Estimation of Fermentation Biomass Concentration by

Transcription

1 APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Feb. 1978, p /78/ $02.00/0 Copyright i 1978 Amerian Soiety for Mirobiology Vol. 35, No. 2 Printed in U.S.A. Estimation of Fermentation Biomass Conentration by Measuring Culture Fluoresene D. W. ZABRISKIEt* AND A. E. HUMPHREY University of Pennsylvania, Department of Chemial and Biohemial Engineering, Philadelphia, Pennsylvania Reeived for publiation 7 September 1977 The fluoresene of a fermentation ulture was studied for its appliation as an estimator of biomass onentration. The measurement was obtained by irradiating the ulture with ultraviolet light (366 nm) through a glass window and deteting fluoresent light at the window surfae at 460 nm. It was estimated that over onehalf of the fluoresent material was interellular redued niotinamide adenine dinuleotide, with the remainder being redued niotinamide adenine dinuleotide phosphate and other unidentified interellular and extraeuular fluorophores. The ulture fluoresene was found to be a funtion of biomass onentration, together with environmental fators, whih presumably at at the ellular metaboli level to modify interellular redued niotinamide adenine dinuleotide pools (e.g., dissolved oxygen tension, energy substrate onentration, and inhibitors). When these environmental onditions were ontrolled, a linear relationship was obtained between the log of the biomass onentration and the log of the fluoresene. Under these onditions, this relationship has onsiderable potential as a method to provide real-time biomass onentration estimates for proess ontrol and optimization sine the fluoresene data is obtained on line. When environmental onditions are variable, the fluoresene data may be a sensitive index of overall ulture ativity beause of its dependene on interellular redued niotinamide adenine dinuleotide reserves and metaboli rates. This index may provide information about the period of maximum speifi produtivity for a speifi mirobial produt. A widely reognized problem in the analysis and ontrol of fermentation proesses is the inability to determine the biomass onentration in the ulture on a real-time basis (9). This fat is a onsequene of the long time periods assoiated with most hemial and physial assays. Methods employing the optial properties of ell suspensions have rapid response times, but suffer from numerous interferenes suh as medium solids, dense medium oloration, myelial ulture morphology, and growth on the optial surfaes, whih are harateristi of many pratial fermentations. Sine interellular redued niotinamide adenine dinuleotide and redued niotinamide adenine dinuleotide phosphate [NAD(P)H] fluorese at 460 nm when ells are irradiated by 366-nm light, this property has been used extensively by biohemists in studying the transitions in redued pyridine nuleotide onentrations in response to hanges in ellular environment. Duysens and Amesz (4) first used this phenomenon to onfirm that starved yeast ells have t Present address: State University of New York at Buffalo, Department of Chemial Engineering, Amherst, NY less redued niotinamide adenine dinuleotide (NADH) than atively growing ells, a onlusion that was first suggested by less sensitive absorption spetrosopy methods (2). Chane et al. (3) used fluoresene to study damped NADH osillations in response to hanges in ell environment. Harrison and Chane (5) applied the tehnique to ontinuous ultures of ells. They demonstrated that interellular NADH inreases in yeast ells during the transition from aerobi to anaerobi growth. The absene of oxygen prevents oxidative phosphorylation from oxidizing high-energy NADH to low-energy niotinamide adenine dinuleotide (NAD), thereby ausing a net inrease in NADH and fluoresene. These results were onfirmed hemially. The purpose of the researh desribed in this paper was to determine the relationship between the onentration of ells and the amount of ulture fluoresene (D. W. Zabriskie, Ph.D. thesis, University of Pennsylvania, Philadelphia, 1976). MATERLA1S AND METHODS Culture fluorometer. The ulture fluorometer used was an adaptation of the eletronis (7) and

