Correlation of Redox Potential with State Variables in Culturesunder Controlled Dissolved Oxygen Concentration and pH

Tae Ho Lee,† Yong Keun Chang,*,† Bong Hyun Chung,‡ and Young Hoon Park‡

Department of Chemical Engineering and BioProcess Engineering Research Center, Korea Advanced Institute ofScience and Technology, Taejon 305-701, Korea, and Korea Research Institute of Biotechnology andBioengineering, Korea Institute of Science and Technology, Taejon 305-600, Korea

In batch cultures for L-ornithine production in which dissolved oxygen concentrationand pH were closely controlled, time changes of redox potential were observed inconnection with the profiles of cell, glucose, and ornithine concentrations. It was foundthat the redox potential profile had four different phases reflecting the physiologicalstate of the culture and that it was closely related to cell concentration change. Effectsof glucose and ornithine on the redox potential were identified in a separate series ofexperiments. On the basis of the experimental results, a correlation of redox potentialto glucose, cell, and ornithine concentrations has been proposed. The proposedcorrelation can be used for on-line estimation of ornithine concentration from on-linedata of redox potential, glucose concentration, and cell concentration.

IntroductionRedox potential (Eh) electrochemically indicates the

relative concentration of reductants to oxidants in anenvironment. Therefore, it is the indicator that repre-sents the abilities of redox couples as electron donors andacceptors (Halling, 1990).

In a number of previous studies, redox potential wasused for estimating extremely low dissolved oxygenconcentration (DOC) under almost anaerobic conditions(below the lower limit of DO probes) due to its strongdependency on DOC (Kjaeergaard, 1976; Wimpenny,1969; Srinivas et al., 1988).

There have been several reports on the relationshipbetween redox potential and the physiological state of amicrobial system. Wimpenny (1969) showed that theredox potential had an effect on the three tricarboxylicacid (TCA) cycle and respiration chain. He reported thatthe redox potential increased under a more aerobiccondition and that it influenced TCA cycle enzymes,cytochrome, ATP pool, growth yield, and hydrogenaseactivity. It was also found that the redox potentialinfluenced the simultaneous biosynthesis of the antibiot-ics levorin A and B (Kjaeergaard, 1976). Levorin A waspreferably produced at a high redox potential, and levorinB compound was produced at a lower redox potential.Shibai et al. (1974) investigated the effects of redoxpotential on inosine production. Different byproductswere produced for different redox potential values, andthere was no accumulation of inosine when the redoxpotential was lower than -160 mV. Jin and Englande(1996) reported that redox potential had a critical effecton the biodegradation of carbon tetrachloride by Pseudo-monas cepacia and Providence stuartii.

Kwong and Rao studied the monitoring of an aminoacid production process by redox potential measurement(Kwong and Rao, 1991). They also reported on the use

of on-line culture redox potential and dissolved oxygenmeasurements to identify metabolic changes in Coryne-bacterium glutamicum cultures under aerobic conditions(Kwong et al., 1992). Metabolic changes (the end of lagphase, substrate exhaustion etc.) were identified byobserving discrepancies in the profiles of redox potentialand DOC. Higareda et al. (1997) reported on the use ofredox potential and oxygen uptake rate for assessingglucose and glutamine depletion. The beginning ofgrowth phase and death phase, and nutrients depletioncould be assessed from the behavior of redox potentialin the culture of hybridoma cell for monoclonal antibodyproduction. However, the direct relationship of redoxpotential to glucose and glutamine concentrations couldnot be obtained.

As shown above, most research has been carried outto investigate the effect of the redox potential on thephysiological activity of cells and byproduct productionor to predict changes in the physiological state of aculture system from the pattern of redox potential.However, if we take a more ambitious approach, majorstate variables could be predicted on-line beyond thesimple assessment of physiological state changes. Ofcourse, for the on-line monitoring, an accurate correlationof the redox potential to the target state variables shouldbe developed beforehand.

In the present study, a correlation of redox potentialto major state variables such as substrate, cell mass, andproduct concentrations was developed and applied toBrevibacterium ketoglutamicum cultures to produce or-nithine, a pharmaceutical amino acid for the treatmentof liver diseases (Lee, T. H., et al., 1996). Our rationalefor this study was that any changes in redox potentialunder conditions of controlled pH and DOC must berelated to changes in the fermentation broth compositiondue to the reactions involved. The characteristic behav-iors of redox potential with metabolic state changes wereobserved in batch cultures at various DOC levels. Onthe basis of these experimental results, a polynomialfunction has been developed to correlate the redox

* To whom all correspondences to be addressed.† Korea Advanced Institute of Science and Technology.‡ Korea Institute of Science and Technology.

