The Conversion of Mannitol to Fructose by Acetobacter suboxydans

Transcription

1 The Conversion of Mannitol to Fructose by Acetobacter suboxydans MERLIN H. PETERSON, WALDO C. FRIEDLAND, FRANK W. DENISON, JR., AND J. C. SYLVESTER Species of the genus Acetobacter are well known for their characteristic incomplete oxidations of polyhydric alcohols, sugars, and polyhydroxy acids (Butlin, 1936; Underkofler and Hickey, 1954). Of particular interest has been the preparation of ketose sugars through the oxidative action of Acetobacter suboxydans on certain polyhydric alcohols (Fulmer and Underkofler, 1947). Of these, the conversion of D-sorbitol to L-sorbose has been most extensively studied with Wells et al. (1937, 1939) ultimately reporting the successful pilot plant production of L-sorbose under submerged fermentation conditions. The conversion of D-mannitol to D-fructose by the oxidative action of A. suboxydans, which was first reported by Bertrand (1904), has also been studied but to a considerably lesser extent. Laboratory-scale studies on this conversion under surfaceculture fermentation conditions were first reported by Fulmer et al. (1939) and more recently by Isbell and Karabinos (1952). With a surface: volume ratio of 1:195, Fulmer et al. (1939) found that mannitol concentrations up to and including 25 per cent weight per volume (w/v) were consistently converted to fructose to the extent of 90 per cent or better of theoretical in 7 days or less. The findings of Fulmer et al. (1939) were confirmed by the work of Isbell and Karabinos (1952), but only with regard to low mannitol concentrations (less than 5 per cent). As far as is known, these surfaceculture laboratory-scale investigations are the only published reports available on the conversion of mannitol to fructose by A. suboxydans. Submerged fermentation studies and pilot plant studies have apparently not been reported prior to this time. The purpose of the work reported here was to study the submerged production of D-fructose from D-mannitol by A. suboxydans in shaken flasks and three sizes of pilot plant equipment, with the ultimate goal of developing a satisfactory fructose fermentation not only from the microbiological viewpoint but also from the viewpoint of satisfactory recovery and purification of the fructose produced. Only recovery details pertinent to the development of the fermentation are covered in this report. The complete details of the fructose recovery studies involved in this project have already been given (Cushing and Davis, 1955). MATERIALS AND METHODS Culture. The NRRL B-72 strain of A. suboxydans was used for all of the studies described here. The permanent Abbott Laboratories, Research Division, North Chicago, Illinois Received for publication June 26, 1956 stock culture of this strain is maintained in lyophilized form. Working stock cultures were inoculated directly from the permanent lyophilized stock culture when necessary, and consisted of test-tube agar-slant cultures on the following medium: glycerol, 5.0 per cent; Difco yeast extract, 0.5 per cent; agar, 1.5 per cent; and tap water. These working stock cultures were maintained indefinitely by bi-weekly serial transfer to fresh slants of the same medium. Recourse to the permanent lyophilized stock culture on a routine schedule was consequently unnecessary. Freshly inoculated working stock cultures were incubated at 28 C for 36 to 48 hr and either were used immediately or were stored in a cold room (4 C) to be used at any time during the ensuing 2-week period. Mannitol. The D-mannitol used throughout this investigation was obtained from the Atlas Powder Company, Wilmington, Delaware. Analyses demonstrated it to have a purity of 99.5 per cent. Effective aeration rates. Effective aeration rates were determined by oxidation of sulfite in the presence of copper catalyst according to the method of Cooper et al. (1944). Fermentation samples. The course of the fermentation of mannitol to fructose by A. suboxydans was followed by removing beer samples at intervals for fructose analyses. Before analysis, the beer samples were clarified by filtration through diatomaceous earth. These clarified beer samples were also used for total soluble solids analyses. Fructose analyses. For fructose analyses, reducingsugar values were determined by the method of Shaffer and Somogyi (1933), and the corresponding fructose levels were read from a standard curve prepared from analyses of pure D-fructose by this method. Total soluble solids analyses. Total solids analyses on clarified beer samples were carried out using the Karl Fischer method (U. S. Pharmacopeia, 1955) to determine total moisture, and then total soluble solids were calculated by difference. This method for total soluble solids will be hereafter designated as the Karl Fischer method. All of the total soluble solids data used in this presentation were obtained using this method. An alternative direct method for determining total solids in clarified fructose beer samples, subsequently found to be more satisfactory than the Karl Fischer method, is the quartz-sand vacuum-drying procedure for fructose syrups of the Association of Official Agricultural 316

