Research Article Metabolic Modeling of Common Escherichia...

Research ArticleMetabolic Modeling of Common Escherichia coli Strains inHuman Gut Microbiome

Yue-Dong Gao1 Yuqi Zhao2 and Jingfei Huang23

1 Kunming Biological Diversity Regional Center of Instruments Kunming Institute of Zoology Chinese Academy of SciencesKunming 650223 China

2 State Key Laboratory of Genetic Resources and Evolution Kunming Institute of Zoology Chinese Academy of Sciences32 Eastern Jiaochang Road Kunming Yunnan 650223 China

3 Kunming Institute of Zoology Chinese University of Hongkong Joint Research Center for Bio-Resources andHuman Disease Mechanisms Kunming 650223 China

Correspondence should be addressed to Yuqi Zhao zhaoyqmailkizaccn and Jingfei Huang huangjfmailkizaccn

Received 21 April 2014 Revised 11 June 2014 Accepted 13 June 2014 Published 13 July 2014

Academic Editor Zhixi Su

Copyright copy 2014 Yue-Dong Gao et alThis is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

The recent high-throughput sequencing has enabled the composition of Escherichia coli strains in the humanmicrobial communityto be profiled en masse However there are two challenges to address (1) exploring the genetic differences between E coli strains inhuman gut and (2) dynamic responses of E coli to diverse stress conditions As a result we investigated the E coli strains in humangut microbiome using deep sequencing data and reconstructed genome-wide metabolic networks for the three most common Ecoli strains including E coli HS UTI89 and CFT073 The metabolic models show obvious strain-specific characteristics both innetwork contents and in behaviors We predicted optimal biomass production for three models on four different carbon sources(acetate ethanol glucose and succinate) and found that these stress-associated genes were involved in host-microbial interactionsand increased in human obesity Besides it shows that the growth rates are similar among the models but the flux distributionsare different even in E coli core reactions The correlations between human diabetes-associated metabolic reactions in the E colimodels were also predictedThe study provides a systems perspective on E coli strains in human gutmicrobiome andwill be helpfulin integrating diverse data sources in the following study

1 Introduction

Escherichia coli (E coli) is themostwidely studied prokaryoticmodel organism and an important species in the fields ofbiotechnology and microbiology E coli constitutes about01 of human gut flora [1] which benefits human beingsby providing supplemental nutrition by enhancing nutri-ent acquisition and by preventing the establishment ofpathogenic bacteria within the intestine [2] The study of thisbacterium is both of importance for applications such asenvironmental testing and metabolic engineering [3] and ofinterest as a fundamental physical problem For example arecent study demonstrated an obvious increase in the numberof E coli in the stool while diarrhea was apparent [4]

In the recent five years the flood of deep sequencingdata has set the latest wave of microbiome research apart

from earlier studies with the ability to enumerate all ofthe cells in a complex microbial community at once [5]For instance using deep sequencing the Human Micro-biome Project (HMP) was launched to characterize themicrobial communities found at several different sites onthe human body and to analyze the role of these microbesin human health and disease [6 7] This switch from thelow-throughput technique culture-based enumeration tothe high-throughput technology of deep sequencing offersseveral advantages including high accuracy culture-freesampling and comprehensive information However thereare still two challenges to address First due to the huge datasize and high complexity of the different algorithms it isdifficult to determine the exact roles of the various speciesin human microbiome let alone strains of the same speciesThe composition of E coli strains is of value to human health

Hindawi Publishing CorporationBioMed Research InternationalVolume 2014 Article ID 694967 11 pageshttpdxdoiorg1011552014694967

2 BioMed Research International

for example changes in theE coli compositionwere observedassociated with intestinal inflammatory disorders in humanand mice [8 9] Second most of the microbiota communitystructures obtained from sequencing were ldquostaticrdquo whilethe human microbiomes are diverse and dynamic The dietchanges individual differences sampling sites and physicalconditions are responsible for the dynamic responses ofhuman microbiome [10ndash12] However the comprehensiveresponses of microbiome to the dynamicmicroenvironmentscan hardly be obtained from one or several samples

To solve these problems considerable efforts have beenmade to develop metabolic networks of E coli [3 13 14]These in silicomodels have been successfully applied inmanyfields For example they were frequently used in predictionof steady-state or dynamic responses of cells to changesin ecosystems [3] In addition the metabolic models canbe easily integrated with other data sources such as DNAsequencing [15] expression profiles [16] proteomics [17] ormetabolomics [18] Goals of such data integration effortsare (1) to gain a better understanding of the observablephenotypes of the cell (2) to predict potential functions ofmolecular signatures and (3) to apply these in silico modelsfor biological discovery and engineering applications As aresult integration of relevant omics datawithmetabolicmod-els as a representative species in the human gut microbiotaelucidates the changes in the gut microbiota

In this study we performed in silico modeling ofmetabolic networks of E coli strains in human gut micro-biome First we determined E coli strains in human gutmicrobiome using 148 fecal metagenomes Next we recon-structed genome-wide metabolic network of common E colistrains in human gut Then the cellular phenotypes werepredicted and validated using the genome variation of Ecoli and diet changes The findings of the study will helpin developing new technologies and tools for computationalanalysis and exploring the relationship between disease andchanges in the human microbiome

2 Materials and Methods

21 Human Gut Metagenomes and Reference GenomesHigh-quality short reads of 148 human gut sampleswere retrieved from Human Microbiome Project (HMPhttpwwwhmpdaccorg) The sequenced and well-annotated E coli genomes (totally 61 genomes) depositedin GenBank were downloaded from NCBI database(httpwwwncbinlmnihgov) to build a reference genomedatabase The reads were aligned against the E coli referencegenome using BLASTN (version 2227+) with 119864 lt 001minimal 99 identity cutoff and considering the reads thatwere aligned onto only a single position in the referencegenome

22 De Novo Assembly and Identification of Genes Thereads of human gut samples were assembled by Newbler(454Roche GS MapperAssembler) following the protocolin HMP [19] The assembled scaffolds were aligned against

E coli genomes using BLASTN with minimal 99 identitycutoff and best hit output

23 Reconstruction of Strain-Specific Metabolic Network TheE coli pan-genome (the union of the gene sets of all thestrains of a species) metabolic network has been generatedin a recent study [20] The strain-specific metabolic modelcould be reconstructed based on the pan-genome metabolicnetwork We generated metabolic networks for the commonE coli strains in human gut microbiome based on the pan-genome metabolic network

In the process we derived the strain-specific metabolicmodels using two commonly used algorithms of top-downmetabolic reconstructions including GIMME [21] and iMAT[22] These two algorithms are different the GIMME is alinear programming procedure while the iMAT is a mixedinteger linear programming procedure

