ARTICLE

Equalizing Growth in High-Throughput Small ScaleCultivations Via Precultures Operated inFed-Batch Mode

Robert Huber,1 Marco Scheidle,1 Barbara Dittrich,2 Doris Klee,2 Jochen Buchs1

1AVT-Aachener Verfahrenstechnik, Biochemical Engineering, RWTH Aachen University,

Worringerweg 1, D-52074 Aachen, Germany; telephone: þ49-241-80-23569;

fax: þ49-241-80-22570; e-mail: jochen.buechs@avt.rwth-aachen.de2Technical Chemistry and Macromolecular Chemistry, RWTH Aachen University,

Aachen, Germany

Received 11 December 2008; revision received 17 March 2009; accepted 30 March 2009

Published online 7 April 2009 in Wiley InterScience (www.interscience.wiley.com). DO

I 10.1002/bit.22349

ABSTRACT: An often underestimated problem when work-ing with different clones in microtiter plates and shake flaskscreenings is the non-parallel and non-equal growth of batchcultures. These growth differences are caused by variances ofindividual clones regarding initial biomass concentration,lag-phase or specific growth rate. Problems arising fromunequal growth kinetics are different induction points inexpression studies or uneven cultivation periods at the timeof harvest. Screening for the best producing clones of alibrary under comparable conditions is thus often impracti-cal or even impossible. A new approach to circumvent theproblem of unequal growth kinetics of main cultures is theapplication of fed-batch mode in precultures in microtiterplates and shake flasks. Fed-batch operation in precultures isrealized through a slow-release system for glucose. Afterdifferently growing cultures turn to glucose-limited growth,they all consume the same amount of glucose due to thefixed feed profile of glucose provided by the slow-releasesystem. This leads to equalized growth. Inherent advantagesof this method are that it is easy to use and requires noadditional equipment like pumps. This new techniquefor growth equalization in high-throughput cultivationsis simulated and verified experimentally. The growth ofdistinctly inoculated precultures in microtiter plates andshake flasks could be equalized for different microorganismssuch as Escherichia coli and Hansenula polymorpha.

Biotechnol. Bioeng. 2009;103: 1095–1102.

� 2009 Wiley Periodicals, Inc.

KEYWORDS: high-throughput; fed-batch; preculture;microtiter plate; equalization; shake flask

Robert Huber and Marco Scheidle contributed equally to this work.

Correspondence to: J. Buchs

Contract grant sponsor: Stiftung Industrieforschung (Koln, Germany)

Contract grant number: S753

Contract grant sponsor: Deutsche Bundesstiftung Umwelt (Osnabruck, Germany)

Contract grant number: 13144

� 2009 Wiley Periodicals, Inc.

Introduction

In the post-genomic era, clone libraries are used formany important applications such as screening fordrug candidates, efficient new biocatalysts or secondarymetabolites as well as media and strain selection (Kumaret al., 2004). The screening of these clone libraries is mainlyconducted in shake flasks and microtiter plates in batchmode. In particular, projects in structural genomics,structural proteomics and directed evolution applicationsare using microtiter plates for high-throughput cultivationof clones for expression studies (Berrow et al., 2006;Graslund et al., 2008; Heddle and Mazaleyrat, 2007).

An often underestimated problem when working withdifferent clones in microtiter plates is the non paralleland non equal growth of batch cultures. These growthdifferences are caused by variances of individual clonesregarding, for example, initial biomass concentrations, lagphases, or specific growth rates. The non-parallel growthin precultures can have a tremendous effect on theperformance of bioprocesses. As an example, the processof inducible protein expression will be discussed in moredetail with respect to unequal growth kinetics in precultures.In such processes, the addition of an inducer at a predefinedtime point is the most common way to initiate recombinantprotein production. It is well known that inducing atdifferent metabolic states or phases of a culture is a criticalfactor regarding protein yield (Donovan et al., 1996; Jenzschet al., 2006). Studier extensively described the problem ofsimultaneously inducing protein expression of differentclones and developed thus an autoinduction medium(Studier, 2005). This medium is a highly sophisticated wayto cope with the problem of different induction points;however, it cannot be applied to all microorganisms andhost/vector combinations. Furthermore, Studier emphasizesthat it is very difficult in high-throughput screening to

