Biotechnol. Prog. 1990, 6,210-213
210
Use of Immobilized Microbial Membrane Fragments To Remove Oxygen and Favor the Acetone-Butanol Fermentation Francesc Godia Unitat d’Enginyeria Quimica, Universitat Autonoma de Barcelona, Beliaterra 08193, Barcelona, Spain
Howard I. Adler,+Charles D. Scott, and Brian H. Davison* Oak Ridge National Laboratory,$ P.O. Box 2008, Oak Ridge, Tennessee 37831
Oxygen-reducing membrane fragments obtained from Escherichia coli were used with Clostridium acetobutylicum (C. acetobylicum) to provide an oxygen-free microenvironment for the conversion of glucose to acetone, butanol, and ethanol (ABE). The batch fermentation of suspended C. acetobutylicum NRRL-B-643 and its ability to produce solvents in the presence of membranes as the oxygen-elimination agent are described and compared with the conventional sparging technique used to maintain anaerobiosis. The use of membrane fragments to remove oxygen for fermentation by C. acetobutylicum was successful and gave slightly improved results over the use of sparging with regard to lag, biomass, and solvent production (e.g., final butanol concentration of 3.25 and 2.7 g/L, respectively). Solvent production is also reported for a continuous columnar reactor with coimmobilized cells and membranes in K-carrageenan gel beads and air-saturated liquid feed.
Introduction The use of anaerobic microorganisms in a fermentation system requires an efficient removal of oxygen in the feed medium and the maintenance of an anoxic environment throughout the fermentation. Currently, the manipulation of anaerobic microorganisms is limited by the use of cumbersome physical and chemical techniques for anaerobiosis. Oxygen can be removed fairly efficiently by sparging with nitrogen, but many liquid media are subject to foaming, which presents mechanical difficulties. Oxygen can also be removed from the liquid media by the addition of reducing agents, but most of them, or their byproducts, are toxic or inhibitory to the microorganisms. It has been previously reported that a partially purified membrane fraction from Escherichia coli contains an active electron transport system that can efficiently reduce dissolved oxygen to water in various fermentation media in the presence of a mild hydrogen donor such as sodium lactate, therefore enabling the manipulation of many anaerobic strains (Adler and Crow, 1981; Adler et al., 1983). The sterile membrane fragments can be used without foaming problems, and they have no inhibitory effects on the growth of bacteria, even when they are present in 10-fold excess concentration (Adler and Crow, 1981). A further advantage of the membrane fraction is that, as an enzymatic system, it can be used continuously (assuming that the necessary lactate concentration is present in the medium). Strict anaerobic con-
* To whom correspondence should be addressed. + Current affiliation:
Oak Ridge Associated Universities, P.O. Box
117, Oak Ridge, TN 37831.
Operated by Martin Marietta Energy Systems, Inc., under Contract DE-AC05-840R21400 with the U.S. Department of Energy. Work performed at Oak Ridge National Laboratory. f
ditions can be maintained even if more oxygen is introduced into the medium. The production of acetone and butanol by fermentation, using different strains of Clostridium sp, which are anaerobic bacteria, has again received attention because of the recent interest in renewable resources as feedstocks to obtain chemical products and fuels by fermentation (Ennis et al., 1986; Jones and Woods, 1986). One of the problems associated with this fermentation is the severe inhibition by the products, especially butanol, which limits the final product concentration and keeps the productivity low (Ennis et al., 1986). The lactate has also been reported to have a deleterious effect (Bahl et al., 1986). The use of immobilized cells to provide high cell densities and productivities may overcome this problem and allow a continuous process (Ennis et al., 1986). In the present work, the incorporation of microbial membrane fragments in K-carrageenan beads entrapping Clostridium acetobutylicum (C. acetobutylicum) was studied. The main objective was to test an effective way to maintain a completely anoxic microenvironment in the beads, thus permitting continuous acetone-butanol, anaerobic fermentations without any need to remove oxygen from the medium by physical or chemical means.
