231 Jeffries, Thomas W. Comparison alternatives for the of to by . In: Lowenstein, Michael Z., ed. Energy applications of biomass: Proceedings of the National Meeting on Biomass R & D for Energy Applications; 1984 October 1-3; Arlington, VA. New York, NY: Elsevier Applied Science Publishers; 1985: 231-252. COMPARISON OF ALTERNATIVES FOR THE FERMENTATION OF PENTOSES TO ETHANOL BY YEASTS

T. W. JEFFRIES*

*U.S.D.A., Forest Service, Forest Products Laboratory, Madison, Wisconsin

SYNPOSIS Hemicelluloses are major components of plant biomass. In hardwoods and agricultural residues, is the principal hemicellulosic sugar. Xylose and other hemicellulosic sugars are recovered from lignocellulose more readily but are fermented with greater difficulty than is glucose. Xylose employs pathways distinctly different from those involved in the utilization of glucose. With most yeasts, xylose metabolism requires air. Aeration results in (as opposed to fermentation) and low ethanol yields. It is possible, however, to suppress respiration by feeding small amounts of glucose during the xylose fermentation. Some yeasts, such as Pachysolen tannophilus, will metabolize xylose anaerobi- cally. Alternately, other yeasts will anaerobically ferment the keto isomer of xylose, xylulose, after it is formed from xylose by the action of . In both instances, the fermentation rates are low. Improved strains of P .tannophilus have been obtained by W mutagenesis followed by enrichment for faster growth in nitrate- broth or by selecting for strains incapable of using ethanol as a carbon source. Several yeasts have been described as superior xylose fermenters, including (in approximate ascending order): Candida tropicalis , Kluyveromyces marxianus , P . tanno- philus, the mutant Candida sp. XF 217, and Candida shehatae (and its sex- ually perfect form, Pichia stipitis). The xylose fermentation rate of C . shehatae is 3 to 5 times higher than that obtained with P. tannophilus, but the yields of ethanol from xylose are. similar with the two organisms. The glucose fermentation rate and ethanol yield are lower with C. shehatae) than 232 with P. tannophilus . Unstable petite and grande strains of C. shehatae have bean obtained on urea + xylitol agar, and some show markedly different fer- mentation rates and products. Further strain improvement and process devel- opment should soon provide commercially practicable technology for the fer- mentation of xylose.

1 INTRODUCTION Within the realm of liquid fuel production from biomass, utilization of lignocellulose has focused largely on the problem of cellulose saccharif- ication. Various approaches have been tried including hydrolysis by extra- cellular streptomycete and fungal cellulases, simultaneous saccharification and fermentation by cellulolytic bacteria, and acid hydrolysis followed by fermentation. To a certain extent, the attention given to cellulose is jus- tifiable. Cellulose comprises about half of the total weight of lignocellu- lose, and its fundamental constituent, glucose, is an excellent fermentation substrate. In a larger context, however, consideration of cellulose to the exclusion of the two other major constituents, hemicellulose and lignin, is futile. One reason is that it is uneconomical to throw away almost half of the feedstock. Another is that cellulose has appreciable commercial value as fiber. Converted to pulp, a ton of cellulose is worth $400 to $700; con- verted to ethanol, it is worth less than $300. In the kraft pulping process, lignin and hemicellulose are extracted under alkaline conditions and then burned to recover chemicals and energy. In some instances, the lignin is recovered for other applications. The hem- icellulose is largely degraded to organic acids prior to combustion and has no current commercial value. Other technologies are being developed that will enable the efficient fractionation of lignocellulose into pulp-grade cellulose, useful lignin derivatives, and useful hemicellulosic sugars including xylose. The objective of the research described in this paper is to improve our knowledge of metabolism in yeasts and to thereby provide the means for more efficient utilization of xylose.

1.1 Sources and Recovery Hemicellulosic sugars are major constituents of wood and agricultural residues. Table 1 shows the average proximate composition of seven commonly 233 occurring hardwood (angiosperm) and softwood (gymnosperm) species (Ref. 1-3) along with a few major U.S. agricultural residues (Refs. 4-6).