2 1956] CONVERSION OF MANNITOL TO FRUCTOSE 317 Chemists (Association of Official Agricultural Chemists, 1945). Fructose yields. Fructose yields from mannitol were determined using two methods. The standard method, which was satisfactory for all shaken-flask fermentations and all fermentations in the 30-L fermentors to which no solid mannitol was fed after inoculation, consists of calculating and expressing as per cent the ratio of the fructose produced and that theoretically possible based on the initial mannitol level corrected for inoculum dilution and inoculum mannitol. This standard method was found to be unsatisfactory, however, when the fermentation was carried out in the 50-gal and 500- gal fermentors due to the inaccuracies in measuring and difficulties in controlling volumes in these equipments. The volume measurements and control problems in these equipments are due not only to the errors in volume control during the initial charging and inoculation but also to the lack of satisfactory corrections for volume changes during sterilization by the steam injection method and during the fermentations themselves. As a result, the fructose analyses on these largerscale fermentations had no quantitative significance since it was not known accurately what the mannitol concentrations were initially or at any time during fermentation. For obvious reasons of a similar nature, this standard method was also found to be unsatisfactory for any fermentations to which mannitol was fed after inoculation. The problem was critical, particularly with regard to the larger-scale fermentations, since recovery and purification studies had disclosed no practical largescale method for the purification of fructose in the presence of mannitol where the ratio of fructose to mannitol is less than 19. It was then felt that the best approach to these problems was to find an assay for mannitol in the presence of fructose. No such assay could be found in the scientific literature, however, and laboratory investigations along this line were also unsuccessful. Therefore, a second fructose yield method, known as the total solids method, was developed for the two fermentation types described above. This method is based on the extent to which fructose makes up the total soluble solids of a beer sample, and consists of calculating and expressing as per cent the ratio of the fructose and total soluble solids present. In support of the soundness of this total-solids method for calculating yields, were recovery studies on completely fermented beers produced from the 18 to 35 per cent mannitol media described below. These studies demonstrate that the total soluble solids in such beers normally consist of fructose to the extent of at least 98 per cent. Comparison of this totalsolids method with the standard method on shaken flask and 30-L fermentor fermentations showed that the two methods give practically identical results on the fermentation media containing 15 or more per cent mannitol described below. RESULTS Fermentations in shaken flasks. Preliminary studies on the oxidation of mannitol to fructose under submerged fermentation conditions were carried out in shaken flasks. Unless indicated otherwise hereafter, shakenflask fermentations were carried out using the following standard procedure. Cotton-plugged 500-ml Erlenmeyer flasks containing 100 ml of mannitol-0.25 per cent Difco yeast extract production medium adjusted to ph 5.8 with dilute H2SO4 were autoclaved 20 min at 15 pounds of steam pressure. After cooling, each flask was inoculated with 5 per cent (by volume) of second-transfer, 24-hour shaken-flask inoculum culture grown in the following medium: mannitol, 7.0 per cent; Difco yeast extract, 0.5 per cent; KH2PO4, 0.1 per cent; dilute H2SO4 to ph 5.8; and tap water. Second-transfer shaken-flask inoculum cultures were previously inoculated with 1 per cent of first-transfer, 24-hr shakenflask