Special Perspectives

Challenges in Engineering Microbes for Biofuels Production

See allHide authors and affiliations

Science  09 Feb 2007:
Vol. 315, Issue 5813, pp. 801-804
DOI: 10.1126/science.1139612


Economic and geopolitical factors (high oil prices, environmental concerns, and supply instability) have been prompting policy-makers to put added emphasis on renewable energy sources. For the scientific community, recent advances, embodied in new insights into basic biology and technology that can be applied to metabolic engineering, are generating considerable excitement. There is justified optimism that the full potential of biofuel production from cellulosic biomass will be obtainable in the next 10 to 15 years.

The idea of converting biomass-derived sugars to transportation biofuels was first proposed in the 1970s. Once again, the idea is being seriously contemplated as a possible substitute for petroleum-based liquid fuels. Economic and geopolitical factors (high oil prices, environmental concerns, and supply instability) have certainly played a role in reviving interest in renewable resources. However, an additional impetus is now provided by scientific and technological advances in biosciences and bioengineering that support increased optimism about realizing the full potential of biomass in the liquid fuels area within the next 10 to 15 years. New approaches to biology are being shaped by the genomics revolution; unprecedented ability to transfer genes, modulate gene expression, and engineer proteins; and a new mind-set for studying biological systems in a holistic manner [systems biology (1)]. We are also seeing advances in metabolic engineering (24), with the goal of overproducing useful compounds by rationally and combinatorially engineering cells and their metabolic pathways (5). Combination of concepts and methods from these fields will create a platform of technologies that are critical for overcoming remaining obstacles in cost-efficient biofuel production from cellulosic biomass.

Figure 1 shows the basic features of a biomass-to-biofuels (B2B) process (6). After harvest, biomass is reduced in size and then treated to loosen up the lignin-cellulose fiber entanglement in a step that can take from a few minutes to many hours. Several methods have been used for this purpose, such as biomass treatment with saturated steam at 200°C, explosion with ammonia, and cooking with warm dilute acid (6). Dilute acid pretreatments are fast (minutes), whereas steam-based treatments can take up to a day. After pretreatment, the solid suspension is exposed to cellulolytic enzymes that digest the cellulosic and hemicellulosic biomass components to release the hydrolysis products, primarily six- and five-carbon sugars, respectively (along with acetic acid and lignin-derived phenolic by-products). The type of pretreatment defines the optimal enzyme mixture to be used and the composition of the hydrolysis products. The latter are fermented by ethanol-producing microorganisms such as genetically engineered yeasts, Zymomonas mobilis (Fig. 2), Escherichia coli, or Pichia stipitis (Fig. 3). Presently, cellulose hydrolysis and fermentation are combined in a single unit, termed the simultaneous saccharification fermentation (SSF) stage. The rationale of combining saccharification (the breaking up of complex carbohydrates into monosaccharides) and fermentation (the conversion of a carbohydrate to carbon dioxide and alcohol) in a single unit was to prevent inhibition of the hydrolytic enzymes by the reaction products (7). The SSF step typically lasts 3 to 6 days, with cellulose hydrolysis being the slow, limiting step. The product of SSF is a rather dilute ethanol stream of 4 to 4.5% from which ethanol is separated by distillation.

Fig. 1.

Schematic of the overall conversion process of an energy crop to ethanol.

Fig. 2.

Zymomonas mobilis, a metabolically engineered bacteria used for fermenting both glucose and xylose to ethanol. [Credit: Zhang, Min; DOE/NREL]

Fig. 3.

An 8000-liter fermentation tank used to start the process of turning cellulosic material into ethanol. [Credit: New Energy Company of Indiana; DOE/NREL]

Biomass pretreatment and hydrolysis are areas in need of drastic improvement. Despite substantial reduction in the cost of cellulolytic enzymes (8), sugar release from biomass still remains an expensive and slow step, perhaps the most critical in the overall process. Intensive research and development in all areas of enzyme production reduced the cost of cellulolytic enzymes by a factor of 10 to 30, down to 20 to 30 cents per gallon of ethanol produced (8, 9). Although this is certainly an important advance, it is estimated that the enzyme cost will have to be further reduced to a level comparable to that of current approaches that produce ethanol from the starch in corn kernels at a cost of 3 to 4 cents per gallon of ethanol. Expression of cellulases in fermenting organisms or transfer of the biofuel-synthesizing pathway into a cellulase-producing organism are being pursued in a process termed the consolidated bioprocess (CBP) (10). CBP, however, is presently hampered by the relative inability of yeast to process recombinant cellulases at high rates through their endoplasmic reticulum and secretory pathways, and the relative (with regard to E. coli and yeast) lag in development of molecular biological methods to manipulate organisms (such as Trichoderma) that secrete cellulases naturally. The fact that glucose suppresses respiration in Saccharomyces cerevisiae reduces the amount of adenosine triphosphate available for protein biosynthesis, which may also render it difficult for enzyme production in yeast to be competitive with enzyme production by aerobic fungi such as Trichoderma or Aspergillus. When realized, CBP will enjoy the benefit of completely eliminating the cost of purifying cellulase and of higher activity of the cell-associated cellulase enzyme. To accomplish this goal, the hydrolysis and fermentation steps will have to be coordinated well inside a single cell, such that neither one limits the overall conversion process to proceed at maximum capacity. Although attainable over a longer time scale, in the near term B2B will benefit from the availability of large amounts of inexpensive and more active cellulases. This opportunity should be pursued by coordinated approaches from protein engineering, fungal overexpression, and bioprocess engineering to take advantage of economies of scale in enzyme production.

