http://jla.sagepub.com/Automation

Journal of the Association for Laboratory

http://jla.sagepub.com/content/12/3/143The online version of this article can be found at:

 DOI: 10.1016/j.jala.2006.10.017

2007 12: 143Journal of Laboratory AutomationZhiyu Zhang, Gerardo Perozziello, Paolo Boccazzi, Anthony J. Sinskey, Oliver Geschke and Klavs F. Jensen

Microbioreactors for Bioprocess Development  

Published by:

http://www.sagepublications.com

On behalf of: 

  Society for Laboratory Automation and Screening

can be found at:Journal of the Association for Laboratory Automation Additional services and information for    

  http://jla.sagepub.com/cgi/alertsEmail Alerts:

 

http://jla.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

What is This? 

- Jun 1, 2007Version of Record >>

by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from by guest on October 11, 2013jla.sagepub.comDownloaded from

Keywords:

microbioreactors,

polymer

microfluidics,

fluid coupler,

microlens optical

interface

Original Report

Microbioreactors for BioprocessDevelopment

Zhiyu Zhang,1 Gerardo Perozziello,2 Paolo Boccazzi,1 Anthony J. Sinskey,1 Oliver Geschke,2

and Klavs F. Jensen1*1Massachusetts Institute of Technology, Cambridge, MA

2Technical University of Denmark (DTU), Lyngby, Denmark

As a step toward high-throughput bioprocess

development, we present design, fabrication, and

characterization of polymer based microbioreactors

integrated with automated sensors and actuators. The

devices are realized, in increasing levels of complexity, in

poly(dimethylsiloxane) and poly(methyl methacrylate) by

micromachining and multilayer thermal compression

bonding procedures. Online optical measurements for

optical density, pH, and dissolved oxygen are integrated.

Active mixing is made possible by a miniature magnetic

stir bar. Plug-in-and-flow microfluidic connectors and

fabricated polymer micro-optical lenses/connectors are

integrated in the microbioreactors for fast set up and easy

operation. Application examples demonstrate the

feasibility of culturing microbial cells, specifically

Escherichia coli, in 150 mL-volume bioreactors in batch,

continuous, and fed-batch operations. ( JALA

2007;12:143–51)

INTRODUCTION

Conventional microbial cell cultivation techniqueshave lagged behind array-based automated toolsfor discovery and genetic manipulation of

*Correspondence: Klavs F. Jensen, Ph.D., Department of ChemicalEngineering, Massachusetts Institute of Technology, Room 66-566, 77 Massachusetts Avenue, Cambridge, MA 02139, USA;Phone: þ1.617.253.4589; Fax: þ1.617.258.8224; E-mail:kfjensen@mit.edu

1535-5535/$32.00

Copyright �c 2007 by The Association for Laboratory Automation

doi:10.1016/j.jala.2006.10.017

biological systems. Bioprocess developments for mi-crobial cultivation and optimization are typicallyperformed in expensive, mechanically complex,and labor intensive, stirred-tank bioreactors. Benchscale, stirred-tank bioreactors, with typical volumesof between 0.5 and 10 L are instrumented with effec-tive control of temperature, pH, and dissolved oxy-gen (DO) levels and yield valuable physiological andmetabolic data. However, the relatively small num-ber of experiments that can be performed limits op-timization of growth conditions and metabolicstudies. Bench scale, stirred-tank reactor technologyhas been addressing these challenges by reducingvolume and increasing the number of reactors oper-ating in parallel. However, throughput remains lim-ited by the mechanical complexity of setting up andperforming experimentsdprocesses such as assem-bly, cleaning, and calibration of sensors scale withthe number of bioreactors. Thus, there is a needfor systems that enable rapid testing, optimization,and bioprocess developments in low volume, paral-lel investigations with setup and run-time effortsthat remain nearly independent of the number ofbioreactors. The SIXFORS benchtop device byInfors AG (Bottmingen, Switzerland) has six fer-mentors operating in parallel. Another recent devel-opment, Cellstation bioreactors by FluorometrixCorp. (Stow, MA) allow 12 miniature stirred-tankbioreactors to be operated in parallel. However,the throughput using these devices is still limited.