Generally, hardwoods have slightly more neutral hemicellulosic sugars and cellulose but less lignin than softwoods. The composition of the hem- icellulose differs in hardwoods and softwoods. In hardwoods, the predomin- ant hemicellulosic sugar is the pentose xylose; in softwoods, the predominant hemicellulosic sugar is the hexose mannose. The xylose content of hardwoods is greater than softwoods. Most agricultural crops are angio- sperms and, like hardwoods, have xylose as the predominant hemicellulosic sugar. As described elsewhere in this symposium, low-grade hardwoods are available in relative abundance in the United States, but low-grade soft- woods are in relatively short supply. The combination of a greater angio- sperm resource and a higher proportion of xylose in that resource make xylose utilization a major concern in production of fuel from biomass. In addition to being relatively abundant, xylose is more readily recovered from hemicellulose than glucose is from cellulose. Dilute acid hydrolysis of hemicellulose yields about 85% to 90% of the xylose present in red oak; dilute acid hydrolysis of cellulose yields only about 50% to 60% of the glucose present. The difference between the xylose and glucose yields can be attributed directly to physical and chemical properties of the two polymers and hence is not readily amenable to process changes. The situa- tion is similar with regard to enzymatic hydrolysis. Although up to 90% of the glucose can be recovered from steam-exploded wood if sufficient cellu- 234

lase is added, at economical enzyme loadings glucose yields are substan- tially lower. Taking the differences in yields of xylose and glucose into account, roughly equivalent amounts of sugar can be recovered from the hem- icellulosic and cellulosic fractions.

1.2 Biochemical Pathways and Fermentative Capacities Xylose can be assimilated by many bacteria, yeasts, and filamentous fungi, but initial steps of assimilation in yeasts and fungi are signifi- cantly different from those in bacteria. In yeasts and fungi, xylose is first reduced to xylitol and then oxidized to xylulose. In bacteria, the conversion from xylose to xylulose is catalyzed by xylulose isomerase in a single step (Ref. 7). This paper considers only the activities of naturally occurring yeasts. Most yeasts use a xylose reductase with a specific requirement for NADPH as a cofactor to reduce xylose to xylitol. Next,xylitol dehydro- genase specific for NAD oxidizes xylitol to xylulose. Consequently, assimi- lation of xylose converts NADPH into NADH. In Candida utilis, the organism best studied in this regard (Ref. 8),NADPH is supplied by the oxidative phase of the pentose phosphate pathway (PPP) in a closed cycle (Fig. 1). Under oxidative conditions, the only mode of fungal xylose assimilation known until 1981, NADH is recycled to NAD by respiration. Under anoxic con- ditions, NAD cannot be regenerated, and xylose assimilation ceases (Ref. 9). NADPH is used primarily in metabolic syntheses, and is generated mainly by the oxidative PPP (Ref. 10). Thus, production of NADPH by the oxidative PPP is thought to provide the means to assimilate xylose for aerobic produc- tion of ethanol by Candida tropicalis and other yeasts (Ref. 12). More recently, certain yeasts capable of fermenting xylose to ethanol in the absence of oxygen (anoxically) (Ref.13) have been shown to possess xylose reductase(s) capable of using either NADH or NADPH as a cofactor (Ref. 14). If NAD(H) can be used for both the reductive and oxidative steps of xylose metabolism, the balance between NAD and NADH can be maintained under anoxic conditions (Fig. 2) and xylose utilization is not dependent on aeration. The observation that P . tannophilus will ferment but not grow anaerobically on D-xylose could be attributed to the insufficient production of metabolic reductant or energy for growth. 235

