Optimization of the microbial synthesis of dihydroxyacetone in a semi-continuous repeated-fed-batch process by in situ immobilization of Gluconobacter oxydans

Optimization of the microbial synthesis of dihydroxyacetone in a semi-continuous repeated-fed-batch process by in situ immobilization of Gluconobacter oxydans
of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Optimization of the microbial synthesis of dihydroxyacetonein a semi-continuous repeated-fed-batch process by insitu immobilization of   Gluconobacter oxydans D. Hekmat*, R. Bauer, V. Neff   Institute of Chemical Engineering, Munich University of Technology, Boltzmannstrasse 15, 85747 Garching, Germany Received 23 March 2006; accepted 9 July 2006 Abstract The effect of in situ immobilization of   Gluconobacter oxydans  on a novel carrier material in a repeated-fed-batch operated packed-bed bubble-column bioreactor for the production of the fine chemical dihydroxyacetone was investigated experimentally. The carrier material werebiocompatible, durable, coated Ralu-rings. The coating was a porous silicone matrix with satisfactory wetting characteristics. Settling of cellswas relatively rapid. The cells were protected from abrasion caused by shear forces. A sufficiently high oxygen supply rate to the immobilized cellswas provided due to the high oxygen permeability of the silicone matrix. The immobilized biomass was estimated to be about 65% of the totalbiomasscontainedinthe bioreactorafter18daysofoperation.Theobservedspace-timeyieldwas approx.76% highercomparedtoa similarprocesswhich was performed without an optimized fermentation medium. Compared to previous experiments with a trickle-bed bioreactor, the space-timeyield was approx. 3.7 times higher. A typical time course of the immobilization process was observed: after an induction phase, a transition phasefollowed which later on gave way to a nearly linear accumulation phase. A stationary phase with regard to the amount of immobilized active cells,however, was not reached. Hence, a higher bioreactor performance than observed could be expected at longer operation times. # 2006 Elsevier Ltd. All rights reserved. Keywords:  Dihydroxyacetone;  Gluconobacter oxydans ; In situ immobilization; Optimization; Repeated-fed-batch process 1. Introduction The conventional process of the industrial microbialproduction of the fine chemical dihydroxyacetone (DHA) via Gluconobacter oxydans  is the fed-batch operation in a stirredtank bioreactor. However, the disadvantages of this operationmode are the necessity of cleaning, sterilization, andinoculation procedures after each fed-batch cycle. A favouredalternative to the fed-batch fermentation is represented by thesemi-continuous repeated-fed-batch operation [1]. The proce-dure for the repeated-fed-batch operation is as follows: Thefermentation with controlled substrate feeding is performeduntil a pre-determined DHA concentration threshold value isreached. This DHA threshold value corresponds to a productconcentration at which the culture is not irreversibly growthinhibited. Then, most of the fermentation broth is removed andthe bioreactor is replenished with fresh medium. The residualbroth volume serves as inoculate for the next cycle. A way tointensify and optimize this process is to increase the amount of active biomass in the reactor via recycling of whole cells. Thisprocedure enables a decoupling of the residence times of cellsand liquid in the bioreactor. The main advantage of cellrecycling is the achievement of a higher space-time yield viathe increased active biomass density in the bioreactor. Suchapproaches using whole cell recycling have already beenpursued for decades for numerous applications in the area of biodegradation. However, in comparison, the number of suchapplications in the area of microbial production processes ismuch smaller. Basically, two concepts exist: (i) the immobi-lization of cells on a suitable carrier material which is containedin the bioreactor or (ii) the recycling of cells in an external loopequipped with a cell separation unit (e.g. a membrane unit or acentrifuge) [2,3]. However, in some of the latter cases,repeatedly observed phenomena like membrane