24 Predictions of Cellular Phenotypes Using Metabolic Net-work Fluxes through reactions in the metabolic models canbe predicted using flux balance analysis (FBA) [23] In theprocess fluxes are constrained by steady-state mass balancesenzyme capacities and reaction directionality which yield asolution space of possible flux values Besides FBA uses anobjective function to identify flux distributions thatmaximize(orminimize) the physiologically relevant predicted solutionCellular growth rate (biomass production in another word)was used as an objective function for FBA analyses performedin this study The same biomass equation growth (GAM)andnongrowth (NGAM) associatedATP requirement valuesand PO (number of ATP molecules produced per pair ofelectrons donated to the electron transport system) ratio wereused for all the E coli models and were the same as thatin iAF1260 model [24] When the metabolic models wereused to simulate the change of carbon source (eg fromglucose to succinate) we obtained the corresponding optimalgrowth rates and flux distributions for all the reactions Ifthe uptakesecretion flux for a reaction in the optimal fluxsolution was reduced or increased by over 10 (flux119909 gt11 times flux119910 or flux119909 lt 09 times flux119910) between two conditionswe defined the reactions to be associated with the diet stress

Uniform random sampling of the solution space for Ecoli metabolic models in any environmental condition isa rapid and scalable way to characterize the structure ofthe allowed space of metabolic fluxes [25] The set of fluxdistributions obtained from sampling can be interrogatedfurther to answer a number of questions related to themetabolic network function In the study we studied howdependent two reactions within the E coli network were oneach other

25 Flux Variability Analysis (FVA) Biological systems oftencontain redundancies that contribute to their robustnessFVA can be used to examine these redundancies by calculat-ing the full range of numerical values for each reaction flux ina network [26] In FVA the process is carried out by optimiz-ing for a particular objective while still satisfying the givenconstraints set on biological systems In the study FVA was

BioMed Research International 3

applied to determine the ranges of fluxes that correspond toan optimal solution of the E colimodels determined throughFBA The maximum value of the objective function is firstcomputed and this value is used with multiple optimizationsto calculate the maximum andminimum flux values througheach reaction

3 Results

31 E coli in Human Gut Microbiome Deep metagenomicsequencing provides us the opportunity to explore the exis-tence of a common set of E coli species in human gutmicrobiome

To obtain this goal we built a nonredundant databaseof 61 sequenced and well-annotated E coli genomes Afteraligning the reads of each human gut microbial sample ontothe reference database we determined the proportion ofthe genomes covered by the reads (Methods) At a 99identity threshold and 10-fold coverage (the genomes of Ecoli strains are 5M on average) we detected one in all gutsamples three in 80 and seven in 60 of the 148 humangut samples (Table 1) We focused on the three common Ecoli strains including E coli HS UTI89 and CFT073 Otherrecent studies support our findings including studies fromhuman [27] and animal models [28]

Besides the genome-guidedmethods the reads were usedto perform de novo assembly which can recover transcriptfragments from regions missing in the genome assembly[29] We first assembled metagenomes in 148 human gutmicrobiome samples using over 10 billion reads Then wemapped the 15 million gut scaffolds to the 293663 genes(target genes) of the 61 E coli genomes in the human gutAt a 99 identity threshold over 60 of the target genes ofthe seven E coli in Table 1 had at least 80 of their lengthcovered by a single scaffold indicating that the genes of theseE coli strains were significantly enriched in the gut scaffolds(Fisherrsquos exact test 119875 lt 10minus10)

32 In SilicoMetabolic Models of E coli Strains We generatedgenome-wide metabolic network of three common E coli (EcoliHS UTI89 and CFT073) frommetabolic model of E colipan-genome using GIMME and iMAT algorithms

The results indicate that the metabolic networks obtainedwith the two algorithms are identical (TEXT S1ndashS3 availableonline at httpdxdoiorg1011552014694967) We thenexplored the differences in network properties among thethree models It shows that these models are different in net-work structure (Figure 1(a) Table S1) For example comparedwith E coli CFT073 and E coli UTI89 E coli HS modelhas 41 specific metabolic reactions catalyzed by 36 genes(Figure 1(b)) These reactions are associated with alternatecarbon metabolism murein recycling nitrogen metabolismand inner membrane transport Most of the reactions tend toform a subnetwork rather than are scattered in an apparentlyrandommanner in the metabolic network We also observed32 different metabolites not included in all the three models(Table 2) Only three of the metabolites (including allantoatetRNA-Ala and tRNA-Phe) can be detected in the human

metabolic model Recon2 [30] suggesting that most of thesedifferent metabolites are not involved in direct interactionsof gut microbiome host However some of these metabolitesare of importance to strain-specific characteristics and closelyrelated to human-microbe interactions For example GDP-L-fucose plays important roles in microbial infection andnumerous ontogenic events [31]

The genome-wide metabolic networks for E coli CFT073and UTI89 have recently been reconstructed based on thecomparative genomics analysis [20] We compared our mod-els (TEXT S1ndashS3) with the previous ones and found thatour models included more metabolic genes because the deepsequencing has been proven to lead to the identification oflarge populations of novel as well as missing transcripts thatmight reflect Hydra-specific evolutionary events [32]

33 Optimal Flux Distributions for E coli Strains In theprevious studies one of the most fundamental genome-scalephenotypic calculations is the simulation of cellular growthusing flux balance analysis (FBA) [25] As a result we definedbiomass composition of the cell as the biomass objectivefunction and performed FBA on the model in order tomaximize the objective function It shows that the optimalbiomass flux for the three models are pretty close (optimalflux = 07287 for CFTO73 while optimal flux = 07367 forHS and UTI89) However the optimal flux distributions areof different in the networks Figure 2 shows the optimalflux distribution map of core metabolic network in threeE coli strains It shows that the fluxes of ACS (acetyl-CoA synthetase) PTAr (phosphotransacetylase) and ACKr(acetate kinase) in CFTO73 model are obviously differentfrom that in the other two models

We then estimated the effect of reducing flux throughmetabolic reactions on biomass production of three mod-els Two reactions ACOAD6F (acyl-CoA dehydrogenasetetradecanoyl-CoA) and PGK (phosphoglycerate kinase)were taken as examples here (Figure S1) It shows thatthe growth rate is sustained near the optimal value over arange of values for PGK in all three models indicating thesame network robustness with respect to flux changes in thereaction (Figure S1A) However the effects of reducing fluxthrough ACOAD6F on growth are different between E coliCFTO73 and the other two models (Figure S1B) Besidesthe growth rate is sharply reduced after reaching the optimalvalue in HS and UTI89 models

34 Dynamic Responses of Metabolic Networks to Changesin Carbon Sources Although a few human gut microbiomeprojects have been launched the interrelationships betweenour diets and the structure and operations of our gut micro-bial communities are poorly understood Here we predictedthe human gut E colirsquos response to diet using metabolicmodeling

We simulated the optimal growth rates for three mod-els on carbon source as acetate ethanol glucose andsuccinate respectively (uptake rate sets all changed to9mmol gDWminus1 hminus1) The average growth rates of threemetabolic models corresponding to four diet conditions are


Table 1 Common Escherichia coli strains in human gut

Escherichia coli strains Samples counta Genome size Gene counts Protein count Genes by de novoassembly

E coliHS 148 46M 4629 4377 3606E coli UTI89 134 50M 5127 5017 3435E coli CFT073 125 52M 5579 5369 3406E coli KO11FL 115 49M 4756 4533 3512E coli NA114 94 50M 4975 4873 3381E coli 536 90 49M 4779 4619 3488E coli O127H6 strE234869 90 50M 4890 4552 3179aThere are 148 individual samples in the analysis