2 PROCESS ALTERNATIVES 2.1 Coupled Isomerization and Fermentation In 1980, Wang, Shopsis, and Schneider (Ref. 15) showed that yeasts are able to ferment xylulose to ethanol under anoxic conditions. This find- ing had immediate implications because the conversion could be carried out readily by using commercial xylose (glucose) isomerase. The discovery was immediately seized upon and became the basis for considerable research and development in this field. As proposed for commercial practice, the tech- nology would employ exogenous, immobilized xylose isomerase (already com- mercially derived from bacteria) to convert xylose to an equilibrium mixture of xylose and xylulose. The xylulose would then be fermented to ethanol and the residual xylose recycled over the xylose isomerase. The process would be continued until all xylose was consumed. Several variations on the basic process are possible and most have been attempted, but the principal remains the same. Xylose isomerase could be incorporated directly into the fermen- tation vessel or the xylulose could be produced exogenously and separated from the xylose prior to fermentation. The process of sequential isomarization and fermentation is affected by several factors. At equilibrium in aqueous solution, xylose isomerase catalyzes the formation of about 17% xylulose from xylose. In comparison 47% fructose is formed from glucose. The lower equilibrium obtained with xylose is offset somewhat by the higher turnover rate of xylose isomerase acting on its native substrate. Other reaction conditions, such as tempera- ture or the inclusion of borate to chelate the xylulose as it is formed, can affect the equilibrium. The literature on xylose isomerase and the xylulose fermentation has been covered in earlier reviews (Refs. 16-19). Cost estimates for the isomerization of xylose have been based on information from the isomerization of glucose to high fructose corn syrup (HFCS). Xylose isomerase is used in both instances. But whereas HFCS pro- duction employs high sugar concentrations and optimal conditions for isomer- izing glucose, these factors must be compromised with those optimal for fer- mentation. Reliable values for the cost of HFCS production are difficult to obtain outside the industry; however, one figure published in 1978 placed the cost of isomerization at 2.3 to 3.7¢/kg of fructose produced (Ref. 20). Given that roughly 5.9 kg of sugar are required to produce 1 gal of ethanol, the isomerization reaction would add about 15¢ to 25¢/gal to the cost of ethanol production as compared to an equivalent fermentation using glucose as the feedstock. 236

Although sequential xylose isomerization and fermentation is tech- nically feasible, it is hampered by several factors: the cost of the enzy- matic isomerization, the formation of xylitol as a by-product, inhibition of xylose isomerase by xylitol, the use of separate optimal pile and tempera- tures for isomerization and fermentation, and the low rate of the xylulose fermentation. Alternatively, new yeasts might be constructed by recombinant DNA techniques to possess xylose isomerase. The approach of employing recombined yeasts suffers from many of the difficulties listed above plus the basic problem of obtaining adequate expression of enzymatic activity. Moreover, there are few inherent advantages in carrying out a multistep pro- cess in a single reactor (or with a single organism) If the process steps have different optimal conditions or if separate organisms are capable of carrying out each of the steps more efficiently.

2.2 Fermentation Rates with Different Sugars The specific xylulose fermentation rate, even with the best strains of yeasts, is appreciably lower than the rate attained with glucose, and in some instances, it is lower than the rate attained with the direct fermenta- tion of xylose. About 60 yeast strains have been screened for their abil- ities to ferment either xylulose or a mixture of xylose and xylulose under equilibrium conditions (Refs. 21-23). Results with some of the best strains are summarized in Table 2. In general, C . tropicalis and Schizosaccharo- myces pombe ferment xylulose most rapidly, but strains of also give better-than-average rates. Volumetric fermentation rates (g ethanol/L•h) are subject to a great deal of variation because cell growth varies under the conditions employed. Indeed, Immobilization of cells can lead to very high volumetric rates because of high cell densities. Note, however,that specific fermen- tation rates (g ethanol/g dry wt of cells•h) generally decrease after cells are immobilized. Although immobilization has been attempted with both the xylulose and xylose , the volumetric fermentation rates obtained do not approach those commonly observed in the fermentation of glucose by free cells of S. cerevisiae or Zymomonas mobilis . It is better to use specific rates when comparing fermentation of different sugars. The highest reported specific Xylulose fermentation rate is about 1/18 of the specific glucose fermentation rate obtained with S . cerevisiae . On the other hand, the highest reported xylose fermentation rate is about 1/6 the specific glucose fermentation rate. For both of these 237 pentoses, llttle is known about the regulatory biochemical steps or the con- ditions optimal for fermentation.