inoculum culture grown in the same medium. Firsttransfer shaken-flask inoculum cultures were inoculated with a suspension of the vegetative growth on working stock culture agar slants prepared as described in the section on Materials and Methods. Ten ml of growth suspension were prepared from each working stock culture slant and two first-transfer inoculum shaken flasks were inoculated from each suspension. All shaken flasks (inoculum and production) were incubated at 28 C on a rotary shaker operating at 240 rpm with an eccentric radius of 2.25 in. Fructose yields for all shaken-flask fermentations were calculated using the standard procedure described in the section on Materials and Methods. The principal consideration in the shaken flask experiments was the establishment of a set of standard submerged fermentation conditions under which mannitol would be consistently converted to fructose by A. suboxydans to the extent of 95 per cent or better of theoretical. These standard conditions have been described and the results in table 1 show them to be satisfactory for mannitol medium concentrations up to approximately 24 per cent (w/v). Due to the solubility limit of approximately 25 per cent for D-mannitol at 28 C, 24 per cent was the highest initial mannitol concentration tried in shaken-flask fermentations. Since the cost of Difco yeast extract would make its use in large-scale fermentation equipment prohibitive, preliminary tests on various other supplementary nutrient materials were carried out in shaken-flask fermentations. The data are presented in table 2. The corrected initial mannitol concentration in the shakenflask production medium was g per 100 ml for all runs. As may be seen, fructose yields greater than 95 per cent were obtained with corn steep liquor and brewers yeast # 2019 (primary-grown dried whole yeast). Both of these supplementary nutrient materials

3 318 PETERSON, FRIEDLAND, DENISON, AND SYLVESTER TABLE 1. Fructose production by Acetobacter suboxydans NRRL B-72 in shaken flasks at various initial mannitol levels Initial Mannitol Fructose Produced Fructose Yieldt Fermentation Level* PrdcdFuts ilt Period g/100 ml g/100 ml % hr * Initial manaitol levels corrected for inoculum dilution and mannitol supplied in the inoculum. t Fructose yields calculated by the standard method: F. Y. theoretical fructose possible based on initial mannitol level. TABLE 2. Effect of various supplementary nutrient materials on fructose production by Acetobacter suboxydans NRRL B-72 in shaken flasks Production Medium Used* Fructose FtaiOn Yieldt Peio % hr Mannitol-0.25% Difco yeast extract (standard) Mannitol-1.0% peptone Mannitol-1.0% brewers yeast Mannitol-0.5% brewers yeast Mannitol-0.25% brewers yeast Mannitol-1.0% corn steep lot # Mannitol-0.5% corn steep lot Mannitol-0.5% corn steep lot * The corrected initial mannitol level was g per 100 ml in all the production media tried. t Fructose yields calculated by the standard method: F.Y. theoretical fructose possible based on initial mannitol level. appeared to be satisfactory replacements for the Difco yeast extract in the shaken-flask production medium. Fermentations in 30-L fermentors. For fermentation development studies in small equipment closely resembling that used in large-scale submerged fermentation operations, 30-L stirred-jar fermentors of the type described by Friedland et al. (1955) were used. Unless indicated otherwise hereafter, fermentations in these 30-L fermentors were carried out using the following standard procedure. Ten liters of production medium adjusted to ph 5.8 with dilute H2SO4 were placed in a fermentor which was then autoclaved 30 min at 15 pounds of steam pressure. After cooling, each fermentor [VOL. 4 was inoculated with 5 per cent of 24-hr inoculum culture grown in an aerated bottle at 28 C on the following medium: mannitol, 7.0 per cent; Difco yeast extract, 0.5 per cent; KH2PO4, 0.1 per cent; dilute H2SO4 to ph 5.8; and tap water. Aerated-bottle inoculum cultures were previously inoculated with 2 per cent of secondtransfer, 24-hour shaken-flask inoculum culture prepared as described in the preceding section on shakenflask fermentations. The inoculated fermentors were incubated at 28 C under the following conditions: agitation, 480 rpm; aeration, 10 L per min, or 1.0 volume of air per volume of medium per min. These agitation-aeration conditions represent an effective aeration rate of 2.85 millimoles of oxygen per L per min. The use of the standard mannitol-0.25 per cent Difco yeast extract shaken-flask production medium