Beyond Growth Rate 0.6: What DrivesCorynebacterium glutamicum toHigher Growth Rates in Defined Medium

Simon Unthan,1 Alexander Grunberger,1 Jan van Ooyen,2 Jochem Gatgens,1

Johanna Heinrich,1 Nicole Paczia,1 Wolfgang Wiechert,1 Dietrich Kohlheyer,1

Stephan Noack1

1Institute of Bio- and Geosciences, IBG-1: Biotechnology, Systems Biotechnology,

Forschungszentrum Julich, 52425, Julich, Germany; telephone: þ49-2461-61-6044;

fax: þ49-2461-61-3870; e-mail: s.noack@fz-juelich.de2Systemic Microbiology, Forschungszentrum Julich, Julich, Germany

Abstract: In a former study we showed that Corynebacteriumglutamicum grows much faster in defined CGXII glucosemedium when growth was initiated in highly dilutedenvironments [Grünberger et al. (2013b) Biotechnol Bioeng].Here we studied the batch growth of C. glutamicum in CGXII ata comparable low starting biomass concentration of OD� 0.005 in more detail. During bioreactor cultivations a bi-phasic growth behavior with changing growth rates wasobserved. Initially the culture grew with m ¼ 0:61�0:02 h�1

before the growth rate dropped to m ¼ 0:46�0:02 h�1. Wewere able to confirm the elevated growth rate for C. glutamicumin CGXII and showed for the first time a growth rate beyond0.6 in lab-scale bioreactor cultivations on defined medium.Advanced growth studies combining well-designed bioreac-tor and microfluidic single-cell cultivations (MSCC) withquantitative transcriptomics, metabolomics and integrativein silico analysis revealed protocatechuic acid as a hidden co-substrate for accelerated growthwithin CGXII. The presentedapproach proves the general applicability of MSCC toinvestigate and validate the effect of single mediumcomponents on microorganism growth during cultivationin liquid media, and therefore might be of interest for anykind of basic growth study.

Biotechnol. Bioeng. 2014;111: 359–371.

� 2013 Wiley Periodicals, Inc.

KEYWORDS: growth rate; Corynebacterium glutamicum;microfluidics; single cell growth; protocatechuic acid;biphasic growth

Introduction

Corynebacterium glutamicum is a Gram-positive bacterium,originally isolated from soil (Kinoshita, 1957), and nowadaysfrequently applied as production host for chemicals,materials, and fuels (Becker and Wittmann, 2012). Thisorganism can utilize a variety of different carbon sources andshows monophasic growth for many substrate mixtures,except for combinations of glucose and glutamate or glucoseand ethanol (Arndt et al., 2008; Kramer et al., 1990).C. glutamicum serves as a platform for Metabolic Engineer-

ing and as a model organism for Systems Biology due to itsimportance for industrial biotechnology and its close relationto the host-pathogen Mycobacterium tuberculosis (Nesvera andPatek, 2011; Varela et al., 2012; Wendisch et al., 2006). Forboth areas a well-defined medium composition for biomassgrowth and production is essential, in order to allow forvaluable data generation and, hence, reliable comparison ofstrain properties and performances.One of the earliest publishedmedia for C. glutamicumwith a

clearly defined composition, contained only glucose ascarbon and energy source as well as different sources fornitrogen, phosphate, sulfur, vitamins, mineral salts, and traceelements (Table I, Cremer et al., 1988). Interestingly, thismedium, denoted as CGVIII, also contained high amounts ofcalcium carbonate (CaCO3), which was shown to bebeneficial for cell population growth.In 1989, Liebl et al. found that in the absence of CaCO3,

iron chelating compounds are required for growth ofC. glutamicum and proposed 3,4-dihydroxybenzoic acid (orprotocatechuic acid, PCA) for standard use in defined media(Liebl et al., 1989). Later Keilhauer et al. (1993) substantiallymodified CGVIII by adding the iron chelator PCA andexchanged CaCO3 by very low amounts of an alternativecalcium source (CaCl2, cf. Table I). This medium iscommonly referred to as CGXII minimal medium and,depending on the cultivation system, either used with or

Simon Unthan and Alexander Gr€unberger contributed equally to this work.

Correspondence to: S. Noack

Contract grant sponsor: Bundesministerium f€ur Bildung und Forschung (BMBF)

Contract grant number: 0316017B

Contract grant sponsor: Helmholtz Nanoelectronic Facility (HNF)

Received 20 June 2013; Revision received 7 August 2013; Accepted 19 August 2013

Accepted manuscript online 31 August 2013;

Article first published online 24 September 2013 in Wiley Online Library

(http://onlinelibrary.wiley.com/doi/10.1002/bit.25103/abstract).

DOI 10.1002/bit.25103

ARTICLE

� 2013 Wiley Periodicals, Inc. Biotechnology and Bioengineering, Vol. 111, No. 2, February, 2014 359

without 3-(N-morpholino)propanesulfonic acid (MOPS)as pH regulating buffer.

Since CGXII had been formulated, it became the standardfor Metabolic Engineering and Systems Biology withC. glutamicum and, until now, the publication by Keilhaueret al. has been cited more than 250 times (Google Scholar).In addition, the usage of other PCA containing minimalmedia is even more widespread in the scientific community,since some reports for C. glutamicum refer to otherpublications (cf. Table I, Bäumchen et al., 2007; Kindet al., 2010).

In a previous study we determined the exponentialgrowth rate of C. glutamicum wild type in CGXII mediumat different cultivation scales. At standard initial cell densitiesa growth rate of 0.39–0.45 h�1 was found, independentfrom the cultivation scale and underlying technicaldevices. In contrast, when growth was initiated in highlydiluted environments a greatly elevated growth rate inthe range of 0.60–0.64 h�1 was reproducibly determined(Grünberger et al., 2013b). In further investigations it wasfound that byproducts such as acetate can impair growthalready at very low concentrations. More generally, wesupposed that any condition leading to sufficient by-productdilution enables cell population growth with the “true”maximum growth rate as predefined by the initial mediumcomposition.

In this study, we investigated the relationship betweeninitial cell densities, medium composition, batch growth andmaximum growth rates in more detail.

Materials and Methods

Medium Preparation

Cultivations were performed on the defined medium CGXII,containing per liter of distilled water: 20 g (NH4)2SO4,1 g K2HPO4, 1 g KH2PO4, 5 g urea, 13.25mg CaCl2 · 2H2O,0.25 g MgSO4 · 7H2O, 10mg FeSO4 · 7H2O, 10mg MnSO4 ·H2O, 0.02mg NiCl2 · 6H2O, 0.313mg CuSO4 · 5H2O, 1mgZnSO4 · 7H2O, 0.2mg biotin and 10 g D-glucose. Theconcentration of protocatechuic acid (PCA) was 30mg L�1

in standard medium, but was varied during growth studiesin the range of 0.6mg L�1

–2.5 g L�1. In selected experimentsPCA was replaced by deferoxamine (DFA) with final con-centration of 0.02mM.Medium for cultivation in bioreactorsor microtiter plates was supplemented with 3% (v/v�1)AF204 antifoam agent or 42 g L�1 MOPS buffer, respectively.During medium preparation, some substances were addedsterile after autoclaving (D-glucose, PCA, biotin, traceelements, and AF204) and 4M NaOH was used toadjust pH 7.0. All chemicals used were purchased fromSIGMA Aldrich (Steinheim, Germany).