fouling, shearforces, and insufficient oxygen supply may adversely affect theprocess. On the other hand, Merck KGaA, Darmstadt, Germany[4] demonstrated in a patent a promising system with cellrecycling via an external loop. The first mentioned approach Biochemistry 42 (2007) 71–76* Corresponding author. Tel.: +49 89 289 15770; fax: +49 89 289 15714. E-mail address: (D. Hekmat).1359-5113/$ – see front matter # 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.procbio.2006.07.026  was pursued using carrier materials such as porous particles[5,6] or gel beads made of biopolymers [7,8]. However, it turned out that the settling of cells on the interior surface of theporous particles was not satisfactory in many cases due toinsufficient oxygen supply into the deeper biofilm layers.Furthermore, many biopolymer gel beads did not exhibit therequired operational stability. Therefore, cross-linking wasexamined in order to improve the stability of the gel beads [9].A review on the feasibility and operational stability of immobilized microbial cell systems was published by Freemanand Lilly [10]. The utilization of simple surface carriermaterials was examined as early as 1979. Here, an improvedmicrobial production process of citric acid using wood chips asthe carrier material of whole immobilized cells was success-fully demonstrated [11]. Comparable in situ immobilized wholecell systems have been examined in long-term experiments andproved to be stable for several months [12,13,14]. Hence, thefeasibility of the concept was proven. In addition to reactorsystems with immobilized whole cells, successful applicationsusing immobilized enzymes were reported in the literature[15,16]. It was shown in a previous work that highproductivities coupled with high enantiomeric excess valueswere reached [17].Due to the large oxygen requirements of the strain  G.oxydans usedinthepresentstudy,itmustbeensuredthatalargeenough mass transfer to the immobilized cells is maintained atall times. In a previous study, Hekmat et al. [1] showed that theinsituimmobilizationof  G.oxydans ina85 lpilot-scaletrickle-bed reactor was feasible. The utilized carrier material in thiswork were hydrophilized uncoated polypropylene Ralu rings.These carriers yielded a high specific surface area, ahomogeneous radial distribution of fluids, and a low pressuredrop. It was expected that the hydrodynamics of the trickle-bedreactor ensured mild conditions, e.g. small shear forces, inorder to facilitate the settling of cells and the formation of abiofilm. However, it was observed that the biofilm formationrate was relatively small and only parts of the carrier materialwere covered by the biofilm. Hence, the immobilization wasunsatisfactory. However, the immobilization procedure still ledtoanincreaseoftheproductformationrateofabout75%duringa 17 days operation [1]. The aim of the present study was todevelop and test a suitable coating for an enhanced carriermaterial which yields a fast and evenly distributed immobiliza-tion of cells. The immobilization matrix should be robustenough to enable an operation in a packed bubble-columninstead of a trickle-bed reactor. The packed bubble-columnbioreactor is favourable since the portion of fermentation brothis significantly larger compared to a trickle-bed reactor, thus,resulting in higher space-time yields. The requirements for asuitable coating material were as follows: (1) chemically inert,biocompatible, and durable; (2) thermally stable in order towithstand 121  8 C; (3) mechanically stable; (4) highly porous tofacilitate cell immobilization as well as mass transfer into thedeeper layers of the matrix. A coating material with suchproperties has been developed by Muscat [18]. This materialwas a silicone mixture with the addition of salt with a givennarrow particle size distribution prior to polymerisation. Thesalt particles were dissolved after polymerisation to generate asilicone matrix with a given porosity. Furthermore, titaniumoxide was added during the fabrication process in order toimprove the wetting characteristics by reducing the contactangle between liquid and solid phase [19]. Titanium oxide isbiocompatible and chemically inert and thereforewell suitable.A further advantage of the chosen carrier matrix was the factthat silicone exhibits a relatively high oxygen permeability sothat oxygen supply is carried out not only via the liquid phasebut also via the solid silicone matrix.According to the work of Bettin [19], the permeationcoefficient of oxygen in silicone was 3.5  10  10 m 2 s  1 . Thisvalue is about five-fold higher than the permeation coefficientofoxygeninwater.Hence,itcouldbeexpectedfromthechosenset-up that the oxygen limitation effect on the immobilizedobligate aerobic cells of   G. oxydans  was reduced. 