959Biotechnol. Prog. 1998, 14, 959−962

10.1021/bp980085w CCC: $15.00 © 1998 American Chemical Society and American Institute of Chemical EngineersPublished on Web 10/29/1998

potential to the state variables, which can later be usedfor the on-line monitoring of ornithine concentration.

Materials and Methods

Strain. The microorganism used in this study was B.ketoglutamicum 1047, which is a L-citrulline auxotrophicmutant. Stock culture was stored in a glycerol solutionat -4 °C.

Culture Medium and Conditions. The growthmedium used for inoculation was a YPD medium (glucose20 g/L, yeast extract 10 g/L, peptone 10 g/L, pH 7.0). Theseed was prepared by growing cells in a 250-mL shakeflask for 12 h. The composition of fermentation mediumwas as follows: glucose, 100 g/L; yeast extract, 10 g/L;(NH4)SO4, 20 g/L; KH2PO4, 0.75 g/L; Na2HPO4, 1.5 g/L;MgSO4‚7H2O, 1.0 g/L; MnSO4‚4H2O, 0.05 g/L; traceelements. Fermentation was carried out in a computer-ized fermentor system (Bio Stat E, B. Braun Biotechnol-ogy, U.S.) with 2.5-L culture volume (Figure 1). Allcultures were grown at 30 °C. The pH was maintainedat 7.0 using NH4OH.

Analytical Methods. To obtain dry cell weight, a 10-mL sample was taken and centrifuged. The solid samplewas washed with distilled water three times. Thewashed sample was dried at 85 °C for 24 h to obtain drycell weight. Cell growth was monitored by measuringoptical density at 600 nm using a spectrophotometer(Model 930, UVCON, U.S.). Glucose concentration wasmeasured using an enzyme kit (Young-Dong Pharma-ceutical Co., Korea). L-Ornithine concentration wasdetermined by the Chinard method (Chinard, 1952).DOC was measured using a membrane-type polaro-graphic DO electrode (Ingold Co., U.S.), and the redoxpotential was measured using a platinum electrode(Ingold Co.). The pH was measured by a glass pHelectrode (Ingold Co.).

Data Acquisition and DO Control Systems. Theon-line data of DOC and redox potential were taken every30 s and stored in the computer (Figure 1). The DOCwas controlled by manipulating the agitation speed (150-1000 rpm). Communication between the fermentorsystem and the computer was done via D/A and A/Dconverters and a RS-232 cable. The control methodemployed for DOC regulation was a digital version of anautotuning proportional-integral-derivative (PID) controlalgorithm developed by Lee et al. (Lee, S. C., et al., 1991,1994). The aeration rate was fixed at 1 vvm. DOC wasmaintained at 10, 30, and 50% of air saturation with (1%fluctuation.

TheoryThe redox potential of fermentation broth is influenced

by the redox couples involved as well as DOC and pH.Therefore, the relationship can be represented by ageneralized function as follows:

where Rei denotes concentrations of various redox couplecomponents.

In our particular case, the pH and DOC are maintainedconstant during the culture. In this case, eq 1 can bereformulated as eq 2, in which Eh0 is an appropriateconstant.

The concentrations of dissolved components which actas redox couples in culture medium will depend on howmuch cell has been formed by consuming the mediumcomponents including the substrate and how muchproduct and byproduct(s) are produced. The substrate(S) and product (P) can be determined quantitatively, butthe other dissolved components whose consumption orformation are also related to cell growth cannot or aredifficult to be characterized and determined quantita-tively. Therefore, eq 2 has been recast to eq 3 by lumpingthe effects of these components into the effect of the cellconcentration (X), a measurable quantity, in which theredox potential (Eh) is represented as a function of X, S,and P only.

By employing polynomial functions for the effects ofX, S, and P, eq 4 can be written as follows:

where Eh0′ (≡ Eh0 + a0), and ai’s, bi’s, and ci’s areappropriate constants.

Results and DiscussionDO-Stat Fermentation at Various DOC Levels.

Figure 2 shows the profile of redox potential in DO-statfermentation at a DOC of 10%. It can be divided intofour phases as (a) rapid increase, (b) stationary phase,(c) slow increase, and (d) rapid increase. These differentbehaviors of redox potential at various phases may beexplained by metabolic state changes of the cells, whichare represented by the profiles of cell, substrate, andproduct concentrations. The trend of redox potentialseemed to be strongly related to cell growth. The redoxpotential in the first phase increased rapidly after aninitial sharp drop for very short time. This initial dropwas explained previously by Kwong et al. (1992) as aphenomenon during the lag phase. The redox potentialwas maintained almost constant at 67 mV, when cellgrowth was retarded because of arginine depletion. Asshown in Figure 2, this situation continued for 15 h. Itwas very interesting that ornithine, the product, beganto be produced significantly at the end of the first phaseas the repression of arginine on its production decreased.