Materials and Methods Microorganism. The strain C. acetobutylicum NRRLB-643 was used in this study. It was maintained at room temperature in a potato medium consisting of 5 g of diced Idaho potato, 0.15 g of CaC03, and 12 mL of water autoclaved in a screw-cap tube. Microbial Membrane Fraction. A detailed procedure for the production of the oxygen-reducing membranes has already been published (Adler et al., 1981). Such membranes are now available as “Oxyrase” from Oxyrase, Inc., Box 899, Ashland, OH 44805.
8756-7938/90/3006-0210$02.50/00 1990 American Chemical Society and American Institute of Chemical Engineers
Biotechnol. Prog., 1990, Vol. 6, NO. 3
Fermentation Medium. The fermentation medium used was a modification of that described by Monot et al. (1982). The composition per liter of distilled water was MgS04*7Hz0,0.2 g; MnS04.H20,0.01 g; F e S 0 ~ 7 H z 0 , 0.01 g; NaC1, 0.01 g; PABA, 0.001 g; biotin, 0.01 mg; thiamine, 0.001 g; KzHP04, 0.25 g; CH3COONH4, 0.55 g; glucose, 40 g; and sodium lactate, 1.86 mL of 60% syrup (10 mM). The sugar and salts are autoclaved separately and combined aseptically. The initial pH was adjusted with KOH, usually to pH 6.5. Analysis Methods. In suspended culture, the cell concentration was determined measuring absorbance at 610 nm with a Schimadzu UV-160 spectrophotometer against a water standard. In the case of immobilized cells, a known number of beads were dissolved in 1% sodium citrate, and the cell concentration was determined microscopically. Glucose was analyzed with use of a YSI Model 27 glucose analyzer. Fermentation products, ethanol, acetone, butanol, and acetic and butyric acids, were determined by gas chromatography in a Hewlett-Packard 5890 by a flame ionization detector with a capillary column (Megabore DB-1; 0.53-mm i.d., 30-m length) and a helium carrier. Injector and detector temperatures were 250 "C. The oven temperature was 40 "C for the first 3 min, and then a ramp of 10 "C/min up to 70 "C was used. Oxygen measurement was carried out to k0.03 mg/L with YSI-5739 probes coupled to a YSI Model 58 dissolvedoxygen meter, except for the probe used in the middle section of the tubular, continuous fermenter described in following text. There a smaller probe, built as indicated by Johnson et al. (1967), was used, coupled to a New Brunswick DO 50 dissolved-oxygen analyzer. All probes were calibrated a t air saturation (6.72 mg/L) at 35 "C. Immobilization Method. Beads of 4% (w/v) K-carrageenan (NSAL 798 from FMC Corp.) were made by a previously described technique (Scott, 1987),under anaerobic conditions. Membrane fragments and/or cells were added to the solution before the fixing or polymerization of the beads. Oxygen Removal with Immobilized Membrane Fragments. In order to characterize the oxygenremoval capacity of membrane fragments entrapped in K-carrageenan beads, several experiments were performed using a 350-mL vessel. Experiments were performed a t various pH and membrane-fragment concentrations. In these experiments, a known number of membrane-containing beads were added to a given volume of air-saturated water containing 10 mM sodium lactate, and oxygen depletion as a function of time was followed over successive challenges of air. Batch Fermentation. Batch fermentation experiments were performed with free cells in the same 350-mL fermentor. These experiments compared the use of free anaerobic membrane fragments to reduce oxygen levels with the conventional method of gas sparging. A control experiment without any attempt to remove oxygen from the medium was also performed. The gas mixture of Nz and 5 % COZ was introduced through a glass-frit air diffuser. The pH was measured throughout the experiments. The temperature was 35 "C, and the agitation speed was 100 rpm. When needed, 7 mL of the membrane fragment suspension was added to the medium for a final concentration of 2 % (v/v) membrane suspension. Continuous Fermentation. Continuous fermentations with cells and oxygen-reducing membrane fragments coimmobilized in K-carrageenan beads were car-
211
Figure 1. Schematic diagram of the experimental setup for continuous fermentation: A, feed tank, stirred; B, feed pump; C, measurement chamber (feed dissolved oxygen and pH); D, column; E, bead retention grid; F, effluent pump; G, effluent tank; D o l , D02, D03, dissolved oxygen probes; pH1, pH2, pH3, pH probes; S1, S2, S3,sampling ports; WI, thermostated water inlet; WO, thermostated water outlet.