Disposable, parallel-operated microbioreactorswith small volumes of cultures and integrated real-time measurements have been proposed as a promis-ing solution for high-throughput bioprocessing. Anearly microbioreactor was developed by Waltheret al.1e3 as a 3 mL continuous bioreactor with

JALA June 2007 143

Original Report

integrated biomass, pH, and temperature microelectronicsensors for yeast cell cultivation in space. It was a self-sustained system with medium flow rate measured by a micro-sensor and controlled by a piezo-electric silicon membranepump. pH was monitored by an ion-selective field effect tran-sistor (ISFET) sensor2 and manipulated by coulometric gen-eration of hydroxyl ions at a titanium electrode. The systemwas aerated through gas-permeable cylindrical silicone tubesand mixing was done by a magnetic stir bar. Maharbizet al.4,5 integrated microtiter plate wells with silicon monitor-ing technology to realize 250 mL microbioreactor arrays withISFET sensors on a commercial printed circuit board. Foraeration, oxygen was generated in the reactor by hydrolysisof water. The microscale device reported by Lampinget al.6 was a scaled-down version of conventional stirred-tankbioreactors machined in Plexiglas and outfitted with airspargers and a stirring baffle.

Optical sensors are typically noninvasive and can be inte-grated into the design. Electrochemical sensors for oxygen,used on the macroscale, are generally not suitable for micro-systems because they consume oxygen that otherwise wouldbe available to the cell culture. Rao and coworkers7e10 pio-neered optical sensing for measurements of optical density(OD) (related to cell density), pH, and DO for microbialbatch cultures in 2 mL cuvettes, which demonstrated the fea-sibility of using optical sensors in microbioreactors to reducemechanical complexity and enhance throughput. Puskeileret al.11 reported a parallel operation of 48 magneticallymixed milliliter scale bioreactors with integrated OD andpH real-time measurements and automation of fed-batch op-eration and pH control. These studies in miniaturization andparallelization of micro-bioreactors, as well as the commer-cial Cellstation (Fluorometrix Corp., Stow, MA) and Sim-Cell (BioProcessors, Woburn, MA) microbioreactorplatforms underscore the potential of this technology indevelopment of bioprocesses.

Our microbioreactors are based on the membrane-aeratedmicrobioreactor previously reported by Zanzotto et al.12 Inthis microbioreactor a thin membrane of poly(dimethylsilox-ane) (PDMS), a highly gas-permeable polymer material,serves as the aeration membrane for cell metabolism. Themembrane also defines the gaseliquid interface to providea sterile, single-phase microbioreactor. With the integrationof polymer-fabrication technologies, magnetic active mixing,temperature control, surface modification, and continuousfluidic system individual microbioreactors have been demon-strated for batch13 and continuous culture14 fermentations ofEscherichia coli in mL-scale microbioreactors with real-timeand in situ optical monitoring of OD, pH, and DO. These de-velopments in microbioreactors allow continuous monitoringand control during microbial culture processes, thus greatlysimplifying the effort per fermentation experiment. Further-more, parallel microbial fermentations in a multiplexed sys-tem demonstrate the potential of microbioreactors forhigh-throughput experimentation.15 Lee et al.16 have recentlydeveloped an integrated array of microbioreactors in PDMS

144 JALA June 2007

with 100 mL working volume and having peristaltic oxygen-ating mixers and microfluidic injectors. The systemsupported eight simultaneous E. coli fermentations to celldensities greater than or comparable to those achieved ina 4 L bench scale stirred-tank bioreactor.

One important issue in high-throughput studies of biopro-cesses is the parallel operation of microbial fermentations indisposable microbioreactors. Replacing the reactors requiresa well-defined interface of the microbioreactor to externalfluid handling, optical measurements, and electronic instru-ments. In this work, plug-in-and-flow microfluidic connec-tors and microfabricated polymer optical lenses and fiberconnectors are integrated in the microbioreactor. The dispos-able microbioreactor, realized in poly(methyl methacrylate)(PMMA) and PDMS by a multilayer thermal-bonding pro-cedure, automatically aligns with external fluidic and opticalcomponents, which addresses the need for rapid set up andease of operation.

MICROBIOREACTOR CONFIGURATIONS

The early generation of microbioreactor (Fig. 1A) was de-signed for batch cultivation ofmicrobial cells. The reactor con-sisted of two PDMS layers (L2 and L3) sandwiched in betweentwoPMMAlayers (L1 andL4 inFig. 1A). ThebottomPMMAlayer, made by using a computer-numerical-controlled (CNC)millingmachinewith dimensional accuracy of 12 mm, includedthe microbioreactor chamber (150 mL volume) and three con-necting channels (M), which are used for inoculation andreplenishment of water.