Biotechnology and Bioengineering, Vol. 103, No. 6, August 15, 2009 1095

obtain all of the cultures in a comparable state of growthbecause of varying lag time or growth rate of differentclones. These variations result, for instance, from differentexpression constructs or the genetic variability of the clones,such as variable plasmid insertions into the genome ofHansenula polymorpha. Therefore, it is very important tohave the same starting conditions for each clone in the maincultivation. To provide these defined starting conditionsthe preculture method is the crucial factor. One strategyfor achieving uniform conditions in the preculture is tocultivate the organisms until the stationary growth phase(saturation) (Studier, 2005). Though, when using thisprinciple it has to be assured that the final pH of the culturesis not too acidic when they reach saturation. Moreover, thestationary growth phase in batch mode is characterized bytremendous structural and physiological effects on bacterialcells (Baev et al., 2006; Hengge-Aronis, 1996; Huismanet al., 1996). Some authors have described that preculturesremaining unequal time periods in the stationary phaseshow variations in the lag phase of main cultures (Hornbaeket al., 2004; Pin and Baranyi, 2008). It is also known thatthe inoculum history is very important for the whole maincultivation process regarding reproducibility of growthkinetics (Ferenci, 1999; Neves et al., 2001; Webb andKamat, 1993). Additionally, in stationary phase the possibleproteolytic degradation of target proteins expressed viaIPTG induction or autoinduction may adversely affectproduct yield (Graslund et al., 2008). Consequently, thegrowth of precultures to stationary phase can have negativeeffects on the following main cultivation. Therefore,screening for the best producing clones of a library is verydifficult with conventional approaches.

Batch cultivations are predominately applied for small-scale cultures because of their easy use, flexibility, low costand lack of alternative methods. However, fed-batch modewould often be superior for producing biomass and productin main cultures. Furthermore, the fed-batch mode providesmore defined physiological conditions and is more oftenapplied in industrial scale than the batch mode. Jenzsch et al.(2006) presented the concept of greatly improving thereproducibility of main cultivations in stirred tank reactorsvia initiating a fed-batch mode already very early in thefermentation. Reproducibility is of utmost importanceespecially for good manufacturing practice (GMP), asrecommended in the PAT initiative from the FDA (Jenzschet al., 2006). The authors showed that using fed-batch modestarting in the early biomass formation phase can lead toidentical growth profiles of differently inoculated maincultures because of the fixed feed profile (Jenzsch et al.,2007). This concept which Jenzsch et al. proposed forthe early phase of a main cultivation served as a basisfor equalizing growth in small-scale precultivations withfed-batch mode. The fed-batch mode can be used fordifferent precultures with diverse growth parameters.Therefore, a fed-batch preculture could improve thecomparability and the selection of clones in screeningapplications.

1096 Biotechnology and Bioengineering, Vol. 103, No. 6, August 15, 2009

Fed-batch systems on a microliter scale for screeningand bioprocess development are increasingly being devel-oped. These techniques comprise automated stirrer-drivenmicrobioreactors (Puskeiler et al., 2005), microfluidic chips(Leeuwen van, 2008) or fed-batch in shake flasks (Ruottinenet al., 2008; Weuster-Botz et al., 2001). As these kinds ofdevices require pumps and additional equipment, they aresomewhat impractical for fed-batch in real high-through-put. Panula-Perala et al. (2008) recently published anenzyme controlled glucose auto-delivery system for micro-titer plates and shake flasks. Jeude et al. (2006) first havedeveloped a slow-release system to feed substrates such asglucose to shake flasks and microtiter plates in a much easierway without the need for additional equipment or enzymes.In these microtiter plates, silicone elastomer depots areimmobilized at the bottom of each well, thereby allowing thecontrolled release of glucose to the culture broth. This workpresents a novel method for equalizing growth kinetics inhigh-throughput precultures in shake-flasks and microtiterplates applying fed-batch mode. The focus of this paper isnot the expression of any product directly, but on thegrowth equalization of precultures. Escherichia coli andHansenula polymorpha are used as model organisms.

Materials and Methods

Organisms

E. coli BL21 pRset eYFP-IL6 was maintained in glycerolstocks at �808C in LB medium with 100mg/mL ampicillin.This strain was described by Samorski et al. (2005) with anadditional plasmid pLysS. Hansenula polymorpha RB11pC10-FMD (PFMD-GFP) (Amuel et al., 2000; Gellissen,2000) was maintained in glycerol stocks at �808C in YNBmedium and was kindly provided by Dr. C. Amuel(Heinrich-Heine University, Department of Microbiology,Dusseldorf, Germany).

Media and Solutions

Modified Wilms and Reuss synthetic medium was used forthe E. coli cultivations (Scheidle et al., 2007; Wilms et al.,2001). The medium consists of 20 g/L glucose; 5 g/L(NH4)2SO4; 0.5 g/L NH4Cl; 3 g/L K2HPO4; 2 g/L Na2SO4;0.5 g/L MgSO4�7H2O; 41.85 g/L 3-(N-Morpholino)-propanesulfonic acid (MOPS); 0.1 g/L ampicillin; 0.01 g/Lthiamine hydrochloride; 1mL/L trace element solution[0.54 g/L ZnSO4�7H2O; 0.48 g/L CuSO4�5H2O; 0.3 g/LMnSO4�H2O; 0.54 g/L CoCl2�6H2O; 41.76 g/L FeCl3�6H2

O; 1.98 g/L CaCl2�2H2O; 33.39 g/L Na2EDTA (Titiplex III)].The pH was adjusted to 7.5 with NaOH.