2. Materials and methods 2.1. Microorganisms and media G. oxydans  is an obligate aerobic Gram-negative bacterium and belongs tothefamilyofaceticacidbacteria.Thesemicroorganismsareabletosynthesizeavariety ofdifferentpolyols,such as L -sorbosefrom D -sorbitolor D -gluconicacidfrom  D -glucose [20]. This work focuses on the production of DHA fromglycerol. The reaction is catalysed by a membrane-bound dehydrogenaseand is carried out for the purpose of energy generation. The  G. oxydans industrial strain M1136 was used which was kindly provided by Merck KGaA,Darmstadt, Germany. In order to use genetically consistent material, the strainwas conserved at   20  8 C in a sorbitol medium containing the followingcomponents: 2.0 kg m  3 (NH 4 ) 2 SO 4 , 0.1 kg m  3 K  2 HPO 4 , 0.9 kg m  3 KH 2 PO 4 , 1.0 kg m  3 MgSO 4  7H 2 O, 2.5 kg m  3 yeast extract, and 10 kg m  3 sorbitol. The culture was transferred from the cryoculture to flask cultures everymonth. Ohrem [21] reported that CaCl 2  had a positive effect on growth of  G. oxydans . In addition, Bauer [22] discovered that the addition of sorbitol asa secondary substrate had a positive effect both on growth on glycerol and on theDHA production rate. Therefore, for the experiments in the laboratory-scalefermenters, a medium as described above was used which contained a controlledamount of 5–20 kg m  3 glycerol and, in addition, 1.5 kg m  3 CaCl 2 . 2.2. Analytical methods Optical density (OD) was measured at 578 nm with a Lambda 2 spectro-photometer (Perkin Elmer, Wiesbaden, Germany). The samples were centri-fuged at 5000 rpm for 5 min. The following relation resulted after calibration:biomass dry weight (kg m  3 ) = 0.247  OD 578 nm . Glycerol and DHA con-centrations were determined using an on-line high performance liquid chro-matograph (HPLC) Beckman System Gold (Beckman Coulter GmbH, Krefeld,Germany). A Rezex RCM Monosaccharide calcium-column (Phenomenex,Aschaffenburg, Germany) was employed for the measurements. The eluentwas bi-distilled water. The elution flow rate was 0.6 ml min  1 [22]. 2.3. Description of bioreactor system The bioreactor system was a laboratory-scale bubble-column with a heightof 300 mm and an inner diameter of 100 mm modified from equipment whichsrcinatedfromBioengineeringAG,Wald,Switzerland.Thetotalvolumeofthebioreactor was 2 l. The experiments were performed in a fully automated wayby using a microcomputer. Bettin [19] investigated the limiting and inhibitingeffects on growth by glycerol. It was found that the optimum glycerol con-centrationrangeforgrowthandproductformationwas5–20 kg m  3 .Therefore,the substrate glycerol was fed to the repeated-fed-batch process by computercontrol in order to keep the glycerol concentration within this range at all times.The liquid volume was kept nearly constant during the repeated-fed-batch  D. Hekmat et al./Process Biochemistry 42 (2007) 71–76  72  experiments by a correctly set concentration of the glycerol feed in order tocompensate the loss of broth volume due to evaporation. The biomass con-centration was measured manually. Inoculation was carried out with 50 ml of pre-culture from shake-flasks. The pH was set to 5.3 and controlled. Thetemperature was controlled at 30  8 C. The specifications for the repeated-fed-batch process with and without immobilization are given in Table 1. Theperformance of the bioreactor was described by the yield coefficient  Y  DHA/Gly ,the product formation rate  ˙r  DHA , and the space-time yield STY. The yieldcoefficient is defined by Y  DHA = Gly  ¼ V  L ð C  DHA ; 1  C  DHA ; 0 Þ m Gly ; feed ;  (1)where  V  L  is the volume of the liquid contained in the bioreactor,  C  DHA,1  and C  DHA,0  are the DHA concentrations at the end and at the start of eachrepeated-fed-batch fermentation cycle, respectively, and  m Gly,feed  is the massof glycerol feed during the fermentation cycle. The product formation rate isdefined by ˙r  DHA  ¼ C  DHA ; 1  C  DHA ; 0 t  1  t  0 ;  (2)where( t  1  t  0 )isthetimespanofeachfermentationcycle.Thespace-timeyieldis defined asSTY ¼ V  L V C  DHA ; 1  C  DHA ; 0 t  1  t  0 ;  (3)where  V   is the total bioreactor volume. 