2 338 ZABRISKIE AND HUMPHREY fermentor assembly desribed earlier (5) and is shown shematially in Fig. 1. The fluorometer was mounted on a vessel observation port loated beneath the ulture surfae. The ulture was irradiated with light near 366 nm by using a fluoresent ultraviolet lamp and an optial filter (CS-7-60, Corning Optial Produts, In., Corning, N.Y.). This light exited the ulture, ausing it to fluorese near 460 nm. The fluoresent light was measured using a photomultiplier that was filtered to sreen out the exiting light wavelengths and other fluoresent peaks (CS-5-57 and CS- 3-73, Coming Optial Produts). Tap water was irulated through the mdunting assembly to remove the heat generated by the lamp. Fermentation onditions. Three aerobi bath fermentations using growth onditions harateristi of industrial proesses were studied. Bakers' yeast (Saharomyes erevisiae ATCC 7754) was grown at 30 C (ph 5.0) on a medium omposed of 2% orn steep liquid, 8.0 g of (NH4)2 HPO4 per liter, and gluose. The sugar was added on demand to prevent gluose repression and exessive onversion to ethanol. Eah pulse was suffiient to bring the broth to a onentration of 1.0 g/liter. A pulse was added when gluose beame limiting, as shown by a sudden drop in the arbon dioxide evolution rate. After the maximum growth rate had been reahed, a period of diauxi growth was initiated by withholding gluose and allowing the assimilation of aumulated ethanol and aetate byproduts. Gluose additions were resumed after the exhaustion of the metabolizable intermediates as determined by a sudden deline in the arbon dioxide evolution rate. The biomass onentration was determined by dry-weight analysis (Zabriskie, Ph.D. thesis). The seond fermentation used a speies of Streptomyes, whih is used in the industrial prodution of enzymes. The ulture was grown at 300C (ph 6.0) on a medium similar to the one used to grow the yeast. Gluose, however, was added bathwise (26.4 g/liter) sine it has no repressive effets on this miroorganism under these onditions. The biomass onentration was determined by dry-weight analysis. l b ; = = - = I ---- i ---= := /- -- \ t = /... g Wall = =- =- F ermentor \ = Optial Filter / /\ COptial Filter passing 366 nm Light passing 460 nm Light Fluoresent Photo "Blak Light, Multiplier FIG. 1. Fermentor fluorometer assembly. APPL. ENVIRON. MICROBIOL. The third fermentation used a thermophili speies of Thermoatinomyes, whih is urrently being investigated for its potential to produe single ell protein from ellulosi materials (1). It was grown at 550C (ph 7.2) in a 5% suspension of ellulose (Aviel, PH 102, FMC, In., Philadelphia, Pa.). The soluble omponents of the defined medium inluded (per liter): NaCl, 1.5 g; ethylenediaminetetraaeti aid, 50 mg; MgSO4 7H20, 0.2 g; ZnSO4 7H20, 8 mg; FeSO4 7H20, 20 mg; MnSO4 4H20, 20 mg; CaCl2, 20 mg; (NH4)2SO4, 31 g; thiamine, 1.0 mg; biotin, 1.0 mg; KH2PO4, 9.1 g; and yeast extrat, 0.1 g. Sine the biomass solids and ellulose solids annot be separated, a dry-weight analysis for biomass onentration determination had to be replaed by hemial assays for materials present in the ells and absent from the ellulose. This was aomplished by using modified miro-kjeldahl and Lowry proedures to measure the onentration of organi nitrogen and protein, respetively, after the ulture solids were washed thoroughly with distilled water (Zabriskie, Ph.D. thesis). All fermentations were grown in an asepti 70-liter bath fermentor. Environmental onditions were monitored and ontrolled by an instrumentation pakage, whih inluded analyzers for dissolved oxygen (Johnson type, New Brunswik Sientifi Co., Edison, N.J.), gaseous oxygen onentration (paramagneti wind type, model 802, Mine Safety Applianes, In., Pittsburgh, Pa.), gaseous arbon dioxide onentration (infrared absorption type, model 303, Mine Safety Applianes), ph, temperature, and gas flow rates. The dissolved oxygen was ontrolled above limiting onditions for all fermentations, exept during the yeast oxygen depletion experiments. RESULTS Response to fluorophore onentration. The response of the ulture fluorometer to fluorophore onentration was evaluated by using solutions of NADH in 0.05 M phosphate buffer or quinine in 0.05 M H2SO4. The fermentor was filled with fluorophore-free buffer. Aliquots of onentrated fluorophore solutions were added, and the fluorometer readings were reorded. Typial results are shown in Fig. 2. The non-linearity of the data is due primarily to the inner filter effet (8). As the exiting wavelength light passes through the sample, the light is absorbed aording to the Beer-Lambert Law: I = Ie (1) where I is light intensity, k the molar extintion oeffiient, C is the onentration of absorbing speies, and L is the path length for light absorption. When the onentration of absorbing speies is high, it leads to a gradient of exiting light intensity within the sample. When the sample is loated far from the light soure, it is effetively filtered from the exiting wavelength, thereby reduing its fluoresent apability. Un-