Bioreactor Cultivations

Bioreactor cultivations of C. glutamicum wild type (ATCC13032) were carried out in 1.5 L reactors (DASGIPAG, Jülich,Germany). Cultivations were performed in batch mode atconstant temperature (30�C) and air flow (1 vvm). Aerobicprocess conditions were controlled via stirrer speed (200–1,200 rpm) to maintain 30% dissolved oxygen concentration(DO). The pH of the culture was regulated to pH 7.0 with4M HCl and 4M NaOH. Online measurements weretaken for pH (405-DPAS-SC-K80/225, Mettler Toledo),DO (Visiferm DO 225, Hamilton) and exhaust gascomposition (GA4, DASGIP AG).

All bioreactor cultivations were inoculated directly fromcryo culture aliquots, which had been prepared from a CGXIIshaking flask culture of exponentially growing C. glutamicum.For preparation the cells were harvested at OD600¼ 10,washed with 0.9% (w/v�1) NaCl and stored at�80�C inNaClsolution containing 20% (v/v�1) glycerol. From these cryoculture aliquots, 0.5mL were added to 1 L bioreactor culti-vations to obtain a comparable low inoculum (OD� 0.005).

Growth of batch cultures at such low initial biomassconcentration was tracked with the Coulter Counter Multi-sizer 3 (Beckmann Coulter GmbH, Krefeld, Germany)equipped with a 30mM capillary. Culture samples werediluted with CasyTon (Roche Diagnostics GmbH, Man-nheim, Germany) if necessary, and measured for cellconcentration as well as distribution of cell volumes. Dataof total biovolume (BV) was moreover used to determinespecific uptake rates in the cultures. These rates were related

Table I. Selected media compositions for cultivation of C. glutamicum

under defined conditions.

Cremeret al. (1988),

CGVIII

Keilhaueret al. (1993),

CGXII

Bäumchenet al.(2007)

Kindet al.(2010)

D-Glucose 40 g 40 g 10 g 10 g(NH4)2SO4 20 g 20 g 10 g 15 gUrea 5 gKH2PO4 0.5 g 1 g 1 g 7.7 gK2HPO4 0.5 g 1 g 2 g 24.98 gMgSO4 · 7H2O 0.25 g 0.25 g 0.25 g 0.2 gMOPS 42 gCaCl2 10mg 10mgFeSO4 · 7H2O 10mg 10mg 10mg 20mgMnSO4 ·H2O 10mg 10mg 10mg 2mgZnSO4 · 7H2O 1mg 1mg 1mg 0.5mgCuSO4 0.2mg 0.2mg 0.2mgNiCl2 · 6H2O 0.02mg 0.2mgBiotin 0.2mg 0.2mg 0.2mg 0.5mgProtocatechuic

acid0.03mg

(later correctedto 30mg)

30mg 30mg

CaCO3 20 gNaCl 1 gNa2B4O7 · 10H2O 0.2mg(NH4)6Mo7O24 · 4H2O 0.1mgThiamine-HCl 1mgCuCl2 · 2H2O 0.2mgCaCl2 · H2O 55mgFeCl3 · H2O 2mg

All information refers to 1 L.

360 Biotechnology and Bioengineering, Vol. 111, No. 2, February, 2014

to cell dry weight (CDW) using the linear regression functionCDW[g]¼ 0.474 · BV[mL].

Microtiter Plate Cultivations

Cultivations in microtiter plates were performed using theBioLector device (m2p-labs, Baesweiler, Germany). TheBioLector allows the time-dense tracking of culture param-eters (pH, DO) and biomass concentrations (backscatter)of 48 parallel cultivations in one FlowerPlate1 (m2p-labs,Baesweiler, Germany). The wells of these FlowerPlates arebaffled to enable high oxygen transfer rates, comparable tobioreactor scale.For cultivations, 1mL inoculated culture was filled in each

well of a Flowerplates, which was closed with a gas permeablesterile foil. The BioLector was operated at 1,000 rpm, 30�Cand 95% humidity. Backscatter signals were measured atgain 20 and later normalized by the blank value of each wellseparately.

Microfluidic Single Cell Cultivations

The in-house developed polydimethylsiloxane microfluidicsystem enables the spatio-temporal analysis of growingmicrocolonies with single cell resolution by automatedhigh resolution time-lapse microscopy. Due to continuousmedia perfusion environmental conditions are maintainedconstant, ideal for perturbation studies. Each chip devicecontains 400 microfluidic cultivation chambers of 1mm� 60mm� 60mm (height� length�width) connected tomedia supply channels. Mass transport inside the cultivationchambers is based on diffusion only. For chip fabricationdetails and further information the reader is referred to thepublications by Grünberger et al. (2012, 2013a).The microfluidic chip was mounted onto a motorized

inverted microscope (Nikon Eclipse Ti, Nikon microscopy,Germany) equipped with an incubator for temperature andatmosphere control. The cell suspension with an OD600

between 0.5 and 1, transferred from the pre-culture atexponential growth phase, was infused into the chip toinoculate the microfluidic cultivation chambers with singlecells. Growth medium was infused at approx. 100 nLmin�1

after cell inoculation. CGXII medium was prepared asdescribed before and additionally sterile filtered to preventparticle agglomeration during microfluidic cultivation.Supernatant samples of the lab-scale bioreactor cultivation

were instantly filtered (0.22mmpore size) to remove cells andprevent degradation, and were stored at �20�C for micro-fluidic experiments.Time-lapse phase contrast microscopy images of the

growing microcolonies were recorded every 10minutes over24 h of microfluidic cultivation. Afterwards, cell numbers,cell length and growth rates were derived by image analysisutilizing the Nikon NIS Elements AR software package.

GC-ToF-MS and LC-MS/MS Analysis

Untargeted metabolome screening in culture supernatantswas performed via an Agilent 6890N gas chromatograph

coupled to aWaters Micromass GCT Premier high resolutiontime of flight mass spectrometer. For details regarding samplepreparation, MS operation and peak identification the readeris referred to (Paczia et al., 2012).PCA concentrations in culture supernatants were mea-

sured by HPLC (X-LC 3000 Series, JASCO GmbH, Gross-Umstadt, Germany) coupled to a mass spectrometer (API4000, ABSciex, Foster City, CA) equipped with a TurboIonspray source. For the analysis a C18 column (Synergy Hydro,Phenomenex, Aschaffenburg, Germany) was used witheluent A (10mM tributylamine aqueous solution adjustedpH to 4.95 with 15mM acetic acid) and eluent B (methanol)at a temperature of 40�C. The elution was isocratic with a A/Bratio of 70:30, a flow rate of 0.45mLmin�1 and an injectionvolume of 10mL. For more details regarding MS operationsee (Paczia et al., 2012).