ried out in the fluidized-bed jacketed columnar reactor shown schematically in Figure 1. The internal diameter in the cylindrical part was 2.54 cm, the tapered section in the bottom reduced to 1.28 cm, and the total liquid volume, including the expansion section, was 500 mL. Oxygen and pH were measured a t the three points. The temperature was 25 "C. Feed medium was pumped to the reactor bottom by a Masterflex peristaltic pump a t 30 mL/h. The average bead diameter was 1.57 mm, and the initial cell concentration was 1.6 X lo9 cells/g of beads. The membrane fragment suspension was incorporated a t 4 ?4 (v/v), or 1.2 g/L of bead. The total bead volume in the column was 100 mL. The feed medium pH was adjusted to 6.5, and its oxygen concentration was 4-4.5 mg/L; no attempt was made to remove any oxygen from the feed.
Results and Discussion Preliminary Tests with Immobilized Membrane Fragments. In preliminary experiments, the optimal membrane concentration was determined to be 2% (v/v) of the membrane suspension in the total working volume (including beads, if any). The beads were stable and active during repeated oxygen challenges. Incomplete oxygen removal occurred a t lower membrane concentrations, and there was no further improvement at higher values. The oxygen removal rate of a 2% membrane suspension was relatively constant above pH 6.5 with a value of 0.8 (mg/L)/min, and it dropped at lower pH to a value of 0.07 (mg/L)/min, a t pH 5.5. Because the optimum pH for C. acetobutylicum is near 5.5 in the fermentation experiments, pH was set at 6.5 before inoculation to allow rapid oxygen consumption by the membranes and then was allowed to fall during the fermentation. Batch Experiments. As a control, a batch fermentation was performed without removal of the initial oxygen from the system by either sparging or membranes (see Figure 2). Notice that, after 10 h, the microorganisms start growing slowly and, after 20 h, the oxygen level is reduced to 0.90 mg/L. It has been reported previously that if the cell density is sufficiently high and only limited amounts of oxygen are present, C. acetobutylicum does have the ability to slowly reduce dissolved oxygen (O'Brian and Morris, 1971). It has been suggested that this is the result of the action of the enzyme NADH
Biotechnol. Prog., 1990, Vol. 6, No. 3
212
10
0
24
48
72
Time (h)
0
0
40
24
72
96
Time (h)
Figure 2. Fermentation results with C. acetobutylicum NRRLB-643at 35 "C,100 rpm, without pH control, with no oxygen removal (experiment A): W, absorbance at 610 nm; V, pH; V, glucose; A, butanol; @, acetone; 0, ethanol; A, acetic acid; 0,
Figure 3. Fermentation results for experiment B (Nz/C02 sparg ing). Symbols as in Figure 2.
butyric acid; and X, oxygen.
oxidase. Final butanol and acetone concentrations of 2.82 and 1.20 g/L, respectively, were reached. Experiments of this type are expected to be variable and depend on the initial cell density and dissolved-oxygen concentration. Results obtained by using inert gas sparging for initial oxygen removal are shown in Figure 3. Before inoculation, the oxygen concentration was reduced to 0.27 mg/L. Gas sparging continued for the first 22 h, until the fermentation gas could maintain anaerobiosis. Oxygen was 0.02 mg/L for the remainder of the fermentation. The pH quickly decreased at the beginning of the fermentation while acetic and butyric acids were produced by increased slightly when solvents were released and acids were consumed. Butanol and acetone levels became significant (2.70 and 1.26 g/L, respectively), and final acid concentrations were low. These results were compared with those obtained when oxygen was removed by using the membranes. The fermenter headspace contained air. The oxygen concentration decreased from 6.7 to 0.15 mg/L within 5 min after the addition of the membranes to the medium; at this point, the inoculum was added. The oxygen level during fermentation remained as low as 0.02 mg/L in spite of the relatively low pH. The results are plotted in Figure 4 and are similar to those obtained with sparging; however, the maximum biomass attained is higher, as are the final butanol and acetone concentrations of 3.25 and 1.49 g/L, respectively. A further experiment was performed to check the influence of the air present in the fermenter headspace (not shown). In this experiment, the headspace gas was Nz, and membrane fragments were used. The results showed that with a maximum biomass concentration slightly lower than in the previous experiment, the final solvent concentrations were higher, 3.48 g/L of butanol and 2.58 g/L of acetone, and the fermentation stopped earlier at 35 h instead of -46 h. The pH followed the same pattern in both cases. The effect of lactate, the proton donor for the reducing membranes, was not examined here. Bahl et al. (1986) tested the effect of lactate for the ABE fermentation of glucose in synthetic medium by C . acetobutylicum. They found a decrease of more than 25% in total solvent pro-
Time (h)
Figure 4. Fermentation results for experiment C under anaerobic conditions provided by use of membrane fragments. Symbols as in Figure 2.