under the standard 30-L fermentor conditions described above gave as good results as were obtained in shaken flasks. As in shaken-flask runs, fructose yields greater than 95 per cent of theoretical were consistently obtained with initial mannitol concentrations up to 24 per cent. Typical fermentation data at two mannitol levels for both 30-L fermentor and shaken-flask standard runs are given in figure 1. The fructose yields given were calculated using the standard procedure described in the section on Materials and Methods. As may be seen, the principal difference between shaken flasks and 30-L fermentors was the greatly increased mannitol oxidation rate obtained in the 30-L fermentors, which correlates with the 4igher aeration efficiency obtained in these fermentors as compared to that normally obtained in shaken flasks. The time required in 30-L fermentors for oxidation of either of the two mannitol levels to the extent of 95 per cent or better of theoretical was approximately one-third that required for the same mannitol level in shaken flasks. The low ph levels at which the major portion of mannitol oxidation occurred in both types of equipment were apparently due to the production of acetic acid, as indicated by the characteristic odor produced. As pointed out previously, the use of initial mannitol levels much above 24 per cent in the shaken flask and 30-L fermentor production medium was impossible due to the solubility limit of approximately 25 per cent for D-mannitol at 28 C. Studies in the 30-L fermentors showed, however, that beers containing up to 35 per cent of fructose with 95 per cent or better fructose yields could be obtained by adding additional solid mannitol batchwise to a standard 20 per cent initial mannitol fermentation after approximately 75 per cent of the initial mannitol charge has been converted to fructose. Fructose yields for 30-L fermentor runs of this type were calculated using the total solids procedure described in the section on Materials and Methods. Beers containing 30 and 35 per cent of fructose with 95

4 1956] CONVERSION OF MANNITOL TO FRUCTOSE 319 Fructose / 8 /// ~ 7 I //.~~ l0 JI 6 6/ 2~~ Hour Hours % Mae.nitol Fructo-se Fermentatio Yield Period % hrs-. ---Sh. Flask L Feom FIG. 1. ph and fructose content of media during fez per cent or better fructose yields were produced by this solid mannitol addition procedure in the 30-L fermentors in 40 and 48 hr, respectively. At this stage in development of the fructose fermentation, recovery and purification studies were initiated. Four major recovery criteria for fructose beers were established. These were as follows: conversion of mannitol to fructose, 95 per cent or better of theoretical as previously indicated; minimal or removable colored substances, or substances producing color under recovery conditions; minimal or removable pyrogenic and antigenic substances; and satisfactory media cost. The fructose beers produced in the normal fermentation periods indicated in the 30-L fermentors u,sing the mannitol-0.25 per cent Difco yeast extract production medium under the standard fermentation conditions previously described were found to meet all these criteria except the last. As previously pointed out in the section on shaken-flask fermentations, the cost of Difco yeast extract would make its use in large-scale fermentarmen // / / _ S ph ~~/ 6 /~~~~~~~~~ phph so 100 Hours 23*NJ Mamnitol Fructose Fermentation Yield Period % hrs.,-sh., F:Lask 9q L Ferm ttations in shaken flasks and the 30-L fermentors. tion equipment prohibitive. Corn steep liquor and brewers yeast ^' 2019 (primary-grown dried whole yeast) appeared to be satisfactory low-cost replacements for Difco yeast extract, since fructose yields greater than 95 per cent were obtained with either of them in shaken-flask fermentations (table 2). More extensive testing of these two materials in the 30-L fermentors, accompanied by recovery and purification studies on the fructose beer produced, demonstrated that brewers yeast S 2019 was the better. Not only were inconsistent fructose yields obtained with different corn steep liquor lots, but color problems in fructose recovery were encountered. Fructose beers produced in the 30-L fermentors using a mannitol-0.5 per cent brewers yeast p2019 production medium and a 7 per cent mannitol- 0.5 per cent brewers yeast N per cent KH2PO4 inoculum medium under the standard fermentation conditions previously described were found to meet all four recovery criteria given above for fructose beers. Pilot plant fermentations. Pilot plant fructose fermen-