Transcriptome Analysis

For transcriptome analysis, cells were cultivated in fourbioreactors in standard CGXII medium as described before.Each culture was harvested either during first or secondgrowth phase at OD¼ 0.2 or OD¼ 2, respectively. Globalgene expression was analyzed independently followingtwo biological replicates for each condition. RNA and thesynthesis of fluorescently labeled cDNA were prepared asdescribed before (Moker et al., 2004).The custom-made DNA microarrays were obtained

from Agilent Technologies (Waldbronn, Germany). Agilent’seArray platform was used to assemble custom 4� 44K60mer microarray designs (https://earray.chem.agilent.com/earray/). The customdesign included oligonucleotides for theannotated protein-coding genes and structural RNA genesof the bacterial genomes from C. glutamicum, Escherichia coli,Gluconobacter oxydans, and Bacillus subtilis for genome-widegene expression analysis. For C. glutamicum, the genomeannotation by Kalinowski et al. (2003) (accession numberNC_006958) was used, listing 3,057 protein coding genes and80 structural tRNA and ribosomal RNA genes. In the customdesign the C. glutamicum genes are represented by one, two, orthree oligonucleotides, which were employed to determinethe relative RNA level. The custom array design also includedthe Agilent’s control spots. Purified cDNA samples to becompared were pooled. The prepared two-color sampleswere hybridized on 4� 44K arrays at 65�C for 17 h usingAgilent’s gene expression hybridization kit, Agilent’s hybrid-ization chamber, and Agilent’s hybridization oven. Afterhybridization, the arrays were washed using Agilent’s washbuffer kit according to the manufacturer’s instructions.Fluorescence of hybridized DNA microarrays was deter-mined at 532 nm (Cy3-dUTP) and 635 nm (Cy5-dUTP) at5mm resolution with a GenePix 4000B laser scanner andGenePix Pro 6.0 software (Molecular Devices, Sunnyvale,CA). Fluorescence images were saved to raw data files in TIFFformat (GenePix Pro 6.0).Quantitative TIFF image analysis was carried out using

GenePix image analysis software and the Agilent’s gene array

Unthan et al.: Biphasic Growth of Corynebacterium glutamicum 361

Biotechnology and Bioengineering

list (GAL) file. The results were saved as GPR-file (GenePixPro 6.0). For background correction of spot intensities,ratio calculation and normalization, as well as generation ofdiagnostic plots for array quality control, GPR-files wereprocessed using the BioConductor R-packages limma andmarray (www.bioconductor.org). For further analysis, theprocessed and loess-normalized data as well as detailedexperimental information according to MIAME (Brazmaet al., 2001) were stored in the in-house DNA microarraydatabase (Polen and Wendisch, 2004). To search the datafor differentially expressed genes by the processed Cy5/Cy3ratio reflecting the relative RNA level, the criteria flags �2(GenePix Pro 6.0) and signal/noise�5 for Cy5 (F635Median/B635Median, GenePix Pro 6.0) or Cy3 (F532Median/B532Median, GenePix Pro 6.0) were used. If the signal/noise of Cy5 and of Cy3 were <3, then signals wereconsidered as too weak to analyze the Cy5/Cy3 ratio of a gene.Furthermore, P values were calculated by a paired Student’st test comparing the relative RNA levels of a gene in thereplicates to the relative RNA levels of all other genes in thereplicates.

Results and Discussion

Bi-Phasic Growth of C. glutamicum in CGXII Medium

The batch growth of C. glutamicum was studied undercontrolled bioreactor conditions in standard CGXII mediumand low initial cell densities. Each cultivation was inoculatedat a comparable low cell concentration of N(t¼ 0)¼ 1.3� 106� 2.1� 105 cell mL�1, which corresponds to an opticaldensity (OD) of OD� 0.005 and a cell dry weight (CDW) ofCDW� 0.002 g L�1. The cultivations were performed in 10biological replicates, in order to allow for a reliable estimationof specific growth rates. Noteworthy, the biomass growth wastracked by a sensitive Coulter Counter method, since thelimit of quantification (LOQ) for the commonly applied ODand CDW measurements are much higher (LOQOD� 0.5,LOQCWD� 1 g L�1, results from preliminary experiments).

As shown in Figure 1A, the batch cultivation follows twogrowth phases with a metabolic switch at t� 10 h and acorresponding OD� 0.5. For both phases, the exponentialgrowth rates were determined. For the first phase theexponential growth rate is estimated as m ¼ 0:61�0:02 h�1,which matches the higher growth rate of C. glutamicum whencultivated in the microfluidic perfusion system (Grünbergeret al., 2013b). In contrast, the exponential growth rate duringthe second phase with m ¼ 0:46�0:02 h�1 is much lower,and close to the growth rate estimate of 0.42� 0.03 h�1

presented in our former study. Noteworthy, in formerexperiments the cells were inoculated at OD¼ 1 andtherefore cultivated at higher densities during exponentialgrowth (see supplementary information in Grünberger et al.(2013b)). In the actual experiments such cell densities arereached after 14 h of cultivation (cf. Fig. 1A).

The growth curves were further analyzed to exclude anymisinterpretation in the distinction between the two

observed growth phases. Firstly, the differential growth ratesalong the cultivationwere determined to evaluate the divisionof the growth curve into two phases (cf. Fig. 1A, bottom). Aclear drop of differential growth rates at 10 h cultivation timesupported the time-dependent metabolic switch. Secondly,we checked if the observed changes in growth rates whichare based on cell number measurements, were distorted byunderlying dynamic changes in cell sizes. Therefore, theoverall cell volume (biovolume) and the mean single cellvolume were followed along all cultivations (see Supplemen-tal Fig. S1). As expected, the biphasic growth behaviorwas again reproduced from the biovolume measurements.Moreover, after inoculation the cells first grew in size (lag-phase with respect to cell division) and then, eventually afterreaching a maximum cell size, continuously became smallerwith decreasing substrate availability. This observation isconsistent with previous results, where C. glutamicum wascultivated in a long-term batch following prolonged carbonlimitation and decreasing cell sizes (Paczia et al., 2012).

The observed metabolic switch in CGXII batch culturesoccurring at a certain threshold cell density that leads todecreasing growth rates, can result from the accumulation ofinhibiting by-products or the limitation of specific mediacomponents. However, to discriminate between both effects,the instantaneous growth phenotype in response to thechanging media composition has to bemeasured and all side-effects resulting from the cultivation history must be safelyexcluded.