Figure 1. Bioreactor and instrumentation system : (1) jarfermentor, (2) agitation motor, (3) redox potential electrode, (4)DO electrode, (5) pH electrode, (6) NH4OH solution, (7) D/Aconverter, (8) personal computer, (9) A/D converter.

Eh ) f(DOC, pH, Rei) (1)

Eh ) Eh0 + g(Rei) (2)

Eh ) Eh0 + g(X,S,P) (3)

Eh ) Eh0 + a0 + a1X + a2X2 + ‚‚‚ + b1S + b2S

2 +

‚‚‚ + c1P + c2P2 + ‚‚‚

) Eh0′ + a1X + a2X2 + ‚‚‚ + b1S + b2S

2 + ‚‚‚ +

c1P + c2P2 + ‚‚‚ (4)

960 Biotechnol. Prog., 1998, Vol. 14, No. 6

After the stationary period, the dry cell weight increasedagain. This was considered to result from the appearanceof revertant cells that had lost the characteristics ofarginine auxotroph under a prolonged arginine-depletedcondition. The redox potential increased again in thethird phase, but the slope was lower than that in the firstphase. After glucose was completely consumed, the redoxpotential began to increase more rapidly in the fourthphase. Such change of redox potential during glucosedepletion was also reported previously without detailedexplanations (Oktyabr’skii and Smirnova, 1986). Theobservation results above implies that we can predict on-line the physiological state of the culture from the redoxpotential profile without on-line monitoring of other key

state variables. The final cell concentration was 18.0 g/L,and the L-ornithine concentration was 14.2 g/L after 42h.

Figures 3 and 4 show the results for 30 and 50% ofDOC. Four distinguished phases were also observed asin the case of 10% DOC. When DOC was 30%, the finalcell concentration was 18.3 g/L and the ornithine con-centration was 19.2 g/L after 45 h of cultivation. WhenDOC was 50%, the final cell concentration and ornithineconcentration were 28.1 and 15.1 g/L, respectively, after52 h.

Correlation of Redox Potential with State Vari-ables. Careful examinations of the profiles of redox

Figure 2. Profiles of cell, glucose, and L-ornithine concentra-tions and redox potential at 10% DOC.

Figure 3. Profiles of cell, glucose and L-ornithine concentra-tions and redox potential at 30% DOC.

Figure 4. Profiles of cell, glucose and L-ornithine concentra-tions and redox potential at 50% DOC.

Figure 5. Effects of L-ornithine and glucose on redox potential(A) ornithine and (B) glucose.

Figure 6. Comparison of estimated values of redox potentialwith measured values: measured values (s); estimated values(b, DOC 10%; 9, DOC 30%; [, DOC 50%).

Biotechnol. Prog., 1998, Vol. 14, No. 6 961

potential, cell, glucose, and ornithine concentrations, andDOC revealed that the redox potential was stronglyrelated to the cell concentration when the ornithineconcentration was low.

On the other hand, when the ornithine concentrationwas fairly high, the relationship was not obvious. Fromthis, we could see that ornithine also had an effect onthe redox potential, which should be identified. Wespeculate that glucose might have some effect on theredox potential also. However, no clear conclusion couldbe made for its effect only with the data given in Figures1-4. Therefore, we carried out a series of experimentsto separately determine the effects of ornithine andglucose, in which the redox potential was measured forvarious concentrations of glucose and ornithine, sepa-rately dissolved in distilled water. It was found that theredox potential could be represented by linear andsecond-order functions of S and P, respectively (Figure5).

From eqs 4-6, eq 7 was obtained.

The regression of (Eh, X) data by using eq 7 for variousDOC levels revealed that the redox potential was linearlydependent on X as in eq 8 and that the coefficients Eh0′

and a1 had different values for different DOC levels. Theregression results are summarized in Table 1. In allcases, satisfactory correlations were obtained as can beseen in Figure 6 and the correlation factors were over0.987.

As mentioned above, the values of Eh0′ and a1 werefound to vary with DOC level. As for Eh0′, it is under-standable that it changes with DOC level because theredox potential is influenced by DOC. The changeaccording to DOC level of the coefficient a1 is consideredto have resulted from the differences in the productionyields of the product and byproducts and the consumptionof the other medium components than glucose at differentlevels of dissolved oxygen concentration.