Hansenula polymorpha was cultivated in Syn6-MESmedium. The Syn6-MES mineral medium consisted of1.0 g/L KH2PO4, 7.66 g/L (NH4)2SO4, 3.3 g/L KCl, 3.0 g/LMgSO4�7H2O, 0.3 g/L NaCl, 27.3 g/L 2-morpholinoethane-sulfonic acid (MES). This aqueous basic solution was

adjusted to pH 6.4. Then the following substances wereadded (per L basic solution): 6.67mL calcium chloridesolution (150 g/L CaCl2 2H2O), 6.67mL microelementsolution (10.0 g/L (NH4)2Fe(SO4)2�6H2O, 0.8 g/L CuSO4

5H2O, 3.0 g/L ZnSO4 7H2O, 4.0 g/L MnSO4�H2O, 10.0 g/LEDTA (Titriplex III)), 6.67mL of vitamin solution (0.06 g/LD-biotin, 20.0 g/L thiamine hydrochloride), 3.33mL oftrace element solution (0.2 g/L NiSO4 6 H2O, 0.2 g/LCoCl2 6H2O, 0.2 g/L boric acid, 0.2 g/L KI and 0.2 g/LNa2MoO4 2H2O). The medium resulted in a final volume of1,023.33mL and no final pH adjustment was necessary(Gellissen, 2004; Jeude et al., 2006).

The medium used for the fed-batch precultivations hadno initial glucose because the immediate release of glucosefrom the slow-release system renders it unnecessary (Jeudeet al., 2006). All reagents were of analytical grade andpurchased from Carl Roth GmbH&Co. KG (Karlsruhe,Germany).

Cultivation

For reproducible inoculation of the example precultures inthis paper, additional first cultures were performed toprovide biomass. These first cultivations were conducted inWilms and Reuss synthetic medium and Syn6-MESmediumwith 20 g/L glucose for E. coli and H. polymorpha,respectively. The following cultivation parameters wereapplied: 350 rpm shaking frequency (n), 50mm shakingdiameter (d0), 10mL filling volume (VL) in 250mL shakeflasks. These first cultures were centrifuged and washed twotimes in 5mL fresh medium and optical densities (OD) weremeasured. The OD values were used for the calculationof the required inoculation volume for each describedexperiment. All cultures in shake flasks were conducted in anin-house made Respiration Activity Monitoring System(RAMOS) for online-monitoring of oxygen transfer rates(OTR), as previously described by Anderlei et al. (2001,2004). A commercial version of this device is available fromHiTec Zang GmbH (Herzogenrath, Germany) or KuhnerAG (Birsfelden, Switzerland). The following cultivationparameters were applied: 350 rpm shaking frequency,50mm shaking diameter, 10mL filling volume in 250mLRAMOS flasks. The applied fed-batch mode in shake flaskswas realized by using three slow-release discs per flask. Theseslow-release discs (denominated as ‘‘FeedBeads’’) containa silicone elastomer matrix in which glucose is embedded.They are available from Adolf Kuhner AG (Birsfelden,Switzerland). In general, the discs with silicon elastomer didnot adversely affect microbial growth (Jeude et al., 2006).

For fed-batch precultivations, deepwell plates withimmobilized silicon elastomer depots at the bottom ofeach well (FeedPlates) were used under the followingconditions: 700mL filling volume, 25mm shaking diameterand 400 rpm shaking frequency. The microtiter plates weresealed with an airpore-sheet (nonwoven sealing foil, HJBioanalytik) and cultivated under a humified aerated hood

H

to minimize evaporation. The substrate released up to thetime t can be described as follows:

glucose� release ¼ 2t0:69½mg=disc� for feedbeads and (1)

glucose� release ¼ 0:8t0:72½mg=well� for feedplates (2)

The glucose release kinetic of Equation (2) is shown inFigure 1D. For subsequent fed-batch precultivations, thefirst cultures were centrifuged and resuspended in glucose-free medium to prevent possible residues of glucose from thefirst cultures. All cultures were incubated at 378C in shakersof type LS-W or ISF-4-W from Adolf Kuhner AG.