Here microfluidic single-cell cultivation (MSCC) is apowerful technique that can be used to access growthphenotypes (including division rates and morphology),which are predominantly resulting from the mediumcomponents in the inflow. The continuous dilution withfresh medium prevents any by-product accumulation in thecultivation chamber and therefore excludes any potentialgrowth impairing effect of such compounds (Grünbergeret al., 2013b). Moreover, MSCC allows direct onlinemonitoring of growing cell populations and therefore higheraccuracy in growth rate estimation.

Biosensing of Culture Supernatants With MSCC

We applied MSCC to determine the growth phenotype ofC. glutamicum in cell free supernatant samples from bioreactorcultivations (Fig. 1B). The supernatant from the beginningof the bioreactor cultivation, which nearly equals theoriginal CGXII medium, resulted in growth rates ofm ¼ 0:65�0:04 h�1, essentially reproducing our formerresults. Within the subsequent supernatant samples, thegrowth rate first decreased until a minimum of m ¼0:46�0:03 h�1 was reached and then increased again tosignificantly higher values. Noteworthy, also the mean cellsize of C. glutamicum follows the same pattern, pointing to alink between medium composition, growth rate and cell size(see Supplemental Fig. S2). The bi-phasic growth character-istics, observed in the bioreactor cultivation, could thereforebe reproduced by MSCC.

362 Biotechnology and Bioengineering, Vol. 111, No. 2, February, 2014

The re-switch to higher growth rates found during MSCCon bioreactor supernatants from late exponential phase couldbe explained by differences in the growth environment andtime-scale between both cultivation systems. In the bioreac-tor the cells grow in batchmode on glucose excess and, hence,the cultivation broth is likely to be enriched with significantamounts of intracellular metabolites (Paczia et al., 2012).During this process the extracellular environment isdynamically changing, which most likely forces the cells tocontinuously adapt their metabolism. This energy demand-ing adaption will directly affect growth rates which are,moreover, the estimated averages from a heterogeneous cellpopulation in different stages of adaption. In contrast, duringMSCC only a few cells are cultivated and the growthmedium,here in form of a culture supernatant, is continuously flushed

to provide a constant extracellular environment. As a result,the cells can adapt for multiple hours to the uniquecomposition of each particular supernatant sample fromthe batch reactor. Hence, even small amounts of co-metabolites might significantly improve cellular growth,after a time-dependent adaption to the complex compositioncould take place. In conclusion, a MSCC approach shows themaximal possible growth rate in a given medium composi-tion and might thereby be significantly higher than found inthe dynamic environment in a batch culture.The observation of a growth rate minimum around the

metabolic switch following MSCC rather excludes anaccumulation of inhibiting by-products like, for example,acetate as reason for lower growth rates during bioreactorbatch cultivation. In such a case one would not expect

I II

A B

µmax = 0.62 h-1

I II

^

^

10x BR

3x MSCC

BR supernatants

µII 0.46 h-1± 0.02

µI 0.61 ± 0.02 h-1

OD 1

^

^

1000

Figure 1. Batch cultivation of C. glutamicum in standard CGXII glucose medium in 1 L bioreactors at low initial cell concentrations reveals a bi-phasic growth behavior. A: Two

distinct growth phases were identified in ten biological replicates, of which the first was characterized by a higher growth rate of m ¼ 0:61�0:02 h�1. For comparison the graph

depicts the initial biomass density (OD� 1) applied in our former bioreactor experiments (Gr€unberger et al., 2013b). B: Filtrated culture supernatants of distinct time-points from the

bioreactor were used as growth medium during MSCC. BR, bioreactor; MSCC, microfluidic single-cell cultivation.

Unthan et al.: Biphasic Growth of Corynebacterium glutamicum 363

Biotechnology and Bioengineering

increasing growth rates on supernatant samples taken fromthe second growth phase. Moreover, in our former studyacetate was already tested as potentially inhibiting by-productand provoked decreased growth rates under MSCC con-ditions when supplemented next to glucose (Grünbergeret al., 2013b).

Consequently, the MSCC data strongly points to alimitation of a media component as the key factor forthe bi-phasic growth behavior. To further test this hypo-thesis, we analyzed the time dependent changes in thecomposition of culture supernatants during bioreactor batchcultivation.

Potential Growth Effectors in Culture Supernatants

In a first approach we applied an untargeted metabolomescreening via GC-ToF-MS, focusing on supernatant samplesfrom the bioreactor cultivation started at low initial celldensity. It was observed that the medium components ureaand PCA decreased extracellular, and PCA completelyvanished from the medium during the first growth phase(Supplementary Fig. S3). While the absence of urea in theCGXII medium had no effect on growth (data not shown), itwas already shown that PCA or a comparable iron chelatorlike, for example, catechol or citrate, is necessary to initiatecell division in C. glutamicum (Liebl et al., 1989; Osten et al.,1989).

Hence, in the following we concentrated on PCA andperformed quantitative, targeted LC-MS/MS analysis of thesame supernatant samples (Fig. 2A). As a result, PCA wascompletely consumed by the cells within 10 h of cultivationand the time point of PCA depletion coincided with thesubsequent start of the second growth phase. In accordanceto this observation, recent reports have broadenedthe potential function of PCA from a sole iron chelatorto an alternative substrate, as C. glutamicum was able to growon PCA as single carbon source (Haussmann and Poetsch,2012; Merkens et al., 2005; Shen et al., 2012).

Differential Transcriptome Analysis

With PCA as potential target compound at hand, wecontinued with a differential transcriptome analysis toquantify gene expression changes, accompanied with thegrowth rate reduction in the second growth phase.

For transcriptome analysis, two bioreactor samples weretaken from each growth phase, that is, 2 h before and 1.5 hafter the drop in the growth rates. The resulting mean ratiosof selected mRNA levels are shown in Figure 2B. A completelist of significantly up or down-regulated genes can be foundin Table S1 of the Supplementary Information.

As a main result, a significant up-regulation of nearly allgenes encoding for the b-ketoadipate pathway was deter-mined during the first growth phase (Fig. 2C and D). Withinthis pathway, PCA is taken up by the 4-hydroxybenzoatetransporter pcaK (Shen et al., 2012) and then degraded withinsix enzymatic steps to the tricarboxylic acid (TCA) cycle

intermediates acetyl-coenzyme-A (AcCoA) and succinyl-CoA (SuCoA) (Brinkrolf et al., 2006).

Interestingly, several genes (iolB, iolC, gntP, pobA, catA1)encoding for catabolic degradation of substances typicallyfound in soil (inositol, gluconate, benzoate, catechol) werealso found to be greatly up-regulated in the first growth phase(cf. Supplementary Table S1). This finding might reflect amore global transcriptomic response of C. glutamicum to thepresence of substrates occurring in its natural habitat.

