolically
engineered E. coli. Journal of Biomolecular Structure
and Dynamics, 35(9), 1874-1889 (2016).
10. NA, D., YOO, S.M., CHUNG, H., PARK, H., PARK,
J.H., LEE, S.Y. Metabolic engineering of Escherichia
coli using synthetic small regulatory RNAs. Nature
Biotechnology, 31(2), 170-4 (2013).
11. YIM, H., HASELBECK, R., NIU, W., PUJOL-
BAXLEY, C., BURGARD, A., BOLDT, J., VAN
DIEN, S. Metabolic engineering of Escherichia coli for
direct production of 1,4-butanediol. Nature Chemical
Biology, 7(7), 445-452 (2011).
12. COOPER, I., DUFFIELD, M. The in silico prediction
of bacterial essential genes. Science Against Microbial
Pathogens, 551-559 (2011).
13. HUANG, D., LI, S., XIA, M., WEN, J., JIA, X.
Genome-scale metabolic network guided engineering
of Streptomyces tsukubaensis for FK506 production
improvement. Microbial Cell Factories, 12, 52 (2013).
14. TEPPER, N., SHLOMI, T. Predicting metabolic engi-
neering knockout strategies for chemical production:
Accounting for competing pathways. Bioinformatics,
26(4), 536-543 (2009).
15. FEIST, A.M., PALSSON, B.Ø. The growing scope of
applications of genome-scale metabolic reconstruc-
tions using Escherichia coli. Nature Biotechnology,
26(6), 659-667 (2008).
Using genome-scale model to predict the metabolic engineering impact on Escherichia coli metabolism
1675
16. KIM, T.Y., SOHN, S.B., KIM, Y.B, KIM, W.J., LEE,
S.Y. Recent advances in reconstruction and applica-
tions of genome-scale metabolic models. Current
Opinion in Biotechnology. 23(4):617-623 (2012).
17. ORTH, J.D., CONRAD, T.M., NA, J., LERMAN,
J.A., NAM, H., FEIST, A.M., PALSSON, B.Ø.
A comprehensive genome-scale reconstruction of
Escherichia coli metabolism – 2011. Molecular
Systems Biology, 7, 1-16 (2011).
18. SAHA, R., CHOWDHURY, A., MARANAS, C.D.
Recent advances in the reconstruction of metabolic
models and integration of omics data. Current Opinion
in Biotechnology, 29(1) 39-45, (2014).
19. PEY, J., VALGEPEA, K., RUBIO, A., BEASLEY,
J.E., PLANES, F.J. Integrating gene and protein
expression data with genome-scale metabolic networks
to infer functional pathways. BMC Systems Biology, 7,
134 (2013).
20. FAULON, J., CARBONELL, P. Systems Metabolic
Engineering. Systems Metabolic Engineering: Methods
and Protocols, Methods in Molecular Biology, 985(1),
149-173 (2013).
21. LEE, S.Y., KIM, H.U., CHAE, T.U., CHO, J.S., KIM,
J.W., SHIN, J.H., JANG, Y.-S. A comprehensive
metabolic map for production of bio-based chemicals.
Nature Catalysis, 2(1), 18-33 (2019).
22. STEPHANOPOULOS, G. Synthetic Biology and
Metabolic Engineering. {ACS} Synth. Biol., 1(11), 514-
525 (2012).
23. FEIST, A.M., ZIELINSKI, D.C., ORTH, J.D.,
SCHELLENBERGER, J., HERRGARD, M.J.,
PALSSON, B.O. Model-driven evaluation of the
production potential for growth-coupled products of
Escherichia coli. Metabolic Engineering, 12(3), 173-
186 (2010).
24. LEWIS, N.E., NAGARAJAN, H., PALSSON, B.O.
Constraining the metabolic genotype–phenotype
relationship using a phylogeny of in silico methods.
Nature Reviews Microbiology, 10(4), 291-305 (2012).
25. BORDBAR, A., MONK, J.M., KING, Z.A,
PALSSON, B.O. Constraint-based models predict
metabolic and associated cellular functions. Nature
Reviews. Genetics, 15(2), 107-120 (2014).
26. ANDREOZZI, S., CHAKRABARTI, A., SOH,
K.C., BURGARD, A., YANG, T.H., VAN DIEN,
S., HATZIMANIKATIS, V. Identification of metabolic
engineering targets for the enhancement of 1,4-butane-
diol production in recombinant E. coli using large-scale
kinetic models. Metabolic Engineering (2016).