Manufacture of Microtiter Plates WithSlow-Release System

To manufacture microtiter plates with the slow-release system,denominated as ‘‘FeedPlates,’’ solvent-free two-componentsilicone SylgardTM184 was used. The ratio between the twocomponents was 10:1 as recommended by the manufacturer.Anhydrous glucose was supplied by Sigma–Aldrich(Crailsheim, Germany) with the highest degree of purity.The glucose was milled with a vibration micromill(SpartanTM, Fritsch, Idar-Oberstein, Germany) in a high-grade steel mortar and then sieved with test sieves (Fritsch).The fraction with particle sizes ranging from 20 to 50mmwas used. First, the two silicone components of theSylgardTM184 and glucose were mixed. Second chloroform(1–5mL chloroform per 10 g silicone-glucose mixture) wasadded to decrease the viscosity of the mixture allowing it toflow more easily. Thereafter, 100mL of the compounds werefilled at the bottom of each cavity of a 2.2mL polypropylenemicrotiter plate (HJ Bioanalytik, Monchengladbach,Germany) using a multipette (Eppendorf, Wesseling-Berzdorf, Germany). The plate was stored at 508C for12 h to aid cross-linking.

Analytical Methods

Optical density of microtiter plate experiments wasmeasured at 600 nm (OD600) with the microtiter platereader Powerwave X340 (Bio-Tek Instruments GmbH, BadFriedrichsthal, Germany) and for shake flask experiments aUvikon 922 spectrophotometer (Kontron, Milano, Italy)was used. Samples were measured in the linear range of ODmeasurements after dilution in fresh medium. Sampling andOD measurement were each conducted twice or thrice. Itwas not feasible to take samples from each well at every pointduring the experiments because of the limited volume of700mL per well. A lower volume after sampling and thecontinuous release of substrate would have led to changingglucose concentrations. Therefore, samples from the fed-batch microtiter plates were withdrawn from different wellswith identical conditions.

uber et al.: Equalizing Growth in High-Throughput Cultivations 1097

Biotechnology and Bioengineering

Figure 1. Comparison of batch and fed-batch modes for cultivating precultures. Variations in lag phase, maximum specific growth rate or initial biomass concentration

were simulated. (A) Growth kinetics in batch mode; (B) substrate kinetics in batch mode; (C) growth kinetics in fed-batch mode; (D) substrate kinetics in fed-batch mode.

(—) reference growth parameters (X0 ¼ 0.1 g/L; tlag ¼ 0.5 h; mmax ¼ 0.5 1/h); (- - -) decreased initial biomass concentration (X0 ¼ 0.02 g/L; tlag ¼ 0.5 h; mmax ¼ 0.5 1/h); (���) increased

lag time (X0 ¼ 0.1 g/L; tlag ¼ 2.0 h; mmax ¼ 0.5 1/h); (- �� -) decreased mmax (X0 ¼ 0.1 g/L; tlag ¼ 0.5 h; mmax ¼ 0.3 1/h); (- � -) total release of substrate from slow-release system in

fed-batch mode (calculated using Eq. 2). The arrow and the bracket indicate the time for inoculation of the main culture from batch and fed-batch preculture, respectively.

Theoretical Background

A simple model for batch and fed-batch cultivations inmicrotiter plates was applied to demonstrate the concept ofequalizing the growth in precultures. The variation of initialbiomass concentrations (inocula), lag phase and specificgrowth rates was chosen to visualize growth and substratekinetics of different precultures. These precultures mayrepresent various clones of a clone library.

Modeling

A simple model for fed-batch cultivations in microtiterplates with standard bioreaction equations based on Monodkinetics was applied (Eqs. 3–7).

dX

dt¼ ðm� KdÞX ½g=L=h� (3)

dS

dt¼ � 1

YX=SmmaxX þ Feedrate ½g=L=h� (4)

1098 Biotechnology and Bioengineering, Vol. 103, No. 6, August 15, 2009

m ¼ mmaxS

ðSþ KSÞrelact ½1=h� (5)

Feedrate ¼ 0:8� 0:72� tð�0:28Þ

V½g=L=h� (6)

relact ¼1

1þ eð2þ4t�tlagtacc

Þ(7)

Equation (6) was obtained by differentiating Equation (2)and referring it to the utilized filling volume in a well.Equation (7) is an equation developed by us to represent thelag and acceleration phase of a culture. The term relativeactivity (relact) in Equations (5) and (7) accounts for the lagphase and acceleration phase and varies from 0 (start ofcultivation) to 1 (cells adapted to the new medium). Theadvantage over existing equations is that the two parameterstlag and tacc can directly be interpreted as the time constantfor the lag and acceleration phase, respectively. A decay termwas introduced to simulate the decrease in biomass due tothe lack of a carbon source and the accumulation of endproducts in the stationary phase of batch cultivations. As

there is no general consensus for mathematical modeling ofa decay term or death rate (Toal et al., 2000), a constantdecay term was assumed in the model (Moser and Steiner1975). Simulations were conducted with Modelmaker(Cherwell Scientific, Oxford, UK).