Two recesses (S, diameter 2 mm, depth 250 mm) at the bot-tom of the bioreactor chamber accommodated DO and pHfluorescence lifetime sensors (DO sensor foil PSt3, and pHsensor solution HP2A, PreSensdPrecision Sensing GmbH,Regensburg, Germany). Details of the fluorescence lifetimemeasurements are summarized elsewhere.12,15 In the centerof the device, a reactor chamber was fabricated in layersL1 to have a cylindrical geometry with a diameter of10 mm and a depth of 1 mm. A ring-shape magnetic stirbar (K) with 6 mm arm length and 0.5 mm thickness (cus-tom-made by Engineered Concepts, Vestavia Hills, AL)was held by a PMMA post (T) and was used for activemixing in the reactor chamber. The mixing efficiency in themicrobioreactor chamber was evaluated through experi-ments and computational fluid dynamic simulations.13

A thin layer (w100 mm in thickness) of spin-coated PDMS(L2; mixing ratio of silicone to curing agent was 10:1. Sylgard184, Dow Corning Corp., Midland, MI, USA) covered thereactor chamber and served as the aeration membrane.PDMS was spin-coated at a speed of 1200 rpm for 25 s andthen baked at 70 �C for 2 h for curing. The PDMS membranebulged upward as a result of a positive pressure in the micro-bioreactor chamber to reach a reactor volume of 150 mL. Tofacilitate device assembly, hermetical sealing, and connectionof microfluidic channels (M), this PDMS layer was held bya 5-mm thick PDMS gasket layer (L3). A top PMMA layer

Original Report

Figure 1. (A) Microbioreactor device used in batch cell cultivation. (B) Device used in pH-control and fed-batch cell cultivation. Micro-channels were used for water- (P1), base- (P2), and acid-feeding (P3), inoculation (P4), and waste exit (P5) purposes, respectively. (C) Ther-mally bonded PMMA device used in continuous cell cultivation. Microfluidic channels allow for water-replenishment (P1), inoculation (P2),and waste exit (P3), and cell outflow (D) before and during chemostat experiments. (D) Cross-section of an integrated microbioreactor.L1eL5, thermally bonded PMMA layers; F PMMA cork used for mechanical assembly of the aeration membrane; H silicone O-ring forsealing; I optical fiber fixed by F; J grid for holding the PDMS membrane above the reactor chamber; K magnetic mixer in the center ofreactor chamber; L recesses in PMMA for accommodating alignment pins P; M microfluidic channels; N small silicone O-rings for fluidicinterconnections; O pH and DO fluorescent sensors; P PDMS alignment pins on optical plugs; S optical sensors.

was used to provide a rigid support for the mechanical as-sembly. Three ports connected the microbioreactor chamberwith external setup via microchannels (M) and served for thepurposes of inoculation (P1), waste replenishment (P2),waste outflow (P3; used during inoculation).

The microbioreactors were adapted to more complex fer-mentation processes, for example, chemostat and fed-batch,by adding additional features. The device in Figure 1B wasused for pH control and fed-batch operations.17 The reactorchamber was made of two PMMA layers (L1 and L2 inFig. 1B;GoodfellowCorp., PA) thermally bonded at a temper-ature of 140 �C for 90 min in a homemade press. In this press,a pair of Belleville disc springs (diameter: 119.0 mm; thickness:1.25 mm; height: 2.80 mm, MSC Industry Supply Co., Inc.,New York) maintained a constant load of 0.12 MPa.

The major feature for this microbioreactor is the abilityof base-, acid-, and glucose-additions during fermenta-tion, closed-loop controlled by external microvalves(INKX0514300A, The Lee Co., Westbrook, CT) and controlcircuits (IECX0501350A, The Lee Co.). Microfluidic chan-nels (M), machined in layer L1 and sealed by layer L2, wereused for water- (P1), base- (P2), and acid-feeding (P3), inoc-ulation (P4), and waste outflow (P5; used during inoculation)purposes, respectively.