Strains Genes Reactions Metabolites Reversiblereactions

Ecoli_CFT073 1149 2226 1621 838Ecoli_HS 1223 2330 1646 841

Ecoli_UTI89 1193 2314 1632 845

(a)

MetaboliteGeneReaction

Murein recycling

Transport inner membraneAlternate carbon metabolism

Nitrogen metabolism

GGPTRCO

GGGABADr

ALAALAD

MMCD

HPPPNDO

HKNDDH

DHCINDO

HKNTDH

GGGABAH

ALDD2x

PPPNDO

HOPNTAL

MMM2

MANGLYCptspp

MALDDH

PPCSCT

OP4ENH

TYROXDApp

42A12BOOXpp

PEAMNOpp

PACCOAL

FRULYSDG

MELIBt2pp

FRULYSt2pp

3HCINNMH

3HPPPNH

PSCLYSt2pp

DHPPD

DHCIND

CINNDOCYNTAHCYNTt2pp

GALCTNLt2pp

XYLt2pp

HCINNMt2rpp

HPPPNt2rpp

ASO3t8pp

ALDD19x

GGPTRCSFRULYSE FRULYSK

(b)

Figure 1 Comparisons of metabolic networks of three E coli strains (a) Basic parameters of metabolic models (b) Strain-specific reactionsin E coliHS model


Table 2 Different metabolites in E coli strains

Metabolites Descriptions Formulas Charges4h2opntn 4-Hydroxy-2-oxopentanoate C5H7O4 minus1acglc-D 6-Acetyl-D-glucose C8H14O7 0acmalt Acetyl-maltose C14H24O12 0alatrna L-Alanyl-tRNA(Ala) C3H6NOR 1all6p D-Allose 6-phosphate C6H11O9P minus2alltt Allantoate C4H7N4O4 minus1allul6p Allulose 6-phosphate C6H11O9P minus2cechddd cis-3-(3-Carboxyethyl)-35-cyclohexadiene-12-diol C9H11O4 minus1cenchddd cis-3-(3-Carboxyethenyl)-35-cyclohexadiene-12-diol C9H9O4 minus1cinnm trans-Cinnamate C9H7O2 minus1dhcinnm 23-Dihydroxicinnamic acid C9H7O4 minus1dhpppn 3-(23-Dihydroxyphenyl)propanoate C9H9O4 minus1dtdp4d6dm dTDP-4-dehydro-6-deoxy-L-mannose C16H22N2O15P2 minus2dtdprmn dTDP-L-Rhamnose C16H24N2O15P2 minus2frulysp Fructoselysine phosphate C12H24N2O10P minus1gdpddman GDP-4-Dehydro-6-deoxy-D-mannose C16H21N5O15P2 minus2gdpfuc GDP-L-Fucose C16H23N5O15P2 minus2gdpofuc GDP-4-oxo-L-Fucose C16H21N5O15P2 minus2gg4abut Gamma-glutamyl-gamma aminobutyric acid C9H15O5N2 minus1ggbutal Gamma-glutamyl-gamma-butyraldehyde C9H16O4N2 0ggptrc Gamma-glutamyl-putrescine C9H20O3N3 1hkndd 2-Hydroxy-6-oxonona-24-diene-19-dioate C9H8O6 minus2hkntd 2-Hydroxy-6-ketononatrienedioate C9H6O6 minus2malt6p Maltose 61015840-phosphate C12H21O14P minus2man6pglyc 2(alpha-D-Mannosyl-6-phosphate)-D-glycerate C9H14O12P minus3op4en 2-Oxopent-4-enoate C5H5O3 minus1pac Phenylacetic acid C8H7O2 minus1phaccoa Phenylacetyl-CoA C29H38N7O17P3S minus4phetrna L-Phenylalanyl-tRNA(Phe) C9H10NOR 1trnaala tRNA(Ala) R 0trnaphe tRNA(Phe) R 0urdglyc (-)-Ureidoglycolate C3H5N2O4 minus1

shown in Figure 3(a) We can see that the growth rates forthree models are similar in different conditions Besides itdemonstrates substantially decreased anaerobic growth ascompared with aerobic (18mmol gDWminus1 hminus1) growth withthe same glucose uptake rate which was supported by recentstudies that E coli requires aerobic respiration to competesuccessfully in the mouse intestine [8 9] For E coli strainsin human gut carbon sources are diverse but glucose is mostsuitable for their growth

These responses of E coli to the diet changes involvemany metabolic genes and pathways We explored the per-turbations in the metabolic networks and found 10 genes(including ADH5 ALDH5A1 DLD FECH GCLC GPTGSR KARS MPST and TST) closely associated with thediet stress (Figure 3(b)) The glycolysis gluconeogenesisand glycerophospholipid metabolism were enriched in themetabolic reactions catalyzed by these genes (119875 lt 10minus3 usingFisherrsquos exact test) Besides we found that these enzymeswere evolutionarily conserved from E coli to human and

were involved in the interactions between human and E coli[14 33] Especially nine out of these 10 genes (except GPTglutamic-pyruvate transaminase) were found to be increasedin human obesity [34]

35 Analyzing Flux Correlations in Diabetes-Associated Path-ways in E coli Using Sampling Assessment and charac-terization of gut microbiota (E coli acts as an integralcomponent) has become amajor research area in human type2 diabetes the most prevalent endocrine disease worldwideA recent metagenomic research identified and validated over400 type-2-diabetes-associated markers in E coli includingover 100 metabolic genes [35] In the study we performeduniform random sampling for three models under glucose-limiting aerobic growth conditions to explore the relation-ships between the diabetes-associated pathways

We detected 158metabolic reactions in E colimodels thatwere associated with human type 2 diabetes (Table S2) Itshows that these reactions participate in many subsystems


+50Scale linear flux

(a) (b)

minus50

Figure 2 Flux balance analysis of metabolic models The figure shows the core metabolic map (a) in E coli and the reactions with differentfluxes (b) among three E colimodels ACS acetyl-CoA synthetase PTAr phosphotransacetylase ACKr acetate kinase

of which over 30 are associated with lipid metabolismand cofactorprosthetic group biosynthesis Correlation shipsbetween some metabolic reactions can be observed inFigure 4 For example PGL (6-phosphogluconolactonase)and GND (phosphogluconate dehydrogenase) fluxes arepositively correlated in the E coli HS model whereas PGLshows negative correlation with RPI (ribose-5-phosphate iso-merase) fluxesThe correlation ships between these diabetes-associated reaction fluxes are the same in other two models

36 Flux Variability Analysis (FVA) of E coli Models FBAreturns a single flux distribution that corresponds tomaximalbiomass production under given growth conditions How-ever alternate optimal solutionsmay exist which correspondto maximal growth As a result we performed FVA for thethree E coli models under glucose-limited aerobic growthconditions (glucose and oxygen were changed to 10 and18mmol gDWminus1 hminus1 resp)