2.4. Carrier material In this work, polypropylene Ralu-rings made by Raschig AG (Ludwig-shafen, Germany) were used as the carrier material. The specifications of theuncoated Ralu-rings were as follows: diameter = height = 18 mm, specificweight = 92 kg m  3 number of pieces per m 3 = 170,000, specific surfacearea = 320 m 2 m  3 , porosity = 0.94. The Ralu-rings were coated with aporous silicone material based on the works of Muscat [18] and Bettin[19]. A two-components silicone (RTV 2, Bayer AG, Leverkusen, Germany)was used for coating. The RTV 2 silicone consisted of the basic polymer(Silopren U1) and the cross-linking agent (Silopren U 260) which weremixed in the ratio of 94:6. A platinum catalyst (Pt L) was added to the two-components mixture in the ratio Pt L:RTV2 = 1:1000. Subsequently, the RTV2 silicone was mixed with NaCl salt (SOLSEL 3356, Solvay Salz GmbH,Rheinberg, Germany) and with titanium oxide (rutile pigment 2210, KronosTitan GmbH, Nordenham, Germany). Then, the Ralu-rings were coated withthe mixture via submersion and the coating was polymerized at 90  8 C forapprox. 1 h. The salt was then washed out by multiple heating in apressurized water-bath at 120  8 C. 3. Results and discussion 3.1. Production of the silicone-coated Ralu-rings The production procedure of the silicone-coated Ralu-ringswas simple enough and reproducible. After sterilization, thecoated Ralu-rings proved to be thermally and mechanicallystable as well as biocompatible and chemically inert. Themacroporous structure and the capability for oxygen permea-tion of the silicone coating offered the advantage that activecellswereabletosettlealsointothedeeperlayersofthecarriermatrix. The resulting porosity corresponded approximately tothe expected one, i.e. the washout procedure of the NaClcrystals was effective. The desired wettability by addition of titanium oxide was achieved. In Fig. 1, uncoated Ralu-ringsand Ralu-rings coated with the porous silicone matrix areshown.  D. Hekmat et al./Process Biochemistry 42 (2007) 71–76   73Table 1Specifications of the repeated-fed-batch process in the laboratory-scale bubble-column bioreactorOperational modeRepeated-fed-batchoperation withoutimmobilizationRepeated-fed-batchoperation withimmobilizationPacking material – Silicone-coatedRalu-ringsReactor volume,  V   (l) 2.0 2.0Liquid volume,  V  L  (l) 1.5 1.2pH 5.3  0.1Temperature ( 8 C) 30  0.1Aeration rate (m N3 h  1 ) 0.13Glycerol concentration(kg m  3 )5–20 (controlled)DHA-threshold value(kg m  3 )82Residual volume of liquidin the bubble column (l)0.03Fig. 1. (A) Uncoated Ralu-rings; (B) Ralu-rings coated with porous siliconematrix.  3.2. Repeated-fed-batch experiments without immobilization At first, experiments were performed in the repeated-fed-batch mode without immobilization of cells. The DHAthreshold value was set to 82 kg m  3 . This value correspondedto a product concentration at which the culture was notirreversibly growth inhibited [23]. Six reproducible cycles overa time period of 140 h were achieved as depicted in Fig. 2. Theculture had reached a steady state with regard to the productformation rate and the space-time yield. As expected, theculture was completely capable for regeneration at the chosenDHA concentration threshold of 82 kg m  3 . The glycerolconcentration was controlled in the range of 5–20 kg m  3 viasubstrate feeding. The oxygen concentration course of everyrepeated-fed-batch cycle passed through a respiratory mini-mum after approx. 