27. HANSEN, A.S.L., LENNEN, R.M., SONNENSCHEIN,
N., HERRGÅRD, M.J. Systems biology solutions for
biochemical production challenges. Current Opinion in
Biotechnology, 45, 85-91 (2017).
28. LEE, J.W., NA, D., PARK, J.M., LEE, J., CHOI,
S., LEE, S.Y. Systems metabolic engineering of
microorganisms for natural and non-natural chemicals.
Nature Chemical Biology, 8(6), 536-546 (2012).
29. MAIA, P., ROCHA, M., ROCHA, I. In silico
constraint-based strain optimization methods: the quest
for optimal cell factories. Microbiology and Molecular
Biology Reviews : MMBR, 80(1), 45-67 (2016).
30. CAMPODONICO, M.A., ANDREWS, B.A., ASENJO,
J.A., PALSSON, B.Ø., FEIST, A.M. Generation
of an atlas for commodity chemical production in
Escherichia coli and a novel pathway prediction
algorithm, GEM-Path. Metabolic Engineering, 25,
140-158 (2014).
31. LISHA, K.P., SARKAR, D. Dynamic flux balance
analysis of batch fermentation: Effect of genetic
manipulations on ethanol production. Bioprocess and
Biosystems Engineering, 37(4), 617-627, (2014).
32. RAMAN, K., CHANDRA, N. Flux balance analysis
of biological systems: Applications and challenges.
Briefings in Bioinformatics, 10(4), 435-449 (2009).
33. CHOI, K.R., SHIN, J.H., CHO, J.S., YANG, D., LEE,
S.Y. Systems Metabolic Engineering of Escherichia
coli. EcoSal Plus, 7(1) (2016).
34. NIELSEN, J., KEASLING, J.D. Engineering Cellular
Metabolism. Cell, 164(6), 1185-1197, (2016).
35. DATSENKO, K.A., WANNER, B.L. One-step inacti-
vation of chromosomal genes in Escherichia coli K-12
using PCR products. Proceedings of the National
Academy of Sciences of the United States of America,
97(12), 6640-6650 (2000).
36. WENDISCH, V.F., LINDNER, S.N., MEISWINKEL,
T.M. Use of Glycerol in Biotechnological Applica-
tions. Biodiesel – Quality, Emissions and By-Products,
305-340 (2011).
37. GARLAPATI, V.K., SHANKAR, U., BUDHIRAJA,
A. Bioconversion technologies of crude glycerol
to value added industrial products. Biotechnology
Reports, 9, 9-14 (2016).
38. YANG, F., HANNA, M.A., SUN, R. Value-added
uses for crude glycerol--a byproduct of biodiesel
production. Biotechnology for Biofuels, 5(1), 13
(2012).
39. WERPY, T., PETERSEN, G. Top Value Added
Chemicals from Biomass Volume I — Results of
Screening for Potential Candidates from Sugars and
Synthesis Gas Top Value Added Chemicals From
Biomass Volume I : Results of Screening for Potential
Candidates (2004).
40. LIANG, L.A., LIU, R.M., MA, J.F., CHEN, K.Q.,
JIANG, M., WEI, P. Increased production of succinic
acid in Escherichia coli by overexpression of malate
dehydrogenase. Biotechnology Letters, 33(12), 2439-
2444 (2011).
41. SALVACHÚA, D., SMITH, H., ST. JOHN, P.C.,
MOHAGHEGHI, A., PETERSON, D.J., BLACK,
B.A., BECKHAM, G.T. Succinic acid production from
lignocellulosic hydrolysate by Basfia succinicipro-
ducens. Bioresource Technology, 214, 558-566 (2016).
42. ILIE, M., MARINESCU, F., SZÉP, R., GHIŢA,
G., DEAK, G., ANGHEL, A.M., URIŢESCU, B.
ZSOLT BODOR et al
1676
Ecological risk assessment of heavy metals in surface
sediments from the Danube River. Carpathian Journal
of Earth and Environmental Sciences, 12(2), 437-445
(2017).