Yeasts ferment glucose, xylose, and xylulose at characteristic rates. In a survey of several different xylose- and Xyluloee-fermenting yeasts, Maleszka and Schneider (Ref. 22) showed that yeasts capable of fer- menting xylose were typically poor xylulose fermenters, and vice versa. Two primary exanples are S. pombe and P. tannophilus . S. pombe is a very good fermenter of glucose, but it does not metabolize xylose at all; on the other hand, it is a good fermenter of xylulose. Pachysolen tannophilus ferments glucose more readily than it does xylose, but it ferments xylulose poorly. The fermentation rate obtained on glucose is still much lower than that attained with S. cerevisiae, S. pombe ,and other yeasts used for commercial alcoholic fermentations. In this work from our laboratory, C . shehatae has been shown to ferment glucose at a lower specific rate than P. tannophilus , 238 even though it ferments xylose much more rapidly (Table 2 and Figs. 3, 4, and 5).

2.3 Incidence of Xyose-Fermenting Yeasts Sixty-four percent of the species listed in Ref. 32 are cited as cap- able of assimilating xylose and 7% are cited as variable) but none is listed as capable of fermenting this sugar. A separate taxonomic treatment by Barnette, Payne, and Yarrow lists P. tannophilus and P. stipitis as capable of fermenting xylose (Ref. 35). This discrepancy stems in part from the inability of these yeasts to grow under anaerobic conditions. Even though P. tannophilus will ferment xylose, no cell growth occurs anaerobically, and because the specific fermentation rate is very low, negative results appear unless high cell densities are employed as the inoculum. In a study specif- ically designed to identify xylose-fermenting yeasts (Ref. 36), 200 species able to ferment glucose anaerobically and to grow on xylose aerobically were tested for their abilities to ferment D-xylose. In most of these species, ethanol production on xylose was negligible. Only 19 species produced between 0.1 and 0.1 g/L of ethanol. Strains of Brettanomyces naardenensis , Candida shehatae , Candida tenuis, Pachysolen tannophilus , Pichia segobien- sis, and Pichia stipitis produced more than 1 g/L ethanol from 2% xylose.

3 PROCESS VARIABLES

3.1 Effects of Aeration Aeration stimlates cell growth and occasionally stimulates fermenta- tion as well. Only a few yeasts are capable of (limited) anaerobic growth. This inability stems in part from a biochemical requirement for molecular oxygen in the synthesis of membrane . However, Schneider and co-workers (Refs. 7,8) have shown that the inability of P. tannophilus to grow anaerobically cannot be overcome by the addition of ergosterol or unsaturated fatty acids. Maleszka and Schneider have also shown that oxygen and mitochondrial function are also required for S . cerevisiae to grow on xylulose (Ref. 39). These observations suggest that the anaerobic metabo- lism of pentoses supplies metabolic energy (ATP) fast enough to satisfy only the basal metabolic demand but does not provide enough ATP to allow cell growth. Aside from affecting growth, aeration strongly affects the specific fermentation of glucose by P. tannophilus . This was first shown by Schef- fers and Wiken (Ref. 40). Unexpectedly, the stimulation does not extend to 239 xylose (Table 3). Aeration does increase the volumetric fermentation rate, but this stimulation can be attributed to increase in cell mass.

Pachysolen tannophilus is not alone in showing a stimulation of the glucose fermentation by aeration. The genus Brettanomyces shows this trait among most of its species (Ref. 40). Aeration is also known to play a role in the fermentation of glucose by Saccharomyces (Ref. 41), but in this instance, it is primarily important in maintaining cell viability and ethanol tolerance (Ref. 42). Aeration decreases the yield of ethanol from xylose by P. tannophi- lus. It is hypothesized that the reduction occurs by virtue of increased ethanol respiration (Ref. 41). Under strictly anaerobic conditions, P. tan- nophilus produces essentially the same net yield of ethanol from xylose-- after correcting for the amount of xylose going into xylitol--as it does from glucose (Table 4). The amount of carbon going into xylitol is deducted from the calculation, because it accumulates early in the fermentation path- way and essentially represents sugar that is not metabolized. The yield of xylitol decreases under aerobic conditions and increases under anaerobic conditions (Ref. 28).