In engineering better microorganisms for biofuels production, combinatorial searches for promising target genes and other lab-scale experiments should be conducted with synthetic media. In terms of identified target genes or cell phenotypes, results obtained with the more convenient complex media (Luria broth or yeast extract) do not usually translate well to industrial conditions that use synthetic media.

Nonenzymatic, physicochemical hydrolytic methods (such as high-temperature pretreatments and hot-acid hydrolysis) are much faster than enzymatic approaches, albeit at the cost of reduced sugar yields due to undesirable side reactions. This is a problem that can be potentially solved by novel bioreactor designs operating at optimal contact times so as to minimize the rate of sugar-degrading side reactions without impairing biomass hydrolysis in the first place. The presence of lignin that effectively accumulates in the solids fraction as the carbohydrates are hydrolyzed away can interfere mechanically with filtration and recycling operations and complicate efforts to optimize the performance of the hydrolysis step. Advanced material-handling methods and new filtration devices specifically addressing the peculiarities of lignin consistency, or sequence-reversing schemes (whereby lignin removal precedes the hydrolysis step) are some possibilities that could exploit the fast rates of physicochemical hydrolysis while minimizing adverse side reactions. Finally, use of novel types of solvents such as those derived from ionic liquids are promising alternatives that should be further evaluated.

The cost competitiveness of a process such as that depicted in Fig. 1 depends on product titer, yield, and productivity. Final product titer is an important cost determinant not only because it affects the downstream purification cost but because it defines the size of the footprint of the entire processing plant. Low product titer is caused by various factors, including the total amount of substrate solids fed to the fermentor, the presence of inhibitory compounds as byproducts of biomass hydrolysis (such as aromatics, furfurals, furan derivatives, and phenolics), and, of course, the toxicity of the final product itself. If, as seems likely, we can increase the solids loading into the SSF unit, then we maybeabletoincrease substantially the final ethanol concentrations. This makes the engineering of ethanol-tolerant strains, which can tolerate the adverse environment in which the process takes place, of the utmost importance. Not much progress has been made on this front, perhaps because of the preconception that a complex phenotype such as ethanol tolerance could be modulated by a single gene, or at most a handful of genes. There is now accumulating evidence that no single gene can endow microbes with tolerance to ethanol and other toxic compounds. On the contrary, tolerance is a multigenic trait that must be elicited by drastically different approaches, such as global transcription machinery engineering (11). This method and its extensions should be systematically explored to identify transcription factor mutants that can increase the tolerance of industrial strains to the final fuel product, as well as other relevant toxic compounds.

Because the cost of a biomass-derived fuel depends critically on the yield of sugar conversion to the final product, much attention has been focused on the engineering of strains to use all sugars released from biomass hydrolysis, in particular the pentose sugars that are products of hemicellulose hydrolysis. Such sugars may constitute 5 to 30% of the total carbohydrates; hence, various strategies have been used to attempt either to introduce the ethanol pathway in natural xylose consumers (12) or to engineer the xylose-catabolizing pathway in natural ethanol producers (13, 14). The state of the art is rather well advanced as far as the engineering of various pathways (including pentose phosphate, glycolytic, ethanologenic, and redox balancing) is concerned. An area that has received relatively limited attention is that of sugar transporters and their regulation. There is evidence that a multitude of such transporters may be in operation (15) and that their activity may depend on signaling defined by the sugar composition of the fermentation medium (16). Elucidation of sugar transport at the molecular level and better characterization of kinetic and regulatory properties, including quorum-sensing mechanisms, should be given high priority because they may provide the basis for the simultaneous use of the sugar mixtures released from biomass hydrolysis as opposed to the slower and suboptimal sequential use characterizing most present operations (sugars are consumed simultaneously in, for example, fermentations by recombinant Z. mobilis, albeit at low rates). It is important to remember that one mole of CO2 is produced for each mole of ethanol, for a total yield of 0.51 g of ethanol per gram of glucose consumed. As carbon oxidation to CO2 is essential for generating the energy and redox equivalents needed to sustain cellular functions and the ethanol pathway itself, an interesting long-term idea is the capture and conversion to liquid fuels of this CO2 by means of hydrogen supplied from carbon-free sources (such as nuclear or solar). This could be accomplished by conventional Fischer-Tropsch processes.

