Fermentation Processes Leading to Glycerol

III. Fermentation Processes Leading to Glycerol Studies on Glycerol Formation in the Presence of Alkalis Imperial Chemical Industries Limited, Nobel Division, Research Department, Stevenston, Ayrshire, Scotland In addition to the modified yeast fermentation of hexoses in the presence of sulfites and bisulfites, which is characterized by dissimilation of a proportion of the hexose molecules by Neuberg's second form of fermentation, glycerol formation is also stimulated by alkalis and alkaline salts. In this case fermentation is in accordance with Neuberg's third equation. This method is frequently referred to as the "alkaline fermentation process" and is associated with the work of Eoff et al. (1919), in which media based on blackstrap molasses and other sources of sugars were fermented in the presence of sodium carbonate. The alkaline fermentation route to glycerol was also studied in Germany during World War I by Neuberg and Fiirber (1917), Neuberg and Hirsch (1919a, b) and Neuberg et al. (1920). Experiments on dissimilation of glucose at controlled ph values in the presence of ammonium hydroxide and sodium hydroxide, respectively, are described by Neish and Blackwood (1951). The rate of fermentation showed a broad optimum from ph 4.0 to 6.6 and as was expected from previous work the yield of glycerol and acetic acid increased with increase of the ph. Further experiments in the presence of alkalis are reported in the present paper. Development of an improved method of glycerol determination has made it possible to carry out an appraisal of the yields of ethanol and glycerol from a large number of yeast strains, in relation to the theoretical yields. The ph changes in the fermenting media have been studied and correlated with the corresponding variations during the course of fermentations in the presence of sulfites and bisulfites. Dissimilation of acetaldehyde, in the presence of alkali and of yeast cells, has been investigated. EXPERIMENTAL METHODS General Fermentation Technique Received for publication June 28, 1956 The general procedure was similar to that described by Freeman and Donald (1957a). The fermentation medium was based on diluted Cuban blackstrap molasses and contained reducing sugars, 16 to 22 g/100 ml as invert sugar, disodium hydrogen phosphate, 0.25 g/100 ml, and ammonium sulfate, 0.14 g/100 ml. The ph was adjusted to ph 6.8 by addition of sodium carbonate solution. The yeast inoculum was prepared by the method previously described. Eoff et al. (1919) suggested that "acclimatization" of the yeast culture to growth in the presence of sodium carbonate led to improved fermentations. In preliminary experiments, growth of yeasts in the presence of alkali (1.2 g Na2CO1 per 100 ml of inoculum medium) was found to be without significant effect on the kinetics of the subsequent fermentation or yields of products. The fermentations were carried out in stainless steel vessels, containing initially 7 L of solution and maintained at 30 C in thermostatically controlled water baths. After inoculation the medium was aerated with air (4 vol/vol/hr) for 0.5 hr to promote yeast growth. Vigorous visible fermentation was allowed to develop (3 to 4 hr) before additions of steering reagent were begun. In the standard procedure, sodium carbonate (30 g/100 ml solution) was added at intervals in a series of 5 unequal portions (as recommented by Eoff et al. 1919) to give a total dosage of carbonate equivalent to :30 per cent (as Na2CO3) of the weight of fermentable sugars. Visible fermentation was inhibited for a period after each addition of alkali and vigorous fermentation was allowed to become re-established before further additions were made. After the final addition, fermentation was allowed to proceed to completion, as determined by periodic determination of reducing sugars. Normally, fermentation was complete in 4 to 5 days. Analytical Methods Determinations of ph, reducing sugars, ethanol, acetic acid, and yeast cell counts were as described by Freeman and Donald (1957a). Acetaldehyde was determined by Ripper's (1900) method. Glycerol determinations were made by the chromatographic-periodate method of Sporek and Williams (1954). (The kerosene distillation method described by Freeman and Donald (1957a) also gave consistent values, which were, however, 8 per cent low.) Effect of Yeast Strain on Yields of Products The yields of products from fermentations in the presence of sodium carbonate varied widely with the nature of the yeast strain. A total of 41 strains, including vine yeasts, brewers' and distillers' yeasts, was 216