It shows that the minimum and maximum fluxes for thereactions in E coli models are different Figure 5 illustratesFVA result for the seven reactions in pyrimidine biosynthesisAll the reactions have different flux range in three networksespecially carbamate kinase and dihydroorotic acid dehydro-genase

4 Discussion and Conclusion

In our study we determined the common E coli strains inhuman gut microbial communities based on HMP datasetsWe applied two widely used algorithms (GIMME and iMAT)to reconstruct genome-wide metabolic models for three

common E coli strains (E coli HS UTI89 and CFT073) andcompared the network characteristics of these models Thesemodels were then used to predict the cellular phenotypesand dynamic responses to the diverse gutmicroenvironmentThe models were also applied in exploring the relationshipsbetween E coli and human diabetes The results will be help-ful in exploring the dynamic responses of gut microbiome tothe environmental perturbations

The E coli strains have been proven to be significantlydifferent among individuals although the species is abundantin human gut [36] Although it is well accepted that thecomposition of E coli strains in human gut flora is associatedwith health status the exact molecular mechanism is stillunclearWe detected the common E coli strains in human gutand systematically compared their functions through in silicomodeling which have two advantages over the traditionalmethods First the sequencing data allows for a much moreaccurate determination of microbiome composition Theadvent of next-generation sequencing (NGS) enabled severalhigh-profile collaborative projects including the HMP Con-sortium (httpwwwhmpdaccorgproject cataloghtml)and MetaHIT Consortium (httpwwwsangeracukresour-cesdownloadsbacteriametahit) which have released awide range of data on the human microbiome Using thesedatasets we applied differentmethods (genome-guidedmap-ping and de novo assembly) to determine the common Ecoli strains making the following study of interconnectivitybetween gut microbiota diet and cell molecular responsesavailable Second the metabolic modeling can allow us tosee how a biological system might respond [37] This willguide the wet lab experiments and avoidmost of themistakesin the process In fact developing computational methods


Aerobic Anaerobic0807060504030201

0

Biom

ass fl

ux (h

minus1)

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

(a)

Glutathione-disulfidereductase

Glutathione

R00094Oxidizedglutathione

3alpha 7alpha-dihydroxy-5beta-cholestan-26-al

GSR

R00115

L-LysineFerrochelatase

HemeThiosulfate

sulfurtransferase FECH

TST

tRNA(Lys)R00310RE1691

34-Dihydroxyphenylethyleneglycol

L-Lysyl-tRNA

KARS

Lysine--tRNAligase

R03658

Alaninetransaminase

R00258L-Glutamate

R03105

GPT Mercaptopyruvate

Pyruvate

RE0702(S)-2-Aminobutanoate hypothiocyanite

RE2034

Protoporphyrin

34-Dihydroxymandelaldehyde

R00894

gamma-L-Glutamyl-L-cysteine

L-Cysteine

3-Mercaptopyruvatesulfur transferase

Glutamate--cysteineligase

GCLC

R04880

R01931 CyanideHydrogencyanide

Thiosulfate

Sulfate

Thiocyanate RE0159

R00209

Succinyl-CoA

CoA

L-Alanine

Dihydrolipoyldehydrogenase

2-Oxobutanoate2-Oxoglutarate

RE0361

Dihydrolipoylprotein

R01698R03815

Dihydrolipoamide

Lipoamide

Lipoylprotein

RE2882

S-(Hydroxymethyl)glutathione

S-(Hydroxymethyl)glutathionedehydrogenase

RE1711

20-oxo-leukotrieneB4

20-OH-LeukotrieneB4

5-Hydroxyindoleacetaldehyde

5-hydroxytryptophol

R06983

RE3011

MPSTD-Glyceraldehyde

3alpha 7alpha 26-trihydroxy-5beta-cholestane

R01036 R04805R00210 Trichloroethanol

R07105

N2-Succinyl-L-glutamate5-semialdehyde

ALDH5A1

20-Hydroxy-leukotrieneE4

Succinate-semialdehydedehydrogenase

R01041Glycerol

20-oxo-LeukotrieneE4

R05049Succinate

Succinatesemialdehyde

N2-Succinyl-L-glutamate

Acetyl-CoA

Lipoate

DLD

R00713

ADH5

Acetaldehyde

R00754

Alcoholdehydrogenase

S-Formylglutathione

Ethanol

ChloralhydrateRE3551

MetabolitesReactionsEnzymes

ReversibleIrreversible

Metabolic genes

(b)

Figure 3 Optimal growth rates for E coli strains on different carbon sources and the associated gene-protein reactions (a) Optimal growthrates for E coli strains on nutrition sources in human gutThe length of each bar represents the average optimal growth rates for three modelson the same carbon source (b) The diet stress-associated metabolic network in gut E coli

capable of predicting metabolic flux by integrating these datasources with a metabolic network is a major challenge ofsystems biology [18] For example the predicted behaviorsof diabetes-associated reactions in E coli (Table S2) can beintegrated with experimental validations to detect the causalgenes in human diabetes

The E coli is regarded as the prototypical pluripotentpathogens capable of causing a wide variety of illnesses ina broad array of species including pyelonephritis diarrheadysentery and the hemolytic-uremic syndrome [38] Inparticlar human gut E coli and its relationship to complexdiseases such as cancer [39] and diabetes [40] has attracted


5 5

0 0 01

0 001 02

04 05 05 0

5

0

53 56

553 56

55 62 69

33 36 39 4

0 0 0 0

7

minus26 minus25 minus24 minus23

GND GNK RBK HSK PGL FBA FUM MME RPI

GN

DG

NK

RBK

HSK

PGL

FBA

FUM

MM

ERP

I

Figure 4 Flux sampling of E coli HS model Flux distribution histograms (diagonal) and pairwise scatterplots (off-diagonal) for diabetes-associated metabolic reactions in E coli HS model The x-axis of the histograms indicates the magnitude of the flux through theparticular reaction The scatterplots on the off-diagonal elements show the relationship between fluxes through two reactions GNDphosphogluconate dehydrogenase CAT catalase GNK gluconokinase RBK ribokinase HSK homoserine kinase TMK thiamine kinasePGL 6-phosphogluconolactonase FBA fructose-bisphosphate aldolase FUM fumaraseMMEmethylmalonyl-CoA epimerase RPI ribose-5-phosphate isomerase

increasing interest in the last few years A question thenarises ldquoHow is it possible for this Jekyll and Hyde species toboth coexist peacefully with its host and cause devastatingillnessrdquo [38] The answer mainly lies in the existence ofdifferent strains of E coli with variable pathogenic potential[41] However we can hardly draw a complete picture of howthe E coli strains respond physiologically to the complexgut microenvironment Our study can provide valuableinformation based on the systematic comparisons of differentE coli strains It shows that although the optimal growthrates are similar for three E coli strains the optimal fluxdistributions are different for three models even in E colicore reactions The detected different reactions such as ACS(acetyl-CoA synthetase) and PTAr (phosphotransacetylase)were approved to be involved in the virulence of E coli andbe associated with human complex diseases [35 42] Theresults can be integrated with other data sets such as humanclinical trials and virulence profiles which will help establishthe extent of commonality between food-source and humangutE coli [43] and estimate the contribution of strain-specificreactions or genes to infections in humans