The developed correlation has an important meaningin the viewpoint of bioprocess engineering. Concentra-tions of cell and glucose can be measured on-line withno difficulty. On the other hand, the on-line measure-

ment of ornithine is very difficult, if not impossible,because its analysis relies on a rather complicated assaymethod or requires time-consuming pretreatment steps.Under such situation, the correlation developed in thisstudy makes it possible to estimate ornithine concentra-tion from the on-line data of redox potential, cell concen-tration, and glucose concentration. However, it has alimitation in that it can be used only for the cases whenthe initial condition and the DOC level are identical tothose used in the present study. For it to be more useful,correlations which can clearly explain the dependence ofthe coefficients Eh0′ and a1 on the initial condition andDOC should be developed.

References and NotesChinard, F. P. Photometric estimation of proline and ornithine.

J. Biol. Chem. 1952, 199, 91-95.Halling, P. J. Sensors for Fermentation. In Fermentation, a

Practical Approach; McNeil, B., Harvey, L. M., Eds.; IRLPress: Oxford, U.K., 1990; pp 143-145.

Higareda, A. E.; Possani, L. D.; Ramirez, O. T. The use of cultureredox potential and oxygen uptake rate for assessing glucoseand glutamine depletion in hybridoma cultures. Biotechnol.Bioeng. 1997, 56, 555-563.

Jin, G.; Englande, A. J., Jr. Redox potential as a controllingfactor in enhancing carbon tetrachloride biodegradation.Water Sci. Technol. 1996, 34, 59-66.

Kjaeergaard, L. Influence of redox potential on glucose catabo-lism of chemostat-grown Bacillus icheniformis. Eur. J. Appl.Microbiol. 1976, 2, 215-220.

Kwong, S. C. W.; Rao, G. The utility of culture redox potentialfor identifying metabolic state changes in the amino acidfermentation. Biotechnol. Bioeng. 1991, 38, 1034-1040.

Kwong, S. C. W.; Randers, L.; Rao, G. On-line assessment ofmetabolic activity based on culture redox potential anddissolved oxygen profiles during aerobic fermentation. Bio-technol. Prog. 1992, 8, 576-579.

Lee, S. C.; Hwang, Y. B.; Chang, H. N.; Chang, Y. K. Dissolvedoxygen concentration regulation using auto-tuning PID con-troller in fermentation process. Biotechnol. Tech. 1991, 5, 85-90.

Lee, S. C.; Hwang, Y. B.; Lee, T. H.; Chang, H. N.; Chang, Y. K.Characteristics and performance of auto-tuning PID control-ler for dissolved oxygen concentration. Biotechnol. Prog. 1994,10, 447-450.

Lee, T. H.; Ryu, W. S.; Chang, Y. K.; Chung, B. H.; Park, Y. H.Effects of medium components on L-ornithine production byBrevibacterium ketoglutamicum. Biotechnol. Bioprocess Eng.1996, 1, 41-45.

Oktyabr’skii, O. N.; Smirnova, G. V. Changes in the redoxpotential of the medium for B. subtilis cultures and theirconnection with the electrical parameters of the cells. Bio-physics 1986, 31, 503-508.

Shibai, H.; Ichizaki, A.; Kobayashi, K.; Hirose, Y. Simultaneousmeasurement of dissolved oxygen and oxidation-reductionpotentials in the aerobic culture. Agric. Biol. Chem. 1974, 38,2407-2411.

Srinivas, S. P.; Rao, G.; Mutharasan, R. Redox potential inanaerobic and microaerobic fermentation. In Handbook onAnaerobic Fermentation; Erickson, L. E., Fung, D. Y. C., Eds.;Marcel Dekker: New York, 1988; pp 1187-1206.

Wimpenny, J. W. The effect of Eh on regulatory processes infacultative anaerobes. Biotechnol. Bioeng. 1969, 11, 623-629.

Accepted September 21, 1998.

BP980085W

Table 1. Results of Regression Using Eq 8

DOC (%) Eh0′ a1

correlationfactor

10 82.15 2.62 0.99720 94.58 4.29 0.99250 124.4 6.35 0.987

Eh ) -0.323S + C1 (for glucose) (5)

Eh ) 0.063P2 - 4.360P + C2 (for L-ornithine) (6)

Eh ) Eh0′ + a1X + a2X2 + ‚‚‚ -0.323S - 4.360P +

0.063P2 (7)

Eh ) Eh0′ + a1X - 0.323S - 4.360P + 0.063P2 (8)

962 Biotechnol. Prog., 1998, Vol. 14, No. 6