Research Article Dynamic Metabolic Footprinting Reveals ...downloads.hindawi.com/journals/ijg/2014/894296.pdfNational Food Institute, Technical University of Denmark, Mørkhøj Bygade,

Research ArticleDynamic Metabolic Footprinting Reveals the Key Componentsof Metabolic Network in Yeast Saccharomyces cerevisiae

Pramote Chumnanpuen1 Michael Adsetts Edberg Hansen2

Joslashrn Smedsgaard3 and Jens Nielsen4

1 Department of Zoology Faculty of Science Kasetsart University Bangkok 10900 Thailand2 Institut Pasteur Korea Sampyeong-dong 696 Bundang-gu Seongnam-si 463-400 Gyeonggi-do Republic of Korea3 National Food Institute Technical University of Denmark Moslashrkhoslashj Bygade 19 2860 Soslashborg Denmark4Department of Chemical and Biological Engineering Chalmers University of Technology Kemivagen 10412 96 Gothenburg Sweden

Correspondence should be addressed to Pramote Chumnanpuen pramoteckuacth

Received 9 August 2013 Revised 31 October 2013 Accepted 13 November 2013 Published 29 January 2014

Academic Editor Wanwipa Vongsangnak

Copyright copy 2014 Pramote Chumnanpuen et alThis is an open access article distributed under the Creative CommonsAttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

Metabolic footprinting offers a relatively easy approach to exploit the potentials of metabolomics for phenotypic characterizationof microbial cells To capture the highly dynamic nature of metabolites we propose the use of dynamic metabolic footprintinginstead of the traditional method which relies on analysis at a single time point Using direct infusion-mass spectrometry (DI-MS)we could observe the dynamic metabolic footprinting in yeast S cerevisiae BY4709 (wild type) cultured on 3 different C-sources(glucose glycerol and ethanol) and sampled along 10 time points with 5 biological replicates In order to analyze the dynamic massspectrometry data we developed the novel analysismethods that allowus to performcorrelation analysis to identifymetabolites thatsignificantly correlate over time during growth on the different carbon sources Both positive and negative electrospray ionization(ESI) modes were performed to obtain the complete information about the metabolite content Using sparse principal componentanalysis (Sparse PCA) we further identified those pairs of metabolites that significantly contribute to the separation From the listof significant metabolite pairs we reconstructed an interactionmap that provides information of how different metabolic pathwayshave correlated patterns during growth on the different carbon sources

1 Introduction

The word metabolism comes from Greek metabole whichmeans change or transformation and in fact the levels ofmost cellular metabolites change with half times of minutesseconds or even faster During the past years the complexityand dynamics nature of themetabolites have become increas-ingly apparent [1ndash6] The comprehensive analysis of a largeset of metabolites now referred to as metabolomics presentin a biological sample has emerged as an important toolin functional genomics and systems biology [7 8] Since itis ldquodownstreamrdquo of central dogma (not like transcriptomeand proteome at higher cascade) the metabolome shouldshow greater effects of genetic or physiological changes [910] and is therefore much closer related to the phenotypeexpressed by an organism [11] Metabolic fingerprinting (an

analysis of intracellular metabolites profiling) and metabolicfootprinting (an analysis of extracellular metabolites pro-filing) have succeeded in experimental characterization ofgenetic mutants on the basis of combined intracellular andextracellularmetabolite datameasured bymass spectrometry(MS) [5 8 11 12] However the measurement of extracel-lular metabolites often referred to as footprinting [4 11]represents several advantages over the analysis of intracellularcompounds often referred to as metabolic fingerprinting[13] for microbial cultures Although the functional analysisusing metabolic fingerprinting can classify the different phe-notypes of several mutants it is difficult to scale up for high-throughput screening of many mutants as it is quite time-consuming and subject to technical difficulties caused by therapid turnover and the need of quenching and extracting ofmetabolites from the extracellular space [5 11]

Hindawi Publishing CorporationInternational Journal of GenomicsVolume 2014 Article ID 894296 14 pageshttpdxdoiorg1011552014894296

2 International Journal of Genomics

Analysis ofmetabolite profiles can be done using differentanalytical techniques but recent technical advances in MShave brought this technology to the forefront amongmethodsfor metabolome analysis due to the high sensitivity andseparation efficiency [13 14] Gas chromatography coupledto MS (GC-MS) is the most extensively used techniquein metabolome analysis [15] as it is effective for resolvingcomplex biological mixtures and hereby enable reliablyidentification of the analyzed compounds However GC-MScan only analyze metabolites which are stable at the hightemperatures present in GC and they have to be volatileor can be made volatile upon chemical derivatization Mostmetabolites are not volatile and many of them are not stableat high temperature or cannot be derivatized Recently DirectInfusion-Mass Spectrometry (DI-MS) has been reported asan alternative to GC-MS and it seems to be an ideal analyticaltool for high-throughput metabolome analysis [16ndash20] Themost significant feature of ESI mode for DI-MS is that it isa very soft ionization technique which in many cases willproduce protonated molecular species (in positive ESI) fora broad range of different compounds with very high sen-sitivity [19] With high sensitivity and atmospheric pressureionization DI-MS can analyze the majority of metabolites inthe sample in a few seconds without any chromatographicseparation [16 17 19 20] An obvious extension of a DI-MS approach is to store the spectra in a database using thedatabase software included with most instruments This willgive a sample identification database rather than compoundidentification as anticipated by the manufactures [15 17 1921 22]

Here we present the use of dynamic metabolic footprint-ing analyzed by DI-MS for phenotypic information of yeastUsing novel data analysis methods we show that it is possibleto extract the key information about the metabolism fromfootprinting data when yeast is grown on different carbonsources The data analysis is based on identification of signif-icantly correlated metabolites over time which correspondsto flux-ratios for different metabolites The advantage of thisapproach is that it is not sensitive to a single time pointand therefore better allows for analysis of mutants havingdifferent growth rates

2 Materials and Methods

21 Reagents All reagents used for metabolite analysis andthe medium for yeast cultures were prepared using analyticalgrade ingredients Methanol acetonitrile and formic acidused for mass spectrometry were obtained from Sigma-Aldrich (St Louis MO USA) MilliQ-purified water wasused during the sample preparation for HPLC and DI-MSanalyses

22 Yeast Strain The S cerevisiae strain BY4709 (MAT120572ura3Δ0) obtained from the European S cerevisiae archivefor functional analysis (EUROSCARF Frankfurt Germany)was used in this study

23 Medium The strain was grown in a minimal syntheticmedium supplemented with a metabolite cocktail accordingto Allen et al [11] (see Table 1) The metabolite cocktail con-tained a selection of amino acids organic acids and organicbases The medium was based on yeast nitrogen base (YNB)without amino acids (BD Difco Franklin Lakes NJ USA)and the metabolite cocktail was composed of L-arginineL-aspartate L-glutamate L-histidine L-leucine L-lysine L-methionine L-serine L-threonine L-tryptophan L-valinecitrate fumarate malate pyruvate succinate cytosine anduracil All compounds from the metabolite cocktail had aconcentration of 1mM in the finalmediumCultivationswereperformed with 3 different carbon sources that is glucoseglycerol and ethanol Each carbon source is dissolved inwater autoclaved separately and finally added to give a finalconcentration 067 C-moleL corresponding to (2000 gLglucose 1534 gL ethanol or 2045 gL glycerol)