We found that the E coli responded distinctly to differentgut diets and the stress-associated genes were closely associ-ated with obesity With the high prevalence of diet-inducedhealth concerns such as diabetes and obesity there remains aneed for approaches that treat the causal factors Among thesefactors gutmicrobiome is drawingmore attention [35 44] forit is suitable as diseasemarkers and drug targets For example

Qin et al carried out a metagenome-wide association studywhich indicated that patients with type 2 diabetes have onlymoderate intestinal dysbiosis but that butyrate-producingbacteria are less abundant and opportunistic pathogens aremore abundant in these individuals than in healthy controls[35] The underlying mechanisms of interactions betweengut microbiome and human health are complicated howeverthe stress-associated pathways (such as the detected thegluconeogenesis and glycerophospholipid metabolism) mayplay important roles in the disease development The dietchanges first induced changes of involved metabolic genes(such as ADH5 alcohol dehydrogenase 5) which trigger thedownstream signaling pathways These signaling pathwaysmainly associated with immune responses and development[44 45] It is commonly accepted that the gut microbiotainteracts with the immune system providing signals topromote the maturation of immune cells and the normaldevelopment of immune functions [46] The dynamic inter-actions between all components of the microbiota and hosttissue over time will be crucial for building predictive modelsfor diagnosis and treatment of diseases linked to imbalancesin our microbiota

In summary the findings here represent a significantlyexpanded and comprehensive reconstruction of the E colimetabolic network in human gut This work will enable awider spectrum of studies focused on microbe-host interac-tions and serve as a means of integrating other omics sets insystems biology


Pyrimidine biosynthesis

cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump

co2ASPCT DHORTSCBMKr

DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108

BidirectionalreversibleUnidirectionalreversible

reverseUnidirectionalirreversible

(b)

Figure 5 FVA of E colimodels Shown is a map of metabolic reactions in pyrimidine biosynthesis pathway of E colimodels Using FVA theminimum (min) and maximum (max) allowable flux values for each reaction were determined The values shown in the table correspond tothe min and max allowable fluxes for each reaction shown in the map The results were further characterized by the direction of predictedflux (bidirectional or unidirectional) computed using FVA The full names of the metabolic reactions are included in TEXT S1ndashS3

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was supported by the National Natural ScienceFoundation of China (Grant no 31123005) and the Instru-ments FunctionDeployment Foundation ofCAS (Grants nosyg2010044 and yg2011057)

References

[1] P B Eckburg E M Bik C N Bernstein et al ldquoMicrobiologydiversity of the human intestinal microbial florardquo Science vol308 no 5728 pp 1635ndash1638 2005

[2] G Reid J Howard and B S Gan ldquoCan bacterial interferenceprevent infectionrdquoTrends inMicrobiology vol 9 no 9 pp 424ndash428 2001

[3] DMcCloskey B Oslash Palsson andAM Feist ldquoBasic and applieduses of genome-scale metabolic network reconstructions ofEscherichia colirdquoMolecular Systems Biology vol 9 no 1 article661 2013

[4] S Nakamura T Nakaya and T Iida ldquoMetagenomic analysisof bacterial infections by means of high-throughput DNAsequencingrdquo Experimental Biology and Medicine vol 236 no8 pp 968ndash971 2011

[5] M S Donia and M A Fischbach ldquoDyeing to learn more aboutthe gut microbiotardquo Cell Host and Microbe vol 13 no 2 pp119ndash120 2013

[6] L Rup ldquoThe human microbiome projectrdquo Indian Journal ofMicrobiology vol 52 no 3 p 315 2012

[7] P J Turnbaugh R E Ley M Hamady C M Fraser-LiggettR Knight and J I Gordon ldquoThe human microbiome projectrdquoNature vol 449 no 7164 pp 804ndash810 2007

[8] S A Jones F Z Chowdhury A J Fabich et al ldquoRespiration ofEscherichia coli in themouse intestinerdquo Infection and Immunityvol 75 no 10 pp 4891ndash4899 2007

[9] S E Winter M G Winter M N Xavier et al ldquoHost-derivednitrate boosts growth of E coli in the inflamed gutrdquo Science vol339 no 6120 pp 708ndash711 2013


[10] I Cho and M J Blaser ldquoThe human microbiome at theinterface of health and diseaserdquo Nature Reviews Genetics vol13 no 4 pp 260ndash270 2012

[11] CHuttenhower DGevers R Knight et al ldquoStructure functionand diversity of the healthy human microbiomerdquo Nature vol486 no 7402 pp 207ndash214 2012

[12] C F Maurice H J Haiser and P J Turnbaugh ldquoXenobioticsshape the physiology and gene expression of the active humangut microbiomerdquo Cell vol 152 no 1-2 pp 39ndash50 2013

[13] A Heinken S Sahoo R M Fleming and I Thiele ldquoSystems-level characterization of a host-microbe metabolic symbiosis inthemammalian gutrdquoGutMicrobes vol 4 no 1 pp 28ndash40 2013

[14] S Shoaie F KarlssonAMardinoglu I Nookaew S Bordel andJ Nielsen ldquoUnderstanding the interactions between bacteria inthe human gut throughmetabolic modelingrdquo Scientific Reportsvol 3 article 2532 2013

[15] C Lozupone K Faust J Raes et al ldquoIdentifying genomicand metabolic features that can underlie early successionaland opportunistic lifestyles of human gut symbiontsrdquo GenomeResearch vol 22 no 10 pp 1974ndash1984 2012

[16] Y Q Zhao and J F Huang ldquoReconstruction and analysis ofhuman heart-specific metabolic network based on transcrip-tome and proteome datardquo Biochemical and Biophysical ResearchCommunications vol 415 no 3 pp 450ndash454 2011

[17] J Zhao C Geng L Tao et al ldquoReconstruction and analysisof human liver-specific metabolic network based on CNHLPPdatardquo Journal of Proteome Research vol 9 no 4 pp 1648ndash16582010

[18] K Yizhak T Benyamini W Liebermeister E Ruppin and TShlomi ldquoIntegrating quantitative proteomics andmetabolomicswith a genome-scale metabolic networkmodelrdquo Bioinformaticsvol 26 no 12 pp i255ndashi260 2010

[19] B A Methe K E Nelson M Pop et al ldquoA framework forhuman microbiome researchrdquo Nature vol 486 no 7402 pp215ndash221 2012

[20] D J Baumler R G Peplinski J L Reed J D Glasner and NT Perna ldquoThe evolution of metabolic networks of E colirdquo BMCSystems Biology vol 5 article 182 2011

[21] S A Becker and B O Palsson ldquoContext-specific metabolicnetworks are consistentwith experimentsrdquoPLoSComputationalBiology vol 4 no 5 Article ID e1000082 2008

[22] H Zur E Ruppin and T Shlomi ldquoiMAT an integrativemetabolic analysis toolrdquoBioinformatics vol 26 no 24 pp 3140ndash3142 2010

[23] J D Orth I Thiele and B O Palsson ldquoWhat is flux balanceanalysisrdquo Nature Biotechnology vol 28 no 3 pp 245ndash2482010