duction from glucose in the presence of 50 mM lactate.
As lactate was seen to have a deleterious effect, this strongly suggests that the improvements observed here are due to the use of oxygen-reducing membrane fragments. From these batch experiments, it is clear that membrane fragments are effective in obtaining and maintaining anoxic conditions during an anaerobic acetone, butanol, and ethanol (ABE) fermentation. When such fragments are used, somewhat higher product and biomass concentrations were obtained than when the conventional approach is used. We have also confirmed that oxygen removal, whether by sparging or by the membranes, is necessary to ensure growth without long lag phases. Finally, initial Nz sparging of the fermenter headspace, when the oxygen is reduced with membranes, provides an additional improvement probably because this avoids the effects of the oxygen transport from the air to the liquid. Continuous Fermentation. The results obtained in the batch fermentation experiments suggest that a columnar, continuous reactor, with cells and membrane fragments coimmobilized in porous beads, would be particularly effective in this case. The medium pH could be set a t 6.5 to ensure a rapid oxygen removal at the bottom of the column, followed by a natural decrease in pH along the fermenter to improve solvent production. Anoxic
213
Biotechnol. Prog., 1990, Vol. 6, No. 3
Solvent Production in a FBR Using Clostridia in Anaerobic Beads
0
24
48
72
96
120
Time (h) Figure 5. Glucose and products concentration at the effluent of the columnar reactor with coimmobilized cells and membrane fragments during a continuous fermentation run. Flow rate is 30 mL/h. Symbols as in Figure 2.
conditions would be maintained throughout by the membranes (especially in the bead microenvironment), and the effect of reintroduction of oxygen from the headspace would be minimized in the startup and completely avoided once gas was produced. A continuous run was carried out in the experimental setup described. The oxygen concentration in the fermenter dropped very quickly when the experiment started. Its value in the liquid at the middle of the column was 0.07 mg/L after 1 h and 0.00 after 2 h. In the top section, it was 0.20 mg/L after 1 h and 0.10 after 2 h. The feed remained oxygenated throughout. These concentrations remained constant throughout the experiment (5 days). Under these conditions the beads have extra oxygen removal capacity. The oxygen removal capacity is estimated to be sufficient to handle a 10-fold increase in flow rate in the reactor. In the steady state, pH decreased from 6.5 at the entry point to -5 at the middle of the fermenter and to -4.9 at the top section. The variation of glucose and products concentrations in the effluent of the column during the continuous run is plotted in Figure 5. It can be seen that the steady state is reached after -70 h. The concentration of butanol (2.76 g/L), acetone (1.57 g/L), and ethanol (0.58 g/L) are on the order of the final concentration in the batch fermentations. As in the batch runs, the glucose was not completely consumed. The hydrodynamic behavior of the bed can be described as a fluidized bed, with some slugs resulting from the fermentation gas. The final cell concentration in the beads was 8.6 X lo9 cells/gram of beads. The total solvent productivity on a volumetric basis was 0.35 g/L.h and was 1.77 g/L.h on the basis of the actual bead volume in the bioreactor. These values, as well as the total solvent production, are lower than some previous results published on this fermentation (Ennis, Gutierrez, and Maddox, 1986; Frick and Schugerl, 1986) and should be increased by an optimization of the strain, feed medium, and especially glucose consumption, which was only partial in the reported experiments.