24 Cultivation Conditions Cultivations were performed in5 replicates in 500mL baffled shake flasks (150 rpm) at 30∘CEach flask contained 100mL of footprinting media preparedaccording to Allen et al [11] and closed with cotton plugsDuring cultivation the culture purity was monitored on aregular basis by phase contrast microscopy and they eastgrowth was monitored by OD measurements at 600 nm

25 Cell Mass Determination Biomass dry weight was deter-mined by filtering a known volume of fermentation brothapproximately 5mL through a dried preweighed nitrocel-lulose filter (Sartorius Stedim Biotech SA) with a pore size045 120583m The residue was washed twice with distilled waterThe filter was dried to constant weight in a microwave ovenat 150W for 10min cooled in a desiccator and the weightwas measured Besides the optical density was determinedat 600 nm using a Genesis 20 (Thermo spectronic) spec-trophotometer Sampleswere dilutedwithwater to obtainODmeasurements in the linear range of 01ndash05 OD units

26 Analysis of Culture Media by HPLC Samples harvestedfrom the cultivation broth were immediately filtered througha 045 120583m pore-size cellulose acetate filter (VWR) and storedat ndash20∘C until analysis Glucose glycerol ethanol succinateand acetate concentrations were determined by HPLC anal-ysis using an Aminex HPX-87H column (Biorad HerculesCA) and all the conditions were set following Zaldivar et al[23] Briefly the separation was performed at 45∘C withsulfuric acid (5mM) at a flow rate of 05mLmin as themobile phase Subsequent determinations of yield coefficientsfor extracellular metabolites as well as biomass were basedon linear regressions of their concentration as a function ofthe residual glucose concentration in the exponential growthphase

27 Sampling Procedure The yeast strain BY4709 was grownon three different carbon sources glucose ethanol andglycerol For each carbon source 5 replicates were carried outand the sampleswere taken every 3 hours from3 hr until 30 hrsuch that they represented different growth phases for growth

International Journal of Genomics 3

Table 1 Media composition used for metabolic footprinting experiment

Group Formula Common name MW gL

Yeast nitrogen base

Nitrogen source (NH4)2SO4 Ammonium sulphate 13213952 50198

Vitamins

C10H16N2O3S Biotin 24431064 000002C18H32CaN2O10 Calcium pantothenate 47653208 0002C19H19N7O6 Folic acid 44139746 0002C6H12O6 Inositol 18015588 001C6H5NO2 Niacin 1231094 00004C7H7NO2 p-Aminobenzoic acid 13713598 00002

C8H11NO3sdotHCl Pyridoxine hydrochloride 20563878 000056C17H20N4O6 Riboflavin 3763639 00002

C12H17ClN4OSsdotHCl Thiamin hydrochloride 33726852 00004

Trace elements

H3BO3 Boric acid 6183302 005CuSO4 Copper sulphate 1596086 000004KI Potassium iodide 16600277 00001

FeCl3 Ferric chloride 162204 00002MnSO4 Manganese sulphate 151000645 00004Na2MoO4 Sodium molybdate 2059171386 00002ZnSO4 Zinc sulphate 1614716 00004

Salts

KH2PO4 Potassium phosphate monobasic 136085542 085K2HPO4 Potassium phosphate dibasic 174175902 015MgSO4 Magnesium sulphate 1203676 05NaCl Sodium chloride 5844276928 001CaCl2 Calcium chloride 110984 01

Metabolite cocktail

Amino acids (all L-form)

C6H14N4O2 Arginine 17420096 01742C4H6NO4K Aspartate (monopotassium salt) 17119304 01712C5H8NNaO4 Glutamate (monosodium salt) 1691110893 01691C6H9N3O2 Histidine 15515456 01552C6H13NO2 Leucine 13117292 01312C6H14N2O2 Lysine (hydrochloride) 14618756 01462C5H11NO2S Methionine 14921134 01492C3H7NO3 Serine 10509258 01051C4H9NO3 Threonine 11911916 01191C11H12N2O2 Tryptophan 20422518 002042C5H11NO2 Valine 11714634 01051

Organic acids

C6H8O7 Citrate 19212352 02101C4H2O4Na2 Fumarate (disodium salt) 1600358186 016C4H6O5 Malate 13408744 01341C3H4O3 Pyruvate 8806206 00881

C4H4Na2O4 Succinate (disodium salt) 1620516986 02701

Base C4H5N3O Cytosine 111102 01111C4H4N2O2 Uracil 11208676 01121


on the different carbon sources Two-milliliter samples werecentrifuged (8000 g 10min) and the supernatant was storedat ndash18∘C until further DI-MS analysis

The supernatants were diluted fivefold with acetonitrileright before they were analyzed by DI-MS

28 Direct Infusion-Mass Spectrometry Analysis The samplepreparation for the dynamic metabolic footprinting analysiswas performed as illustrated in the metabolic footprintingpipeline in Figure 1 The supernatants were diluted fivefoldwith acetonitrile right before the injection The DI-MSanalysis was performed on a system setup with an Agi-lent 1100 microflow HLPC pump LC-Packings autosamplercoupled to a Micromass (Waters Manchester) Q-tof systemwith an electrospray ionization interface The instrumentwas tuned for maximal sensitivity at low flow rate andminimal fragmentation using leucine-enkephalin followedby external calibration using a mixture of PEG200 and 400in acetonitrile-water The samples were diluted fivefold inacetonitrile (with 1 120583g120583L leucine-enkephalin as an internalstandard mass reference) and centrifuged at 10000 g for1min The supernatants were transferred into 200 120583L HPLCvial inserts and the vials were placed in the autosamplerThe sequence of samples was randomly injected in orderto decrease the effects from instrumental bias The sampleswere analyzed by infusion of 5 120583L sample into the ESI sourceof Q-tof MS at a flow rate of 20120583Lmin A carrier flow ofmethanol was used at a rate 15 120583Lmin from the LC-pumpthrough the autosampler just before the ion source a flow of2 formic acid in water was fed into the solvent stream froma syringe pump at a rate of 5 120583Lmin using a t-piece givinga combined flow of 20 120583Lmin of 75 methanol-water with05 formic acid going into the source Mass spectra wereacquired in both positive and negative mode and data werecollected for 3minsample between 50 and 1000Dae at a rateof one continuum scansecond The Q-tof conditions werethe following capillary voltage 3000V in positive mode and2600V in negative mode desolvation temperature 150∘Cdry gas desolvation gas at 300 Lhr nebulizer flow 20 Lhrsource temperature 90∘C and cone voltage optimized tominimal fragmentation approximately 40V in positive modeand 30V in negative mode

29 Data Analysis Initially the raw data were processed acc-ording to the method described by Hansen and Smedsgaard[24]TheMSdata (as shown in Figure S1A)were preprocessedin the following way for each sample the elution profile weredetected 40 continuum scans (1 sscan) were summarizedto a single continuum spectrum to reduce noise followedby background subtraction and calculating the centroidspectrum (Figure S1B) Then all data were normalizedbased on the ion count of leucine-enkephalin (mz 5562771)internal standard to reduce technical variability from theinstrumental bias [24] The result was a matrix where eachrow corresponded to a sample and each of the columns to ametabolite The width of each column roughly correspondedto 05mz

In the following two criteria were applied for selecting(cherry-picking) the metabolites