3.2 Effect of Glucose on Ethanol Yields from Xylose The aerobic ethanol yield from xylose can be improved by adding small amounts of glucose during the fermentation (Table 5). By using this approach, a high rate of ethanol production can be achieved with relatively little ethanol loss. The improvement in yield is not observed under anaero- bic conditions, and control experiments show that adding glucose at the low concentrations employed does not affect the rate of xylose assimilation. So 240

the observed improvement in yield is attributed to a decrease in the rate of ethanol respiration (Ref. 28).

3.3 Effects of Nitrate on Ethanol Production Nitrate lncreases the levels of PPP enzymes in yeasts, fungi, and plant cells (Refs. 10,45-57). The enhancement occurs because the PPP is the primary source of NADPH and because nitrate reductase requires large amounts of NADPH for nitrogen assimilation. It was for this reason that we examined the ability of nitrate to stimulate the rate of xylose fermentation in P . tannophilus (Ref. 48). Although nitrate stimulated the specific aerobic xylose fermentation rate, cells grew slower on nitrate, and under anaerobic conditions, the specific rate of ethanol production of nitrate-grown cells was appreciably lower. The anaerobic effect was dependent on both pregrowth on nitrate and the presence of nitrate in the medium. 241

3.4 Effects of Nitrate and Xylitol on Strain Selection in P. tannophilus Xylitol and nitrate were used in an indirect enrichment and selection method to obtain improved xylose fermenters. These restrictive carbon and nitrogen sources were used to help select vigorous rather than crippled mutants. P . tannophilus tends to accumulate xylitol during growth on xylose, so it was used as a sole carbon source on the assumption that xylitol utilization is a rate-limiting step. P. tannophilus grows slower on nitrate than on other more readily assimilated nitrogen sources,and nitrate-grown cells exhibit higher specific aerobic fermentation rates than ammonia-grown cells. Moreover,nitrate is knowm to induce higher levels of PPP enzymes; therefore,by using it as a nitrogen source, the cells are fully induced for PPP enzymes. Any faster-growing mutant would have meta- bolic capacities beyond the normal adaptative range of the parent. For these reasons, cells able to grow well on nitrate should be capable of gen- erating NADPH at an elevated rate. Hence, nitrate was chosen as the sole nitrogen source. Taken together, these restrictive conditions slowed growth so that a minimum of 7 to 10 days was required for significant growth to occur in liquid or on solid media (Ref. 49). Strains capable of relatively rapid growth on nitrate + xylitol media were generally much better xylose fermenters than the parent strain or mutants obtained under less restrictive conditions (Fig. 6). The strains derived from nitrate + xylitol enrichment produce ethanol twice as fast and in 30% better yield than the parent strain under aerobic conditions. More- over, they have a specific fermentation rate 50% greater under anaerobic conditions (Fig. 7). These strains are stable under repeated subculture, and the enrichment and selection method has been successfully employed several times with P. tannophilus . Other approaches to obtaining Improved mutants of xylose-fermenting yeasts have been attempted, including selecting strains of Candida sp. for relative growth rates on xylose and xylitol media (Ref. 50) and selecting strains of P . tannophilus for low rates of ethanol assimilation (Ref. 51). Both of these methods have led to improved xylose-fermenting strains.

3.5 Candida shehatae as a Rapid Xylose Fermenter Although mutation and selection methods have been successful in obtaining incremental improvements in the xylose fermentation rates of laboratory strains, enrichment and screening of yeast strains from natural sources has led to the identification of C . shehatae as a species capable of fermenting xylose at four to five times the specific rate of P . tannophilus 242

(Ref. 31). Candida shehatae produces up to 3.8% (w/w) ethanol from 16% D-xylose (Fig. 8) and about 5% ethanol from 16% D-glucose (Fig. 9). In comparison to P. tannophilus ,which forms much more ethanol on D-glucose than on D-xylose, with C . shehatae the final ethanol concentrations on these two sugars and the ethanol yields (after deducting xylitol production) are about the same (Table 6).