Typical values for precultivation parameters of E. coliwere introduced into the model (KS¼ 0.2 g/L, tacc¼ 0.5 h,Kd¼ 0.01 1/h, YXS¼ 0.5 g/g, S0¼ 15 g/L for batch, S0¼ 0 g/Lfor fed-batch). All the other model parameters were selectedaccording to the caption of Figure 1. The volume ofthe precultivation was set at 700mL as a typical value forsmall-scale cultivations in deepwell microtiter plates. Withinthe simulation time of 34 h, the initial glucose concentration(substrate S0) in the batch model was comparable to thetotal amount of glucose fed in the fed-batch mode. Theterm Feedrate in Equation (4) was omitted for the batchsimulations.

Batch Cultivation

After the lag phase, the different batch precultivations growexponentially (Fig. 1A) until the substrate is exhausted(Fig. 1B). Then, the stationary phase begins and the decay ofviable biomass becomes apparent. This applies for all foursimulations which results in different growth kinetics andphysiological states at any time in batch mode. Moreover,non-optimal conditions for inoculation of the main cultureare realized in the stationary phase (see arrow in Fig. 1A),because the clones are nutrient depleted in the stationaryphase for different times. When these cells are used as apreculture for screening experiments, differences in growthkinetics in the main cultivation are most probable.

Figure 2. Batch cultivation of E. coli BL21 pRset eYFP-IL6 in RAMOS flasks with

various initial biomass concentrations: ODto ¼ 0.5 (&), ODto ¼ 0.3 (*), ODto ¼ 0.1 (~).

Wilms-MOPS medium with 15 g/L glucose, T¼ 378C, VL¼ 10 mL, n¼ 350 rpm,

d0 ¼ 50 mm.

Fed-Batch Cultivation

The fed-batch mode allows predefined growth behavior ofthe preculture via feeding and, thus, offers the advantage of amore controlled process than the batch mode. In fed-batchsimulations, shown in Figure 1C and D, the addition ofglucose to the applied medium was unnecessary, because ofthe immediate release of glucose from the release system.This is also apparent in Figure 1D, which shows the totalrelease of glucose throughout the cultivation. The substrateaccumulates at the beginning of the fermentation whenthe cell concentration is still too low to consume thereleased glucose. After 5–10 h the biomass of the simulatedprecultures consume more glucose than provided bythe slow-release system and, consequently, the substrateconcentration decreases (Fig. 1D). The preculture turnsfrom a batch to a fed-batch phase after 8–15 h (Fig. 1C). Inthis phase the organisms are growing in a substrate-limitedfashion and show a growth rate predefined by the feed rateof the slow-release system (Eq. 6). It is obvious that the fed-batch mode equalizes the different growth behavior of allprecultures. The cells are constantly supplied with substrate,thereby resulting in constant growth and a definedmetabolic state from 15 h onwards (Fig. 1C). In this way,

H

the adverse effects of nutrient starvation or accumulation ofoverflow metabolites are minimized.

This system is self regulating concerning the consumptionof glucose of each preculture. After the different bacterialclones switched over to glucose-limited growth, they allconsume equal amounts of glucose and, hence, produceequal cell densities. The simulation demonstrates that fed-batch fermentations with defined feeding rates can equalizeprecultures that have different inocula, specific growth ratesand lag phases. Moreover, the time for inoculation of maincultures is, in contrast to batch-precultures, no longerimportant. Even if the transfer of inocula is postponedrelative to a fixed schedule, for example, due to somepractical reasons, the precultures do not suffer from carbonsource depletion (see bracket in Fig. 1C) and preserve theirmetabolic activity.

Results and Discussion

RAMOS Cultivations

Batch Mode

Batch precultures of the recombinant strain E. coli BL21pRset eYFP-IL6 were cultivated as reference at threedifferent initial optical densities (ODto) from 0.1 to 0.5.They were inoculated from the same first culture in order toprovide defined starting conditions. The different ODto

represent a range commonly applied when inoculatingprecultures. The oxygen transfer rates are depicted inFigure 2 against the fermentation time. The oxygen transferrate (OTR) signal shows typically exponential growth andoxygen is limited only for a short period (plateau) upon

uber et al.: Equalizing Growth in High-Throughput Cultivations 1099

Biotechnology and Bioengineering

attaining an OTR of approximately 60mmol/L/h.When glucose is exhausted, the OTR decreases sharply(e.g., at 12 h for the culture with ODto¼ 0.1). Due tooverflow metabolism, acetate is formed during theexponential growth phase; its assimilation marks the secondpeak in the OTR curve of each culture. These phenomenahave been previously described for this strain (Scheidle et al.,2007). The different inocula resulted in a time variation ofapproximately 4 h in growth. The results demonstrateclearly the significant difference in growth kinetics of themicroorganisms caused only by different initial biomassconcentrations.