Figure 1C was used for continuous culture of bacterial cellsto obtain chemostat data (data shown in Fig. 1B).14 Themicrobioreactor chamber (10 mm in diameter and 2 mm indepth, with a total volume of 150 mL) and four connectingchannels were fabricated in three thermally bonded bottom

PMMA layers (layers L1 through L3; Goodfellow Corp.).The thin PDMS aeration membrane was covered with an ad-ditional layer of stainless steel grid (B-PMX-062, Small PartsInc., Miami, FL) fixed by a homemade PDMS O-ring toprovide a perforated membrane structure and to avoid mem-brane bulging. Microfluidic channels (M) were machined inboth sides of layer L2 and sealed by layers L1 and L3. Thesemicrofluidic channels allow for medium addition (P1), inocu-lation (P2), waste exit (P3, used during inoculation), and celloutflow (P4) before and during chemostat experiments. Anexternal syringe pump (PhD2000, HarvardApparatus, Hollis-ton, MA) connected to P1 provided a steady media feedingduring continuous culture experiments.

The latest microbioreactors (Figs. 1D and 2) improve in-terfacing of the microbioreactor to the external instruments(fluidic handling, optical monitoring, electronics, and com-puter control) without modifying the biological function ofprevious designs. The devices are made of five thermallybonded PMMA layers. Precise thermal bonding of PMMAlayers with different glass transition (TG) temperatures wasperformed in a mechanical press in two steps. First, the threebottom layers (layers L1 through L3; Goodfellow Corp.)were bonded at a temperature of 140 �C for 90 min. Second,the resulting composite was bonded with the top two layers(A and B; MSC Industrial Supply) at a lower temperatureof 120 �C for 60 min. The reactor chamber, fabricated inlayers L2 and L3, had the same shape and size as in the pre-vious designs. On the top of reactor chamber, the thin PDMSaeration layer was held by a 3-mm thick PDMS gasket layer

JALA June 2007 145

Original Report

Figure 2. (A) Overview of individual parts for the microbioreactor shown in Figure 1D. (B) Top view photograph of assembled andbonded microbioreactor.

(part G in Fig. 2, 20 mm inner diameter, 25 mm outer diam-eter, fabricated in a polycarbonate mold) to facilitate deviceassembly and sealing.

A PMMA ‘‘cork’’ (Fig. 1D) (F) with an outer diameterslightly larger (13 mm larger in diameter) than the inner diam-eter of PMMA housing frame (machined in A and B) createda seal by compressing the PDMS (G) and a Sylastic RTV sil-icone elastomer (Dow Corning) O-ring (H, inner diameter of10 mm, outer diameter of 20 mm, and height of 3 mm madewith a polycarbonate mold). A small hole in the cork alsoaligned the optical fiber (I) for the OD transmission mea-surements. This cork allowed the microbioreactor to becleaned and reused by replacing PDMS membranes afterexperiments, a useful feature in exploratory studies, but toolabor intensive for parallel investigations. Therefore, in thedisposable bioreactor version the PDMS membrane was per-manently fixed between PMMA layers after thermal bondingand the stabilizing grid structure was molded in the PMMA

146 JALA June 2007

layer B. Similar to previous microbioreactors (Fig. 1AeC),the fluorescence lifetime sensors are integrated at the bottomof device chamber (Fig. 1D) for pH and DO online measure-ments. Indentations beneath these sensors (L1 and L2) nowcouple optical fibers (O) to external fluorescence detectorsand lifetime measurements for better alignment.

The fluidic interface between a microbioreactor andexternal fluidic units was composed of custom-made elastomerO-rings (N) integrated in the PMMA device (Fig. 1D). Theyallowed aseptic self-sealing, ‘‘plug-in-and-run’’ functionality,between external fluid handling and internal microfluidicchannels for inoculation, reagent feed, sampling, and wasteoutflow. The O-rings were cast in Sylastic RTV silicone elasto-mer (Dow Corning) from stainless steel molds with an outerdiameter of 4.2 mm, an inner diameter of 0.2 mm, and adepth of 4.6 mm. The stainless steel molds were fabricatedby using a 2-mm diameter ball-head endmill (MSC IndustrialSupply) at a high rotation speed of 8000 rpm and subsequent

Original Report

electro-polishing to obtain a smooth surface. TheO-rings werecured at room temperature for more than 12 h to obtaina Young’s modulus of 4.5 MPa. They were then embeddedinto housings machined in a thick PMMA layer (L4, Fig. 1B)andwere fixed by a cover PMMAlayer (A)when the two layerswere thermally bonded. The housing for each O-ring hada slightly smaller diameter (4.1 mm) and depth (4.45 mm) sothat the O-ring was compressed and the center hole sealed;1.5-mm diameter through-holes in the covering PMMA layercorresponded to the center positions of the O-rings. Whenstainless steel tubes (12 mm long, 23 gages, Small Parts, Inc.)were inserted, the elastomer O-ring expanded and making afluidic connection. The process was reversible and the seal leaktight up to pressures of w0.6 MPa.