Pichia stipitis is the sexually perfect stage of C . shehatae . Although no published study has yet made a detailed comparison of the fer- mentative capacities of various strains of these two forms, work in this laboratory has shown that for the most part, they are very similar. As much variation exists among strains of each form as between the anomorph and the teleomorph. In other research,separate studies have compared fermentation characteristics of P . tannophilus with either C . shehatae (Ref. 50) or P . stipitis (Ref. 53) and found C . shehatae or P . stipitis to be the better fermenter in each case. Strain improvement Is proceeding with C. shehatae . One of the first approached tried was to apply the same enrichment and selection method used successfully with P .tannophilus . According to conventional taxonomic tests, C . shehatae is unable to use nitrate as a nitrogen source (nitrate negative). We have found that some strains will grow to a limited extent In nitrate + xylitol medium, but this approach has not been successful with this organism. The fastest xylose-fermenting strain we have obtained to date is an unstable petite-like variant derived from C. shehatae ATCC 22984 by selection on urea + xylitol medium (Ref. 54). Certain strains of C . shehatae exhibit marked small colonies remin- iscent of the petite mutation in Saccharomyces when they are grown on urea + xylitol agar. Conventionally, the petite designation refers to strains 243

showing small colonies on glucose and deficiencies in respiratory metabo- lism. Many of the petite-like colonies of C. shehatae ATCC 22984 on xylitol agar show diminished respiratory capacity as judged by the tetrazolium over- lay method (Ref. 55). The petite-like colonial morphology of C. shehatae is expressed on both xylose and glucose. However,strains designated grande on xylitol exhibit slightly smaller colonial diameters when growing on glu- cose. Conversely, strains designated petite on xylitol show larger colonial diameters when growing on glucose. The transitions between small and large colonial sizes occur in both directions (Table 7).

The xylose fermentation characteristics of petite and grande strains are related to respiratory activities. When a tetrazolium agar overlay is applied to colonies growing on urea + xylitol agar, five different colony types can be distinguished (Table 8). Some of these strains, occurring in low frequency, exhibit a small colony diameter and a strong tetrazolium reaction on xylitol. These strains are poor ethanol producers, but they form other products. The petite colonies showing a weak tetrazolium reac- tion on xylitol tend to produce less xylitol and glycerol, but the higher overall yield of all products formed by grande tetrazolium-positive strains tends to suggest that these strains have lower endogenous respiratory activ- ity when growing fermentatively. Neither grande nor petite strains show significant tetrazolium reactions on xylose. Preliminary studies in my laboratory have shown that a similar petite-like variation occurs in strains of P . stipitis . An improved understanding of this petite-like variation should eventually contribute to the isolation of yeast strains capable of fermenting xylose economically under practical conditions. 244

3.6 Comparison of Various Xylose-Fermenting Yeast Strains Various researchers have used many different media and cultural con- ditions in studying different yeast strains for their abilities to ferment xylose. While these strains doubtless possess different optima for ethanol production, it is useful to compare them under a single set of fermentation conditions. My lab has recently done such a comparison. Results show that although the mutant strains of P. tanophilus and Canida sp. performed better than their parent strains, all strains of C .shehatae were better than any other strain tested. These results show that further strain development with C . shehatae as well as enrichment and selection of new isolates should continue.

4 CONCLUSIONS 1. Xylose is widely available in angiosperm residues and more readily recoverable than glucose from lignocellulosic materials. 2. Although a two-stage isomerization and fermentation of xylose is feasible, direct fermentation of xylose to ethanol can proceed at a higher specific rate and is more likely to have a lower overall cost. 3. Xylose fermentation rates and ethanol yields are still much lower than commercial glucose fermentation, but they are improving. For special- ized situations where waste xylose streams constitute a disposal problem, fermentation to ethanol may be economical. 245

4. Biochemical, genetic, and strain selection studies have only recently been undertaken, and it is expected that they should result in better strains and fermentation conditions.

5 ACKNOWLEDGEMENT The author wishes to thank Henry Schneider and his coworkers at the National Research Council, Ottawa, Ontario for useful discussions and for sharing references and preprints of unpublished data.

6 REFERENCES 246 247 248 249 250 251 252