[24] A M Feist C S Henry J L Reed et al ldquoA genome-scalemetabolic reconstruction for Escherichia coli K-12 MG1655that accounts for 1260 ORFs and thermodynamic informationrdquoMolecular Systems Biology vol 3 article 121 2007

[25] S A Becker A M Feist M L Mo G Hannum B OslashPalsson andM J Herrgard ldquoQuantitative prediction of cellularmetabolism with constraint-based models the COBRA Tool-boxrdquo Nature Protocols vol 2 no 3 pp 727ndash738 2007

[26] R Mahadevan and C H Schilling ldquoThe effects of alternateoptimal solutions in constraint-based genome-scale metabolicmodelsrdquoMetabolic Engineering vol 5 no 4 pp 264ndash276 2003

[27] S L Chen M Wu J P Henderson et al ldquoGenomic diversityand fitness of E coli strains recovered from the intestinal andurinary tracts of women with recurrent urinary tract infectionrdquoScience Translational Medicine vol 5 no 184 2013

[28] J S Ayres N J Trinidad and R E Vance ldquoLethal inflam-masome activation by a multidrug-resistant pathobiont uponantibiotic disruption of the microbiotardquo Nature Medicine vol18 no 5 pp 799ndash806 2012

[29] P Jain N M Krishnan and B Panda ldquoAugmenting tran-scriptome assembly by combining de novo and genome-guidedtoolsrdquo PeerJ vol 1 article e133 2013

[30] I Thiele N Swainston R M T Fleming et al ldquoA community-driven global reconstruction of human metabolismrdquo NatureBiotechnology vol 31 no 5 pp 419ndash425 2013

[31] D J Becker and J B Lowe ldquoFucose biosynthesis and biologicalfunction in mammalsrdquo Glycobiology vol 13 no 7 pp 41Rndash53R2003

[32] Y Wenger and B Galliot ldquoRNAseq versus genome-predictedtranscriptomes a large population of novel transcripts identi-fied in an Illumina-454 Hydra transcriptomerdquo BMC Genomicsvol 14 no 1 article 204 2013

[33] Z Qi andM R OrsquoBrian ldquoInteraction between the bacterial ironresponse regulator and ferrochelatase mediates genetic controlof heme biosynthesisrdquo Molecular Cell vol 9 no 1 pp 155ndash1622002

[34] S Greenblum P J Turnbaugh and E Borenstein ldquoMetage-nomic systems biology of the human gut microbiome revealstopological shifts associated with obesity and inflammatorybowel diseaserdquo Proceedings of the National Academy of Sciencesof the United States of America vol 109 no 2 pp 594ndash599 2012

[35] J J Qin Y R Li Z M Cai et al ldquoA metagenome-wideassociation study of gut microbiota in type 2 diabetesrdquo Naturevol 490 no 7418 pp 55ndash60 2012

[36] M Martinez-Medina X Aldeguer M Lopez-Siles et alldquoMolecular diversity of Escherichia coli in the human gut newecological evidence supporting the role of adherent-invasive Ecoli (AIEC) in Crohnrsquos diseaserdquo Inflammatory Bowel Diseasesvol 15 no 6 pp 872ndash882 2009

[37] J R Karr J C Sanghvi D N MacKlin et al ldquoA whole-cellcomputational model predicts phenotype from genotyperdquo Cellvol 150 no 2 pp 389ndash401 2012

[38] M Donnenberg Escherichia coli Pathotypes and Principles ofPathogenesis Academic Press New York NY USA 2013

[39] H Tlaskalova-Hogenova R Tpankova H Kozakova et al ldquoTherole of gut microbiota (commensal bacteria) and the mucosalbarrier in the pathogenesis of inflammatory and autoimmunediseases and cancer Contribution of germ-free and gnotobioticanimal models of human diseasesrdquo Cellular and MolecularImmunology vol 8 no 2 pp 110ndash120 2011

[40] N Larsen F K Vogensen F W J van den Berg et al ldquoGutmicrobiota in human adults with type 2 diabetes differs fromnon-diabetic adultsrdquo PLoS ONE vol 5 no 2 Article ID e90852010

[41] R M Robins-Browne ldquoTraditional enteropathogenicEscherichia coli of infantile diarrheardquo Reviews of InfectiousDiseases vol 9 no 1 pp 28ndash53 1987

[42] K Guan and Y Xiong ldquoRegulation of intermediarymetabolismby protein acetylationrdquo Trends in Biochemical Sciences vol 36no 2 pp 108ndash116 2011

[43] J R Johnson M A Kuskowski K Smith T T OrsquoBryan and STatini ldquoAntimicrobial-resistant and extraintestinal pathogenicEscherichia coli in retail foodsrdquo Journal of Infectious Diseasesvol 191 no 7 pp 1040ndash1049 2005

[44] J C Clemente L K Ursell L W Parfrey and R Knight ldquoTheimpact of the gut microbiota on human health an integrativeviewrdquo Cell vol 148 no 6 pp 1258ndash1270 2012


[45] P D Gluckman K A Lillycrop M H Vickers et al ldquoMetabolicplasticity during mammalian development is directionallydependent on early nutritional statusrdquo Proceedings of theNational Academy of Sciences of the United States of Americavol 104 no 31 pp 12796ndash12800 2007

[46] J Chow S M Lee Y Shen A Khosravi and S K MazmanianldquoHost-bacterial symbiosis in health and diseaserdquo Advances inImmunology vol 107 pp 243ndash274 2010

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Anatomy Research International

PeptidesInternational Journal of


Hindawi Publishing Corporation httpwwwhindawicom

International Journal of

Volume 2014

Zoology


Molecular Biology International

GenomicsInternational Journal of


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


BioinformaticsAdvances in

Marine BiologyJournal of



Signal TransductionJournal of


BioMed Research International

Evolutionary BiologyInternational Journal of



Biochemistry Research International

ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Genetics Research International


Advances in

Virolog y

Hindawi Publishing Corporationhttpwwwhindawicom

Nucleic AcidsJournal of

Volume 2014

Stem CellsInternational



Enzyme Research



Microbiology


for example changes in theE coli compositionwere observedassociated with intestinal inflammatory disorders in humanand mice [8 9] Second most of the microbiota communitystructures obtained from sequencing were ldquostaticrdquo whilethe human microbiomes are diverse and dynamic The dietchanges individual differences sampling sites and physicalconditions are responsible for the dynamic responses ofhuman microbiome [10ndash12] However the comprehensiveresponses of microbiome to the dynamicmicroenvironmentscan hardly be obtained from one or several samples

To solve these problems considerable efforts have beenmade to develop metabolic networks of E coli [3 13 14]These in silicomodels have been successfully applied inmanyfields For example they were frequently used in predictionof steady-state or dynamic responses of cells to changesin ecosystems [3] In addition the metabolic models canbe easily integrated with other data sources such as DNAsequencing [15] expression profiles [16] proteomics [17] ormetabolomics [18] Goals of such data integration effortsare (1) to gain a better understanding of the observablephenotypes of the cell (2) to predict potential functions ofmolecular signatures and (3) to apply these in silico modelsfor biological discovery and engineering applications As aresult integration of relevant omics datawithmetabolicmod-els as a representative species in the human gut microbiotaelucidates the changes in the gut microbiota