(1) The first criteria dealt with screening for the singlemetabolites in each injected sample showing properchanges over time (either decreasing or increasing ora combination of both) Since the changing patterncan either be linear or nonlinear (eg exponentialor sigmoidal pattern) the data smoothness step wasrequired for the nonlinear function by fitting to the(standardized) time profiles In the later step thoseselected metabolites which have the actual trend inthe profiles (noise excluded) will further be analyzed

(2) For the next criteria we identified metabolite pairswhich show the covariance across time Eachmetabo-lite was paired up and plotted against each other toinvestigate the correlation pattern of individual pairand the regression line was fitted to the data Finallythe correlation (1198772 of the regression line) and the 119875value were calculated

The intercept slope (which is equivalent to what is oftenreferred to as yield coefficients) 1198772 and the 119875 value of theregression line for each metabolite were stored in a matrix ofdimension as the number of metabolites

Since many of the identified metabolites are related andcovary we used only the selected metabolites to analyze thevariance present in data Whereas ordinary Principal Com-ponent Analysis (PCA) [25] is widely used in data processingand dimensionality reduction PCA in general suffer fromthe fact that each component is a linear combination ofall the original variables Even if some variables containrandom noise those will be assigned weights (loadings) inthe linear combination Thus it is often difficult to interpretthe results As for ordinary PCA Sparse PCA [26] finds sets ofSparse vectors (having a relatively small number of nonzeroelements) for use as weights (loadings) in the linear combina-tions while still explainingmost of the variance present in thedata For all samples based on the 119875 value calculated for eachpair of ions we could generate a symmetric matrix of corre-lations with the ions along both axes The upper triangularpart of the matrix was extracted and unwrapped (Figure S2)Based on the matrix WT (1) times ESI-mode (2) times Source (3) timesRep (5) rows and (119873minus1)1198732 columns where119873 is the numberof ions present we investigated the variation (informational)content by Sparse Principal Component Analysis and furtheridentifiedmetabolites that additionally correlated in responseto time for growth on the different carbon sources

3 Results and Discussions

Metabolic footprinting is traditionally based on analysis ata single time point Therefore the analysis becomes verydependent on the growth rate of the organism and also onfactors such as inoculum size In order to circumvent thisproblem we developed a pipeline for dynamic metabolicfootprinting (Figure 1) This involved both HPLC and MSanalysis of extracellular samples during the fermentation


Yeast cultivation and sampling

Metabolic footprinting

analysisand

correlation analysis

PCA analysisKey metabolites

identification andnetwork analysis

Cherry-picking

Figure 1 The pipeline for our dynamic metabolic footprinting process Different perturbations are imposed on the microorganism forexample growth on different carbon sources or parallel analysis of different mutants and the growth profile is recorded using at least 10samples The samples are processed and analyzed using DI-MS The resulting spectra are processed to obtain correlation between differentmetabolites analyzed

00 5 10 15 20 25 30

20

40

60

80

Inte

nsity

Time (hour)

Glucose positivemdashmz 14412 (95)

(a)

0

Inte

nsity

0 5

5

10

10

15

15

20 25 30Time (hour)


(b)

0

5

0 20

10

15

40 60 80

Δx

Δy

Correlation = 099Yield =

Δy

Δx

Note∙ The numbers beside the dotsindicate the sampling time (hour)

∙ The number in blanket refers to theorder of peaks found in MS spectra

mz

214

09

(165

)

mz14412 (95)

log 10(P value)

(c)

Figure 2 The cherry-picking process illustrated Two upper figures show two selected metabolites during the first step and the lower figureshows their corresponding correlation in time Both metabolites will be picked Slopes (yields) in the matrix that had a 119875 value less than 001were set to zero

Table 2 Specific rates for the 3 sets of conditions

C-sources 120583max (hminus1) 119903

119904

(g substrategDWh) 119903119904

(C-mmolegDWh)Glucose 0391 0304 10133Ethanol 0121 0027 1117Glycerol 0129 0048 1564


0

1

2

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Time (hour)

Glucose

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

(a)

Time (hour)

0

1

2

3

4

5

6

7

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Ethanol

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

(b)

Time (hour)

0

1

2

3

4

5

0

2

4

6

8

10

12

14

16

18

20

22

0 3 6 9 12 15 18 21 24 27 30 33

Glycerol

GlucoseGlycerolSuccinate

ODEthanol

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

(c)

Figure 3 Profiles of growth (OD600

) and metabolite concentrations glucose glycerol succinate and ethanol of yeast BY4709 during thefermentationmeasured byHPLCThe plots show fermentation profile grown in different carbon sources glucose (a) ethanol (b) and glycerol(c)

31 Cultivation Profile Based on HPLCmeasurements of thecarbon sources and metabolic products the overall kineticsof substrate uptake could be determined For the glucosecultures the cells took approximately 12 hours to finish theirexponential phase with a glucose consumption rate of 0304 gglucosegDWh (see Table 2 and Figure 3) In contrast thecells growing on a nonfermentative carbon source took 21hours to finish their exponential phase with a consumptionrate of 0027 g ethanolgDW h and 0048 g glycerolgDWhrespectively Even though they were grown in different initialconcentrations in terms of g substrateL the C-mole amountswere exactly the same (067 C-moleL) for all three carbonsources Comparing the specific consumption rate in termsof C-mole (see Figure 3 and also Table 2) the cells could

consume glucose approximately 10 times faster than ethanoland glycerol (ie 10133 1117 and 1564 C-mmole of car-bongDWh for glucose ethanol and glycerol resp) Not sur-prisingly yeast cells preferred glucose which is a fermentablecarbon source more than the 2 nonfermentable carbonsources

32 Dynamic Metabolic Footprinting and DI-MS AnalysisThere is as mentioned in the introduction an interest inusingmetabolic footprinting as the turnover rate for extracel-lular metabolites is much longer than that for intra-cel-lular metabolites (fingerprinting) and footprinting does notrequire any complicated procedure for the quenching andextraction steps However using a single data point can


PCA labelled by source

0 20 400

0

50

50

02040

GlucoseEthanolGlycerol

minus50

minus50

minus60 minus40 minus20

minus40

minus60

minus20

PCq

PCp

(a)

0

0

2000 4000 6000 8000 10000 12000

0

0005

001

0015

002PCA loading (PC1)

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus001

minus0005

(b)

0 2000 4000 6000 8000 10000 12000

0

0005

001

0015

002

0025

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus0025

minus001

minus0005

PCA loading (PC2)

(c)

Figure 4 Standard PCA and loading show the clustering of different C-source cultures (including both ESI modes) and 5 replicates basedon yieldslope of metabolite profile

cause problems with capturing the phenotype of differentmutants and the approach proposed here may thereforeallow for wider use of metabolic footprinting Here we dem-onstrate high-throughput metabolic footprinting using DI-MS which allow us to cover more than 2000 ion peaks (bothfrom positive and negative ESI mode) which cover manymetabolites in different pathways to be analyzed at the sametime From the DI-MS analysis results and its integratedsignal spectra (Figures S1A and B) we pointed out how thetechniquewas so consistent giving such a high reproducibilityin all 5 biological replicates Even though it is complicatedto interpret biological meaning from thousands of ion peakswe here demonstrate the cherry-picking criteria to identify alist of single metabolites that showed significant change over

time Thus our method is an unbiased strategy to get to thekey metabolites

The DI-MS analysis showed that amino acids were bestdetected in positive mode probably due to the amino groupThe typical base peak ion was [M + H]+ and [M + CH