5 320 PETERSON, FRIEDLAND, DENISON, AND SYLVESTER [VOL. 4- TABLE 3. Fructose fermentations in 50-gal fermentors Solid Harvest Total Soluble Fructose at Fructose Antifoam Usedt Remarks Run No. Production Medium Used Mannitol Age Solids at Harvest Yieldt Fed* Ae Harvest Hret Yed % hr g/100 ml g/100 ml % 1 Mannitol-20% None None Excessive foaming 2 Mannitol-18% None DC antifoam emulsion Foaming controlled B.Y. # 2019{-0.5% 3 Mannitol-24% None CLRS Excessive foaming 4 Mannitol-20% DC antifoam emulsion Foaming controlled B.Y. # % * Solid mannitol fed in run # 4 at 23 hr after inoculation. t Fructose yields calculated by the total solids methods: F.Y. = totalpsoluble solids t DC antifoam emulsion is a silicone-emulsion type antifoam made by Dow Corning Corp., Midland, Michigan. CLRS is a lipidtype antifoam made by Biocompounds Company, Chicago, Illinois. tations were studied in 50-gal and 500-gal stirred-tank fermentors. The satisfactory standard conditions described previously for fermentations in the 30-L fermentors were duplicated as closely as possible. Since all of the fructose produced in these pilot plant fermentations was recovered and purified, if possible, for production purposes, and since recovery color problems were minimized by the initiation of recovery processing as soon after completion of fermentation as possible, a method more rapid than those previously described for ascertaining the exact time of completion of fermentation was necessary in these pilot plant fermentations operations. The method used was simple, and consisted merely of following the course of fructose production by measuring optical rotation on clarified beer samples. Fermentations were harvested when consecutive hourly polarimeter readings indicated approximately the theoretical beer fructose concentration expected and differed by less than 0.4 degrees. Actual fructose yields were, of course, calculated later, using the less-rapid total solids procedure described in the section on Materials and Methods. The criterion of good 50-gal fermentor inoculum cultures to be used in either 50-gal or 500-gal fermentor production runs evolved through experience. As indicated below, the medium used for all 50-gal fermentor inoculum cultures contained 18 per cent mannitol. Satisfactory normal production runs were consistently obtained when the 50-gal fermentor inoculum cultures grown on this 18 per cent mannitol medium showed fructose levels of 10 to 15 per cent at the age of use (approximately 21 hr). Serious foaming problems were encountered in both the 50-gal and 500- gal fermentor runs. Conventional antifoams of the lipidtype were completely ineffective. Satisfactory foam control was obtained, however, with a silicone-emulsion type antifoam (Dow Corning Antifoam AF Emulsion).' For fructose fermentations in the 50-gal fermentors, the following fermentation conditions were used. Thirty gal of production medium adjusted to ph 5.8 with dilute H2SO4 were sterilized in each 50-gal fermentor at 122 C for 30 min. After cooling, each fermentor was inoculated with 3 gal of 21-hr inoculum culture grown in another 50-gal fermentor on the following medium: mannitol, 18.0 per cent; brewers yeast # 2019, 0.5 per cent; KH2PO4, 0.1 per cent; dilute H2SO4 to ph 5.8; and tap water. Fifty-gal fermentor inoculum cultures were previously inoculated with 10 L of 24-hr aerated bottle inoculum culture prepared as described in the preceding section on fermentations in 30-L fermentors. All 50-gal fermentors (inoculum and production) were incubated at 28 C under the following conditions: antifoam (if used) 0.06 per cent nonsterile Antifoam AF Emulsion added manually in 0.02 per cent batches at 0 hr (post-sterilization, pre-inoculation), 4 hr, and 7 hr, or sterile CLRS2 added automatically as needed; agitation, 250 rpm; aeration, 3.0 cfm or 0.68 volume of air per volume of medium per min. These agitationaeration conditions represent an effective aeration rate in these equipments of 2.75 millimoles of oxygen per L per min. Data for typical runs in the 50-gal fermentors are presented in table 3. These data confirm the results obtained with the mannitol-0.5 per cent brewers yeast # 2019 production medium in the 30-L fermentors. Satisfactory fructose beers were obtained at initial mannitol levels of 18, 20, and 24 per cent. In addition, a satisfactory 26 per cent fructose beer was obtained from run 4 by using the same solid mannitol feeding tech- Dow Corning Corporation, Midland, Michigan. 'Biocompounds Company, Chicago, Illinois.