1957] FERMENTATION PROCESSES LEADING TO GLYCEROL. III 217 examined. Nine of the cultures were grown from single cell colonies isolated from a local brewery yeast (strain B.63 in the laboratory collection). Data from a series of representative fermentations are summarized in table 1. Glycerol yields varied from 10.5 to 24.2 per cent (in terms of total fermentable sugar) and the corresponding ethanol yields from 36 to 27 per cent. The observed ethanol yields were compared with the corresponding theoretical values by the following method. On the assumption that the whole of the hexose molecules were fermented in accordance with either equation (1) or (2), a graph (not reproduced here) was drawn relating the yields of glycerol, ethanol and acetic acid. C6H1206 = 2C2H5OH + 2CO2 (1) 2C6H1206 + H20 = 2C3H803 + CH3COOH + C2H5OH + 2CO2 (2) The observed glycerol yield (in terms of total fermentable hexose) was converted to terms of hexose fermented by dividing by the observed attenuation. The corresponding theoretical ethanol yield was obtained from the graph and converted to the same basis as the observed ethanol yield by multiplying by the attenuation value. Observed ethanol yields were, in general, 2 to 5 per cent lower than the theoretical values calculated from the corresponding glycerol yields, although different yeast strains varied considerably in TABLE 1. this. respect. Three strains, B.75/1, B.75/2 (culture) from single cell colonies of a Californian wine yeasts, and B.92 (distillers' yeast) gave the highest observed glycerol yield of 24.2 per cent with correspondingethanol yields of 28.7 to 32 per cent, as compared with the theoretical values of 31.3 to 31.8 per cent. Several other strains gave glycerol yields of 20 to 23 per cent. A number of strains, not included in table 1, gave very slow and incomplete fermentations in the presence of alkali; these included tokai, sauterne and Herrliberg wine yeasts and a mead yeast. Effect of Mode of Addition of Sodium Carbonate Addition -f sodium carbonate, equivalent to 30 per cent of the fermentable sugars, to the fermenting liquor at the stage when vigorous visible fermentation was first observed raised the reaction of the medium to ph 9 and caused irreversible inhibition of fermentation. Addition of the carbonate in a series of portions was therefore necessary. Eoff et al. (1919) recommended that this should be done in five parts of the solid corresponding to 0.125, 0.187, 0.313, 0.25, and 0.125 of the total. The following methods of sodium carbonate (Na2CO3) addition have been investigated: (a) 15 equal portions of solid at intervals of 0.75 to 1 hr, (b) the same as 30 per cent (wt/vol) solution, (c) five unequal portions of solid and (d) the same as 30 per cent (wt/vol) solution. In all cases, vigorous fermentation was Influence of yeast strain on time to completion of fermentation and yields of products Nature and Source of Yeast Culture Observed Yields (Per Cent of Theoretical Time to Reducing Fermentable Hexose) Ethanol Completion of Sugar Laboratory Description Source Yield Fermentation Attenuation Collection No. ~~~~~~~~~~~~~~Glycerol Ethanol % days B.63 Brewers' yeast Local brewery 20.6 28.3 34.2 4 97 B.63/1 Single-cell isolate from B.63 23.1 29.4 31.2 5 95 B.63/3 Single-cell isolate from B.63 19.1 30.7 33.1 5 93 B.63/9 Single-cell isolate from B.63 16.4 32.3 37.3 4 97 B.75/1 Single-cell isolates from Californian N.C.T.C., No. 2404 24.2 28.7 31.3 5 97 B.75/2 wine yeast 24.2 29.6 31.3 5 97 B.90 Saccharomyces cerevisiae Hansen N.C.Y.C., No. 82 10.5 9.9 41.8 7 97 Carlsberg strain B.92 S. cerevisiae Hansen, distillers' N.C.Y.C., No. 90 24.2 32.0 31.8 5 98 yeast B.105 Brewers' yeast C.R., No. 49 14.5 30.8 35.1 8 90 B.111 Super attenuative strain from C.R., No. 33 23.0 30.1 31.7 5 96 brewers' yeast B.71 Industrial alcohol distillery yeast 13.6 33.2 38.8 4 96 The fermentations were carried out by the standard method. Total sodium carbonate addition was 30 per cent of the fermentable sugar. Initial reducing sugar concentration was 20 g/100 ml as invert sugar. The method of calculation of the theoretical ethanol yield is described in the text. N.C.T.C. = National Collection of Type Cultures. N.C.Y.C. = National Collection of Yeast Cultures. C.R. = Cultures received from Dr. C. Rainbow, Department of Applied Biochemistry, The University, Birmingham.