3CN +

H]+

33 Key Metabolites (Cherries) Picked up from MS SpectraAfter conversion and alignment of the data according toHansen and Smedsgaard [24] each of the metabolites wasfirst analyzed for trends over time The resulting matrixhad 300 rows (one for each sample) and approximately1500 columns After binning the data as part of the initialprocedure columns containing no data were removed


Sparse PCA labelled by source

0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10Sparse PCA loading (PC1)

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

Sparse PCA loading (PC2)

(c)

Figure 5 Sparse PCA and loading show the clustering of different C-source (including both ESI modes) and 5 replicates based on yieldslopeof metabolite profile

The first cherry-picking criteria identified a list of singlemetabolites that showed the proper change over time Afterthis step the number of columns in thematrix was reduced to145 metabolites Two examples of identified metabolites canbe seen in the top two figures in Figure 2

As the next criteria the degree of covariance over timewas calculated Figure 2 (bottom picture) illustrates the pro-cess where the two metabolites were picked during the firstselection step and shows covariance across time For all met-abolite pairs the intercept slope (yields) 1198772 and the 119875 valueof the regression line for each metabolite were stored in amatrix of dimension as the number of metabolites

34 Standard and Sparse PCA Analysis In order to evaluatethe information present in the matrix of metabolite pairs weinitially analyzed the variation in yields by PCA to identifythe key metabolites which are significantly different betweenthe sample groups and to how well they can separate growthon the different carbon source from each other

The upper triangular part of the matrix of the yields wasextracted and unwrapped Based on the matrix columns asshown with the cherry-picking method (Figure 2) we calcu-lated the PCA loading for PCA analysis Figure 4 illustratesthe scores and loadings for the first two principal componentsusing ordinary PCA As can be seen from the plots the yieldsof the selected metabolite pairs show a nice separation of


GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP

T3P1T3P2GNMHISPDLASPPYROX

GL3P

Figure 6 Key metabolite network on PC1 there are 32 key metabolites linking to each other with 4 hubs that is glucose cyclic AMP cyclicdAMP and NAMN in PC1 networkThese key metabolites can be used as PC1 to separate the sample taken from glucose culture from ethanoland glycerol cultures Since there are 3 different patterns of changing in metabolite profile these nodes are shown in 3 different colors red(increasing) (decreasing) and blue (nonmonotonous or constant)

the carbon sources Still when looking at the loadings we seethat all metabolite pairs are assigned weights

Sparse PCA was used to obtain sets of Sparse vectors forweights (loadings) in the linear combinations while explain-ing most of the variance present in the data Figure 5 illus-trates the effect of using of Sparse PCA rather than ordinaryPCA on the loadingsWe see thatmost of themetabolite pairshave been assigned zeroweight whereas only a few pairs havebeen assigned a weight Although most of the weights in theloadings are zero we still see the same grouping into carbonsources as when using ordinary PCA

As can be seen from the Sparse PCA a few pairs of met-abolites can separate the data into distinct carbon sourcesNext we wish to investigate those metabolites in furtherdetail From the reconstruction of the correlation networksit was found that pairs from PC1 are most important asthey generate the largest network PC1 segregates the signalfrom glucose from the other two carbon sources (ethanoland glycerol) The size of key metabolites network representsthe level of differences in dynamic pattern among the samplegroups [6 21]

35 Identification of the High Correlated Ion Pairs Pairs ofMS ions that correlated over time were identified as thosethe Sparse PCA were assigned weights Only the metabolitepairs from the loadings corresponding to the Sparse PC1and Sparse PC2 were extracted There were a total of 48pairs which corresponds to correlation between 44 ions askey metabolites 24 pairs between 17 metabolites for PC1and 3 pairs between 3 metabolites for PC2 in positive ESImode (see Table 3) and 17 pairs between 19 metabolites forPC1 and 4 pairs between 5 metabolites for PC2 in negativeESI mode (see Table 4) A few metabolites were identifiedboth in positive and negative ESI mode and the number ofmetabolites identified was 45 For each ion the correspondingmetabolite was identified using the database developed byHoslashjer-Pedersen [17] and a list of the identified metabolitesfor the two modes are given in Tables 5 and 6 Moreoverthe fragmentation patterns from ESI-MS of each metabolitewere also carefully considered and compared according to theGolmmetabolome databases [15] to confirm the possible cal-culatedmass per charge Formost of the ions there is a uniqueidentification of the corresponding metabolite but for someions more than one metabolite has the corresponding mass


Table 3 List of major metabolites (significantly correlate response over time in the different C-source) by positive ESI mode

Positive ESI pairs PChomoserinethreonine Carnitine4-aminobutanoate

1

Methylglyoxal 3-IndoleacetonitrileSuccinate 3-IndoleacetonitrileValine 31015840 51015840 cyclic AMPLeucine 31015840 51015840 cyclic AMP4-Aminobutyric acidaspartate 31015840 51015840 cyclic AMPValine 31015840 51015840 cyclic AMPHomoserinethreonine 31015840 51015840 cyclic AMPThymine4-guanidino-butanamide 31015840 51015840 cyclic AMPLeucine 31015840 51015840 cyclic AMPsn-Glycerol-3-phosphate3-dehydroshikimate 31015840 51015840 cyclic AMPGlutamine 31015840 51015840 cyclic AMPBut-1-ene-124-tricarboxylate78 diaminononanoate 31015840 51015840 cyclic AMPValine 31015840 51015840 cyclic AMPMethylglyoxal 31015840 51015840 cyclic AMPhomoserinethreonine 31015840 51015840 cyclic AMPValine 31015840 51015840 cyclic AMPSuccinate 31015840 51015840 cyclic AMPN-Acetyl-L-glutamate 31015840 51015840 Cyclic AMPLeucine Nicotinate-D-ribonucletide4-Aminobutyric acidaspartate Nicotinate-D-ribonucletideGlyceronelactate Nicotinate-D-ribonucletidesn-Glycerol-3-phosphate3-dehydroshikimate Nicotinate-D-ribonucletideBut-1-ene-124-tricarboxylate78 diaminononanoate Nicotinate-D-ribonucletideLeucine O-Acetyl-L-serine

2Aspartate O-Acetyl-L-serineLeucine O-Acetyl-L-serine

for example for the ion 1790353mz there are three potentialcandidates for the corresponding metabolite namely 3-(4-hydroxyphenyl) pyruvate glucose and myo-inositol In thisparticular case glucose is the most likely metabolite and itwas chosen but cases where there could not be made a clearassignment both metabolite names were used

PC1 ions primarily separate the glucose samples from theethanol and glycerol samples and PC2 mainly separates theethanol samples from the glycerol samples The metabolitepairs from PC1 represent a network as illustrated in Figure 6and the metabolite pairs from PC2 represent two networks asillustrated in Figure 7 In these networks the edges representcorrelations between the corresponding key metabolites It isobserved that most of the key metabolites are amino acidsthat are linked to cyclic AMP or cyclic dAMP but glucoseis also clearly correlated to several metabolites The colorcode indicates the slope with green indicating decreasingconcentration with time and red increasing concentrationwith time A few metabolites do not change much in concen-tration profile or do not have a monotonous concentrationprofile during the whole fermentation for example theirconcentration are increasing in one part of the fermentationand decreasing in another part of the fermentation and theseare marked blue in the networks It is interesting to note

that there is a positive correlation between cAMP and mostamino acids whereas the glucose concentration is negativelycorrelated with the concentration of several metabolites forexample glycerol