6 1956] CONVERSION OF MANNITOL TO FRUCTOSE 321 TABLE 4. Fructose fermentations in 500-gal fermentors ~~~~~~~~Total Soluble Fructose at Fructose Atfa sd Harvest Run Run No. Production No. Medium Used Harvest Age Solids at Harvest Yield* AntifamUsedf Remarks hr g/100 ml g/100 ml % 1 Mannitol-21% CLRS Excessive foaming 2 Mannitol-21% CLRS Excessive foaming B.Y. S % 3 Mannitol-18% CLRS Excessive foaming 4 Mannitol-18% DC antifoam emulsion Foaming controlled 5 Mannitol-18% DC antifoam emulsion Foaming controlled B.Y. # 2019{-0.5% 6 Mannitol-18% DC antifoam emulsion Foaming controlled a silicone-emul- * Fructose yields calculated by the total solids method: F.Y. = total soluble solids t CLRS is a lipid-type antifoam made by Biocompounds Company, Chicago, Illinois; DC antifoam emulsion is sion type antifoam made by Dow Corning Corp., Midland, Michigan. nique successfully employed in producing high fructoselevel beers in the 30-L fermentors. The only other serious problem encountered was the excessive foaming which the silicone-emulsion type antifoam controlled very effectively. For fructose fermentations in the 500-gal fermentors, the following fermentation conditions were used. Three hundred gal of production medium adjusted to ph 5.8 with dilute H2SO4 were sterilized in each 500-gal fermentor at 122 C for 30 min. After cooling, each fermentor was inoculated with 30 gal of 21-hr, 50-gal fermentor inoculum culture prepared as described under fermentations in the 50-gal fermentors. The 500-gal fermentors were incubated at 28 C under the following conditions: antifoam, 0.06 per cent nonsterile Antifoam AF Emulsion added manually in 0.02 per cent batches at 0 hr (post-sterilization, pre-inoculation), 4 hr and 7 hr, or sterile CLRS added automatically as needed; agitation, 300 rpm; aeration, 48 cfm or 1.09 volumes of air per volume of medium per min. These agitation-aeration conditions represent an effective aeration rate in these equipments of 3.0 millimoles of oxygen per L per min. Data for typical runs in the 500-gal fermentors are presented in table 4. Satisfactory results were not obtained immediately with the mannitol-0.5 per cent brewers yeast 2019 production medium at an initial mannitol level of 21 per cent. In run 1 the mannitol oxidation rate was quite low compared to those obtained under comparable conditions in the 30-L and 50- gal fermentors. The normal fermentation period for a 21 per cent initial mannitol level in this equipment is 26 to 30 hr. Generally speaking, fructose fermentations with low mannitol oxidation rates, even though showing satisfactory oxidation at harvest, were found to give excessive fructose recovery color problems. In addition to having a low mannitol oxidation rate, run 2 stopped completely at 36 hr with a fructose yield of only 81.5 per cent. The beers from both these runs were consequently unsatisfactory for fructose recovery and purification. At this point, recognizing that these 500-gal fermentors do not have as good performance characteristics as the 50-gal fermentors, it was felt that an initial production medium mannitol level lower than 21 per cent might be handled more efficiently in the 500-gal fermentors. To test this possibility, the initial mannitol level for the production medium for run 3 was lowered to 18 per cent. As may be seen, the results obtained were good; the fructose yield was 99.4 per cent and the run was completed in 26 hr, a period of time practically identical to that required under standard conditions in the 30-L and 50-gal fermentors for the same initial mannitol level. In addition, the resulting fructose beer gave no recovery color problems. The excellent results obtained with. the 18 per cent mannitol-0.5 per cent brewers yeast # 2019 production medium in run 3 were confirmed in subsequent runs (4, 5, and 6) and no further medial variations were tried. With regard to foaming, the silicone-emulsion type antifoam was found to be as effective for fructose fermentations in the 500- gal fermentors as in the 50-gal fermentor runs. ACKNOWLEDGMENTS The authors wish to express their appreciation to Irving C. West for assistance in carrying out the 50-gal