3 VOL. 35, 1978 FERMENTATION BIOMASS CONCENTRATION Quinine Conentration (PPB) FIG. 2. Fermentor fluorometer response to quinine onentration. der these onditions, the total fluoresene, f, generated in all diretions, is given by (8): f=i(1-e-kcl)f (2) where of is the fluoresent effiieny of the fluorophore. Therefore, fluoresene and fluorophore onentration are only approximately linear at low fluorophore onentrations. Generally, fluorometers do not measure the total amount of fluoresene generated by a sample as given by the previous equation, but, rather, the amount ontained in a omparatively small solid angle. Therefore, the geometry of the detetor system determines a fluorometer's response to onentration with inner filter effet aberrations. Log-log plots have been shown to be useful in orrelating fluoresene and onentration data from instruments that irradiate and measure the fluoresene through the same surfae on the sample holder (surfae detetor type [8]). The physial situation for the fermentor fluorometer is muh more omplex. This onfiguration does not have a defined boundary for the sample as do fluorometers with sample uvettes. The exiting light penetrates the sample to whatever distane it an before it is ompletely absorbed or sattered. This penetration depth varies with the fluorophore onentration. The presene of ells and solids satters the inident light out of the area measured by the fluorometer. The heterogeneity of the ulture medium favors a seondary inner filter effet in whih the fluoresent light is absorbed by the sample before it passes out of the sample to the detetor. This heterogeneity also an ause quenhing phenomena, whih divert light energy absorbed by a fluorophore into hannels other than the emission proesses responsible for luminesene. These inlude internal energy onversion, intersystem rossing, energy transfer, and the deativation aused by olliding with other moleules of solutes (8). Sine the fermentor fluorometer is a surfae detetor type, the data of Fig. 2 were replotted on log-log oordinates (Fig. 3). Despite the system's omplexity, the results were effetively linearized by the oordinate transformation suggested by the elementary surfae detetor system, yielding a orrelating equation of the form: 9 :g 7 C = [exp(-b )fnet]l/a (3) Constants a and b are the respetive slope and interept of the linear least squares line through the data plotted on the oordinates shown in Fig. 3. The quantity [net is the differene between the observed fluoresene and the bakground signal obtained in the absene of fluorophore. Environmental onditions affeting fluoresene. Several experiments were performed to evaluate the responses of the fluorometer to hanges in ulture onditions that were shown by other investigators to influene yeast ell fluoresene (4, 5). In the experiments desribed below, (net is the differene between the observed ulture fluoresene and the value obtained before inoulation of the fermentor. Figure 4 shows that dereasing the dissolved oxygen below 45% saturation, where oxygen beomes growth limiting, aused an inrease in fluoresene, as observed by Harrison and Chane (5). When the dissolved oxygen was restored to above 45%, the fluoresene dereased to its original value. Cell starvation redued ulture fluoresene. Figure 5 presents data from an atively growing yeast fermentation in whih the energy substrate, gluose, was permitted to deplete. As it beame limiting, the CO2 evolution rate delined orresponding to a dereasing respiration rate and was followed losely by dereasing ulture fluoresene. The inability of the ells to pro _ Natural Log of Conentration (PPB) FIG. 3. Fermentor fluorometer response to quinine onentration (log-log oordinates).