The PCA results both ordinary PCA and Sparse PCAshowed that the main variance in the data is caused by the3 different C-sources glucose ethanol and glycerol TheSparse PCA finds sets of Sparse vectors for use as weightsin the linear combinations while still explaining most ofthe variance present in the data It is built on the factthat PCA can be written as a regression-type optimizationproblem with a quadratic penalty the lasso penalty (via theelastic net) can then be directly integrated into the regressioncriterion leading to a modified PCA with Sparse loadings[26] According to the results we obtained fromboth ordinaryPCA (Figure 3) and Sparse PCA (Figure 4) it is clearlyshown that the Sparse PCA gives a better clustering amongthe groups compared to the normal PCA results Focusingon the consistency of this analysis the results from PCAindicated that the main variance in the data is caused by thedifferent C-sources with high reproducibilityThe 5 replicatesare separated into the same cluster based on the dynamicmetabolic profile which is promising


Table 4 List of major metabolites (significantly correlate response over time in the different C-source) by negative ESI mode

Negative ESI pairs PCGlyoxylate(s)-lactaldehyde 4-Hydroxybenzoate

1

Ser ItaconateN-acetylputrescine4-Aminobutanoate Xanthine4-Aminobutanoate CarnitineGlyceronelactate 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositol(s)-Lactaldehydeglycerol 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositolGlycine 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositol4-Hydroxybenzoate 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositol2-Oxoglutarategamma-amino-gamma-cyanobutanoate2-dehydropantoateglutaminelysine 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositol

ItaconateN-acetylputrescine 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositol4-Hydroxy-L-threonineadeninehomocysteine2-phenylacetamide3-hydroxyanthranilate 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositol

3-Hydroxyanthranilate 3-(-4-hydroxyphenyl)pyruvateglucosemyoinositol(s)-lactaldehydeglycerol cdAMPGlycinesn-glycerol 3-phosphate cdAMPItaconateN-acetylputrescine cdAMPD-Glyceraldehyde 3-phosphateglyceronephosphateguanineN-methyl-L-histidinepyridoxaminephosphatepyridoxine

cdAMP

3-Hydroxyanthranilate cdAMPSulfide Glycerol

2Sulfide UracilGlycerol PhosphocholineMet Phosphocholine

Table 5 List of key metabolites in positive ESI mode

Abbreviation Metabolites 119898119911 SuggestedACNL 3-Indoleacetonitrile 1981135 +CH3CNASP Aspartate 880879 minusCHOOH

cAMP Cyclic AMP 31248573300861 minusH2O+H

DANNA 78 Diaminonanoate 1890893 +HDHSK 3-Dehydroshikimate 1731333 +HGBAD 4-Guanidinine-butanide 1271331 minusH2OGL3P sn-Glycerol-3-phosphate 1731333 +HGLN Glutamine 1880823 +CH3CNHSER Homoserine 1020638 minusH2O

LEU Leucine 861026 and 1321146 minusNminusCHOOH

MTHGXL Methylglyoxal 730893 +HNAGLU N-Acetyl-L-glutamate 190069 +HNAMN Nicotinate-D-ribonucleotide 3780777 +CH3CNOAHSER O-Acetyl-L-Serine 1364695 +HSUCC Succinate 1190985 +H

THR Threonine 1024362 and 1204362 minusH2O+H

THY Thymine 1271331 +H

VAL Valine 720881 and 1180985 minusCHOOH+H


Table 6 List of key metabolites in negative ESI mode

Abbreviation Metabolites 119898119911 Suggested4HBZ 4-Hydroxybebzoate 1190248 minusH2O4HLT 4-Hydroxy-L-threonine 1343312 minusHAD Adenine 1343312 minusHAKG 2-Oxoglutarate 1270434 minusH2OGABA 4-Aminobutanoate 1024149 minusH

GL Glycerol 730310 and 910174 minusH2OminusH

GLC Glucose 1790353 minusHGLX Glyoxalate 730310 minusHGLY Glycine 760346 minusHGN Guanine 1503323 minusHHAN 3-Hydroxyanthranilate 1520691 minusHHCYS Homocysteine 1343312 minusHLACAL Lactaldehyde 730310 minusHLYS Lysine 1270434 minusH2OMET Methionine 1480470 minusHMVL Mevalonate 1290356 ndashH2OMHIS N-Methyl-L-histidine 1503223 minusH2OPCHO Choline Phosphate 1830650 minusHPDLASP Pyridoxamine phosphate 1503223 minusH3PO4

cdAMP Cyclic dAMP 2150070 minusH3PO4

PYRDX Pyridoxine 1503223 minusH2OSER Serine 1041285 minusHH2SO3 Sulfide 809743 minusHT3P1 D-Glyceraldehyde-3-phosphate 1503223 minusH2OT3P2 Glyceronephosphate 1503223 minusH2OURA Uracil 930162 minusH2OXAN Xanthine 1510958 minusH

36 Key Components of Metabolic Network of Yeast Grownon Different C-Sources The key metabolites network on PC1which contain more metabolite nodes compared to thaton PC2 show the higher difference between glucose andethanol or glucose and glycerol cultures while there are lessdifferences between glycerol and ethanol cultures In otherwords the dynamic pattern of the glucose culture is moreunique than the rest So we can obviously see that thereare largest dynamic changes in metabolite concentrationsduring growth on glucose which can be explained by the factthat glucose is a fermentative carbon source whereas there ispurely respiratory growth on ethanol and glycerol

The 4 main hubs of the network are glucose cyclic AMPcyclic dAMP and nicotinate-D-ribonucleotide (NAMN) asillustrated in the key metabolites network on PC1 (Figure 6)

Cyclic AMP and cyclic dAMP have been shown toregulatemany different nutrient responsesThe levels of thesemetabolites are linked to the biosynthesis of many aminoacids and also affected by the concentrations of glucose

and also amino acids [27] Since fermentable sugars arespecific stimulators for cAMP synthesis in yeast cells [28] theRascAMP pathway is activated by both growth signal (egglucose) and stress signals (eg UV radiation and starvation)[29] Our finding from the correlation analyses is consistentwith this as it shows that glucose and cAMP have the samedynamic pattern at all conditions

Since low cAMP concentration stimulates the uptake ofand L-leucine [30] we can see that the uptake rate of leucinewas increasing (the extracellular leucine level was decreas-ing) while the extracellular cAMP level was decreasing Fromthe keymetabolites network on PC2 (Figure 7) we also foundthe link between methionine glycine 4-phospho-hydroxy-L-threonine and pyridoxine As the methionine is requiredfor synthesis of phosphatidylcholine (PC) via methylation ofphosphatidylethanolamine (PE) [31] these two metaboliteshave high correlation (sharing the same pattern) in corre-lation analysis result In a similar case 4-phospho-hydroxy-L-threonine (4HLT the precursor to pyridoxal 51015840-phosphate


MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3

Figure 7 Key metabolites network on PC2 there are 8 key metabolites in PC2 network 5 and 3 nodes in each subnetwork Thesekey metabolites can be used as PC2 to separate the sample grouptaken from ethanol culture from glycerol culture Since there are 3different patterns of changing in metabolite profile these nodes areshown in 3 different colors red (increasing) (decreasing) and blue(nonmonotonous or constant)

biosynthesis) was also linked to glycine serine and threoninemetabolism and also pyridoxal phosphate synthesis pathwayvia cyclic AMP and cyclic dAMP

In conclusion we here show that through dynamic met-abolic footprinting it is possible to identify correlationsbetween metabolites that can be used to provide new insightinto possible regulatory structures Furthermore our correla-tion analysis may be used for identification of key biomarkersfor specific phenotypes during dynamic growth conditions

Conflict of Interests

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

Acknowledgments

This researchworkwas supported by theCenter forMicrobialBiotechnology (CMB)Department of SystemsBiology Tech-nical University of Denmark Part of our research was alsosupported by Preproposal Research Fund from the Faculty ofScience Kasetsart University The authors also thank HanneJakobsen for valuable assistance with running the DI-MS

References

[1] D B Kell R M Darby and J Draper ldquoGenomic computingExplanatory analysis of plant expression profiling data usingmachine learningrdquo Plant Physiology vol 126 no 3 pp 943ndash9512001

[2] S Krug G Kastenmuller F Stuckler et al ldquoThe dynamicrange of the humanmetabolome revealed by challengesrdquo FASEBJournal vol 26 no 6 pp 2607ndash2619 2012

[3] V Behrends T M D Ebbels H D Williams and J G BundyldquoTime-resolved metabolic footprinting for nonlinear modelingof Bacterial substrate utilizationrdquo Applied and EnvironmentalMicrobiology vol 75 no 8 pp 2453ndash2463 2009

[4] D B Kell M Brown H M Davey W B Dunn I Spasic andS G Oliver ldquoMetabolic footprinting and systems biology themedium is the messagerdquo Nature Reviews Microbiology vol 3no 7 pp 557ndash565 2005

[5] V Mapelli L Olsson and J Nielsen ldquoMetabolic footprint-ing in microbiology methods and applications in functionalgenomics and biotechnologyrdquo Trends in Biotechnology vol 26no 9 pp 490ndash497 2008

[6] R J Bino R D Hall O Fiehn et al ldquoPotential of metabolomicsas a functional genomics toolrdquo Trends in Plant Science vol 9no 9 pp 418ndash425 2004

[7] O Fiehn ldquoMetabolomicsmdashthe link between genotypes andphenotypesrdquo Plant Molecular Biology vol 48 no 1-2 pp 155ndash171 2002

[8] S G Villas-Boas J Nielsen J Smedsgaard et al MetabolomeAnalysis An Introduction Wiley 2007

[9] A B Canelas N Harrison A Fazio et al ldquoIntegrated mul-tilaboratory systems biology reveals differences in proteinmetabolism between two reference yeast strainsrdquo Nature Com-munications vol 1 no 9 article 145 2010

[10] J K Kim T Bamba K Harada E Fukusaki and A KobayashildquoTime-course metabolic profiling in Arabidopsis thaliana cellcultures after salt stress treatmentrdquo Journal of ExperimentalBotany vol 58 no 3 pp 415ndash424 2007

[11] J Allen H M Davey D Broadhurst et al ldquoHigh-throughputclassification of yeast mutants for functional genomics usingmetabolic footprintingrdquoNature Biotechnology vol 21 no 6 pp692ndash696 2003

[12] J I Castrillo A Hayes S Mohammed S J Gaskell and SG Oliver ldquoAn optimized protocol for metabolome analysis inyeast using direct infusion electrospray mass spectrometryrdquoPhytochemistry vol 62 no 6 pp 929ndash937 2003

[13] S G Villas-Boas S Mas M Akesson J Smedsgaard and JNielsen ldquoMass spectrometry in metabolome analysisrdquo MassSpectrometry Reviews vol 24 no 5 pp 613ndash646 2005

[14] S G Villas-Boas S Noel G A Lane G Attwood and ACookson ldquoExtracellular metabolomics a metabolic footprint-ing approach to assess fiber degradation in complex mediardquoAnalytical Biochemistry vol 349 no 2 pp 297ndash305 2006

[15] J Kopka N Schauer S Krueger et al ldquoGMDCSBDB theGolm metabolome databaserdquo Bioinformatics vol 21 no 8 pp1635ndash1638 2005

[16] W B Dunn S Overy and W P Quick ldquoEvaluation ofautomated electrospray-TOF mass spectrometry for metabolicfingerprinting of the plant metabolomerdquo Metabolomics vol 1no 2 pp 137ndash148 2005


[17] J Hoslashjer-Pedersen J Smedsgaard and J Nielsen ldquoThe yeastmetabolome addressed by electrospray ionization mass spec-trometry initiation of a mass spectral library and its appli-cations for metabolic footprinting by direct infusion massspectrometryrdquoMetabolomics vol 4 no 4 pp 393ndash405 2008

[18] J Smedsgaard and J C Frisvad ldquoUsing direct electrospray massspectrometry in taxonomy and secondary metabolite profilingof crude fungal extractsrdquo Journal of Microbiological Methodsvol 25 no 1 pp 5ndash17 1996

[19] J Smedsgaard and J Nielsen ldquoMetabolite profiling of fungi andyeast from phenotype to metabolome by MS and informaticsrdquoJournal of Experimental Botany vol 56 no 410 pp 273ndash2862005

[20] S Mas S G Villas-Boas M E Hansen M Akesson and JNielsen ldquoA comparison of direct infusion MS and GC-MS formetabolic footprinting of yeast mutantsrdquo Biotechnology andBioengineering vol 96 no 5 pp 1014ndash1022 2007

[21] M Stitt R Sulpice and J Keurentjes ldquoMetabolic networks howto identify key components in the regulation ofmetabolism andgrowthrdquo Plant Physiology vol 152 no 2 pp 428ndash444 2010

[22] M Arita ldquoAdditional paper computational resources formetabolomicsrdquo Briefings in Functional Genomics amp Proteomicsvol 3 no 1 pp 84ndash93 2004

[23] J Zaldivar A Borges B Johansson et al ldquoFermentationperformance and intracellularmetabolite patterns in laboratoryand industrial xylose-fermenting Saccharomyces cerevisiaerdquoApplied Microbiology and Biotechnology vol 59 no 4-5 pp436ndash442 2002

[24] M A E Hansen and J Smedsgaard ldquoAutomated work-flow forprocessing high-resolution direct infusion electrospray ioniza-tion mass spectral fingerprintsrdquo Metabolomics vol 3 no 1 pp41ndash54 2007

[25] I T Jolliffe Principal Component Analysis Springer New YorkNY USA 2nd edition 2002

[26] H Zou T Hastie and R Tibshirani ldquoSparse principal compo-nent analysisrdquo Journal of Computational and Graphical Statis-tics vol 15 no 2 pp 265ndash286 2006

[27] I Leclerc and G A Rutter ldquoAMP-activated protein kinasea new 120573-cell glucose sensor Regulation by amino acids andcalcium ionsrdquo Diabetes vol 53 no 3 pp S67ndashS74 2004

[28] K Mbonyi L van Aelst J C Arguelles A W H Jans and J MThevelein ldquoGlucose-induced hyperaccumulation of cyclicAMPand defective glucose repression in yeast strains with reducedactivity of cyclic AMP-dependent protein kinaserdquo Molecularand Cellular Biology vol 10 no 9 pp 4518ndash4523 1990

[29] D Engelberg C Klein H Martinetto K Struhl and M KarinldquoThe UV response involving the Ras signaling pathway andAP-1 transcription factors is conserved between yeast andmammalsrdquo Cell vol 77 no 3 pp 381ndash390 1994

[30] S K Sharma and G P Talwar ldquoAction of cyclic adenosine 3rsquo5rsquo-monophosphate in vitro on the uptake and incorporation ofuridine into ribonucleic acid in ovariectomized rat uterusrdquoTheJournal of Biological Chemistry vol 245 no 7 pp 1513ndash15191970

[31] H Jamil Z Yao and D E Vance ldquoFeedback regulationof CTP phosphocholine cytidylyltransferase translocationbetween cytosol and endoplasmic reticulum by phosphatidyl-cholinerdquoThe Journal of Biological Chemistry vol 265 no 8 pp4332ndash4339 1990

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


Analysis ofmetabolite profiles can be done using differentanalytical techniques but recent technical advances in MShave brought this technology to the forefront amongmethodsfor metabolome analysis due to the high sensitivity andseparation efficiency [13 14] Gas chromatography coupledto MS (GC-MS) is the most extensively used techniquein metabolome analysis [15] as it is effective for resolvingcomplex biological mixtures and hereby enable reliablyidentification of the analyzed compounds However GC-MScan only analyze metabolites which are stable at the hightemperatures present in GC and they have to be volatileor can be made volatile upon chemical derivatization Mostmetabolites are not volatile and many of them are not stableat high temperature or cannot be derivatized Recently DirectInfusion-Mass Spectrometry (DI-MS) has been reported asan alternative to GC-MS and it seems to be an ideal analyticaltool for high-throughput metabolome analysis [16ndash20] Themost significant feature of ESI mode for DI-MS is that it isa very soft ionization technique which in many cases willproduce protonated molecular species (in positive ESI) fora broad range of different compounds with very high sen-sitivity [19] With high sensitivity and atmospheric pressureionization DI-MS can analyze the majority of metabolites inthe sample in a few seconds without any chromatographicseparation [16 17 19 20] An obvious extension of a DI-MS approach is to store the spectra in a database using thedatabase software included with most instruments This willgive a sample identification database rather than compoundidentification as anticipated by the manufactures [15 17 1921 22]

Here we present the use of dynamic metabolic footprint-ing analyzed by DI-MS for phenotypic information of yeastUsing novel data analysis methods we show that it is possibleto extract the key information about the metabolism fromfootprinting data when yeast is grown on different carbonsources The data analysis is based on identification of signif-icantly correlated metabolites over time which correspondsto flux-ratios for different metabolites The advantage of thisapproach is that it is not sensitive to a single time pointand therefore better allows for analysis of mutants havingdifferent growth rates

2 Materials and Methods

21 Reagents All reagents used for metabolite analysis andthe medium for yeast cultures were prepared using analyticalgrade ingredients Methanol acetonitrile and formic acidused for mass spectrometry were obtained from Sigma-Aldrich (St Louis MO USA) MilliQ-purified water wasused during the sample preparation for HPLC and DI-MSanalyses

22 Yeast Strain The S cerevisiae strain BY4709 (MAT120572ura3Δ0) obtained from the European S cerevisiae archivefor functional analysis (EUROSCARF Frankfurt Germany)was used in this study

23 Medium The strain was grown in a minimal syntheticmedium supplemented with a metabolite cocktail accordingto Allen et al [11] (see Table 1) The metabolite cocktail con-tained a selection of amino acids organic acids and organicbases The medium was based on yeast nitrogen base (YNB)without amino acids (BD Difco Franklin Lakes NJ USA)and the metabolite cocktail was composed of L-arginineL-aspartate L-glutamate L-histidine L-leucine L-lysine L-methionine L-serine L-threonine L-tryptophan L-valinecitrate fumarate malate pyruvate succinate cytosine anduracil All compounds from the metabolite cocktail had aconcentration of 1mM in the finalmediumCultivationswereperformed with 3 different carbon sources that is glucoseglycerol and ethanol Each carbon source is dissolved inwater autoclaved separately and finally added to give a finalconcentration 067 C-moleL corresponding to (2000 gLglucose 1534 gL ethanol or 2045 gL glycerol)

24 Cultivation Conditions Cultivations were performed in5 replicates in 500mL baffled shake flasks (150 rpm) at 30∘CEach flask contained 100mL of footprinting media preparedaccording to Allen et al [11] and closed with cotton plugsDuring cultivation the culture purity was monitored on aregular basis by phase contrast microscopy and they eastgrowth was monitored by OD measurements at 600 nm

25 Cell Mass Determination Biomass dry weight was deter-mined by filtering a known volume of fermentation brothapproximately 5mL through a dried preweighed nitrocel-lulose filter (Sartorius Stedim Biotech SA) with a pore size045 120583m The residue was washed twice with distilled waterThe filter was dried to constant weight in a microwave ovenat 150W for 10min cooled in a desiccator and the weightwas measured Besides the optical density was determinedat 600 nm using a Genesis 20 (Thermo spectronic) spec-trophotometer Sampleswere dilutedwithwater to obtainODmeasurements in the linear range of 01ndash05 OD units

26 Analysis of Culture Media by HPLC Samples harvestedfrom the cultivation broth were immediately filtered througha 045 120583m pore-size cellulose acetate filter (VWR) and storedat ndash20∘C until analysis Glucose glycerol ethanol succinateand acetate concentrations were determined by HPLC anal-ysis using an Aminex HPX-87H column (Biorad HerculesCA) and all the conditions were set following Zaldivar et al[23] Briefly the separation was performed at 45∘C withsulfuric acid (5mM) at a flow rate of 05mLmin as themobile phase Subsequent determinations of yield coefficientsfor extracellular metabolites as well as biomass were basedon linear regressions of their concentration as a function ofthe residual glucose concentration in the exponential growthphase

27 Sampling Procedure The yeast strain BY4709 was grownon three different carbon sources glucose ethanol andglycerol For each carbon source 5 replicates were carried outand the sampleswere taken every 3 hours from3 hr until 30 hrsuch that they represented different growth phases for growth




Yeast nitrogen base


Vitamins




Trace elements



Salts


Metabolite cocktail



Organic acids



















analysisand




Cherry-picking


00 5 10 15 20 25 30

20

40

60

80

Inte

nsity

Time (hour)


(a)

0

Inte

nsity

0 5

5

10

10

15

15

20 25 30Time (hour)


(b)

0

5

0 20

10

15

40 60 80

Δx

Δy


Δy

Δx



mz

214

09

(165

)

mz14412 (95)

log 10(P value)

(c)




119904




0

1

2

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Time (hour)

Glucose

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

(a)

Time (hour)

0

1

2

3

4

5

6

7

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Ethanol

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

(b)

Time (hour)

0

1

2

3

4

5

0

2

4

6

8

10

12

14

16

18

20

22

0 3 6 9 12 15 18 21 24 27 30 33

Glycerol


ODEthanol

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

(c)








0 20 400

0

50

50

02040


minus50

minus50


minus40

minus60

minus20

PCq

PCp

(a)