218 allowed to become established before a further addition was made. The comparisons were made with yeast strain B.71. They showed that addition of sodium carbonate solution instead of the solid in five unequal portions had no significant effect on the kinetics of fermentation other than a slight increase of glycerol yield. Similarly, addition of the steering reagent in 15 equal portions, either as solid or solution, did not significantly affect the kinetics of fermentation or yields of products as compared with addition in five unequal portions. The ph changes in the fermenting solutions were, however, markedly dependent on the mode of addition of the reagent. The curves in figure 1 show the effect on the ph of the fermenting medium of addition of sodium carbonate (a) in 15 equal portions, total dosage 30 per cent mean of experiments with solid and dissolved reagent, which did not differ significantly, (b) in 5 unequal portions of the solution, total dosage 30 per cent. In (a) the ph of the fermenting liquor rose smoothly from ph 6 to ph 7.5 and finally rose to ph 7.6 to 7.8 in the concluding stages of the fermentation. When the carbonate was added in large portions, the ph of the medium increased rapidly to a peak, which in one case was as high as ph 8.2, and then fell again to ph 7.0 to 7.3, before the next alkali addition. The equilibrium ph at the completion of fermentation was not markedly affected by the mode of addition of the alkali. Corresponding curves were obtained when the total dosage was 50 per cent. Effect of Sodium Carbonate Dosage The effect of variation of the dosage of sodium carbonate was investigated in a series of fermentations in which the dosage was varied between zero and 49 per cent, based on fermentable sugar (table 2). Two fermentations were carried out at zero carbonate dosage, one in which the initial ph was 5.2 (corresponding approximately to that of industrial ethanol fermentations) and the other with an initial ph of 6.6 (corresponding to the remainder of the series). Glycerol and acetic acid yields increased with increase of carbonate dosage, whereas that of ethanol decreased progressively. The time to completion of fermentation also increased markedly at alkali dosages above 40 per cent. Sodium hydroxide and sodium bicarbonate were also examined as steering reagents but showed no advantages as compared with the carbonate. Effect of ph and Presence of Yeast Cells on the Dismutation of Acetaldehyde (a) Effect of ph in absence of yeast cells. The rate of dismutation of acetaldehyde in dilute aqueous solutions was investigated at ph values 3.4 and 8.1. Aqueous acetaldehyde solutions (1.1 g/100 ml, ph 3.4, 200 ml) were incubated in stoppered flasks at 35 C for 6 days. Similar solutions were adjusted to ph 8.1 by addition of NaOH and incubated. No detectable loss of acetaldehyde occurred at ph 3.4, but at ph 8.1 acetaldehyde concentration had fallen to 0.8 g/100 ml after incubation and the ethanol concentration was 0.5 g/100 ml. (b) Dissimilation in the presence of yeast. To four acetaldehyde solutions with initial concentrations of 0.1, 0.15, 0.20 and 0.50 g/100 ml at ph 8.0 (500 ml) was added commercial pressed yeast (10 g) and the suspensions were incubated at 30 C for 48 hr. ph values and acetaldehyde concentrations were determined at intervals (table 3). Although the ph of the solutions quickly fell to less than ph 4, rapid dismutation of the acetaldehyde took place. In a similar experiment in which the reaction of the medium was restored to ph 8.0 by addition of sodium hydroxide at hourly intervals the rate of dissimilation of acetaldehyde was slightly increased. 0 10 20 30 40 50 Tt- (Houqs) FIG. 1. Variation of ph of molasses medium during typical fermentations in the presence of sodium carbonate. (The arrows indicate times at which addition of alkali was completed.) TABLE 2. [VOL. 5 Variation in yields of products with sodium carbonate dosage Yields of Products (Per Cent of Sodium Fermentable Hexose) Time to Com- Reducing Carbonate pletion of Sugar Dosage Acetic Fermentation Attenuation Glycerol Etbanol acid % hr % 0* 5.1 41.2 1.8 144 94.3 Ot 6.5 39.0 3.0 144 98.1 5 15.1 38.5 3.5 120 96.5 10 17.1 36.5 4.2 120 97.7 20 19.8 32.7 5.1 120 97.7 25 18.4 33.3 97 100 30 22.0 32.2 5.7 120 98.4 35 21.4 30.2 100 98.2 40 20.2 29.8 121 97.4 45 23.3 28.9 295 94.6 49 22.0 24.4 295 78 In a series of fermentations the sodium carbonate dosage was varied from zero to 49 per cent of the total fermentable sugar. Yeast strain B.63. Alkali additions were made in five unequal portions. * Initial ph 5.2. t Initial ph 6.6.