EXPERIMENTAL SETUP AND OPTICAL INTERFACE

DO, pH, and OD600 nm were measured by the optical sensingmethods already described by Zanzotto et al.12 so only a briefsummary is given. The experimental setup for a single fer-mentation run is illustrated in Figure 3. Fermentations werecarried out by placing the microbioreactor in an aluminumchamber maintained at 37 �C by flowing heated waterthrough the chamber base. Bifurcated optical fibers

(custom-made by RoMack Fiber Optics, Williamsburg,VA) led into the chamber from both the top and the bottomand connected to LEDs and photodetectors (PDA-55, Thor-labs, Newton, NJ) to perform the optical measurements. Bio-mass was followed by OD600 nm data obtained froma transmission measurement using an orange LED (EpitexL600-10 V, 600 nm, Kyoto, Japan). The bifurcated branchprovided a reference signal to compensate for any intensityfluctuations of the orange LED. Both DO and pH were mea-sured using phase modulation lifetime fluorimetry. The DOand pH sensors were excited with a blue-green LED(505 nm, NSPE590S, Nichia America Corporation, Mount-ville, PA) and a blue LED (465 nm, NSPB500S, Nichia),respectively. Excitation band pass filters (Omega OpticalXF1016 and XF1014) and emission long pass filters (OmegaOptical XF 3016 and XF 3018) separated the respective exci-tation and emission signals to minimize cross-excitation.Data switches (8037, Electro Standard Laboratories, Cran-ston, RI) multiplexed the output signal and the input signalof the function generator (33220A, Agilent Technologies,Palo Alto, CA) and the lock-in amplifier (SR 830, StanfordResearch Systems, Sunnyvale, CA). LabVIEW software (Na-tional Instruments Corp., Austin, TX) enabled automatedand real-time measurement of the parameters.

Figure 3. Illustration of the measurement setup for microbioreactor. Dashed lines indicate optical fibers, and solid lines show electronicwires and fluid tubes.

JALA June 2007 147

Original Report

Based on this design, a multiplexed system for paralleloperation of four microbioreactors was reported by Szitaet al.15 In this system, a stepper motor carried an opticalbracket to scan over the microbioreactors in stop-and-go se-quences executed by computer-controlled algorithms. Theprocess parameters were measured and recorded for each re-actor using lifetime fluorescence and absorbance methods ina sequential mode. The number of experiments was increasedfor the same external instruments, such as the lock-in ampli-fier and function generator. However, the scanning/readingspeed for the individual microbioreactor reactor ultimatelylimited the possible number of parallel experiments.

As an alternative, a self-aligning strategy was developedfor the optical interface between external measurement setupand disposable fluorescence sensors inside of the micro-bioreactor. Integrated microlens and optical connectors(Fig. 4A) were molded out of PDMS in an aluminum moldfabricated by conventional milling using a 2-mm diameterball-head endmill (MSC) and mechanically polished usinga shaft grinder kit (Dremel, MSC), cotton swabs, and polish-ing paste (Novus Plastic Polish, MSC). The aluminum moldwas composed of two parts. The bottom part of the mold en-sured the replication of the negative shape of the microlens

148 JALA June 2007

and four pillars aligning the plug to the microbioreactor.The upper part contained 10-mm long columns aligned con-centrically to the lenses for casting the PDMS plug housing,guiding, and fixing the optical fibers.

For pH or DO measurements two optical fibers (plasticfibers 1 mm and 0.6 mm in diameter, BFL 37-1000 fromThorlabs, Inc. and FVA500550590 from Polymicro Technol-ogy LLC, Phoenix, AZ, respectively) were housed in the cor-respondent cavity fabricated in the PDMS plug (Fig. 4A).The fibers were secured by PDMS added after fiber insertionand held in place by a support. The PDMS was cured and fi-nally epoxy was cast on top to give mechanical stability tothe plug. These fibers replaced the expensive, customer-madebifurcated optical fibers used in earlier work.12,15 At the samefabrication process, one fiber (0.6 mm in diameter, PolymicroTechnology LLC) was fixed in the center of plug for the ODreading. The optical fibers were placed at specific axial dis-tances from the hemispherical outer surface shape to focuson the optical sensors in the bioreactor. Different focalpoints were designed in the optical plug for the differentDO, pH, and OD measurements. Calibration and optimiza-tion of positions for pH and DO reading maximized theintensity of light from the fluorescent sensor to improve the