0

0

2000 4000 6000 8000 10000 12000

0

0005

001

0015


Ion pair number

Load

ing

valu

e

minus002

minus0015

minus001

minus0005

(b)

0 2000 4000 6000 8000 10000 12000

0

0005

001

0015

002

0025

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus0025

minus001

minus0005

PCA loading (PC2)

(c)





3CN +

H]+




0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08


Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02


(c)







GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology




Yeast nitrogen base


Vitamins




Trace elements



Salts


Metabolite cocktail



Organic acids



















analysisand




Cherry-picking


00 5 10 15 20 25 30

20

40

60

80

Inte

nsity

Time (hour)


(a)

0

Inte

nsity

0 5

5

10

10

15

15

20 25 30Time (hour)


(b)

0

5

0 20

10

15

40 60 80

Δx

Δy


Δy

Δx



mz

214

09

(165

)

mz14412 (95)

log 10(P value)

(c)




119904




0

1

2

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Time (hour)

Glucose

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

(a)

Time (hour)

0

1

2

3

4

5

6

7

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Ethanol

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

(b)

Time (hour)

0

1

2

3

4

5

0

2

4

6

8

10

12

14

16

18

20

22

0 3 6 9 12 15 18 21 24 27 30 33

Glycerol


ODEthanol

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

(c)








0 20 400

0

50

50

02040


minus50

minus50


minus40

minus60

minus20

PCq

PCp

(a)

0

0

2000 4000 6000 8000 10000 12000

0

0005

001

0015


Ion pair number

Load

ing

valu

e

minus002

minus0015

minus001

minus0005

(b)

0 2000 4000 6000 8000 10000 12000

0

0005

001

0015

002

0025

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus0025

minus001

minus0005

PCA loading (PC2)

(c)





3CN +

H]+




0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08


Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02


(c)







GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology
















analysisand




Cherry-picking


00 5 10 15 20 25 30

20

40

60

80

Inte

nsity

Time (hour)


(a)

0

Inte

nsity

0 5

5

10

10

15

15

20 25 30Time (hour)


(b)

0

5

0 20

10

15

40 60 80

Δx

Δy


Δy

Δx



mz

214

09

(165

)

mz14412 (95)

log 10(P value)

(c)




119904




0

1

2

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Time (hour)

Glucose

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

(a)

Time (hour)

0

1

2

3

4

5

6

7

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Ethanol

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

(b)

Time (hour)

0

1

2

3

4

5

0

2

4

6

8

10

12

14

16

18

20

22

0 3 6 9 12 15 18 21 24 27 30 33

Glycerol


ODEthanol

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

(c)








0 20 400

0

50

50

02040


minus50

minus50


minus40

minus60

minus20

PCq

PCp

(a)

0

0

2000 4000 6000 8000 10000 12000

0

0005

001

0015


Ion pair number

Load

ing

valu

e

minus002

minus0015

minus001

minus0005

(b)

0 2000 4000 6000 8000 10000 12000

0

0005

001

0015

002

0025

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus0025

minus001

minus0005

PCA loading (PC2)

(c)





3CN +

H]+




0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08


Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02


(c)







GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology




analysisand




Cherry-picking


00 5 10 15 20 25 30

20

40

60

80

Inte

nsity

Time (hour)


(a)

0

Inte

nsity

0 5

5

10

10

15

15

20 25 30Time (hour)


(b)

0

5

0 20

10

15

40 60 80

Δx

Δy


Δy

Δx



mz

214

09

(165

)

mz14412 (95)

log 10(P value)

(c)




119904




0

1

2

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Time (hour)

Glucose

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

(a)

Time (hour)

0

1

2

3

4

5

6

7

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Ethanol

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

(b)

Time (hour)

0

1

2

3

4

5

0

2

4

6

8

10

12

14

16

18

20

22

0 3 6 9 12 15 18 21 24 27 30 33

Glycerol


ODEthanol

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

(c)








0 20 400

0

50

50

02040


minus50

minus50


minus40

minus60

minus20

PCq

PCp

(a)

0

0

2000 4000 6000 8000 10000 12000

0

0005

001

0015


Ion pair number

Load

ing

valu

e

minus002

minus0015

minus001

minus0005

(b)

0 2000 4000 6000 8000 10000 12000

0

0005

001

0015

002

0025

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus0025

minus001

minus0005

PCA loading (PC2)

(c)





3CN +

H]+




0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08


Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02


(c)







GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


0

1

2

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Time (hour)

Glucose

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

(a)

Time (hour)

0

1

2

3

4

5

6

7

0

2

4

6

8

10

12

14

16

18

20

0 3 6 9 12 15 18 21 24 27 30 33

Ethanol

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

(b)

Time (hour)

0

1

2

3

4

5

0

2

4

6

8

10

12

14

16

18

20

22

0 3 6 9 12 15 18 21 24 27 30 33

Glycerol


ODEthanol

Con

cent

ratio

n of

gly

cero

l and

acet

ate (

gL)

Con

cent

ratio

n of

C-s

ourc

es (g

L)

OD600

OD600

(c)








0 20 400

0

50

50

02040


minus50

minus50


minus40

minus60

minus20

PCq

PCp

(a)

0

0

2000 4000 6000 8000 10000 12000

0

0005

001

0015


Ion pair number

Load

ing

valu

e

minus002

minus0015

minus001

minus0005

(b)

0 2000 4000 6000 8000 10000 12000

0

0005

001

0015

002

0025

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus0025

minus001

minus0005

PCA loading (PC2)

(c)





3CN +

H]+




0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08


Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02


(c)







GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology



0 20 400

0

50

50

02040


minus50

minus50


minus40

minus60

minus20

PCq

PCp

(a)

0

0

2000 4000 6000 8000 10000 12000

0

0005

001

0015


Ion pair number

Load

ing

valu

e

minus002

minus0015

minus001

minus0005

(b)

0 2000 4000 6000 8000 10000 12000

0

0005

001

0015

002

0025

Ion pair number

Load

ing

valu

e

minus002

minus0015

minus0025

minus001

minus0005

PCA loading (PC2)

(c)





3CN +

H]+




0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08


Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02


(c)







GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology



0

0

minus20

minus20

minus2

minus15

minus15

minus15

minus15

minus05

minus05

minus1

minus1minus10

minus10

minus5

minus5

minus2

PCq

PCp


(a)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08


Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02

(b)

0 2000 4000 6000 8000 10000 12000

0

02

04

06

08

10

Ion pair number

Load

ing

valu

e

minus10

minus08

minus06

minus04

minus02


(c)







GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


GLX AKGLYS

LACAL

GLY

SER

GLC

HAN

GL

MVL

GLYNLAC

NAMN

cdAMP

XAN GABA

CAR

HSERTHR

LEU

ASP

DANNADHSK

GLN

NAGLU

VAL

THYGBAD

SUCC

ACNLMTHGXL

4HBZ4HLTADHCYS

cAMP


GL3P









1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology




1










1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology




1




cdAMP

























MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology














MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






Acknowledgments


References








































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


MET

PCHO

GL

URA

ASP

OAHSER

LEUH2SO3






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








Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology

Documents

Research Article Dynamic Metabolic Footprinting Reveals ...downloads.hindawi.com/journals/ijg/2014/894296.pdfNational Food Institute, Technical University of Denmark, Mørkhøj Bygade,