10 h due to the increasing inhibition of DHA. The suspended biomass concentration was measured atthe end of each cycle and reached a nearly constant value of 1.1  0.13 kg m  3 . This low value was a result of the largeoxygen requirement of the obligate aerobic species  G. oxydans .Almost no production of DHAwas observedat the beginningof all fermentation cycles due to lag-phases lasting for about4 h. After these lag-phases, the product formation wasnearly linear with time and reproducible. The yield coefficient Y  DHA/GlY  was 0.85  0.04 kg kg  1 , the product formationrate was 3.7  0.06 kg m  3 h  1 , and the space-time yield was2.8  0.06 kg m  3 h  1 . 3.3. Repeated-fed-batch experiments with in situimmobilization Theseexperimentswereperformedwithapackedbedof130pieces of silicone-coated Ralu-rings with a total displacementvolume of 0.3 l. This corresponded to approx. 25% of the liquidvolume. The silicone-coated Ralu-rings were fixed as anirregular packing above the aeration device of the bubble-column reactor. During cultivation, the silicone-coated Ralu-rings were immersed completely in the culture medium. 50 mlof a shake-flask culture were used as inoculate. The DHAthreshold concentration for the first five cycles was set to65 kg m  3 . These lower DHA values accelerated the immo-bilization process of the cells due to a reduced productinhibition effect. Over a time period of 430 h, 26 cycles wereperformed. The concentration profiles for DHA, glycerol, andoxygen are shown in Fig. 3. Again, reproducible cycles wereobtained. Because of HPLC maintenance operations, concen-tration measurements were not performed between the 320thand the 370th hour. Instead of active fermentation control, aconstant presetreducedfeedrateofglycerolwasappliedduringthis time. The active control of fermentation was started afterthe commencement of the HPLC measurements. The followingfermentation cycles were again reproducible. After the start-upof the process, the lag-phase lasted approx. 4 h. However, thelag-phases decreased with increasing number of cycles. Fromthe 7th cycle on, no apparent lag-phases were observedanymore. The dissolved oxygen concentration decreasedslightly with increasing number of cycles. The reason for thiswas an increasing oxygen demand because of an increasedamount of immobilized active  G. oxydans  cells. During the23rd to the 25th cycle, the suspended biomass concentrationwas measured in the culture medium. At the end of the cycles,the suspended biomass reached a concentration of 0.6  0.04 kg m  3 . Thus, compared to the repeated-fed-batchcycles without cell immobilization, the suspended biomassconcentration was reduced by about 54%. The amount of immobilized cells was estimated using the following assump-tions: (i) the product formation rate was proportional to thebiomass concentration, and (ii) the product formation rates of suspended cells and immobilized cells were identical. Usingthese assumptions, the concentration of immobilized cells after18daysofoperationamountedto1.14 kg m  3 which wasabout65% of the total biomass. Again, this low value was a result of the large oxygen requirement of the obligate aerobic species G. oxydans . The yield coefficient  Y  DHA/GlY  was 0.87  0.05 kg kg  1 and was nearly constant during the cultivationtime. During 18 days of operation, the product formation rateincreased from 3.5 to 5.9 kg m  3 h  1 and the space-time yield  D. Hekmat et al./Process Biochemistry 42 (2007) 71–76  74Fig. 2. Concentration profiles of DHA, glycerol, and oxygen of the repeated-fed-batch process without cell immobilization in the laboratory-scale bubble-column.Fig. 3. Concentration profiles of DHA, glycerol, and oxygen of the repeated-fed-batch process with cell immobilization in the laboratory-scale bubble-column (maintenance of the HPLC between the 320th and the 370th hour).  increased from 2.1 to 3.7 kg m  3 h  1 . Compared to theexperiments without immobilization, the space-time yieldincreased by approx. 