In addition, a twofold up-regulation was observed forthe genes nifS2 (cg1761), sufC (cg1762), sufB (cg1764) and aputative transcriptional regulator (cg1765) which build theoperon of essential [Fe–S]-cluster biosynthesis (Fontecaveand Ollagnier-de-Choudens, 2008). Noteworthy, different[Fe–S]-cluster containing enzymes are linked to PCAdegradation, its further downstream catabolism and, moregeneral, are required for fast growth. This holds true forprotocatechuate dioxygenase (PcaGH in Fig. 2D), the TCA-cycle enzymes aconitase, succinate-dehydrogenase as well asthe cytochrome bc1 complex, involved in respiratory energymetabolism (Eggeling and Bott, 2005).

In contrast, during the second growth phase no distincttranscriptomic response with regard to the absence of PCAwas found and especially the expression of genes related tothe iron starvation response in C. glutamicum (Wennerholdand Bott, 2006) was unaltered.

Growth Response to Varying PCA Concentration

The results from metabolome and transcriptome analysisstrongly indicated that PCA is actively catabolized byC. glutamicum in standard CGXII medium, what can onlybe observed at comparable low cell densities. In the following,the effect of PCA availability on the growth of C. glutamicumwas investigated in more detail by a specially designed seriesof batch and MSCC experiments.

C. glutamicumwas grown in two batch cultivations in CGXIImedium with and without PCA inoculated at a standardcell density of OD� 0.5 (Fig. 3). Both cultures showedgrowth demonstrating that C. glutamicum can initiate celldivision also in the absence of PCA, when inoculated atsufficient cell densities. However, the PCA containing cultureimmediately grew exponentially, while the PCA-freeculture showed a delayed growth phase in the beginning.This observation could either point to a growth enhancingeffect of PCA during early culture stages or simply indicatethat C. glutamicum needs to adapt to PCA-free medium (lag-phase). A potential adaptionwould, however, not be expectedwhen the cells were grown in a PCA-free pre-culture.

Therefore, a series of six repetitive batch cycles wasperformed in which cells from the previous PCA-free culturewere used to inoculate subsequent cultivations with andwithout PCA (cf. Fig. 3). As a result, all PCA containingcultures immediately grew exponentially, while the PCA-freecultures showed a conserved delayed growth behavior.Hence, it can be concluded that the amount of PCA that isinitially available per single cell directly influences early cell

364 Biotechnology and Bioengineering, Vol. 111, No. 2, February, 2014

1000

C

µI=0.59 h-1^

I II

µII=0.46 h-1^

PobA

PCA

PcaGPcaH

PcaB PcaC PcaD PcaIPcaJ

+

PcaF4-HBA

ACCOASUCCOA

PcaKextracellular

D

2.9 1.9 0.5 1.10.33.1 2.01.6

0.92.2

M-value 0.50.5

pcaHpcaG

catA1

pcaBpcaC pobA

nifS2 sufBsufC

cg1765

cg3321

pcaIpcaJ

PCA

pO2

4000 6000 8000 10000 12000

pcaH(cg2631)

pcaG(cg2630)

pcaB(cg2629)

pcaC(cg2628)

pcaO(cg2627)

pcaD(cg2626)

pcaF(cg2625)

pcaR(cg2624)

pcaI(cg2623)

pcaJ(cg2622)

pobA(cg1226)

pcaK(cg1225)

BA

LC-MS/MS 4x DNA-chip

Figure 2. In-depth analysis of bi-phasic growth behavior of C. glutamicum when cultivated in CGXII glucose medium. A: LC-MS/MS analytics revealed a decreasing

concentration of the iron chelator PCA in the supernatant. After PCA completely vanished (t� 9 h), the switch from first to second growth phase was observed. B,C: In a differential

transcriptomics study it was shown, that multiple genes encoding for the b-ketoadipate pathway were higher expressed in the first growth phase. D: Illustration of the PCA

catabolism in C. glutamicum, resulting in the TCA intermediates succinyl-CoA and acetyl-CoA.

Unthan et al.: Biphasic Growth of Corynebacterium glutamicum 365

Biotechnology and Bioengineering

population growth. In the later growth phase all repetitivebatch cultures showed the standard exponential growthrate (m ¼ 0:46�0:01 h�1) independently from initial PCAavailability.

Next, we studied the effect of PCA surplus whenC. glutamicum is grown in CGXII medium in batch cultureswith glucose inoculated at OD� 0.5 (Fig. 4A). In a culturewith 16.3mM PCA (i.e., 83 times the standard concentra-tion) the elevated growth rate of m ¼ 0:61 h�1 was nowmaintained up to higher cell densities of OD� 3. Inaccordance with the faster growth during the first phase,the dissolved oxygen (DO) concentration decreased muchfaster until the PCAwas fully consumed. After complete PCAconsumption a step-increase in the DO-signal was observed,which clearly indicated a metabolic switch to the PCA-freephase and standard exponential growth on glucose.

To further evaluate the role of PCA as an additionalC-source the final biomass density (backscatter signal) wasanalyzed with respect to carbon availability in the usedgrowth media (Fig. 4B). While glucose concentration waskept constant, the total carbon fraction was altered from PCAvariation in the different medium compositions. The final

biomass densities in those growth media showed a pro-portional increase to elevated PCA concentrations, support-ing once more the role of PCA for biomass generation.

Finally, microfluidic experiments confirmed thatC. glutamicum can grow on PCA as single carbon sourceand, most interestingly, the resulting maximum growth ratesare directly depending on the amount of PCA supplied via thecontinuous feeding rate (Fig. 5A).

Moreover, a strong correlation between PCA supply,specific growth rate and cell size was found for growth inCGXII medium with and without glucose throughout allperformed experiments (Fig. 5B). This result is in agreementwith the cell size pattern observed during the biosensingexperiments (cf. Supplementaty Fig. S2), and generallystrengthens the hypothesis of a direct linkage betweennutrient availability, cell size and growth rate in micro-organisms (Volkmer and Heinemann, 2011; Yao et al., 2012).

In Silico Analysis of PCA Catabolism

To specify the carbon source function of PCA, we performedflux variability analysis (FVA) using a genome scale model of

CGXII+ PCA

CGXII- PCA

Figure 3. Growth of C. glutamicum in microtiter plates containing CGXII medium with and without PCA. Cells from the PCA free culture were used to inoculate subsequent

cultivations in both medium variants. In PCA free medium a delayed growth pattern was observed. However, later growth rates were comparable to cultures in PCA containing

medium. In six following repetitive batch cycles the growth pattern was found unchanged and thereby, no cellular adaption to PCA free medium was found.