In this study we performed in silico modeling ofmetabolic networks of E coli strains in human gut micro-biome First we determined E coli strains in human gutmicrobiome using 148 fecal metagenomes Next we recon-structed genome-wide metabolic network of common E colistrains in human gut Then the cellular phenotypes werepredicted and validated using the genome variation of Ecoli and diet changes The findings of the study will helpin developing new technologies and tools for computationalanalysis and exploring the relationship between disease andchanges in the human microbiome

2 Materials and Methods

21 Human Gut Metagenomes and Reference GenomesHigh-quality short reads of 148 human gut sampleswere retrieved from Human Microbiome Project (HMPhttpwwwhmpdaccorg) The sequenced and well-annotated E coli genomes (totally 61 genomes) depositedin GenBank were downloaded from NCBI database(httpwwwncbinlmnihgov) to build a reference genomedatabase The reads were aligned against the E coli referencegenome using BLASTN (version 2227+) with 119864 lt 001minimal 99 identity cutoff and considering the reads thatwere aligned onto only a single position in the referencegenome

22 De Novo Assembly and Identification of Genes Thereads of human gut samples were assembled by Newbler(454Roche GS MapperAssembler) following the protocolin HMP [19] The assembled scaffolds were aligned against

E coli genomes using BLASTN with minimal 99 identitycutoff and best hit output

23 Reconstruction of Strain-Specific Metabolic Network TheE coli pan-genome (the union of the gene sets of all thestrains of a species) metabolic network has been generatedin a recent study [20] The strain-specific metabolic modelcould be reconstructed based on the pan-genome metabolicnetwork We generated metabolic networks for the commonE coli strains in human gut microbiome based on the pan-genome metabolic network

In the process we derived the strain-specific metabolicmodels using two commonly used algorithms of top-downmetabolic reconstructions including GIMME [21] and iMAT[22] These two algorithms are different the GIMME is alinear programming procedure while the iMAT is a mixedinteger linear programming procedure

24 Predictions of Cellular Phenotypes Using Metabolic Net-work Fluxes through reactions in the metabolic models canbe predicted using flux balance analysis (FBA) [23] In theprocess fluxes are constrained by steady-state mass balancesenzyme capacities and reaction directionality which yield asolution space of possible flux values Besides FBA uses anobjective function to identify flux distributions thatmaximize(orminimize) the physiologically relevant predicted solutionCellular growth rate (biomass production in another word)was used as an objective function for FBA analyses performedin this study The same biomass equation growth (GAM)andnongrowth (NGAM) associatedATP requirement valuesand PO (number of ATP molecules produced per pair ofelectrons donated to the electron transport system) ratio wereused for all the E coli models and were the same as thatin iAF1260 model [24] When the metabolic models wereused to simulate the change of carbon source (eg fromglucose to succinate) we obtained the corresponding optimalgrowth rates and flux distributions for all the reactions Ifthe uptakesecretion flux for a reaction in the optimal fluxsolution was reduced or increased by over 10 (flux119909 gt11 times flux119910 or flux119909 lt 09 times flux119910) between two conditionswe defined the reactions to be associated with the diet stress

Uniform random sampling of the solution space for Ecoli metabolic models in any environmental condition isa rapid and scalable way to characterize the structure ofthe allowed space of metabolic fluxes [25] The set of fluxdistributions obtained from sampling can be interrogatedfurther to answer a number of questions related to themetabolic network function In the study we studied howdependent two reactions within the E coli network were oneach other

25 Flux Variability Analysis (FVA) Biological systems oftencontain redundancies that contribute to their robustnessFVA can be used to examine these redundancies by calculat-ing the full range of numerical values for each reaction flux ina network [26] In FVA the process is carried out by optimiz-ing for a particular objective while still satisfying the givenconstraints set on biological systems In the study FVA was



3 Results


















Ecoli_UTI89 1193 2314 1632 845

(a)


Murein recycling


Nitrogen metabolism

GGPTRCO

GGGABADr

ALAALAD

MMCD

HPPPNDO

HKNDDH

DHCINDO

HKNTDH

GGGABAH

ALDD2x

PPPNDO

HOPNTAL

MMM2

MANGLYCptspp

MALDDH

PPCSCT

OP4ENH

TYROXDApp

42A12BOOXpp

PEAMNOpp

PACCOAL

FRULYSDG

MELIBt2pp

FRULYSt2pp

3HCINNMH

3HPPPNH

PSCLYSt2pp

DHPPD

DHCIND


GALCTNLt2pp

XYLt2pp

HCINNMt2rpp

HPPPNt2rpp

ASO3t8pp

ALDD19x


(b)












(a) (b)

minus50











0

Biom

ass fl

ux (h

minus1)

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

(a)


Glutathione



GSR

R00115


HemeThiosulfate


TST



L-Lysyl-tRNA

KARS

Lysine--tRNAligase

R03658

Alaninetransaminase

R00258L-Glutamate

R03105


Pyruvate


RE2034

Protoporphyrin


R00894


L-Cysteine



GCLC

R04880


Thiosulfate

Sulfate

Thiocyanate RE0159

R00209

Succinyl-CoA

CoA

L-Alanine



RE0361


R01698R03815

Dihydrolipoamide

Lipoamide

Lipoylprotein

RE2882



RE1711


20-OH-LeukotrieneB4


5-hydroxytryptophol

R06983

RE3011




R07105


ALDH5A1



R01041Glycerol


R05049Succinate



Acetyl-CoA

Lipoate

DLD

R00713

ADH5

Acetaldehyde

R00754


S-Formylglutathione

Ethanol




Metabolic genes

(b)





5 5

0 0 01

0 001 02

04 05 05 0

5

0

53 56

553 56

55 62 69

33 36 39 4

0 0 0 0

7



GN

DG

NK

RBK

HSK

PGL

FBA

FUM

MM

ERP

I








cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump


DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108



(b)




Acknowledgments


References
























































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology



3 Results


















Ecoli_UTI89 1193 2314 1632 845

(a)


Murein recycling


Nitrogen metabolism

GGPTRCO

GGGABADr

ALAALAD

MMCD

HPPPNDO

HKNDDH

DHCINDO

HKNTDH

GGGABAH

ALDD2x

PPPNDO

HOPNTAL

MMM2

MANGLYCptspp

MALDDH

PPCSCT

OP4ENH

TYROXDApp

42A12BOOXpp

PEAMNOpp

PACCOAL

FRULYSDG

MELIBt2pp

FRULYSt2pp

3HCINNMH

3HPPPNH

PSCLYSt2pp

DHPPD

DHCIND


GALCTNLt2pp

XYLt2pp

HCINNMt2rpp

HPPPNt2rpp

ASO3t8pp

ALDD19x


(b)












(a) (b)

minus50











0

Biom

ass fl

ux (h

minus1)

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

(a)