4 340 ZABRISKIE AND HUMPHREY Es 28r 24 Time (minutes) FIG. 4. Response of ulture fluoresene to dissolved-oxygen depletion in a bakers'yeast fermentation. 20.Oo 16 0 C.) 812 L 0 x 8 4L _ 4w C> - 2n o C-) LA 1, b Time (minutes) FIG. 5. Response of ulture fluoresene to depletion of energy substrate (gluose) in an aerobi bakers' yeast fermentation. due energy by respiration during gluose starvation aused redutions in the high-energy NADH pools of the ell. When gluose additions were resumed, the respiration rate and fluoresene returned to their original levels. Other fators affeting fluoresene inlude temperature and ph. Fluoresene inreased dramatially as temperature dereased (temperature oeffiient = per 10 C), whih emphasizes the need for preise temperature ontrol in fluoresent sample measurement. Fluoresene dereased with inreasing ph. Levels of ph are thought to affet fluoresene by hanging the ioni state of some fluorophores or by ating on the solvent and other absorbing speies in the sample to alter the inner filter effet. The response of the ulture fluorometer was generally insensitive to hanges in agitation and aeration rate, provided the dissolved oxygen remained above limiting levels (-45% saturation). Gas bubbles ontributed to the overall noise levels in the signal. This noise was minimized by using an ative filter on the output of the fluorometer and by irradiating the ulture and measuring the fluoresene over a large surfae area (13/8 inh-diameter disk). Quantitation of NADH ontent of yeast ulture fluorophores. To approximate the NADH portion of the fluorophores responsible for ulture fluoresene, iodoaetate was added to a ulture of atively growing yeast with a biomass onentration of 12.6 g/liter. The onentration of iodoaetate was 1.5 mm, whih was suffiient to blok the Embden-Meyerhof- Parnas pathway and prevent any further prodution of interellular energy. This auses a depletion of interellular NADH reserves followed by ell death. The effet of adding this potent inhibitor aused a 12.5% derease in ulture fluoresene (Fig. 6), similar in magnitude to the derease obtained in the gluose starvation experiment. The estimate of NADH ontent of the ulture fluorophores is obtained from the alibration urve shown in Fig. 7. If it is assumed that the fluorophores present are diretly proportional to the amount of biomass, then the absissa of Fig. 7 may be thought of in terms of ulture fluorophore onentration. It an then be seen that a 12.5% redution in ulture fluoresene aused by the exhaustion of interellular NADH orresponds with a 50% redution in fluorophore onentration. This observation suggests that 50% of the ulture fluorophore ontent is interellular NADH. The atual NADH ontribution to uli > 140' 13, _ o i2.0 APPL. ENVIRON. MICROBIOL Time (minutes) FIG. 6. Response of ulture fluoresene to the inhibition of the Embden-Meyerhof-Parnas metaboli pathway by iodoaetate in an aerobi bakers' yeast fermentation.

5 VOL. 35, 1978 FERMENTATION BIOMASS CONCENTRATION r - 12 E _ O/ 50% 310 :>8 4 I? 2 %125% Biomass Conentration (g/l) FIG. 7. Estimation of the interellular NADHportion of the ulture fluorophores. ture fluoresene may exeed 50% sine the bloking of the ellular energy-generating apparatus probably affets mitohondrial NADH pools. Therefore, NADH present in the ell ytoplasm is not aounted for by this approximation proedure (6). The soures of non-nadh fluoresene are open to speulation. Many biologial ompounds whih are found in ellular material fluorese (e.g., deoxyribonulei aid, ribonulei aid, proteins, ytohromes, some vitamins, hormones, amino aids, and nuleotides). However, only a few are known to fluorese under the onditions seleted to optimize NADH fluoresene used in these experiments. Some of this residual fluoresene is due to interellular NADPH. Some medium nutrients fluorese, ausing the medium to fluorese before inoulation. Other fluorophores may be released into the medium by growing ells, suh as pyridoxal enzymes and flavins, suggested by other investigators (6). The presene of extraellular fluorophores was onfirmed by measuring the fluoresene of ell-free broth and whole-broth samples. A modifiation of the instrument desribed above, whih ould aommodate a standard optial uvette, was used to make these measurements. Diret omparison of these results with those from the orginal instrument annot be made, and, therefore, the measurements are expressed in arbitrary units (perent span, Fig. 8). The data for the whole-broth and ell-free broth samples have similar trends with respet to fermentation time (Fig. 8). The removal of the ells was aompanied by a derease in the fluoresent intensity of the sample. Biomass onentration results. Typial results of ulture fluoresene and biomass onentration for a yeast fermentation are presented in Fig. 7. The nonlinear relationship was similar L6 to that obtained from experiments using soluble fluorophores (Fig. 2) and suggested that a loglog representation of the data might be appropriate. Figure 9 shows the results for another S. erevisiae fermentation in whih oxygen and arbohydrate nutrients were present in exess to minimize non-biomass-related fluoresene phenomena. Culture ph and temperature were ontrolled at onstant levels. The linearity of the data suggested that equation 3 ould be used to orrelate the biomass onentration and ulture fluoresene data. After determining the slope, a, and interept, b, of the line on the log-log plots (Fig. 9), equation 3 was used to ompute estimates for biomass onentration from the fluoresene data. The estimates and laboratory data were ompared for the yeast fermentation (Fig. 10) and a Streptomyes sp. fermentation (Fig. 11). The respetive dispersions of the estimates from the biomass onentration data were 15 and 36%, alulated as oeffiients of variation B- * 70 o 60 LL. 50 o Complete Broth Fluoresene A Cell-Free Broth Fluoresene FIG. 8. Comparison of the fluoresene ofomplete ulture samples with the fluoresene of ell-free medium samples. 24 E a, 0 z Natural Log of Biomass Conentration (g/x) FIG. 9. Natural log of ulture fluoresene versus natural log of S. erevisiae biomass onentration.

6 342 ZABRISKIE AND HUMPHREY '12_ E o Biomass Conentration Data (Dry Weight) A Biomoss Conentration Estimates (Fluoresene) t FIG. 10. S. erevisiae biomass onentration data and estimates derived from ulture fluoresene measurements. n 1 o Biomass Conentration Data (Dry Weight) A Biomass Conentration Estimates (Fluoresene) FIG. 11. Streptomyes sp. biomass onentration data and estimates derived from ulture fluoresene measurements. The fluoresene behavior of the ellulose fermentation (Fig. 12) was quite different from the previous experiments. At hours 13 to 15, the ulture fluoresene dereased, even though biomass onentration ontinued to inrease. Sine the ph was maintained onstant and dissolved oxygen was ontrolled above 50% saturation, the infletion period appears to reflet a metaboli shift in the ulture, perhaps initiated by a substrate limitation or inhibition by a produt or intermediate. Although this trend did not permit biomass onentration estimation as in the previous experiments, the reproduibility of the phenomenon suggests that it may reflet an important metaboli transition that may have impliations of its own in the overall ontrol and optimization of the proess. Real-time estimation of biomass onentration. The potential appliation of this tehnique is its use to estimate the biomass onentration in real time, using parameters a and b derived from a previous experiment, with equation 3 and the on-line ulture fluoresene data. However, experiments whih were onduted to evaluate this proedure did not provide aurate approximations. The ause of this problem was found to be the poor stability of the fluoresent ultraviolet lamp in the instrument. A deay in lamp intensity hanges the values of a and b during the ourse of a long experiment and from one experiment to another. A series of experiments was onduted to study the long-term stability harateristis of the lamp using samples of onstant fluorophore onentration. The signal generated and drifted up and down the sale, with a gradual overall downward exursion orresponding to the aging of the lamp. Drift rates of 0.24 to 0.72 mv/day were ommon. Fermentations of the duration of these studies produed total fluoresene hanges of 2 to 15 mv. It an be seen that the instrument drift an be a substantial portion of the overall signal hange during a fermentation. This effet beomes espeially important at high ell densities when a small hange in fluoresene orresponds to a large equivalent hange in biomass onentration due to the form of equation 3. DISCUSSION These results suggest that the measurement of ulture fluoresene will be able to provide aurate real-time estimates for fermentation biomass onentration under ontrolled environmental onditions. Before this potential an be ompletely evaluated, ompensation must be made for the drifting intensity of the exiting ultraviolet lamp. A new instrument is being developed in our laboratory at the State University of New York at Buffalo, whih will eliminate '2r 10 8 C- ' 6 o 4 E 2 0L E _) 0 o- _34 15r APPL. ENVIRON. MICROBIOL. FIG. 12. Culture fluoresene and Thermoatinomyes sp. biomass onentration data.

7 VOL. 35, 1978 FERMENTATION BIOMASS CONCENTRATION 343 lamp stability effets 'by using a double-beam approah, analogous to the ratio measurement priniple used in double-beam spetrophotometers. This new instrument will enable a muh more detailed analysis of the fluoresene signal with biomass onentration and ell metaboli ativity. It has also been shown that fluroesene is a strong funtion of the mirobial environment. Although this poses a problem in using fluoresene to estimate biomass onentration in fermentations in whih the environment annot be ontrolled, this harateristi may prove to be a valuable attribute. Evidene suggests that ulture fluoresene is a measure of ulture metaboli ativity, i.e., the produt of biomass onentration and the relative rate of metaboli ativity of eah ell. It is, therefore, reasonable to speulate that ulture ativity may prove to be a more important parameter to monitor and regulate than biomass onentration, when optimizing and ontrolling fermentations that produe produts related to ell metaboli proesses. ACKNOWLEDGMENTS This researh was supported by grants from the Ford Foundation (421219), the National Siene Foundation (ENG ), and the General Eletri Co. (GE037-D23001), and by New Brunswik Sientifi Co., Edison, N.J., and Assoiated Bio-engineers and Consultants In., Lehigh Valley, Pa. LITERATURE CITED 1. Armiger, W. B., D. W. Zabriskie, A. E. Humphrey, S. E. Lee, A. Moreira, and G. Joly Computer ontrol of ellulose fermentations: SCP prodution by thermophili Atinomyes. AIChE Symp. Ser. 158: Chane, B Spetrophotometry of intraellular respiratory pigments. Siene 120: Chane, B., R. W. Estabrook, and A. Ghosh Damped sinusoidal osillations of ytoplasmi redued pyridine nuleotide in yeast ells. Pro. Natl. Aad. Si. U.S.A. 51: Duysens, L. N. M., and J. Amesz Fluoresene spetrophotometry of redued phosphopyridine nuleotide in intat ells in the near-ultraviolet and visible region. Biohim. Biophys. Ata 24: Harrison, D. E. F., and B. Chane Fluorometri tehnique for monitoring hanges in the level of redued niotinamide nuleotides in ontinuous ultures of miroogranisms. Appl. Mirobiol. 19: Maneshin, S. K., and A. A. Arevshatyan Change in the fluoresene intensity of NAD-H2 in Candida guilliermondii in the transition from an anaerobi to an aerobi state. Prikl. Biohkim. Mikrobiol. 8: Mayer, D. H., J. R. Williamson, and V. Legallais A sensitive filter fluorometer for metabolite assays. Chem. Instrum. 1: Udenfriend, S Fluoresene assay in biology and mediine. Aademi Press In., New York. 9. Zabriskie, D. W., W. B. Armiger, and A. E. Humphrey Appliations of omputers to the indiret measurement of biomass onentration and growth rate by omponent balaning. In R. P. Jefferis III (ed.), Computer appliations in fermentation tehnology, Gesellshaft fur Biotehnologishe Forshung Monogr. series no. 3. Verlag Chemie, New York. Downloaded from on Deember 24, 2018 by guest