32%. The time course of the space-timeyield with increasing cycle number of the repeated-fed-batchprocess with immobilization is shown in Fig. 4. This course canbe divided into three sections: from the start-up to the 5th cycle,no significant change of the space-time yield was observed.After a transition phase of about another five cycles, anapproximately linear increase of the space-time yield wasobserved.Thisbehaviorwaswellinaccordancewithpreviouslyreportedexperimental observations by Flemming[24].Thefirstphase was the induction phase where the colonies of cellsstarted to settle in the immobilization matrix. This phase lastedabout 100 h. The second phase was the transition phase whichwas governed by adaptation of the cells due to theimmobilization process and due to the increase of the DHAthreshold value from 65 to 82 kg m  3 . The accumulation phasebegan after the 10th cycle. The nearly linear increase of thespace-time yield during this phase was due to the biofilmaccumulation. According to Flemming [24], the accumulationphase was followed by a so-called stationary phase, where theamount of accumulating cells in the biofilm is equal to theamount of cells lost by inactivation, lysis, shear forces, or otherphenomena. However, this phase was not reached during thepresented experiments. Nevertheless, the immobilization of   G.oxydans  led to a significant improvement of the space-timeyield of the process. In Table 2, a comparison of the achievedbioreactor performances with previous experimental results isgiven. As expected, the experiments with the trickle-bedbioreactor led to the lowest space-time yields. The DHAthreshold had been set to 60 kg m  3 in these experiments.Bauer [22] had performed experiments in an identicallaboratory-scale packed-bed bubble-column bioreactor as usedin the present study, again with a DHA threshold value of 60 kg m  3 . However, the pH was set to 4.5 instead of 5.3 asusedin the present experiments. The lower pHapparentlyled toa higher yield coefficient  Y  DHA/GlY  of 0.95 kg kg  1 . Further-more, the medium had not been optimized by the addition of CaCl 2 andthesecondarysubstratesorbitol.Theresultingspace-time yield was 2.1 kg m  3 h  1 . The present experiments wereperformed using a higher DHA threshold value of 82 kg m  3 .Here, the highest space-time yield of 3.7 kg m  3 h  1 wasmeasured usingsilicone-coated Ralu-rings. The results indicatethat the optimization of the medium had a significant effect.Hence, compared to the experiments without optimizedmedium, the space-time yield was raised by about 76%. 4. Conclusions Inthe present work, the effect ofinsitu immobilization of   G.oxydans  on a carrier matrix in a repeated-fed-batch operatedpacked-bed bubble-column bioreactor was studied. A noveltype of silicone-coated Ralu-ring carrier was used. Advantagesof the porous carrier matrix were: (i) ample space for thesettling of microorganisms, (ii) the protection of the cells fromabrasion caused by shear forces, (iii) a sufficiently high oxygensupply rate due to the high oxygen permeability of the utilizedsilicone matrix, and (iv) satisfactory wetting characteristics dueto the addition of titanium oxide during the fabrication process.The carrier matrix proved to be biocompatible, durable,mechanically stable, and chemically inert. The experimentalresults indicate that the immobilized biomass amounted toapprox. 65% of the total biomass contained in the packed-bedbubble-column bioreactor after 18 days of operation. Thespace-time yield was approx. 76% higher compared to asimilar process which was performed without an optimizedfermentation medium. Compared to previous experiments with  D. Hekmat et al./Process Biochemistry 42 (2007) 71–76   75Fig. 4. Increase of the space-time yield of the repeated-fed-batch process withcell immobilization in the laboratory-scale bubble-column with increasingcycle number.Table 2Comparison of the achieved bioreactor performances with previous experiments with and without cell immobilizationOperational modePilot-scale trickle-bedbioreactorLaboratory-scale bubble-columnbioreactor[1] [1] [22] This study This studyDuration of operation (days) 30 17 17 6 18Optimized medium No No No Yes YesType of immobilization – UncoatedRalu-ringsSilicone-coatedRalu-rings– Silicone-coatedRalu-ringsFinal DHA concentration,  C  DHA,1  (kg m  3 ) 60 60 60 82 82Yield coefficient,  Y  DHA/GlY  (kg kg  1 ) 0.80 0.80 0.95 0.85 0.87Maximum product formation rate,  ˙r  DHA  (kg m  3 h  1 ) 1.6 2.8 3.4 3.7 5.9Maximum space-time yield, STY (kg m  3 h  1 ) 0.57 1.0 2.1 2.8 3.7
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks