APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Apr. 1988, p. 917-922

Vol. 54, No. 4

0099-2240/88/040917-06$02.00/0

Copyright © 1988, American Society for Microbiology

Selection of Ethanol-Tolerant Yeast Hybrids in pH-Regulated Continuous Culture JUAN JIMENEZ* AND TAHIA BENiTEZ

Departamento de Genetica, Facultad de Biologia, Universidad de Sevilla, Apartado 1095, E41080 Seville, Spain Received 15 October 1987/Accepted 9 January 1988

Hybrids between naturally occurring wine yeast strains and laboratory strains were formed as a method of increasing genetic variability to improve the ethanol tolerance of yeast strains. The hybrids were subjected to competition experiments under continuous culture controlled by pH with increasing ethanol concentrations over a wide range to select the fastest-growing strain at any concentration of ethanol. The continuous culture system was obtained by controlling the dilution rate of a chemostat connected to a pH-meter. The nutrient pump of the chemostat was switched on and off in response to the pH of the culture, which was thereby kept near a critical value (pHc). Under these conditions, when the medium was supplemented with ethanol, the ethanol concentration of the culture increased with each pulse of dilution. A hybrid strain was selected by this procedure that was more tolerant than any of the highly ethanol-tolerant wine yeast strains at any concentration of ethanol and was able to grow at up to 16% (vol/vol) ethanol. This improvement in ethanol tolerance led to an increase in both the ethanol production rate and the total amount of ethanol produced.

Ethanol tolerance in Saccharomvces species is known to be genetically controlled by a large number of genes (1, 7, 9). Consequently, it is very difficult to obtain highly ethanoltolerant yeast mutants, and the isolation of such mutants usually requires long-term selection techniques in continuous culture (4). The genetic analysis of ethanol-tolerant wine yeast strains has also led to the conclusion that many genes limit cell proliferation in the presence of ethanol (9). These genes are different in nonisogenic strains, indicating that nonisogenic hybridization (genetic complementation) might generate yeast hybrids more tolerant than their parental strains (7, 9). In fact, yeast hybrids have been found to be highly ethanol tolerant (9), and hybridization has been the usual method of generating ethanol-tolerant strains (7, 9, 14). Genes that limit growth are different at different ethanol concentrations (9), which indicates that the kinetics of ethanol-inhibited growth is the result of the inhibition of different cellular functions at increasing ethanol concentrations (9). Therefore, the selection of ethanol-tolerant strains should be carried out in a wide range of ethanol concentrations, since selection at any fixed concentration does not necessarily lead to an increase in tolerance at other levels. This consideration may be particularly relevant to ethanol production, where the fermenting strains need to grow and ferment under constantly increasing concentrations of ethanol rather than at any particular set value. In this study, a large number of wine-laboratory yeast strain hybrids were formed to increase genetic variability, generating highly ethanol-tolerant strains, and a selection method was developed that subjected the hybrids to continuous culture conditions with increasing ethanol concentrations. This selection method, which is extensively described, provides a useful long-term selection system because Saccharomyces spp. and other yeasts acidify the medium during growth on fermentable carbon sources (16). Ethanol tolerance and ethanol production by the selected hybrids are discussed below. *

MATERIALS AND METHODS Organisms. The yeast strains used in this study have been described. The wine strains IF182, IF1256, ACA4, and ACA21 and the "flor" strains FJF206 and FJF414, which are highly ethanol tolerant (3, 10), were selected from among 632 naturally occurring wine yeast strains (3) and characterized for ethanol production (8). The less tolerant laboratory Saccharomyces cerevisiae strains D273-11A (MATTa adel hisi), D517-4B (MATa ade2 lys9) (both provided by J. Conde, La Cruz del Campo, Seville, Spain), and MMY1 (MA To ura3-A52 cyhR) (provided by R. Bailey, Solar Energy Research Institute, Golden, Colo.) were also used. Wine strains were sporulated, and their meiotic products were crossed with haploid laboratory strains (9). The 25 winelaboratory yeast hybrids formed between IF1256 and either MMY1 (two hybrids) or D517-4B (two hybrids); between IF182 and either MMY1 (two hybrids) or D517-4B (two hybrids); between ACA4 and either MMY1 (three hybrids) or D517-4B (two hybrids); between ACA21 and either MMY1 (two hybrids) or D517-4B (two hybrids); between FJF206 and either D273-11A (two hybrids) or D517-4B (two hybrids); and between FJF414 and either D273-1lA (two hybrids) or D517-4B (two hybrids) have been described (9). Media. Complete medium YPD (1% yeast extract, 2% peptone, and 2% glucose), solid YPD (YPD plus 2% agar), solid minimal medium SD (0.17% yeast nitrogen base [Difco Laboratories] without amino acids or ammonium sulfate, 0.5% ammonium sulfate, 2% glucose, and 2% agar), and solid sporulation medium SPO (0.1% yeast extract [Difco], 1% potassium acetate, 0.05% glucose, and 2% agar) were used. When the YPD medium was supplemented with ethanol, the final concentration is indicated in the text as YPDE (percent ethanol). To supplement auxotrophic requirements, the appropriate amino acids or bases were added to SD (15). Culture conditions. (i) Growth in batch cultures. For later experiments in continuous culture, growth rate and pH changes were first determined in batch cultures. The strains used were grown at 30°C in 10-ml tubes containing 3 ml of YPD to the stationary phase. At this point, 0.5 ml was inoculated into 50-ml Erlenmeyer flasks containing 25 ml of either YPD or YPDE with the appropriate ethanol concen-

Corresponding author. 917

918

APPL. ENVIRON. MICROBIOL.

JIMENEZ AND BENiTEZ

1) results in a "semicontinuous" culture, since a pulse of dilution is followed by a pulse of batch growth. However, it will be referred to as a pH-controlled continuous culture because no differences exist when the duration of the dilution pulses is very short. The culture vessel with 1.5 liters of the appropriate medium was inoculated with 15 ml of an early-stationary-phase culture of the strain(s) used and kept at 30°C, with magnetic stirring (300 rpm) and low aeration (0.5 U in the fermentor scale). To isolate colonies, samples were taken periodically, each sample was appropriately diluted in water, and 0.1 ml was plated on YPD. Analytical procedures. Glucose was measured by injecting 20 ,ul of a culture sample into a YSI 27 glucose analyzer (Yellow Spring Instruments, Yellow Spring, Ohio). Ethanol concentrations were determined by the method of Kaplan and Giotti (11). pH was continuously registered with a standard recorder; more accurate measurements were obtained in a pH-meter (Crison Instruments, S.A., Madrid, Spain).

FIG. 1. Chemostat (schematic) that automatically controls culture density with a pH-meter. The nutrient pump of the chemostat, connected to the automatic pH controller, was switched on (D > >) and off (D = 0) in response to the pH of the culture, which was

therefore kept near culture system.

a

critical value, pHc, resulting in

a

continuous

tration, and the cultures were incubated at 30°C in a shaker. After inoculation, samples were taken periodically, and the absorbance at 660 nm (A660), the remaining glucose concentration, and the pH were determined until the glucose in the medium was exhausted. The pHs obtained were further used to fix the pH range in the continuous culture. An exponential increase in A660 between 0.1 and 0.5 was used to determine the growth rate, ,u. A660 was determined in a Spectronic 2000 spectrophotometer (Bausch & Lomb, Rochester, N.Y.). Previously, a linear relationship among cell number, dry weight, and A660 ranging from 0.1 to 0.5 was established. (ii) Ethanol production in batch culture. Samples (1.5 ml) of an early-stationary-phase culture of the selected hybrid and the wine strains ACA4 and IFI256 were inoculated into 100-ml flasks with 28.5 ml of YPD supplemented with either 30 or 45% glucose. The flasks, provided with a water trap to remove the CO2 produced, were incubated at 30°C in a shaker. Samples were taken periodically, and the remaining glucose and the ethanol produced, as well as the A660, were determined. Continuous culture controlled by pH. A Bioflo model C30 fermentor (New Brunswick Scientific Co., Inc., Edison, N.J.) with a 1.5-liter culture vessel was used. The fermentor was equipped with a pH electrode in the culture vessel, connected to an automatic pH controller (New Brunswick model pH-40). The nutrient pump of the fermentor was regulated to provide a dilution rate, D, of 0.5 h-1, higher than the highest growth rate of the strains used (0.47 h-1 for strain ACA4 in YPD at 30°C). The pump was connected to the pH controller so that when growth increased and the pH fell below a fixed value (pHc), the pH controller switched on the nutrient pump and the culture was diluted with fresh medium (D > p), increasing the pH. When, after dilution, the pH rose above pHc, the pH controller switched off the nutrient pump (D = 0). This feedback system (shown in Fig.

RESULTS AND DISCUSSION Steady state under continuous culture controlled by pH. To fix the pHs to be used in the continuous culture system, the initial (maximal) and final (minimal) pHs were measured in batch cultures of the FJF strains. The minimal pH was the pH measured in a culture after the glucose, the limiting substrate, had been exhausted. For strains FJF in YPD, the minimal pH oscillated between 4.9 and 5.0 (stationary-phase culture) compared with the initial pH of 6.0 (inoculated culture). Even in the presence of ethanol, the pH, which was initially above 6 (6.2 and 6.3 in YPDE with 5% and 10% ethanol, respectively), fell below 5 (4.9 and 4.7 in YPDE with 5% and 10% ethanol, respectively). Accordingly, the pH control described above was regulated to operate at a critical pH, pH,, between 5.0 and 6.0. When growing the FJF414 strain in the pH-controlled chemostat in YPD or YPDE medium, continuous acidification of the culture medium resulted from the growth and fermentation of the yeast cells (Fig. 2). As shown in the sketch of the control system (Fig. 1), the nutrient pump of the chemostat was switched on when the pH fell below the pHc (arrow in Fig. 2), thereby diluting the culture (D > i). The nutrient pump was switched off (D = 0) when the pH rose above pHc after dilution. This feedback control provided a long-term steadystate continuously growing culture, subjected to alternate pulses of dilution and growth (shown by the zigzag pH trace in Fig. 2). During this steady state, the glucose concentration in the culture vessel was almost constant, varying between 1.7 and 2.0% under all conditions used (Table 1), and very similar to the initial concentration (2.2%), which indicates that the cells were growing at their highest specific growth rate because glucose, the limiting substrate, was in excess (13). Since in this equilibrium the constant pH variation gave rise to a proportional turbidity variation (Fig. 2, Table 1), the TABLE 1. Maximal and minimal pH, A660 values, and glucose concentrations at steady state of a continuous culture controlled by pH of strain FJF414 Glucose

pH

Medium

YPD YPDE (5% ethanol) YPDE (10% ethanol)

A6w

concn (%, wt/vol)

Initial

Final

Initial

Final

Initial

Final

5.67 5.62 5.88

5.66 5.59 5.84

0.15 0.18 0.11

0.12 0.16 0.09

1.9 1.9 2.0

1.8 1.8 2.0

VOL. 54, 1988

SELECTION OF ETHANOL-TOLERANT YEAST HYBRIDS

919

6.2

6.0 F

Q

pHc5.8

t A

I

0

10

20 Time (hr)

30

FIG. 2. Time course traces of pH in a continuous culture controlled by pH of strain FJF414 growing in YPDE with 10% ethanol. Once a (pH,) was reached (arrow), the culture was subjected to dilution pulses whenever the pH went below the pH,.

set pH

acidification rate, A (decrease of pH per unit of time when D = 0) obtained after each dilution pulse was proportional to the specific growth rate, I. (increase in turbidity per time unit when D = 0); that is, A = k,u (k = empiric constant). In theory, if the range of pH variations is shortened, the duration of the dilution pulses is also shortened and growth and dilution are equilibrated without apparent fluctuations, and the system results in a turbidostatic continuous culture controlled by pH; however, the continuous culture shown in Fig. 2 has some advantages, as described below. The alternate sequence of dilution and growth under the steady state of the feedback control described introduces a very simple method of increasing, pulse by pulse, the ethanol concentration of the culture. This allows the effects of a particular ethanol concentration obtained after each pulse on the acidification rate of the yeast culture to be studied. Once equilibrium is reached in the culture with the initial concentration of ethanol (ro), the ethanol concentra-

6.0

ro r, r2 r3

r4

r5

12.8

13.5

14.2

10.0 11.1

5.8

tion in the fresh medium (that of the reservoir vessel) can be increased to a higher value, R. In consequence, the next dilution pulse, caused by pumping this medium for a time t at a dilution rate D, will cause the ro concentration of ethanol in the culture to increase to a higher value, r, = R(1 - e-DI) + ro(eD%). In general, after n pulses the ethanol concentration in the culture increases up to the value rn = R(1 - e Dtn) + ro(e Dtn). The effects of the increased ethanol concentration on growth and fermentation are immediately reflected in the subsequent acidification rate. For instance, as Fig. 3 shows, when the FJF414 strain was grown in continuous culture in YPDE with 10% ethanol (ro = 10), an initial equilibrium was reached. The ethanol concentration in the reservoir vessel was then increased to 18% (R = 18). After the next pulse, the ethanol concentration of the culture was increased and a reduction in the acidification rate was observed. The ethanol concentration of the culture increased after each pulse until, after the fifth pulse, the acidification rate was nil (Fig. 3).

12.0

t

20

30

40

50

Time (hr) FIG. 3. Time course traces of pH under steady-state conditions in a continuous culture controlled by pH of strain FJF414 growing in YPDE with 10% ethanol. Arrow indicates the time that the medium supply was supplemented until the concentration of ethanol reached 18%. The r, values indicate the ethanol concentration generated in the culture vessel after n pulses of dilution, according to the general formula given in the text.

APPL. ENVIRON. MICROBIOL.

JIMENEZ AND BENfTEZ

920

A

IL

*LL

dt 50

50

LiL

0

0

20

10

T i me

100

(hr)

FIG. 4. Variation in the percentage of CFU of strains FJF414 and D517-4B, subjected to continuous culture controlled by pH in YPD (A), YPDE with 5% ethanol (U), or YPDE with 10% ethanol (0).

According to the above equation, for initial ro of 10, R of 18, and after five pulses (n = 5) of 0.3 h per pulse (t = 0.3) at a dilution rate of 0.5 h-1, the ethanol concentration of the culture should have been r5 = 14.2%, which was the maximal ethanol concentration that permitted growth and fermentation. When the actual ethanol concentration was measured, the results were slightly different from the theoretical values, probably due to the loss of evaporated ethanol, the real value being 12.5%. Therefore, the theoretical expressions for rn should be considered only approximate in the case of ethanol. Since A = k,u when an initial ro of 0 is used, the overall ethanol inhibition kinetics for growth may be obtained from the recorded pH trace in a wide range of ethanol concentrations. Selection of ethanol-tolerant strains. Figure 4 shows the evolution of a continuous culture in which a mixture of strains FJF414 and D517-4B were grown. The proportion of cells of each type was established by plating an appropriately diluted sample of the culture on YPD. Both strains were previously characterized in batch cultures for their ability to grow in the presence (YPDE) and absence (YPD) of ethanol and proved to be ideal for simulation of competition studies under conditions for selection of the highest growth rate (pu) in the presence of ethanol. This is because FJF414 has a higher growth rate than D517-4B when grown in YPDE with more than 5% ethanol (p, = 0.14 h'- and 0.06 h-1 for FJF414 and D517-4B, respectively, in batch cultures of YPDE with 10% ethanol at 30°C) but has a lower growth rate when grown in a lower ethanol concentration (pu = 0.29 and 0.35 h-' for FJF414 and D517-4B, respectively, in batch cultures with YPD at 30°C). In addition, colonies of FJF414 and D517-4B were easily distinguishable because the latter develop a pink color, conferred by the ade2 marker. As shown in Fig. 4, the initial 1:1 proportion of cells from each strain changed to nearly 100% FJF414 when grown in YPDE with 10% ethanol, but to almost 100% D517-4B when grown in YPD. In YPDE with 5% ethanol, there was a slight increase in the proportion of FJF414 cells. This increase was very gradual, in accordance

with the slight difference previously found in the growth rate between both strains under these conditions (,. = 0.27 and 0.25 h-' for FJF414 and D517-4B respectively, in batch cultures with YPDE with 5% ethanol at 30°C). In consequence, the above in vivo simulation showed that, although other factors could also be at work in this competition experiment, the system described selected the strain able to grow more rapidly under the established conditions. Twenty-five different wine-laboratory strain hybrids were mixed and used in competition experiments under continuous culture controlled by pH, in YPD, subjecting them to increasing ethanol concentrations to select the best strains at any ethanol concentration. After 8 days of incubation, when the ethanol concentration was 16.1% (vol/vol), the acidification rate of the remaining strains was nil, indicating that no strain could continue to grow at higher concentrations of ethanol. A sample from the culture with 16.1% ethanol was plated on YPD, and 10 of the resulting colonies were characterized. The flor characteristic is a dominant feature exclusively shown by the FJF strains and all their meiotic products (9). These wine-laboratory hybrids were able to grow as a thin layer on the liquid surface when maintained in YPDE (8% ethanol) for several days (characteristic of the flor yeasts) (8, 9). Their meiotic products showed a 2:2 segregation of the ade and lys markers (characteristics of the laboratory strain D517-4B; see Materials and Methods). The results therefore indicate that the most ethanol-tolerant hybrids selected under a wide range of ethanol concentrations are those formed between meiotic products of either strain FJF414 or FJF206 and laboratory strain D517-4B. This continuous culture controlled by pH provides a precise method for the study of the kinetics of growth inhibition by ethanol and a long-term selection system in the presence of this inhibitor. Ethanol production by selected hybrids. The strains used in industrial fermentation need to work under constantly increasing concentrations of ethanol. Optimal strains should therefore display an ethanol tolerance which is near the maximal tolerance established for any concentration of ethanol from 0 to 16%. The selection system described allowed the isolation of the most ethanol-tolerant strains in a wide range of ethanol concentrations up to 16%. To study the improvement in ethanol production obtained in the selected ethanol-tolerant hybrids, glucose consumption and ethanol production were determined for one of these hybrids (FDHS) and for strains ACA4 and IF1256, the best-fermenting wine yeast strains previously selected by conventional screening methods (3) and characterized for ethanol production (8). At a 30% initial glucose concentration (Fig. 5A, B, and C), the wine yeast strains were unable to ferment all the sugar; the remaining glucose concentration was 3.0% with strain ACA4 and 3.4% with strain IF1256. The hybrid FDHS consumed all the glucose in the medium, producing a final ethanol concentration of 15.2%, whereas IFI256 and ACA4 produced nearly 14% (13.8 and 13.9%, respectively). The maximal rate of glucose consumption (during the first 35 h, Fig. 5) was also obtained with the hybrid FDHS, this rate being 0.76 g/100 ml per h, whereas IF1256 consumed 0.71 g/100 ml per h and ACA4 consumed 0.66 g/100 ml per h. This result indicates that, as a consequence of the increase in ethanol tolerance, the hybrid produced more ethanol and at a higher rate than the best-fermenting strains selected previously. At a 45% initial glucose concentration (Fig. SD, E and F),

VOL. 54, 1988

SELECTION OF ETHANOL-TOLERANT YEAST HYBRIDS

921

&CS' E

-iw 40

CD 0

HOURS FIG. 5. A660 (O), glucose consumption (0), and ethanol production (A) of the hybrid strain FDHS (A, D) and the wine strains IF1256 (B, E) and ACA4 (C, F) in YPD supplemented with 30% (A, B, C) or 45% (D, E, F) glucose.

the remaining glucose concentration after 5 days of incubation was 7% when the fermenting strain was the hybrid FDHS, but 10% when this strain was IF1256 and 9% when the strain was ACA4, the concentrations of ethanol produced being 16.1, 14.1, and 14.4%, respectively. The rate of ethanol production in 45% glucose was always slightly lower than in 30% glucose, probably due to the synergistic effects of the ethanol produced and the remaining substrate, whose inhibitory effects are well known (12). Growth and fermentation are partly linked parameters, so that fermentation rate increases proportionally to a given increase in growth rate. In accordance with these results, the increase in ethanol-tolerant growth for the selected strain gave rise to a proportional increase in the rate of ethanol production (Fig. 5). In addition, yeast cells are usually able to continue producing ethanol after growth has stopped (2, 5). The fact that the ethanol concentration that completely inhibits fermentation is higher than that able to inhibit growth (5, 6) indicates that the glycolytic enzymes are more ethanol tolerant than other enzymes involved in cell growth. However, the maximal ethanol concentration which completely inhibited growth of the selected hybrid was similar to the maximal ethanol concentration that could be produced from glucose, both concentrations being ca. 16%. This result suggests that at extremely high ethanol concentrations, 16%, many cellular functions, including the glycolytic enzymes, may be affected to such an extent that both growth and ethanol production are inhibited.

In Saccharomyces strains, there are many genes and, for each of these genes, several alleles able to limit ethanoltolerant growth (1, 9). The best ethanol-tolerant strain selected is the result of gathering in a single hybrid the alleles most able to permit growth at ethanol concentrations at which highly ethanol-tolerant wine yeast strains do not grow. ACKNOWLEDGMENTS We thank A. F. Estefane for skillful technical assistance, J. Conde and I. L. Calder6n for useful discussions, and J. I. Zoltowski and R. Rhett for correcting the manuscript. This work was supported by the CAICYT, Programa I + D, project number 24/AG.

1. 2.

3.

4.

5.

LITERATURE CITED Aguilera, A., and T. Benitez. 1986. Ethanol-sensitive mutants of Saccharomyces cerevisiae. Arch. Microbiol. 143:337-344. Aguilera, A., and T. Benitez. 1985. Role of mitochondria in ethanol tolerance of Saccharomyces cerevisiae. Arch. Microbiol. 142:389-392. Benitez, T., L. Del Castillo, A. Aguilera, J. Conde, and E. Cerda-Olmedo. 1983. Selection of wine yeast for growth and fermentation in the presence of ethanol and sucrose. Appl. Environ. Microbiol. 45:1429-1436. Brown, S. W., and S. G. Oliver. 1983. Isolation of ethanoltolerant mutants of yeast by continuous selection. Eur. J. Appl. Microbiol. Biotechnol. 16:116-122. Casey, G. P., and W. M. Ingledew. 1986. Ethanol tolerance in

922

JIMENEZ AND BENITEZ

yeasts. Crit. Rev. Microbiol. 13:219-280. 6. Ingram, L. O., and T. M. Buttke. 1984. Effects of alcohols on microorganisms. Adv. Microb. Physiol. 25:253-300. 7. Ismail, A. A., and A. M. M. Ali. 1971. Selection of high ethanol-yielding Saccharomyces. II. Genetics of ethanol tolerance. Folia Microbiol. 16:350-354. 8. Jimenez, J., and T. Benitez. 1986. Characterization of wine yeasts for ethanol production. Appl. Microbiol. Biotechnol. 25: 150-154. 9. Jimenez, J., and T. Benitez. 1987. Genetic analysis of highly ethanol-tolerant wine yeasts. Curr. Genet. 12:421-428. 10. Jimenez, J., and N. van Uden. 1985. Use of extracellular acidification for the rapid testing of ethanol tolerance in yeasts. Biotechnol. Bioeng. 27:1596-1598. 11. Kaplan, N. O., and M. M. Giotti. 1957. Enzymatic determina-

APPL. ENVIRON. MICROBIOL. tion of ethanol. Methods Enzymol. 3:253-255. 12. Moulin, G., P. Boze, and P. Galzy. 1981. A comparative study of the inhibitory effect of ethanol and substrate on the fermentation rate of the parent and a respiratory-deficient mutant. Biotechnol. Lett. 3:351-356. 13. Pirt, S. T. 1975. Principles of microbe and cell cultivation. Blackwell Scientific Publications, Oxford. 14. Seki, T., S. Myoga, S. Limtong, S. Vedono, J. Kumnuanda, and M. Taguchi. 1983. Genetic construction of yeast strains for high ethanol production. Biotechnol. Lett. 5:351-356. 15. Sherman, F., G. R. Fink, and C. W. Lawrence. 1979. Methods in yeast genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 16. Watson, T. G. 1972. The present status and future prospects of the turbidostat. J. Appl. Chem. Biotechnol. 22:229-243.