Glutathione



GSR

R00115


HemeThiosulfate


TST



L-Lysyl-tRNA

KARS

Lysine--tRNAligase

R03658

Alaninetransaminase

R00258L-Glutamate

R03105


Pyruvate


RE2034

Protoporphyrin


R00894


L-Cysteine



GCLC

R04880


Thiosulfate

Sulfate

Thiocyanate RE0159

R00209

Succinyl-CoA

CoA

L-Alanine



RE0361


R01698R03815

Dihydrolipoamide

Lipoamide

Lipoylprotein

RE2882



RE1711


20-OH-LeukotrieneB4


5-hydroxytryptophol

R06983

RE3011




R07105


ALDH5A1



R01041Glycerol


R05049Succinate



Acetyl-CoA

Lipoate

DLD

R00713

ADH5

Acetaldehyde

R00754


S-Formylglutathione

Ethanol




Metabolic genes

(b)





5 5

0 0 01

0 001 02

04 05 05 0

5

0

53 56

553 56

55 62 69

33 36 39 4

0 0 0 0

7



GN

DG

NK

RBK

HSK

PGL

FBA

FUM

MM

ERP

I








cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump


DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108



(b)




Acknowledgments


References
























































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology







Ecoli_UTI89 1193 2314 1632 845

(a)


Murein recycling


Nitrogen metabolism

GGPTRCO

GGGABADr

ALAALAD

MMCD

HPPPNDO

HKNDDH

DHCINDO

HKNTDH

GGGABAH

ALDD2x

PPPNDO

HOPNTAL

MMM2

MANGLYCptspp

MALDDH

PPCSCT

OP4ENH

TYROXDApp

42A12BOOXpp

PEAMNOpp

PACCOAL

FRULYSDG

MELIBt2pp

FRULYSt2pp

3HCINNMH

3HPPPNH

PSCLYSt2pp

DHPPD

DHCIND


GALCTNLt2pp

XYLt2pp

HCINNMt2rpp

HPPPNt2rpp

ASO3t8pp

ALDD19x


(b)












(a) (b)

minus50











0

Biom

ass fl

ux (h

minus1)

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

(a)


Glutathione



GSR

R00115


HemeThiosulfate


TST



L-Lysyl-tRNA

KARS

Lysine--tRNAligase

R03658

Alaninetransaminase

R00258L-Glutamate

R03105


Pyruvate


RE2034

Protoporphyrin


R00894


L-Cysteine



GCLC

R04880


Thiosulfate

Sulfate

Thiocyanate RE0159

R00209

Succinyl-CoA

CoA

L-Alanine



RE0361


R01698R03815

Dihydrolipoamide

Lipoamide

Lipoylprotein

RE2882



RE1711


20-OH-LeukotrieneB4


5-hydroxytryptophol

R06983

RE3011




R07105


ALDH5A1



R01041Glycerol


R05049Succinate



Acetyl-CoA

Lipoate

DLD

R00713

ADH5

Acetaldehyde

R00754


S-Formylglutathione

Ethanol




Metabolic genes

(b)





5 5

0 0 01

0 001 02

04 05 05 0

5

0

53 56

553 56

55 62 69

33 36 39 4

0 0 0 0

7



GN

DG

NK

RBK

HSK

PGL

FBA

FUM

MM

ERP

I








cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump


DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108



(b)




Acknowledgments


References
























































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology











(a) (b)

minus50











0

Biom

ass fl

ux (h

minus1)

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

(a)


Glutathione



GSR

R00115


HemeThiosulfate


TST



L-Lysyl-tRNA

KARS

Lysine--tRNAligase

R03658

Alaninetransaminase

R00258L-Glutamate

R03105


Pyruvate


RE2034

Protoporphyrin


R00894


L-Cysteine



GCLC

R04880


Thiosulfate

Sulfate

Thiocyanate RE0159

R00209

Succinyl-CoA

CoA

L-Alanine



RE0361


R01698R03815

Dihydrolipoamide

Lipoamide

Lipoylprotein

RE2882



RE1711


20-OH-LeukotrieneB4


5-hydroxytryptophol

R06983

RE3011




R07105


ALDH5A1



R01041Glycerol


R05049Succinate



Acetyl-CoA

Lipoate

DLD

R00713

ADH5

Acetaldehyde

R00754


S-Formylglutathione

Ethanol




Metabolic genes

(b)





5 5

0 0 01

0 001 02

04 05 05 0

5

0

53 56

553 56

55 62 69

33 36 39 4

0 0 0 0

7



GN

DG

NK

RBK

HSK

PGL

FBA

FUM

MM

ERP

I








cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump


DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108



(b)




Acknowledgments


References
























































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology



0

Biom

ass fl

ux (h

minus1)

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

Acet

ate

Etha

nol

Glu

cose

Succ

inat

e

(a)


Glutathione



GSR

R00115


HemeThiosulfate


TST



L-Lysyl-tRNA

KARS

Lysine--tRNAligase

R03658

Alaninetransaminase

R00258L-Glutamate

R03105


Pyruvate


RE2034

Protoporphyrin


R00894


L-Cysteine



GCLC

R04880


Thiosulfate

Sulfate

Thiocyanate RE0159

R00209

Succinyl-CoA

CoA

L-Alanine



RE0361


R01698R03815

Dihydrolipoamide

Lipoamide

Lipoylprotein

RE2882



RE1711


20-OH-LeukotrieneB4


5-hydroxytryptophol

R06983

RE3011




R07105


ALDH5A1



R01041Glycerol


R05049Succinate



Acetyl-CoA

Lipoate

DLD

R00713

ADH5

Acetaldehyde

R00754


S-Formylglutathione

Ethanol




Metabolic genes

(b)





5 5

0 0 01

0 001 02

04 05 05 0

5

0

53 56

553 56

55 62 69

33 36 39 4

0 0 0 0

7



GN

DG

NK

RBK

HSK

PGL

FBA

FUM

MM

ERP

I








cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump


DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108



(b)




Acknowledgments


References
























































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


5 5

0 0 01

0 001 02

04 05 05 0

5

0

53 56

553 56

55 62 69

33 36 39 4

0 0 0 0

7



GN

DG

NK

RBK

HSK

PGL

FBA

FUM

MM

ERP

I








cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump


DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108



(b)




Acknowledgments


References
























































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology



cbp

asp-L h

picbasp

h2o

dhor-S

h

co2

atpnh4

adph

q8

orot

q8h2

mqn8 mql8

ppi

orot5p

prpp h

ump


DHORD2

DHORD5

ORPT OMPDC

(a)

Reactions

CBMKr

ASPCT

DHORTS

DHORD2

DHORD5

ORPT

OMPDC

CFTO73

minFlux maxFlux

minus415

025

minus198

0

0

minus196

025

220

196

minus025

198

198

minus025

198

HS

minFlux maxFlux

minus426

026

minus203

0

0

minus203

026

226

203

minus026

203

203

minus026

203

UTI89

minFlux maxFlux

minus724

022

minus108

0

0

minus108

022

320

108

minus022

108

108

minus022

108



(b)




Acknowledgments


References
























































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology















































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology

Research Article Metabolic Modeling of Common Escherichia...

Documents

Transcript of Research Article Metabolic Modeling of Common Escherichia...