Figure 4. (A) Picture of microlens and alignment ring around the microlens (top). Cross-section showing three microlenses assembled withoptical fibers. P represents the PDMS alignment pins. (B) Focusing by a PDMS microlens. Light intensities at different lateral location froma cleaved fiber (green line) and an optical microlens (blue line). Smooth curves from two-dimensional ray tracing model. (C) Photograph ofoptical microlenses assembled with optical fiber housing. Fibers shown are 0.25 mm in diameter. Epoxy and external mechanical support arenot shown.

Original Report

signal-to-noise ratio of measurements. For the OD measure-ment, the design aimed to obtain parallel light transmissionthrough the reactor and good detection on the other sideof the device. To guide the design, a two-dimensional modelwas developed using ray optics theory. Images of the lightdistribution showed good agreement with the model(Fig. 4B). The measurements show a maximum light intensityfrom the plugs, which is 50% higher than from the cleavedoptical fibers and is significantly more focused with respectto lateral distribution of light. Even in presence of slight mis-alignments away from the center, focusing by the microlensesmade coupling of light more efficient compared to butt-endcoupling techniques.

Another important feature of the connector, the align-ment ring around the microlens (cf. Fig. 4A) was fabricatedwith a size comparable to the size of the cavity beneath themicrobioreactor (1.6 mm from the center, 1 mm high, and0.4 mm wide). The mold was fabricated using a 0.4-mmdiameter ball-head endmill, TR-2-0130-BN, PerformanceMicro Tool, Janesville, WI). The alignment ring helped posi-tioning of optical fibers to optical sensors, protection micro-lens, and made setup procedures significantly easier andfaster (Fig. 4C).

BIOLOGICAL EXPERIMENTS IN MICROBIOREACTORS

Applications ofmicrobioreactors in batch,13e15 continuous cul-ture,13e15 and fed-batch17 operations have been demonstratedand are briefly summarized here. In the experiments shown inFigures 5e7, E. coli FB21591 (thiC::Tn5-pKD46, KanR,University of Wisconsin) was used as the model organism.

Figure 5 shows reproducible E. coli batch culture13 inLuriaeBertani (LB) medium containing 8 g/L glucose (Mal-linckrodt, Hazelwood, MO, USA), 100 mg/L kanamycin(Sigma-Aldrich), and 0.1 mol/L 2-(N-morpholino)ethanesul-fonic acid (MES) (Sigma-Aldrich). At the beginning, theE. coli cells started to grow immediately without observablelag phase. Consistent with the high growth rate of cells dur-ing exponential growth phase, the high demand for oxygenwas reflected by the rapid decrease in DO level in the culture

Figure 5. E. coli FB21591 batch culture in LB medium containing8 g/L 100 mg/L kanamycin, and 0.1 mol/L MES (Fig. 1A microreactors).

medium. The DO in the microbioreactor completely depleteat around 1.5 h of the fermentation and does not recover un-til after 5 h, when the cells enter late-stationary phase and thenutrients become limited. During this period, as a conse-quence of anaerobic fermentation the pH level in the mediumdecreases significantly until cells reach stationary phase.

In Figure 6, a fed-batch experiment17 started with the in-oculum of E. coli FB21591 in LB medium containing 2 g/Lglucose and 0.1 mol/L MES. Delivery of 40-g/L glucoseand 1-mol/L NaOH solutions were feed-back controlledbased on DO and pH readings, respectively. For glucoseaddition, when the DO value in the microbioreactor, repre-sented by a phase shift reading, became higher than the37% saturation, a constant volume of 1.34 mL glucose solu-tion was fed into the microbioreactor. In the early stage ofthe experiment, cells used glucose and LB, sequentially, asthe carbon source to support growth, which made the pHvalue first decrease (compensated by base feeding) and thenincrease. This fermentation process lasted about 10 h beforeDO recovered. After 10 h in the experiment, a periodic

Figure 6. Glucose feeding fed-batch cultivation of E. coli in LBmedium containing 8 g/L 100 mg/L kanamycin, and 0.1 mol/LMES (Fig. 1B microreactors). Feeding medium contained 40 g/Lglucose and the DO setpoint was 37% of air saturation level.