1957] FERMENTATION PROCESSES LEADING TO GLYCEROL. III 219 The experiment was repeated on a larger scale to permit determination of the reaction products. Calcium carbonate, as buffer, (200 g) and commerical pressed yeast (100 g) were added to an aqueous acetaldehyde solution (0.5 g/100 ml; 5000 ml) and the mixture incubated at 30 C with mechanical stirring to maintain the yeast cells and buffer in suspension. As the reaction proceeded further quantities of acetaldehyde were added at intervals to increase the concentration of products without exceeding an acetaldehyde concentration of 0.5 g/100 ml, previously shown to permit yeast fermentation of glucose without excessive inhibition (Freeman and Donald, 1957b). Reaction ceased after 54 hr with the death of the yeast cells but the experiment was allowed to proceed for 120 hr. In a control experiment undertaken in parallel with a mixture containing yeast, water, and calcium carbonate relatively small amounts of acetaldehyde, ethanol, and acetic acid were produced by autolysis of the yeast and allowance was made for these. In the experiment acetaldehyde (32.0 g) underwent dismutation with formation of ethanol (16.2 g) and acetic acid (15.3 g). The theoretical production of these products from acetaldehyde (32.0 g) is ethanol (16.7 g) and acetic acid (21.8), which are in satisfactory agreement with the quantities found. DISCUSSION Yeast fermentation in the presence of sodium carbonate, as first described by Eoff et al. (1919), is often referred to as the "alkaline fermentation" of hexoses. Study of the ph changes during the course of the fermentation showed that the ph values were unexpectedly low, particularly when the steering reagent was added at 0.75 hr intervals. The ph values were of the same order as those obtaining during the sulfite fermentation (Freeman and Donald, 1957a), and both fermentations took place at ph values much more alkaline than those of the normal alcoholic fermentation of hexose at ph 4 to 5. During the period of vigorous fermentation in the presence of the steering reagent the ph was in the range 7.1 to 7.4 (8 to 85 hr) TABLE 3. Dissimilation of acetaldehyde in presence of yeast Flask No. ph Value Acetaldehyde Concentration o hr I hr 24 hr 48 hr O hr I hr 24 hr 48 hr g/100 ml 1 8.0 4.0 4.2 3.9 0.10 0.05 0.01 0.01 2 8.0 4.0 3.8 3.8 0.15 0.10 0.04 0.01 3 8.0 4.0 3.8 3.8 0.20 0.14 0.03 0.01 4 8.0 4.0 3.8 3.5 0.50 0.38 0.23 0.12 Commercial pressed yeast (10 g) was added to four acetaldehyde solutions (initial concentrations of 0.1, 0.15, 0.20 and 0.50 g/100 ml) at ph 8.0 (500 ml). The suspensions were incubated at 35 C. ph values and acetaldehyde concentrations were and the corresponding values for the sulfite fermentation are 6.7 to 7.3. These observations lead to interesting speculations on the mechanism of the fermentation. The accepted interpretation of fermentation in alkaline media (Neuberg's third form of fermentation) suggests that acetaldehyde undergoes dismutation with oxidation of one molecule to acetic acid and simultaneous reduction of another molecule to ethanol (cf Werkman and Schlenk, 1951). Lawrie (1928) states that in alkaline media acetaldehyde is transformed to ethanol and acetic acid by a Cannizzaro reaction. It is well known that conversion of aromatic aldehydes takes place readily in the presence of strong alkalis with formation of equimolecular proportions of the corresponding carbinol and acid in accordance with Cannizzaro's reaction, although aliphatic aldehydes do not readily undergo this reaction. Experiments in which acetaldehyde was incubated at ph 3 and ph 8 for 6 days showed that there was very little fall of aldehyde concentration in the alkaline medium and none at ph 3. In the presence of living yeast cells, however, rapid dismutation of acetaldehyde took place with production of approximately equimolecular proportions of ethanol and acetic acid. It is clear, therefore, that the conversion of acetaldehyde to ethanol and acetic acid is due to an enzymic reaction within the yeast cell and is not simply a chemical reaction undergone by an end product of the fermentation. A further point of interest is that, in the sulfite fermentation