Abstract
Conversion of plant cell walls to ethanol constitutes second generation bioethanol production. The process consists of several steps: biomass selection/genetic modification, physiochemical pretreatment, enzymatic saccharification, fermentation and separation. Ultimately, it is desirable to combine as many of the biochemical steps as possible in a single organism to achieve CBP (consolidated bioprocessing). A commercially ready CBP organism is currently unreported. Production of second generation bioethanol is hindered by economics, particularly in the cost of pretreatment (including waste management and solvent recovery), the cost of saccharification enzymes (particularly exocellulases and endocellulases displaying kcat ~1 s−1 on crystalline cellulose), and the inefficiency of co-fermentation of 5- and 6-carbon monosaccharides (owing in part to redox cofactor imbalances in Saccharomyces cerevisiae).
- cellulose
- co-fermentation
- economics
- hemicellulose
- lignin
- physiochemical pretreatment
INTRODUCTION
Economical conversion of the polysaccharides that constitute cell walls of plants to second generation bioethanol may seem a straightforward exercise in chemical conversion. However, over the last several years the processes involved in the conversion of cellulose and hemicellulose, which compose the bulk of the biomass feedstock, have been closely associated with the word ‘recalcitrance’ [1], to emphasize obstacles that can impede the conversion. Recalcitrance of the conversion stems from the key word ‘economical’, as each step in the conversion can be costly and drive production costs to exceed those of its transportation fuel competitors, such as those derived from fossil fuels (e.g. gasoline, diesel and natural gas) or those derived from starch, sucrose and vegetable oils (e.g. first generation bioethanol and biodiesel). The combined strength of the glycosidic bonds of cellulose [2] and its associated crystal structure dictate application of either harsh physiochemical conditions or use of several specific enzymes. In comparison with production of first generation bioethanol (derived from corn starch and cane sugar), the harsh physiochemical conditions add considerable expense to construction of saccharification reactors and downstream processing, and the use of enzymes adds the expense of preparation of numerous enzymes, particularly the cellulases. In recent years, protein production efficiency of cellulases has been increased more than 10-fold, and this alone makes enzymatic saccharification more economical than physiochemical methods [3–5].
Conversion of cell walls into bioethanol can be viewed as occurring in three operational steps. (i) First there is biomass production and pretreatment. The feedstock of biomass will depend on local availability to determine whether excess crop stover (e.g. corn, wheat and sorghum) or a dedicated energy crop (e.g. switchgrass and Miscanthus) is harvested. Pretreatment consists of mechanical, heat and chemical conditions that aim at maximizing exposure of the chemical bonds of biomass to the glycoside hydrolases or esterases that mediate their hydrolysis. Barring pretreatment, the yields of monosaccharides from native biomass hydrolysed with cellulases, xylanases and associated enzymes are only on the order of 20%. An efficient pretreatment will increase the yield to over 80%. (ii) Next there is enzymatic saccharification. This step requires several enzymes to release the major fermentable monosaccharides. The requisite enzymes should display superior hydrolysis kinetics on natural substrates and good stability properties. (iii) Finally there is fermentation. This step needs to be robust, and capable of rapid and efficient fermentation of all the sugars available (i.e. both 5- and 6-carbon sugars). The second and third steps may be combined or run separately to achieve the following processes: SHF (separate hydrolysis and fermentation), where cellulosic and hemicellulosic components of cell walls are separately hydrolysed prior to presenting them to the fermenting organism; SSF (simultaneous saccharification and fermentation), where enzymatic cellulose hydrolysis and hexose fermentation occur concurrently in the same vessel; SSCF (simultaneous saccharification and co-fermentation), where cellulose and hemicellulose hydrolysis, and hexose and pentose fermentation occur concomitantly in the same vessel; and CBP (consolidated bioprocessing), where a single organism carries out saccharification and fermentation. Considerable effort has been applied to engineering an ideal CBP organism (reviewed in [6]), but currently there are no reports of a commercially viable CBP organism suitable for the operation.