366 Biotechnology and Bioengineering, Vol. 111, No. 2, February, 2014

C. glutamicum (Woo et al., 2010). In short, FVA is applied tosimulate the minimum and maximum flux for each reactionin the network while maintaining the maximal possiblegrowth rate under the predefined network constraints. For

more details regarding the FVA method the reader is referredto Schellenberger et al. (2011).The specific uptake rates for glucose ðuptGLC ¼ 4:42�

0:54mmol gCDW�1 h�1Þ and PCA ðuptPCA ¼ 2:04�0:20mmol

A

µI 0.65 h-1

µII 0.45 h-1

I II42x

µI 0.61 h-1=

=

µII 0.41 h-1

I II83x

µI 0.61 h-1

µII 0.43 h-1

I II21x

µII 0.41 h-1=

std

Amount of PCA

195 µM

16.3 mM

8.2 mM

std

83x

42x

21x 4.1 mM

B

substrate supply [C mmol]

maximum backscatter [-]0 100 200 300

std

21x

42x

83x

0.2 0.4 0.6

PCAglucosebackscatter

==

=

=

Figure 4. Influence of PCA surplus on growth of C. glutamicum during batch cultivation in CGXII medium with 10 g L�1 glucose. A: Microtiter plates were inoculated at OD� 0.5

and the standard (std) PCA concentration of 195mM was increased step-wise up to a final concentration of 16.3 mM (83�). Growth was monitored online via backscatter and

dissolved oxygen (pO2) measurements. At higher PCA concentrations the phase of elevated growth rates could be prolonged and thereby established until higher biomass

concentrations of OD� 3. B: Correlation between maximum backscatter from batch cultivation and total carbon supply in the medium, which was changed by increasing PCA

concentrations.

Unthan et al.: Biphasic Growth of Corynebacterium glutamicum 367

Biotechnology and Bioengineering

gCDW�1 h�1Þwere calculated fromone batch experiment using

the standard CGXII medium composition (cf. Fig. 2A) andtaken as network constraints for FVA.

Figure 6 depicts one possible intracellular flux distributionwithin the central carbon metabolism of C. glutamicum thatresults in amaximum growth rate prediction ofm ¼ 0:55 h�1

and thus resembles the situation during the first growth phase(cf. Fig. 1A and Supplementary Fig. S1). In contrast, whenonly glucose is supplied as carbon and energy source, themaximum growth rate is predicted as m ¼ 0:39 h�1, which isin the range of the experimentally determined value duringthe second growth phase.

In both cases the absolute value of the simulatedgrowth rate is lower than the measured one, whichmight be due to the underlying network parameterizationincluding precursor and ATP demand for biomasssynthesis. As the accuracy of absolute rate estimation using

a pure stoichiometric network approach is limited, wefocused on the interpretation of the relative changes in thefollowing.

The predicted increase in growth rates from additionalconsumption of PCA (Dm� 0.15 h�1) is in good correlationto the increase in growth rates found experimentally. Forsome selected reactions the relative flux ranges allowing forthe two optimum solutions are represented. As expected, themetabolization of PCA leads to a significantly higher carbonflux along the TCA-cycle, providing not only a surplus ofthe biomass precursor a-ketoglutarate and oxaloacetate, butalso of energy and reduction equivalents. With regard to theNADPH generation, the flux along the pentose phosphatepathway tends to be down-regulated in response to the higherTCA-cycle flux and a potential activity of the malic enzyme.

Clearly, the latter finding needs further confirmationby trying more elaborate flux estimations via 13C-metabolic

B

std

10x ~1.77 mµ

~1.54 mµ- GLC

25x

std ~2.40 mµ

1/5 ~2.14 mµ

~3.01 mµ

+ GLC

- GLC

1/4 10xstdPCA concentration

0

A

Amount of PCA

39 µM

4.9 mM

1.95 mM

1/5

25x

10x

std 195 µM

1/4 49 µM

cell

fract

ion

[- ]

cell

fract

ion

[-]

Figure 5. Growth of C. glutamicum on PCA as sole carbon source and its impact on cell morphology. A: Growth in CGXII without glucose was observed during microfluidic

cultivation when PCA was added and the observed growth rates were a direct function of PCA concentration. B: Correlation between PCA supply and cell size of C. glutamicum

during MSCC in CGXII with and without glucose at varying PCA concentrations.

368 Biotechnology and Bioengineering, Vol. 111, No. 2, February, 2014

flux analysis (13C-MFA, van Ooyen et al., 2012). Theanaplerosis of C. glutamicum comprises 5 reactions thatpotentially operate in parallel (Eggeling and Bott, 2005),and thereby cannot be resolved in a quantitative manner

even with 13C-MFA (Noack et al., 2011). This is againreflected in the broad flux range estimated for allanaplerotic reactions when applying the FVA approach(cf. Fig. 6).

ACN

ICD

ODHC

SCS

SQO

FUMA

MQO

CS

PK

PDHC

MS ICL

MDH

PEPCk

PTS ATPaseO2tCO2t

BM

CYTBCA

PcaGH

PcaB

PcaC

ME

c6

CIT

ICIT

AKG

AKG

SUCCOA

FUM

MAL

OAA

OAA

SUC

ACCOAACCOA

PYR

PYR

PYR

GLC[e]

PEP

PEP

PEP

G6P

G6P

GLX

ADP

ADP

ADP

ADP

ADP

ADP

ADP

ADP

ATP

ATP

ATP

ATP

ATP

ATP

ATP

ATP

NAD

NAD

NAD

NAD

NAD

NADH

NADH

NADH

NADH

NADH

NADP

NADP

NADP

NADP

NADPH

NADPH

NADPH

NADPH

MQ

MQ

MQ

MQH2

MQH2

MQH2

F6P

G3P

3PG

P5P

E4PCO2

CO2

CO2

CO2

CO2

CO2

CO2

CO2

CO2CO2

COA

COA

COA

COA

COA

COA

COA

H

H[e]

HCO3

O2

O2[e]CO2[e]

NH3

PCA[e]

PCA

OAPCOA OAP

OAPEL

CODFA

CCMFBP

PCx

PEPCx

c3

c5

PCAt

PcaD

PcaIJ

PcaF

[15, 21][3, 12]

[166, 168]

[89, 91][144, 147]

[113, 121][122, 137]

[36, 63][48, 59]

[0, 25][0, 4]

[288, 289]

[542, 544]

[80, 466][162, 620]

[0, 7][0, 3]

[-2, 1][-3, 8]

[0, 3][5, 20]

[0, 25][0, 4]

[78, 80][175, 177]

[70, 79][126, 130]

100 0[489, 490] 100 46

[791, 794]

[159, 162]

0.390.55

Figure 6. Coarse-grained simulation of metabolic fluxes in C. glutamicum during co-metabolism of PCA and glucose. The values represent the minimum and maximum fluxes

resulting from flux variability analysis (FVA). For FVA the measured uptake rates for both substrates were fixed during the simulation and biomass growth was applied as optimization

criteria. The simulation was repeated for growth on sole glucose (assuming no PCA uptake) and the corresponding flux ranges are represented in the lower boxes.

Unthan et al.: Biphasic Growth of Corynebacterium glutamicum 369

Biotechnology and Bioengineering

Role of PCA as Iron Chelator

The in silico results can only support the carbon function ofPCA but not its role for iron uptake, since until now thecomplex machinery of iron utilization in bacteria is not fullyunderstood and hence also not covered by the C. glutamicummodel.