Figure 7. Continuous cultivation of E. coli in microbioreactors(Fig. 1C microreactors). Feeding MOPS medium contained 2 g/L100 mg/L kanamycin, and 0.1 mol/L MES.

JALA June 2007 149

Original Report

Figure 8. Schematic of multiplexed bioreactor system for high-throughput bioprocess developments. Microbioreactors are plugged in toplatform in the middle chamber. Solid lines denote optical fibers and thick dashed lines indicate electronic wiring. Signal conditioning elec-tronics, control units, and power supplies are in the bottom portion.

pattern of depletion-recovery in the DO indicated the effectby glucose feeding on cell metabolism in the microbioreactor.In each cycle, DO recovered when carbon sources in the cul-ture medium were depleted and quickly dropped to zero afterthe glucose-feeding valve was actuated and glucose was fedinto the microbioreactor.

Figure 7 demonstrates the continuous culture14 of E. coliin MOPS minimal medium (Teknova, Inc., Hollister, CA)containing 2 g/L glucose, 100 mg/L kanamycin, and0.1 mol/L MES. Steady states were reached when mediumfeeding rates were increased from 0.5 mL/min to 1 mL/minand to 1.5 mL/min, sequentially. Steady state conditions weremaintained for at least eight turnovers at each dilution rate.Steady DO levels observed were about 94%, 77%, and 56%,respectively. Lower DO levels at higher dilution rates are di-rect indications of faster growth and metabolism rates. Aer-obic metabolism in the microchemostat makes pH level inthe culture medium relatively stable at different dilution ratesdue to sufficient oxidative catabolism and the pH buffercapacity from phosphates in the MOPS medium. As the mea-surement for biomass concentration, OD level remained ata stable level of w1 (biomass concentration of w0.46 g celldry weight/L).

CONCLUSIONS

The goal of this study was to develop a practical, user-friendly bench-scale system to meet the needs of high-throughput bioprocess developments. The microbioreactorplatform was designed as two components: disposable micro-bioreactors and a fixed housing device, which includes expen-sive optical components and instruments. Having previouslydemonstrated a multiplexed microbioreactor platform formicrobial cell cultivation,15 the present efforts are focusedon the packaging, integration, and optimization of the micro-bioreactor to make it compatible with the high-throughputapproach.

Microfluidic connectors and PDMS microlens opticalconnectors were designed and integrated to serve as ‘‘plug-in-and-run’’ interfaces with external fluidic and optical

150 JALA June 2007

instruments, respectively. These connectors not only stream-line and simplify the setup procedures for multiple micro-bioreactors, but also help to increase the accuracy andreproducibility in measurements. The bioreactor chips con-sist of multiple thermally bonded PMMA layers with an em-bedded PDMS aeration membrane. The bioreactor design,with its aeration membrane, magnetic stirring, local temper-ature control, and fluorescent sensors, provides flexibilityfor different bioprocessing applications, including batch,fed-batch, and continuous operations. With the plug-in-and-run interface, the disposable bioreactor could be imple-mented in a next generation multiplexed reactor platformthat avoids the moving parts associated with the previous de-sign.15 Such a parallel system could be envisioned to consistof a platform containing fluid handling, optical measure-ments, data acquisition, and power supplies (Fig. 8). The dis-posable microbioreactor would plug into this platform as theonly replacement part. Previous experience with individualor multiplexed microbioreactors has shown that experimentswith w100 mL volumes reproduce growth kinetics as well asgene expression profiles observed in 0.5 L bench-scale sys-tems.18,19 Thus, microreactor technology promises to enablefaster, less labor intensive bioprocess development, withmuch reduced reagent usage and waste generation.

ACKNOWLEDGMENT

The authors acknowledge the DuPont-MIT Alliance (DMA) for funding as

well as Ruben Kolfschoten and Patrick Boyle for help with the data in

Figure 6.

REFERENCES

1. Walther, I.; van der Schoot, B. H.; Boillat, M.; Cogoli, A. Microtechnol-

ogy in space bioreactors. Chimia 1999, 53, 75e80.

2. Walther, I.; van der Schoot, B. H.; Boillat, M.; Cogoli, A. Performance

of a miniaturized bioreactor in space flight: microtechnology at the ser-

vice of space biology. Enzyme. Microb. Technol. 2000, 27, 778e783.