PLANT CELL WALL STRUCTURE
Plant tissue varies widely in structure and composition and, as might be expected, in its response to pretreatment and enzymatic saccharification [7]. Plants comprise primary and secondary cell walls, both of which are fortified by cellulose microfibrils. Primary cell walls typically contain cellulose, hemicellulose (xyloglucans), pectin and proteins. In grasses, glucuronoarabinoxylan, which is cross-linked by diferulate, substitutes for the pectin [8]. Cellulose microfibres are criss-crossed within the cell wall with closer alignment and spacing in primary cell walls than in secondary cell walls. Secondary cell walls are composed of cellulose, hemicellulose and lignin and constitute the majority of cell wall mass; for example, 70–80% weight of corn stover is present in secondary cell walls [9].
Cellulose
Cellulose, the most abundant biopolymer, is formed by β-(1,4)-linked D-glucoses, where adjacent D-glucoses are flipped making cellobiose the fundamental repeating unit (Figure 1) [10]. The linear flat polymer allows for extensive hydrogen bonding within and between chains, as well as van der Waals stacking interactions between chains [11]. Glucan chains occur in hexagonal arrays of 36 (3 nm×5 nm width) with exceptionally high DPs (degrees of polymerization) [12]. For corn stover, the mean DP is 6000 in primary walls and 14000 in secondary walls [13]. Formation of the arrays is controlled by transmembrane assembly complexes, termed rosettes, which ensure that individual glucans are aligned into crystallized fibrils to maximize tensile strength and to fit tightly enough to exclude water. The lattice structure is polymorphic, where native cellulose is defined as crystal form I, but following chemical treatment can be transformed to crystal forms II, III, V or X [14]. Crystallinity is thought to influence biomass recalcitrance and varies widely among plant cell walls from approximately 40–50% in plant cellulose to 65–80% in bacterial and algal cellulose (reviewed in [15]). Crystallinity measurements can vary depending upon the analytical technique used for measurement and, in the case of pretreated plant cell walls, upon the removal of low-crystalline components (e.g. xylan) [14].
Cellulose consists of repeating β-(1,4)-linked D-glucose residues. Adjacent D-glucose residues are flipped, making cellobiose the fundamental repeating unit. The major hemicelluloses are shown. Xylan consists of repeating β-(1,4)-linked D-xylose residues with the potential for L-arabinose or acetyl substitutions at either the 2-O or 3-O positions or both; 2-O substitution with 4-hydroxy- or 4-methyl-glucuronic acid; and added complexity from substitutions (hexose, pentose and/or phenolics) on the L-arabinose side-chain residues [18,189–192]. Mannan consists of a β-(1,4)-linked D-mannose backbone, whereas the glucomannan backbone has both D-mannose and D-glucose residues with β-(1,4) linkages. Xyloglucan has a β-(1,4)-linked D-glucose backbone with D-xylose side chains, and the mixed-linkage β-glucan backbone has both β-(1,4)- and β-(1,3)-linked glucose residues.
Hemicellulose
Hemicelluloses constitute 20–30% of the biomass of dicotyledonous plants, such as trees, up to 50% for some tissues of monocotyledonous plants, and approximately 25% of the available biomass of the bioenergy-specific crops Miscanthus, switchgrass, fescue and fibre sorghum [16–19]. Hemicellulose bonds to the surface of the cellulose microfibrils and forms a matrix between fibres, where it plays the dual role of keeping fibres from aggregating and adding flexion to the cell wall [20]. Hemicellulose polysaccharides are shorter than those of cellulose (DP of 200 or less) and they are often branched, with short chains containing other sugars, acetyl groups and phenolic groups (Figure 1). Structural heterogeneity is a hallmark of hemicelluloses, arisen due to the physical benefits of resistance to environmental degradation.