Fed-Batch Mode

For fed-batch precultures in shake flasks, FeedBeads wereapplied, and the fermentation was monitored online withthe RAMOS device (Fig. 3A).

The exponential growth of the batch phase in thebeginning is followed by substrate limitation and a sharpOTR decrease. It is noteworthy that this decrease does notfall to zero but rather to a defined value of approximately3mmol/L/h. This value reflects the constant release of thesubstrate from the FeedBeads during the fed-batch phase.This growth pattern is obvious for all of the three differentlyinoculated precultures. The OTR peak of the batch phaseincreases for precultures with less inoculum (from 10mmol/L/h for ODto¼ 0.5 to 17mmol/L/h for ODto¼ 0.1; Fig. 3A).

Figure 3. Fed-batch precultivation of E. coli BL21 pRset eYFP-IL6 in

RAMOS flasks with various initial biomass concentrations: ODto ¼ 0.5 (~), ODto ¼ 0.3

0.3 (*), ODto¼ 0.1 (&). (A) Oxygen transfer rate and (B) calculated total oxygen

consumption. Wilms-MOPS medium with no additional glucose, three FeedBeads per

flask, T¼ 378C, VL ¼ 10 mL, n¼ 350 rpm, d0¼ 50 mm.

1100 Biotechnology and Bioengineering, Vol. 103, No. 6, August 15, 2009

This increase is caused by the fact that more glucoseaccumulates in the flasks with lower inoculum before thefed-phase starts (see also Fig. 1C). Nevertheless, as soon as allprecultures have reached the fed-batch mode (11 h), they allhave consumed the same total amount of glucose. In fed-batch mode neither acetate formation nor oxygen limitationcan be observed due to the lower initial glucose concentra-tion and subsequent glucose-limited growth as comparedwith the batch mode (compare Figs. 3A and 2). All theprecultures are found to be in a defined and similarmetabolic state beginning at 11 h and are actively growing.At the end of the experiment, the OD of the precultures withan initial biomass of ODto of 0.1, 0.3, and 0.5 were 4.4, 4.5,and 4.2, respectively. This demonstrates equalized growth.

Figure 3B shows the total oxygen consumed by the cells atany given time obtained by the integration of the OTR. Byassuming a constant yield coefficient of biomass to oxygenduring fermentation, the total oxygen consumption reflectsthe increase in biomass and is in good agreement with thesimulated data (Fig. 1C). Variations in the total oxygenconsumption during the fed-batch phase (Fig. 3B) resultfrom errors in integrating curves with only few data points.

Fed-Batch in Microtiter Plates

The newly developed fed-batch system for microtiter plateswas used. Two microorganisms were tested to equalize thevarious precultures. Initial optical densities of 0.05, 0.1 and0.3 were applied to simulate variable growth kinetics. Thedifferent precultures were inoculated from the same firstculture in order to provide defined starting conditions. ForE. coli, varying the inoculum from the highest to the lowestinitial biomass concentration yielded a ca. 10 h delay ingrowth (Fig. 4). In fed-batch mode the precultures turned

Figure 4. Fed-batch precultivation of E. coli BL21 pRset eYFP-IL6 in a fed-batch

deepwell plate with various initial biomass concentrations: ODto ¼ 0.5 (~), ODto ¼ 0.3

(*), ODto ¼ 0.1 (&). Wilms-MOPS medium with no additional glucose, T¼ 378C,

VL ¼ 700mL, n¼ 400 rpm, d0 ¼ 25 mm.

one after another from an exponential growth to acontrolled nearly linear increase in biomass concentration.After approximately 15 h, all precultures were equalized andattained the same biomass concentration of OD 6.5 at theend of the experiment. As previously mentioned, this effectwas caused by the same total amount of glucose beingreleased per well.

The same behavior as for E. coli can be observed forHansenula polymorpha precultures. Here, a 10 h delay ingrowth is observed for precultures with high and lowinoculum concentration (Fig. 5). The applied fed-batchmode enables equalization of all precultures after 24 h. At50 h the different precultures still have the same biomassconcentration and are actively growing.

Conclusions

A new technique for growth equalization in high-throughput precultivations by applying fed-batch modewas simulated and verified experimentally. Growth ofdifferently inoculated precultures in shake flasks andmicrotiter plates could be equalized and, therefore,the feasibility of this concept could be demonstrated. Theconcept worked for procaryotic and eucaryotic micro-organisms. Consequently, this technique seems to be ofgeneral applicability.

The inherent advantages of this method are that it is easyto use as it requires no additional equipment for fed-batchprecultivations on a small-scale. Furthermore, there is noneed for permanent and laborious offline-monitoring ofprecultures to determine the right time of transfer to amain cultivation. Moreover, the exact time for inoculationof main cultures is, in contrast to batch precultures, no

Figure 5. Fed-batch precultivation of Hansenula polymorpha in a fed-batch

deepwell plate with various initial biomass concentrations: ODto ¼ 0.5 (&), ODto ¼ 0.3

(D), ODto¼ 0.1 (&). Wilms-MOPS medium with no additional glucose, T¼ 378C,

VL ¼ 700mL, n¼ 400 rpm, d0 ¼ 25 mm.