3. Walther, I.; van der Schoot, B. H.; Jeanneret, S.; Arquint, P.; de Rooij,

N. F.; Gass, V.; Bechler, B.; Lorenzi, G.; Cogoli, A. Development of

a miniature bioreactor for continuous culture in a space laboratory. J.

Biotechnol. 1994, 30, 21e32.

Original Report

4. Maharbiz, M. M.; Holtz, W. J.; Howe, R. T.; Keasling, J. D. Microbior-

eactor arrays with parametric control for high-throughput experimenta-

tion. Biotechnol. Bioeng. 2004, 85, 376e381.

5. Maharbiz, M. M.; Holtz, W. J.; Sharifzadeh, S.; Keasling, J. D.; Howe,

R. T. A microfabricated electrochemical oxygen generator for high-

density cell culture arrays. J. MEMS 2003, 12, 590e599.

6. Lamping, S. R.; Zhang, H.; Allen, B.; Shamlou, P. A. Design of a proto-

type miniature bioreactor for high throughput automated bioprocessing.

Chem. Eng. Sci. 2003, 58, 747e758.

7. Harms, P.; Kostov, Y.; Rao, G. Bioprocess monitoring. Curr. Opinion

Biotechnol. 2002, 13, 124e127.

8. Kermis, H. R.; Kostov, Y.; Rao, G. Rapid method for the preparation

of a robust optical pH sensor. Analysts 2003, 128, 1181e1186.

9. Kostov, Y.; Harms, P.; Randers-Eichhorn, L.; Rao, G. Low-cost micro-

bioreactor for high-throughput bioprocessing. Biotechnol. Bioeng. 2001,

72, 346e352.

10. Rao, G., Bioreactor and bioprocessing technique. 2002/0025547 A1, 2002.

11. Puskeiler, R.; Kaufmann, K.; Weuster-Botz, D. Development, paralleli-

zation, and automation of a gas-inducing milliliter-scale bioreactor for

high-throughput bioprocess design (HTBD). Biotechnol. Bioeng. 2005,

89, 512e523.

12. Zanzotto, A.; Szita, N.; Boccazzi, P.; Lessard, P.; Sinskey, A. J.; Jensen,

K. F. A membrane-aerated microbioreactor for high-throughput bio-

processing. Biotechol. Bioeng. 2004, 85, 376e381.

13. Zhang, Z.; Szita, N.; Boccazzi, P.; Sinskey, A. J.; Jensen, K. F. A well-

mixed, polymer-based microbioreactor with integrated optical measure-

ments. Biotechnol. Bioeng. 2006, 93, 286e296.

14. Zhang, Z.; Boccazzi, P.; Choi, H.-G.; Perozziello, G.; Sinskey, A. J.; Jen-

sen, K. F. Microchemostatdmicrobial continuous culture in a polymer-

based, instrumented microbioreactor. Lab Chip 2006, 6, 906e913.

15. Szita, N.; Boccazzi, P.; Zhang, Z.; Boyle, P.; Sinskey, A. J.; Jensen, K. F.

Development of a multiplexed microbioreactor system for high-through-

put bioprocessing. Lab Chip 2005, 5, 819e826.

16. Lee, H. L. T.; Boccazzi, P.; Ram, R. J.; Sinskey, A. J. Microbioreactor

arrays with integrated mixers and fluid injectors for high-throughput ex-

perimentation with pH and dissolved oxygen control. Lab Chip 2006,

6(9), 1229e1235.

17. Zhang, Z. Microbioreactors for Bioprocess Development. PhD, Massa-

chusetts Institute of Technology, 2006.

18. Boccazzi, P.; Zhang, Z.; Kurosawa, K.; Szita, N.; Bhattacharya, S.; Jen-

sen, K. F.; Sinskey, A. J. Differential gene expression profiles and real-

time measurements of growth parameters in Saccharomyces cerevisiae

grown in microliter-scale bioreactors equipped with internal stirring. Bi-

otechnol. Prog. 2006, 22(3), 710e717.

19. Boccazzi, P.; Zanzotto, A.; Szita, N.; Bhattacharya, S.; Jensen, K. F.;

Sinskey, A. J. Gene expression analysis of Escherichia coli grown in min-

iaturized bioreactor platforms for high-throughput analysis of growth

and genomic data. Appl. Microbiol. Biotechnol. 2005, 68(4), 518e532.

JALA June 2007 151