A procedural definition of hemicellulose is the polysaccharides that are extractable from plant cell walls by alkaline solution (e.g. 4–24% potassium hydroxide). Hemicellulose composition varies with plant species and tissue type and it can be classified into four major groups [21] (Figure 1) based on polysaccharide composition [18,22,23]. (i) Mannans comprise galactomannan, glucomannan and galactoglucomannan. Galactomannans have β-1,4-mannose backbones with α-1,6-galactose branches; glucomannans contain both mannose and D-glucose β-1,4-linked backbones; and galactoglucomannans have β-1,4-mannose and β-1,4-glucose backbones with α-1,6-galactose branches attached to the mannose backbone. For example, the secondary cell walls of conifers (i.e. softwoods) consist of galactoglucomannan [10–30% (w/w)]. (ii) Mixed-linkage glucans comprise a backbone of D-glucose residues having both β-1,3 and β-1,4 linkages. For example, the primary cell walls of grasses contain 2–15% (w/w) of mixed-linkage glucans. (iii) Xylans have β-1,4-linked D-xylose backbones that may include arabinan and glucuronic acid side chains. For example, in grasses, the primary cell walls contain 20–40% (w/w) glucuronoarabinoxylan, whereas the secondary cell walls contain 40–50% (w/w) glucuronoarabinoxylan. In dicots (e.g. hardwoods), the secondary cell walls contain 20–30% glucuronoxylan. (iv) Xyloglucan has a β-1,4-glucan backbone with xylose-containing branches that can contain other monosaccharide substitutions, such as galactose, arabinose and fucose. For example, the primary cell walls of conifers (i.e. softwoods) contain 10% (w/w) xyloglucans and the primary cell walls of dicots (e.g. hardwoods) contain 20–25% (w/w).
Xylans are the most abundant class of hemicelluloses, with glucuronoarabinoxylan being the main target for enzymatic saccharification for renewable bio-feedstock production. Glucuronoarabinoxylan (e.g. from corn stover) is composed of a β-(1,4)-linked D-xylose polymer backbone (xylan) with L-arabinose and glucuronic acid side chains [24]. Extensive acetylation may occur and the L-arabinose side chains can be esterified with ferulic acid that in turn cross-links glucuronoarabinoxylan, hemicelluloses and lignin (Figure 1). This structural heterogeneity indicates six requisite enzyme activities for complete glucuronoarabinoxylan saccharification: endoxylanase, β-xylosidase, α-arabinofuranosidase, α-glucuronidase, acetylxylan esterase and ferulic acid esterase.
Lignin
Lignin principally consists of p-hydroxyphenyl-, guaiacyl- and syringyl-phenylpropanoid units that are polymerized by radical chemistry coupling reactions to form covalent ether and alkyl linkages [9]. Grasses and hardwoods contain largely guaiacyl- and syringyl-phenylpropanoid units, with only trace amounts of p-hydroxyphenyl-phenylpropanoid units; softwoods largely contain syringyl-phenylpropanoid units. Most pretreatments do not extensively degrade lignin because of the nature of its linkages. Studies on lignin-reduced mutants of alfalfa and sorghum plants have demonstrated that lower lignin content is positively correlated with enzymatic digestibility following pretreatment [25,26].
Pectin
Pectin is a complex heteropolysaccharide that hydrates and further cements the primary cell wall matrix. It accounts for 30–40% of non-cellulosic polysaccharides in the primary cell walls of herbaceous dicotyledons and non-graminaceous monocots with significantly lesser amounts found in grasses, woody tissue and secondary cell walls [27,28]. Pectin consists of long homogalacturonan chains of α-(1-4)-linked D-galacturonic acid and is often esterified with methyl or acetyl groups. Homogalacturonan is interspersed with the branched polysaccharides rhamnogalacturonan I (primarily), rhamnogalacturonan II and xylogalacturonan [29]. It is desirable to hydrolyse pectin because it blocks cellulases, xylanases and xylan-debranching enzymes from reaching their substrates. Also, pectin is an important aspect in the conversion of citrus waste and sugar beet pulp into ethanol, where the polysaccharides are abundant [30,31].