To further clarify the impact of iron chelating agentslike PCA we tested the alternative chelator deferoxamine(DFA) performing MSCC experiments. It was found thatC. glutamicum can grow on glucose CGXII mediumsupplemented with DFA at a growth rate of m ¼0:57�0:03 h�1 (Supplementary Fig. S4). Surprisingly, repro-ducible growth was also observed on DFA as single carbonand energy source (m ¼ 0:09�0:01 h�1). DFA is a quitecomplex molecule and to the best of our knowledge noinformation on a potential degradation mechanism for thiscompound in any microorganism exists so far. In summarythese results showed that it is very difficult to clearlydifferentiate between the function as carbon source and ironchelator of a substance, since its uptake is likely to beaccompanied by its catabolization.

With regard to this experimental restriction, we evaluatedthe potential iron chelating function of PCA based on ourformer results on PCA uptake and the iron demand forC. glutamicum biomass formation. Liebl et al. (1989) measuredthe total content of iron in C. glutamicum cells grown in CGXIImedium in a range of 0:3� 0:5mgFe g

�1CDW. In our bioreactor

experiments the amount of biomass produced until PCAdepletion was estimated as DCDW� 0.2 g L�1 (compareFig. 2A). By assuming that each mole of PCA (from an initialconcentration of 195mM) transports one mole of iron intothe cell, the total iron content at the end of the first growthphase would add up to� 55mgFe g

�1CDW. As this value is more

than two orders of magnitude higher compared to the irondemand, we argue that the predominant function of PCA isits direct utilization as carbon and energy source.

Moreover, we showed experimentally that C. glutamicumgrows at rates around m� 0.45 h�1 in PCA depleted culturesas well as in PCA-freemedium (cf. Figs. 2A and 3). Therefore,one might reconsider previous assumptions about the roleof PCA as iron chelator and, more generally speaking, ifcultivation media for C. glutamicum must be supplementedwith iron chelating agents.

Conclusions

In general, biphasic growth of microorganisms is notexpected when cultivated in a defined minimal medium.For cultivation of C. glutamicum, CGXII is the commonlyused defined medium, which is assumed to contain onlyone growth limiting carbon source namely D-glucose.

Our results prove that protocatechuic acid is co-utilizednext to glucose and provokes rapid cell division during theinitial phase of cell population growth. Moreover, weconclude that the channeling of carbon from PCA into theTCA-cycle is the predominant cause for elevated growth rates

of C. glutamicum in CGXII medium. Consequently, one has tobe careful in speaking of “glucose as the sole carbon andenergy source” in CGXII.

In a batch cultivation started at standard inoculum(OD� 0.5) the initial amount of PCA is consumed veryfast and thus no significant impact on the results of aphenotypic characterization is expected. However, theresulting growth can be significantly influenced by the co-utilization of both substrates when C. glutamicum is cultivatedat low cell density or grown under conditions of continuousmedia supply (e.g., fed-batch and chemostat cultures).

As another result we found that C. glutamicum grew in sizewith rising PCA concentrations. Changes in cell size andmorphology will result in different specific cell volumesand must thereby be kept in mind when aiming for thequantification of intracellular metabolite concentrations(Tillack et al., 2012).

From our data we cannot rule out a combined effect on theobserved growth rates of PCA catabolism and iron transportinto the cell. Nevertheless, when taking the cellular demandof iron and the comparably high PCA uptake rate intoaccount, the effect of iron transport should be negligible.

As a variety of other benzoic acids and phenols arecatabolized in theb-ketoadipate pathway, we see the potentialof new substrate mixtures to improve total biomassproductivity. However, these co-substrates need to bepurchased at reasonable costs, as industrial products ofC. glutamicum are often bulk chemicals that achieve relativelylow prices. Here, lignin hydrolyzates from lignocelluloticbiomass might be a promising candidate, which containmultiple aromatic compounds like vanillic, p-coumaric aswell as protocatechuic acid (Martin et al., 2002).

Combining bioreactor and microfluidic single-cell culti-vation experiments, we could rapidly narrow the key factorfor bi-phasic growth down to the culture supernatant. Thisshows the potential of the MSCC approach to validatemedium compositions, as the fluid is exchanged at a highdilution rate and therefore the cells are analyzed in a quasi-stationary state without significant changes in their environ-ment (accumulation or consumption of components). Viceversa, when performing such microfluidic experimentswhere the overall growth strongly depends on the startingmedium composition, one carefully has to prepare themedium to prevent the loss of any essential nutrient, forexample by precipitation.

This work was partly funded by the Bundesministerium für Bildungund Forschung (BMBF, grant. no. 0316017B). This work was partlyperformed at Helmholtz Nanoelectronic Facility (HNF) of For-schungszentrum Jülich GmbH. The authors would like to thank fortheir help and support.

References

Arndt A, Auchter M, Ishige T, Wendisch VF, Eikmanns BJ. 2008. Ethanolcatabolism in Corynebacterium glutamicum. J Mol Microbiol Biotechnol15(4):222–233.

370 Biotechnology and Bioengineering, Vol. 111, No. 2, February, 2014

Bäumchen C, Knoll A, Husemann B, Seletzky J, Maier B, Dietrich C,Amoabediny G, Büchs J. 2007. Effect of elevated dissolved carbondioxide concentrations on growth of Corynebacterium glutamicum onD-glucose and L-lactate. J Biotechnol 128(4):868–874.

Becker J, Wittmann C. 2012. Bio-based production of chemicals, materialsand fuels—Corynebacterium glutamicum as versatile cell factory. Curr OpinBiotechnol 23(4):631–640.

Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C,Aach J, Ansorge W, Ball CA, Causton HC, Gaasterland T, Glenisson P,Holstege FC, Kim IF,Markowitz V,Matese JC, ParkinsonH, Robinson A,Sarkans U, Schulze-Kremer S, Stewart J, Taylor R, Vilo J, Vingron M.2001.Minimum information about amicroarray experiment (MIAME)-toward standards for microarray data. Nat Genet 29(4):365–371.

Brinkrolf K, Brune I, Tauch A. 2006. Transcriptional regulation of catabolicpathways for aromatic compounds in Corynebacterium glutamicum. GenetMol Res 5(4):773–789.

Cremer J, Treptow C, Eggeling L, Sahm H. 1988. Regulation of enzymesof lysine biosynthesis in Corynebacterium glutamicum. J Gen Microbiol134(12):3221–3229.

Eggeling L, Bott M. 2005. Handbook of Corynebacterium glutamicum. BocaRaton, Florida: CRC Press, Tailor and Francis Group.

Fontecave M, Ollagnier-de-Choudens S. 2008. Iron-sulfur cluster biosyn-thesis in bacteria: Mechanisms of cluster assembly and transfer. ArchBiochem Biophys 474(2):226–237.

Grünberger A, Paczia N, Probst C, Schendzielorz G, Eggeling L, Noack S,Wiechert W, Kohlheyer D. 2012. A disposable picoliter bioreactor forcultivation and investigation of industrially relevant bacteria on singlecell level. Lab Chip 12:2060–2068.