Conclusions The feasibility and efficiency of the use of microbial membrane fragments to provide anoxic conditions in immobilized cell reactors has been demonstrated. In this way, the manipulation of continuous anaerobic fermenters can be made easier because there is no need to provide strict anoxic conditions in the liquid feed medium, and even if
some oxygen is present in the fermenter liquid, the microenvironment in the immobilized cell particles is kept anaerobic. The ABE fermentation with C. acetobutylicum NRRL-B-643, with the use of such membrane fragments, gave somewhat better results in solvent production than when conventionalgas sparging was used. These improvements occurred in spite of the deleterious effect of lactate (Bahl et al., 1986). More importantly, a continuous run with bacteria and membrane fragments coimmobilized in K-carrageenan and using oxygenated feed was successful in providing an anoxic environment for solvent production. The use of microbial membrane fragments to reduce oxygen and maintain anaerobic conditions in fermentations has unexplored potential. The advantages are particularly strong in continuous systems with oxygenated feed and when coimmobilized with anaerobic cells for an anoxic microenvironment. The relative costs of this procedure are not known; however, the membrane fragments have been produced in multiliter quantities, and their costs should drop with the increases in scale. Speculatively, the use of nonoptimum pH in this case might be avoided by making the membrane fragments from an organism with a lower pH optimum.
Acknowledgments F.G. thanks NATO Scientific Committee for a research grant to the Oak Ridge National Laboratory, and all the authors thank NATO for a collaborative Research Travel Grant. Work was performed under the Energy Conversion and Utilization Technologies Biocatalysis Program of the US. Department of Energy. Literature Cited Adler, H. I.; Crow, W. D. A Novel Approach to the Growth of Anaerobic Microorganisms. Biotechnol. Bioeng. Symp. 1981, 11,533-540.
Adler, H. I.; Carrasco, A.; Crow, W. D.; Gill, J. S. Cytoplasonic Membrane Fraction that Promotes Separations in an Escherichia coli Ion Mutant. J . Bacteriol. 1981, 147, 326332.
Adler, H. I.; Crow, W. D.; Hadden, C. T.; Hall, J.; Machanoff, R. New Techniques for Growing Anaerobic Bacteria and Experiments with Clostridium butyricum and Clostridium acetobutylicum. Biotechnol. Bioeng. Symp. 1983,13, 153-161. Bahl, H.; Gottwald, M.; Kuhn, A.; Rale, V.; Andersch, W.; Gottschalk, G. Nutritional Factors Affecting the Ratio of Solvents Produced by Clostridium acetobutylicum. Appl. Environ. Microbiol. 1986,52, 169-172. Ennis, B. M.; Gutierrez, N. A,; Maddox, I. S. The Acetone-Butanol Fermentation: A Current Assessment. Process Biochem. 1986,21,131-147. Frick, C. H.; Schugerl, K. Continuous Acetone-Butanol Production with Free and Immobilized Clostridium acetobutylicum. Appl. Microbiol. Biotechnol. 1986, 25, 186-193. Johnson, M. J.; Borkowski, J.; Engblom, C. Stream Sterilizable Probes for Dissolved Oxygen Measurement. Biotechnol. Bioeng. 1967,9,635-639. Jones, D. T.; Woods, D. R. Acetone Butanol Fermentation Revisited. Microbiol. Rev. 1986,50, 484-524. Monot, F.; Martin, J. R.; Petitdemange, H.; Gay, R. Acetone and Butanol Production by Clostridium acetobutylicum in Synthetic Media. Appl. Enuiron. Microbiol. 1982,44, 13181324.
O’Brian, R. W.; Morris, J. G. J. Oxygen and the Growth and Metabolism of Clostridium acetobutylicum. J . Cen. Microbiol. 1971, 68, 307-318. Scott, C. D. Techniques for Producing Mondispersed Biocatalyst Beads for Use in Columnar Bioreactors. Ann. N . Y. Acad. Sci. 1987,501,487-493. Accepted April 11, 1990.