H

longer important. The system is self-regulating; the cells arecontinuously supplied with substrate and are in a definedmetabolic state. This represents another strategy forachieving uniform conditions in the growth of differentclones as recommended by Studier (2005), with theexception that the microorganisms are not in the stationarygrowth phase. Possible oxygen limitations and adverseeffects of the batch mode can also be avoided in thefed-batch mode.

This technique might be especially useful for micro-organisms exhibiting decreasing viability in the stationaryphase and in which synchronous growth of distinctprecultures is very important. Furthermore, precultures infed-batch mode microtiter plates can generate more relevantdata in screening processes (Jeude et al., 2006), becausethe starting conditions for all strains under study are equal.A possible disadvantage of the introduced equalizationtechnique may be the expression of toxic products in hostswhich are de-repressed by low glucose concentrationsduring the fed-batch phase. Further investigations have tobe performed to prove this method with different clonelibraries and the impact of equalized precultures on productformation (e.g., recombinant proteins or amino acids) insubsequent main cultivations.

Nomenclature

d0

uber et

shaking diameter (mm)

Feedrate

feeding rate (g/L/h)

Kd

decay term (1/h)

KS

half-saturation constant (g/L)

n

shaking frequency (rpm)

OD600

optical density at 600 nm

ODt0

start optical density at 600 nm

relact

relative activity

S

substrate concentration (g/L)

S0

initial substrate conc. (g/L)

T

temperature (8C)

t

time (h)

tacc

time of acceleration phase (h)

tlag

time of lag phase (h)

VL

filling volume per well (L)

X

biomass concentration (g/L)

X0

initial biomass conc. (g/L)

Yx/s

yield coefficient (g/g)

m

specific growth rate (1/h)

mmax

maximal sp. growth rate (1/h)

We are grateful for the financial support of the Stiftung Industrie-

forschung (Koln, Germany) and the Deutsche Bundesstiftung Umwelt

(Osnabruck, Germany).We thank Holger Gross for technical support.

References

Amuel C, Gellissen G, Hollenberg CP, Suckow M. 2000. Analysis of heat

shock promoters in Hansenula polymorpha: The TPS1 promoter, a

al.: Equalizing Growth in High-Throughput Cultivations 1101

Biotechnology and Bioengineering

novel element for heterologous gene expression. Biotechnol Bioprocess

Eng 5:247–252.

Anderlei T, Buchs J. 2001. Device for sterile online measurement of the

oxygen transfer rate in shaking flasks. Biochem Eng J 7(2):157–162.

Anderlei T, Zang W, Papaspyrou M, Buchs J. 2004. Online respiration

activity measurement (OTR, CTR, RQ) in shake flasks. Biochem Eng

J 17(3):187–194.

Baev MV, Baev D, Radek AJ, Campbell JW. 2006. Growth of Escherichia coli

MG1655 on LB medium: Determining metabolic strategy with

transcriptional microarrays. Appl Microbiol Biotechnol 71(3):323–

328.

Berrow NS, Bussow K, Coutard B, Diprose J, EkbergM, Folkers GE, Levy N,

Lieu V, Owens RJ, Peleg Y, et al. 2006. Recombinant protein expression

and solubility screening in Escherichia coli: A comparative study. Acta

Crystallogr D Biol Crystallogr 62(Pt 10):1218–1226.

Donovan RS, Robinson CW, Glick BR. 1996. Review: Optimizing inducer

and culture conditions for expression of foreign proteins under the

control of the lac promoter. J Ind Microbiol 16(3):145–154.

Ferenci T. 1999. ‘Growth of bacterial cultures’ 50 years on: Towards an

uncertainty principle instead of constants in bacterial growth kinetics.

Res Microbiol 150(7):431–438.

Gellissen G. 2000. Heterologous protein production in methylotrophic

yeasts. Appl Microbiol Biotechnol 54(6):741–750.

Gellissen G. 2004. Production of Recombinant Proteins. Wiley-VCH,

Weinheim (1).

Graslund S, Nordlund P, Weigelt J, Hallberg BM, Bray J, Gileadi O, Knapp

S, OppermannU, Arrowsmith C, Hui R, et al. 2008. Protein production

and purification. Nat Methods 5(2):135–146.

Heddle C, Mazaleyrat SL. 2007. Development of a screening platform for

directed evolution using the reef coral fluorescent protein ZsGreen as a

solubility reporter. Protein Eng Des Sel 20(7):327–337.