Grünberger A, Probst C, Heyer A, Wiechert W, Frunzke J, Kohlheyer D.2013a.Microfluidic picoliter bioreactor for microbial single cell analysis:Fabrication, system setup and operation. JoVE (in press).

Grünberger A, van Ooyen J, Paczia N, Rohe P, Schendzielorz G, Eggeling L,Wiechert W, Kohlheyer D, Noack S. 2013b. Beyond growth rate 0.6:Corynebacterium glutamicum cultivated in highly diluted environments.Biotechnol Bioeng 110(1):220–228.

HaussmannU, Poetsch A. 2012. Global proteome survey of protocatechuate-and glucose-grown Corynebacterium glutamicum reveals multiple physio-logical differences. J Proteomics 75(9):2649–2659.

Kalinowski J, Bathe B, Bartels D, Bischoff N, Bott M, Burkovski A, Dusch N,Eggeling L, Eikmanns BJ, Gaigalat L, Goesmann A, Hartmann M,Huthmacher K, Kramer R, Linke B, McHardy AC, Meyer F, Mockel B,Pfefferle W, Puhler A, Rey DA, Ruckert C, Rupp O, Sahm H, WendischVF, Wiegrabe I, Tauch A. 2003. The complete Corynebacterium glutamicumATCC 13032 genome sequence and its impact on the production ofL-aspartate-derived amino acids and vitamins. J Biotechnol 104(1–3):5–25.

Keilhauer C, Eggeling L, Sahm H. 1993. Isoleucine synthesis in Corynebacteri-um glutamicum: Molecular analysis of the ilvB-ilvN-ilvC operon. JBacteriol 175(17):5595–5603.

Kind S, Jeong WK, Schroder H, Wittmann C. 2010. Systems-wide metabolicpathway engineering in Corynebacterium glutamicum for bio-basedproduction of diaminopentane. Metab Eng 12(4):341–351.

Kinoshita S. 1957. Studies on the amino acid fermentation. Part I. Productionof L-glutamic acid by various microorganisms. J Gen Appl Microbiol3:193–205.

Kramer R, Lambert C, Hoischen C, Ebbighausen H. 1990. Uptake ofglutamate in Corynebacterium glutamicum. 1. Kinetic properties andregulation by internal pH and potassium. Eur J Biochem 194(3):929–935.

LieblW, Klamer R, Schleifer KH. 1989. Requirement of chelating compoundsfor the growth of corynebacterium-glutamicum in synthetic media.Appl Microbiol Biotechnol 32(2):205–210.

Martin C, Galbe M, Nilvebrant NO, Jonsson LJ. 2002. Comparison of thefermentability of enzymatic hydrolyzates of sugarcane bagasse pretreated

by steam explosion using different impregnating agents. Appl BiochemBiotechnol 98–100:699–716.

Merkens H, Beckers G, Wirtz A, Burkovski A. 2005. Vanillate metabolism inCorynebacterium glutamicum. Curr Microbiol 51(1):59–65.

Moker N, Brocker M, Schaffer S, Kramer R, Morbach S, Bott M. 2004.Deletion of the genes encoding the MtrA-MtrB two-component systemof Corynebacterium glutamicum has a strong influence on cell morphology,antibiotics susceptibility and expression of genes involved in osmopro-tection. Mol Microbiol 54(2):420–438.

Nesvera J, Patek M. 2011. Tools for genetic manipulations in Corynebacteriumglutamicum and their applications. Appl Microbiol Biotechnol 90(5):1641–1654.

Noack S, Nöh K, Moch M, Oldiges M, Wiechert W. 2011. Stationary versusnon-stationary (13)C-MFA: A comparison using a consistent dataset.J Biotechnol 154(2–3):179–190.

Osten CH, Gioannetti C, Sinskey AJ. 1989. Design of a defined medium forgrowth of Corynebacterium glutamicum in which citrate facilitates ironuptake. Biotechnol Lett 11(1):11–16.

Paczia N, Nielgen A, Lehmann T, Gätgens J, Wiechert W, Noack S. 2012.Extensive exometabolome analysis reveals extended overflow metabo-lism in various microorganisms. Microb Cell Factories 11(1).

Polen T,Wendisch VF. 2004. Genomewide expression analysis in amino acid-producing bacteria using DNA microarrays. Appl Biochem Biotechnol118(1–3):215–232.

Schellenberger J, Que R, Fleming RM, Thiele I, Orth JD, Feist AM, ZielinskiDC, Bordbar A, Lewis NE, Rahmanian S, Kang J, Hyduke DR, PalssonBO. 2011. Quantitative prediction of cellular metabolism withconstraint-based models: The COBRA Toolbox v2.0. Nat Protoc 6(9):1290–1307.

Shen XH, Zhou NY, Liu SJ. 2012. Degradation and assimilation of aromaticcompounds by Corynebacterium glutamicum: Another potential forapplications for this bacterium? ApplMicrobiol Biotechnol 95(1):77–89.

Tillack J, Paczia N, Nöh K, Wiechert W, Noack S. 2012. Error propagationanalysis for quantitative intracellular metabolomics. Metabolites 2(4):1012–1030.

van Ooyen J, Noack S, Bott M, Reth A, Eggeling L. 2012. Improved L-lysineproduction with Corynebacterium glutamicum and systemic insight intocitrate synthase flux and activity. Biotechnol Bioeng 109(8):2070–2081.

Varela C, Rittmann D, Singh A, Krumbach K, Bhatt K, Eggeling L, Besra GS,Bhatt A. 2012.MmpL genes are associated withmycolic acidmetabolismin mycobacteria and corynebacteria. Chem Biol 19(4):498–506.

Volkmer B, Heinemann M. 2011. Condition-dependent cell volume andconcentration of Escherichia coli to facilitate data conversion for systemsbiology modeling. PLoS ONE 6(7):e23126.

Wendisch VF, Bott M, Kalinowski J, Oldiges M, Wiechert W. 2006. EmergingCorynebacterium glutamicum systems biology. J Biotechnol 124(1):74–92.

Wennerhold J, BottM. 2006. TheDtxR regulon of Corynebacterium glutamicum.J Bacteriol 188(8):2907–2918.

WooHM,Noack S, SeiboldGM,Willbold S, Eikmanns BJ, BottM. 2010. Linkbetween phosphate starvation and glycogen metabolism in Corynebacte-rium glutamicum, revealed by metabolomics. Appl Environ Microbiol76(20):6910–6919.

Yao Z, Davis RM, Kishony R, KahneD, Ruiz N. 2012. Regulation of cell size inresponse to nutrient availability by fatty acid biosynthesis in Escherichiacoli. Proc Natl Acad Sci USA 109(38):E2561–E2568.

Supporting Information

Additional supporting information may be found in theonline version of this article at the publisher’s web-site.

Unthan et al.: Biphasic Growth of Corynebacterium glutamicum 371

Biotechnology and Bioengineering