43. KERESZTESI, Á., PETRES, S., GHITA, G.,
DUMITRU, F.D., MONCEA, M.A., OZUNU, A.,
SZÉP, R. Ammonium neutralization effect on
rainwater chemistry in the basins of the Eastern
Carpathians-Romania. Revista de Chimie (2018).
44. SZÉP, R., BODOR, Z., MIKLÓSSY, I., NITA, I.A.,
OPREA, O.A., KERESZTESI, Á. Influence of peat
fires on the rainwater chemistry in intra-mountain
basins with specific atmospheric circulations (Eastern
Carpathians, Romania). Science of the Total Environ-
ment, 647, 275-289 (2019).
45. SZÉP, R., KERESZTES, R., CONSTANTIN, L. Multi-
model assessment of tropospheric ozone pollution
indices of risk to human health and crops, and ozone
deposition in Ciuc Depression – Romania. Revista de
Chimie, 67(3), 408-413 (2016).
46. SZÉP, R., KERESZTES, R., DEAK, G., TOBA, F.,
GHIMPUSAN, M. The dry deposition of the PM10 and
PM2.5 to the vegetation and its health effect in the Ciuc
basin. Revista de Chimie (Bucharest, Romania) (2016).
47. SZÉP, R., KERESZTES, R., KORODI, A., TONK,
S., NICULAE, A.G., BIRLOIU, A.M. Dew point –
Indirect particulate matter pollution indicator in the
Ciuc Basin – Harghita, Romania. Revista de Chimie,
67(10), 1914-1921 (2016).
48. SZÉP, R., MATEESCU, E., NECHIFOR, A.C.,
KERESZTESI, Á. Chemical characteristics and
source analysis on ionic composition of rainwater
collected in the Carpathians “Cold Pole,” Ciuc basin,
Eastern Carpathians, Romania. Environmental Science
and Pollution Research, 24(35), 27288-27302 (2017).
49. SZÉP, R., MATEESCU, E., NIȚĂ, I.A., BIRSAN,
M.V., BODOR, Z., KERESZTESI, Á. Effects of the
Eastern Carpathians on atmospheric circulations and
precipitation chemistry from 2006 to 2016 at four
monitoring stations (Eastern Carpathians, Romania).
Atmospheric Research, 214(April), 311-328, (2018).
50. SZÉP, R., MÁTYÁS, L. The role of regional
atmopsheric stability in high-PM10 concentration
episodes in Miercurea Ciuc (Harghita). Carpathian
Journal of Earth and Environmental Sciences, 9(2),
241-250, (2014).
51. BECKER, S.A., FEIST, A.M., MO, M.L., HANNUM,
G., PALSSON, B.Ø., HERRGARD, M.J. Quantitative
prediction of cellular metabolism with constraint-based
models: the COBRA Toolbox. Nature Protocols, 2(3),
727-738 (2007).
52. SCHELLENBERGER, J., QUE, R., FLEMING,
R.M.T., THIELE, I., ORTH, J.D., FEIST, A.M.,
PALSSON, B.Ø. Quantitative prediction of cellular
metabolism with constraint-based models: the COBRA
Toolbox v2.0. Nature Protocols, 6(9), 1290-1307
(2011).
53. VARMA, A., PALSSON, B.O. Stoichiometric flux
balance models quantitatively predict growth and meta-
bolic by-product secretion in wild-type Escherichia
coli W3110. Applied and Environmental Microbiology,
60(10), 3724-3731 (1994).
54. GHOSH, A., ZHAO, H., PRICE, N.D. Genome-
scale consequences of cofactor balancing in engineered
pentose utilization pathways in Saccharomyces cere-
visiae. PLoS ONE, 6(11) (2011).
55. MAHADEVAN, R., SCHILLING, C.H. The effects of
alternate optimal solutions in constraint-based genome-
scale metabolic models. Metabolic Engineering, 5(4),
264-276 (2003).
56. LUAN, T.G., YU, K.S.H., ZHONG, Y., ZHOU, H.W.,
LAN, C.Y., TAM, N.F.Y. Study of metabolites from
the degradation of polycyclic aromatic hydrocarbons
(PAHs) by bacterial consortium enriched from man-
grove sediments. Chemosphere, 65(11), 2289-2296
(2006).
57. JEONG, H., LEE, S.J., KIM, P. Procedure for Adaptive
Laboratory Evolution of Microorganisms Using a
Chemostat, (June), 1-5 (2016).