Hengge-Aronis R. 1996. Regulation of gene expression during entry into

stationary phase. Escherichia coli and Salmonella: Cellular and

molecular biology, 2nd edition. Washington, DC: ASM Press.

pp. 1497–1512.

Hornbaek T, Nielsen AK, Dynesen J, Jakobsen M. 2004. The effect of

inoculum age and solid versus liquid propagation on inoculum quality

of an industrial Bacillus licheniformis strain. FEMS Microbiol Lett

236(1):145–151.

Huisman G, Siegele D, Zambrano M, Kolter R. 1996. Morphological

and physiological changes during stationary phase. Escherichia coli

and Salmonella: Cellular and molecular biology, 2nd edition.

Washington, DC: ASM Press, pp. 1672–1682.

Jenzsch M, Gnoth S, Beck M, Kleinschmidt M, Simutis R, Lubbert A. 2006.

Open-loop control of the biomass concentration within the growth

phase of recombinant protein production processes. J Biotechnol

127(1):84–94.

Jenzsch M, Gnoth S, Kleinschmidt M, Simutis R, Lubbert A. 2007. Improv-

ing the batch-to-batch reproducibility of microbial cultures during

1102 Biotechnology and Bioengineering, Vol. 103, No. 6, August 15, 2009

recombinant protein production by regulation of the total carbon

dioxide production. J Biotechnol 128(4):858–867.

Jeude M, Dittrich B, Niederschulte H, Anderlei T, Knocke C, Klee D, Buchs

J. 2006. Fed-batch mode in shake flasks by slow-release technique.

Biotechnol Bioeng 95(3):433–445.

Kumar S, Wittmann C, Heinzle E. 2004. Minibioreactors. Biotechnol Lett

26(1):1–10.

Leeuwen VanM. 2008. Development of a bioreactor with integrated on-line

sensing for batch and fed-batch cultivation on a 100 mL-scale.

Dissertation, TUD Technische Universiteit Delft.

Moser A, Steiner W. 1975. The Influence of the term kd for endogenous

metabolism on the evaluation of monod kinetics for biotechnological

processes. Eur J Appl Microbiol 1:281–289.

Neves AA, Vieira LM, Menezes JC. 2001. Effects of preculture variability on

clavulanic acid fermentation. Biotechnol Bioeng 72(6):628–633.

Panula-Perala J, Siurkus J, Vasala A, Wilmanowski R, Casteleijn MG,

Neubauer P. 2008. Enzyme controlled glucose auto-delivery for high

cell density cultivations in microplates and shake flasks. Microb Cell

Fact 7:31.

Pin C, Baranyi J. 2008. Single-cell and population lag times as a function of

cell age. Appl Environ Microbiol 74(8):2534–2536.

Puskeiler R, Kusterer A, John GT, Weuster-Botz D. 2005. Miniature

bioreactors for automated high-throughput bioprocess design

(HTBD): Reproducibility of parallel fed-batch cultivations with Escher-

ichia coli. Biotechnol Appl Biochem 42(Pt 3):227–235.

RuottinenM, BollokM, Kogler M, Neubauer A, Krause M, Hamalainen ER,

Myllyharju J, Vasala A, Neubauer P. 2008. Improved production of

human type II procollagen in the yeast Pichia pastoris in shake flasks by

a wireless-controlled fed-batch system. BMC Biotechnol 8:33.

Samorski M, Muller-Newen G, Buchs J. 2005. Quasi-continuous combined

scattered light and fluorescence measurements: A novel measurement

technique for shaken microtiter plates. Biotechnol Bioeng 92(1):61–68.

Scheidle M, Klinger J, Buchs J. 2007. Combination of on-line pH and

Oxygen Transfer Rate Measurement in Shake Flasks by Fiber Optical

Technique and Respiration Activity MOnitoring System (RAMOS).

Sensors 7:3472–3480.

Studier FW. 2005. Protein production by auto-induction in high density

shaking cultures. Protein Expr Purif 41(1):207–234.

ToalME, Yeomans C, KillhamK,Meharg AA. 2000. A review of rhizosphere

carbon flow modelling. Plant Soil 222:263–281.

Webb C, Kamat S. 1993. Improving fermentation consistency through

better inoculum preparation. World J Microbiol Biotechnol 9(3):308–

312.

Weuster-Botz D, Altenbach-Rehm J, Arnold M. 2001. Parallel substrate

feeding and pH-control in shaking-flasks. Biochem Eng J 7(2):163–170.

Wilms B, Hauck A, Reuss M, Syldatk C, Mattes R, Siemann M, Alten-

buchner J. 2001. High-cell-density fermentation for production of

L-N-carbamoylase using an expression system based on the Escherichia

coli rhaBAD promoter. Biotechnol Bioeng 73(2):95–103.