7 322 PETERSON, FRIEDLAND, DENISON, AND SYLVESTER [VOL. 4 and 500-gal fermentor runs, to Reece Cantrell, Lois Durkin, Ursula Biermacher, Denise Wing, and Helen Helgren for carrying out the necessary analyses, and to Doctor William Brownell for advice in the use of various analytical procedures. SUMMARY The submerged production of D-fructose from D-mannitol by Acetobacter suboxydans NRRL B-72 has been studied in shaken flasks, 30-L stirred-jar fermentors, 50- gal stirred-tank fermentors, and 500-gal stirred-tank fermentors. Shaken-flask studies demonstrated that fructose yields in excess of 95 per cent can be obtained consistently at 28 C with initial mannitol concentrations up to 24 per cent. Either dried yeast extract plus KH2PO4 or primary-grown dried whole yeast plus KH2PO4 were found to be satisfactory supplementary nutrient sources. Trials in 30-L and 50-gal fermentors at an effective aeration rate of 2.75 millimoles of oxygen per L per min resulted in satisfactory duplication of the shaken-flask results. Mannitol oxidation rates in the 30-L and 50- gal fermentors were practically identical, and approximately 3 times greater than those obtained in shaken flasks. Initial mannitol levels of 15, 20, and 24 per cent in the 30-L and 50-gal fermentors gave fructose yields in excess of 95 per cent in approximately 18, 26, and 34 hr, respectively. The use of initial mannitol levels above 24 per cent was not possible due to the solubility limit of approximately 25 per cent for D-mannitol at 28 C. Beers containing up to 35 per cent fructose with yields in excess of 95 per cent were obtained in the 30-L and 50-gal fermentors, however, by adding additional solid mannitol, batch-wise, to standard 20 per cent initial mannitol fermentations. Fructose fermentations in the 500-gal fermentors on an 18 per cent mannitol (initial)-0.5 per cent dried whole yeast medium under the optimum fermentation conditions for 30-L and 50-gal fermentor runs consistently gave fructose yields in excess of 95 per cent in 26 hours. The beers produced were found to be satisfactory for fructose recovery and purification. REFERENCES Association of Official Agricultural Chemists 1945 Official and Tentative Methods of Analysis, 6th Edition, BERTRAND, G Biochemistry of the sorbose bacterium. Ann. Chim. et Phys., 3, (VIII), BUTLIN, K. R The biochemical activities of the acetic acid bacteria. Gr. Brit. Dept. Sci. Ind. Research, Special Rept. No. 2, H. M. Stationary Office, London. COOPER, C. M., FERNSTROM, G. A., AND MILLER, S. A Performance of agitated gas-liquid contactors. Ind. Eng. Chem., 36, CUSHING, I. B. AND DAVIS, R. V The purification of fructose produced by the oxidative fermentation of mannitol by Acetobacter suboxydans. Abstracts of Papers, Division of Carbohydrate Chemistry of American Chemical Society, Sec. A, 10D. FRIEDLAND, W. C., PETERSON, M. H., AND SYLVESTER, J. C A fermentor design for small scale submerged fermentations. Abstracts of Papers, Division of Agricultural and Food Chemistry of American Chemical Society, Sec. B, 21A. FULMER, E. I., DUNNING, J. W., AND UNDERKOFLER, L. A The effect of the concentration of mannitol upon the production of levulose by the action of Acetobacter suboxydans. Iowa State College J. Science, 13, FULMER, E. I. AND UNDERKOFLER, L. A Oxidation of polyhydric alcohols by Acetobacter suboxydans. Iowa State College J. Science, 21, ISBELL, H. S. AND KARABINOS, J. V Preparation of D- mannitol-c14 and its conversion to D-fructose-l-(and 6)- C'4 by Acetobacter suboxydans. J. Research Nat. Bur. Standards, 48, SHAFFER, P. A. AND SOMOGYI, M Copper-iodometric reagents for sugar determination. J. Biol. Chem., 100, UNDERKOFLER, L. A. AND HICKEY, R. J Industrial fermentations, Vol. II, Chemical Publishing Company, New York. U. S. Pharmacopeia, The Pharmacopeia of the United States of America, 1955 Lippincott Publishing Co., Philadelphia, Pennsylvania, 15, WELLS, P. A., LocKwoOD, L. B., STUBBS, J. J., ROE, E. T., PORGES, N., AND GASTROCK, E. A Sorbose from sorbitol. Semi-plant-scale production by Acetobacter suboxydans. Ind. Eng. Chem., 31, WELLS, P. A., STUBBS, J. J., LOCKWOOD, L. B., AND ROE, E. T Sorbose from sorbitol. Production by submerged growths of Acetobacter suboxydans. Ind. Eng. Chem., 29,