where approximately the same ph range is involved, this mechanism does not appear to operate since acetic acid is not formed as an end product when the sulfite dosage exceeds 50 per cent (of the fermentable hexose) although about half of the acetaldehyde molecules escape fixation with bisulfite. The end products of these fermentations may be dominated by the relative rates of certain enzyme reactions which are influenced by the ph of the medium and the presence of steering reagents. Experiments with isolated enzymes under these conditions would be of interest. The relative proportions of hexose molecules fermented by Neuberg's 3 forms of fermentation in the presence of various sulfite dosages were reported by Freeman and Donald (1957a). By similar methods it was established, from the data of table 2, that the following percentages of hexose molecules were fermented by the first and third routes when the sodium carbonate dosages were 0, 10, and 30 per cent: Sodium Carbonate Dosage determined at intervals. * Initial ph, 5.2. Hexose Molecules Fermented by Neuberg's Forms of Fermentation First Third O* 86.3 13.7 10 65.7 34.3 30 55.3 44.7

220 SUMMARY The yields of products and kinetics of fermentation of 41 strains of wine, brewers' and distillers' yeasts were compared in molasses fermentations in the presence of sodium carbonate. Glycerol yields varied from 10.5 to 24.2 (in terms of total fermentable hexose) and the corresponding ethanol yields from 36 to 27 per cent. Cultures from single cell colonies of a Californian wine yeast and a distillers' yeast gave the highest glycerol and ethanol yields of 24.2 and 29 to 32 per cent, respectively. The yields of ethanol closely approached the corresponding theoretical values of 31 to 32 per cent. Increase of sodium carbonate dosage from 0 to 30 per cent (in terms of total fermentable sugar) gave a progressive increase in yields of glycerol and acetic acid and a corresponding decrease in ethanol yield. ph changes during the fermentation were studied. The effect of ph and presence of yeast cells on dismutation of acetaldehyde have been investigated. When the aldehyde was incubated for 6 days at ph 8 there was very little fall of concentration and none at ph 3. In the presence of living yeast cells, the aldehyde was rapidly dissimilated with production of approximately equimolecular proportions of ethanol and acetic acid. REFERENCES EOFF, J. R., LINDER, W. V., AND BEYER, G. F. 1919 Production of glycerine from sugar by fermentation. Ind. Eng. Chem. (Ind.), 11, 842-845. [VOL. 5 FREEMAN, G. G. AND DONALD, G. M. S. 1957a Fermentation processes leading to glycerol. I. The influence of certain variables on glycerol formation in the presence of sulfites. Appl. Microbiol., 5, 197-210. FREEMAN, G. G. AND DONALD, G. M. S. 1957b Fermentation processes leading to glycerol. II. Studies on the effect of sulfites on viability, growth, and fermentation of Saccharomyces cerevisiae. Appl. Microbiol., 5, 211-215. LAWRIE, J. W. 1928 Glycerol and the Glycols. Chemical Catalog Co., New York City, New York. NEISH, A. C. AND BLACKWOOD, A. C. 1951 Dissimilation of glucose by yeast at poised hydrogen ion concentrations. Can. J. Technol., 29, 123-129. NEUBERG, C. AND FXRBER, E. 1917 tvber den Verlauf der alkoholischen Garung bei alkalisher Reaktion. I. Zellfreie Garung in alkalischen Losungen. Biochem. Z., 78, 238-263. NEUBERG, C. AND HIRSCH, J. 1919a Uber den Verlauf der alkoholischen Garung bei alkalisher Reaktion. II. Garung mit lebender Hefe in alkalischen Losungen. Biochem. Z., 96, 175-202. NEUBERG, C. AND HIRSCH, J. 1919b Der dritte Vergarungsform des Zuckers. Biochem. Z., 100, 304-322. NEUBERG, C., HIRSCH, J., AND REINFURTH, E. 1920 Die drei Vergarungsformen des Zuckers, ihre Zusammenhiange und Bilanz. Biochem. Z., 105, 307-336. RIPPER, M. 1900 Eine allgemeine anwendbare massanalytische Bestimmung der Aldehyde. Monatsh. Chem., 21, 1079-1084. SPOREK, K. AND WILLIAMS, A. F. 1954 Chromatographic determination of glycerol in fermentation solutions. Analyst, 79, 63-69. WERKMAN, C. H. AND SCHLENK, F. 1951 Chap. IX of Bacterial Physiology, (Ed. by C. H. Werkman and P. W. Wilson). Academic Press, Inc. New York City, New York.