Process productivity is a principal determinant of capital cost. For cellulosic ethanol, the capital cost is estimated at ∼$4 per gallon, contributing 20 to 25% of the ethanol manufacturing cost (17). However, these figures, as well as the ones quoted in the following paragraph, vary considerably from source to source and are also time dependent. They should be viewed only as preliminary estimates that need to be validated by detailed empirical and analytic work. Furthermore, costs contributed by the process units are interrelated and cannot be assessed in isolation. Overall system analysis is critical for assessing the relative importance of the various process units and their interactions. Thus, reliable simulation packages for the integrated system operation must be developed for overall system analysis, optimization, and sensitivity studies (18).

The capital cost must be reduced by more than half for an economical process (along with a similar reduction in the feedstock cost, which will come primarily from yield improvements of an energy crop, and a 15 to 25 cents per gallon reduction in cellulolytic enzyme costs). Achieving the above goal or, equivalently, doubling process productivity, requires a coordinated approach for improving all units of the process and, in particular, the biomass pretreatment-hydrolysis steps mentioned earlier, because these are apparently the process rate-limiting steps. After that, the volumetric productivity of fermentation must be improved (presently between 1.5 and 2.0 grams of ethanol produced per hour and fermentor volume), which is the product of the specific productivity (grams of product produced per gram of cells per hour) and the total cell concentration that can be sustained in the fermentor. The latter, again, is limited by the presence of the same inhibitory compounds; hence, use of more tolerant strains will affect the total process productivity as well. Additionally, specific productivity must be increased.

Various approaches have been suggested for increasing the specific ethanol productivity, such as increasing the amount of “rate-limiting” enzymes, enzyme deregulation, cofactor replenishment, and increase of precursor supply. Although some of these approaches are valid, some others are grossly misguided. A rather obvious approach to increase the flux through a pathway is by increasing the activity of every single enzyme in the pathway. Although this is acceptable for modest flux enhancements, it has not been attempted for large, order-of-magnitude scale flux increases on the grounds that it will cause large perturbations in the metabolites and, hence, the physiology of the organism. Yet, this will not happen if all enzymes in the pathway are similarly amplified, because the same steady state with respect to metabolite levels will be preserved. Simultaneous increase of the activity of all enzymes by a factor f will not affect metabolite levels, while allowing pathway flux to increase by the same factor f. The only limitation in such a scheme is the cell volume, which may not be able to accommodate drastically increased amounts of all enzymes of a pathway (it is estimated that the enzymes of glycolysis make up 10 to 15% of the total cellular protein). However, this problem can be overcome by engineering more active enzymes. Pathwayfluxamplification by coordinated activity enhancement of the pathway enzymes (19) has been successfully used in lysine biosynthesis (20), aromatic amino acid production (21), and polyhydroxybutyrate synthesis in E. coli (22), among other systems. Determination of flux split ratios at key metabolic branch points (23), guided by flux determination methods (24, 25), can aid this research, along with advanced fermentor feeding strategies that control metabolic activity (26).

Product separation for ethanol, the main biofuel currently produced, is carried out by distillation. Although mature and well optimized, it remains an energy-intensive and overall expensive step contributing 17 to 20 cents per gallon (17, 27). In light of accumulating reports describing configurational changes in materials in response to small environmental changes (pH, temperature, or ionic strength), it may be useful to evaluate such phenomena with respect to their potential to facilitate ethanol separation. One can envision, for example, processes in which ethanol adsorbs preferentially on some material and desorbs when the material changes configuration after a small environmental change. Separation in such schemes would be entropically (as opposed to enthalpically) driven, could be less energy intensive than current operations, and possibly could be consolidated with fermentation in a single step.

As mentioned, ethanol is not the sole or optimal fuel to be produced from cellulosic biomass. Butanol is currently attracting attention because of its potential superior properties with respect to corrosiveness, volatility, energy density, and ease of separation (28). Aside from butanol, other higher alcohols, alkanes, and various types of oils are possible biochemically derived biofuels. It is not clear yet which one(s) will be the ideal biofuel, and the answer to this question may well depend on additional factors, such as the type of biomass available, particular climatic conditions, and composition of engine emissions. The ability to clone, transfer, and control genes from different organisms, including plants, has reached the point at which researchers will be able to engineer pathways that take advantage of a variety of conditions with a great degree of confidence. Additionally, genome sequences now provide a straightforward supply of genes to be tested in tentative pathway constructs. Nevertheless, it is important to develop technologies for the synthesis and separation of these alternative fuels, because it is yet unclear what additional requirements such technologies will pose in the design of a robust, cost-efficient, commodity-scale process.

In assessing the potential of current and projected technologies to develop cost-efficient B2B processes, it is important to bear in mind that the present state of affairs was reached by minimal investment directly in biofuels research. The major biosciences and bioengineering infrastructure was developed in the process of exploring medical applications of biology and biotechnology. Although this platform is the basis for the present optimism surrounding the use of biosciences for biofuel production from renewable resources, a number of problems still remain in realizing this potential. These problems must be addressed directly and adequately in the immediate future.

References and Notes

View Abstract

Navigate This Article