PRETREATMENT PROCESSES
Pretreatment can include physical, chemical and thermal processes, where most pretreatment conditions combine all three effects. For example, biomass can be treated in a steam explosion reactor with dilute sulfuric acid as a catalyst. The biomass is mixed with dilute sulfuric acid, heated with direct steam and quenched by rapid depressurization; the final step increases surface area and reduces particle size. Meanwhile the dilute acid hydrolyses the hemicellulose and reduces the DP of the lignin. Microscopic studies of the affect of pretreatment on cell wall properties emphasized that only subtle changes are needed to expose microfibrils sufficiently for cellulase action and that collapsing cell wall structure either leads to the same or worse D-glucose yields [32]. Since pretreatment can have negative consequences, owing to the chemicals released from the native plant material or as side products of the pretreatment, that impede enzyme hydrolysis [33–39] and microbial fermentation [40–42], less harsh treatments are called for. For example, switchgrass pretreated with dilute ammonium [8% (w/v), 180°C for 20 min] was hydrolysed to D-glucose at 80% efficiency using commercial cellulases (B.S. Dien, unpublished work). Many additional pretreatments have been applied, including ammonia fibre explosion [43,44], alkaline peroxide [45–47], liquid hot water [44,48], sulfuric acid/sulfite solutions [49,50] and cellulose solvents (i.e. concentrated phosphoric acid and room temperature liquid ionic solutions) [51–54]. Recent pretreatment studies [54a,54b] have reinforced previous work [55,56] which indicates that the major barrier to cellulase activity on pretreated biomass is mass transfer limitations of cellulases binding to the microfibrils, a feature termed porosity. The significance of modifications to the fibril structure, while observed to benefit conversion of pure samples of cellulose, remains uncertain for more complex lignocellulosic samples, with the exception of pretreatments that dissolve cellulose.
THE ENZYMES
Enzymes that catalyse deconstruction of cellulose
Deconstruction of crystalline cellulose can be achieved by the activity of three enzymes: cellobiohydrolase (exocellulase, E.C. 3.2.1.91 and E.C. 3.2.1.−), endoglucanase (endocellulase, E.C. 3.2.1.4) and β-glucosidase (E.C. 3.2.1.21). The first two enzymes act on cellulose. In comparison with many other glycoside hydrolases acting on their natural substrates, the cellulases are known for their low catalytic activity, particularly on crystalline cellulose [3–5]. Several of the endoglucanases and cellobiohydrolases from Trichoderma reesei display kcat values of 2–20 s−1 on amorphous cellulose at 25°C [57], corresponding to 6-fold lower rates on crystalline cellulose [58]. A similar rate, 3.5 s−1, has been calculated for T. reesei CEL7A cellobiohydrolase acting on crystalline cellulose [59–61]. Endoglucanases display rates of similar magnitude on crystalline cellulose: 0.22 s−1 at 45°C [62], 0.5 s−1 at 80°C [63] and 1.4 s−1 at ~90°C [64]. In recent years, the protein production efficiency has been improved by at least 10-fold, which has been variously reported to decrease the cost of the cellulases to approximately $0.10 [3], $0.20 [4] or $0.50 [5] per gallon of ethanol produced.
Cellobiohydrolases catalyse the release of cellobiose from either the non-reducing end or the reducing end of cellulose, depending on the cellobiohydrolase (Figure 2A). They do so through a double-displacement mechanism, where the stereochemistry at the anomeric centre is retained, or through a single displacement mechanism, where the stereochemistry at the anomeric centre is inverted [60]. Cellobiohydrolases belong to GH (glycoside hydrolase) families 5, 6, 7, 9 and 48, according to the CAZy database (http://www.cazy.org), which groups carbohydrate-active enzymes according to their amino acid sequence homology [65]. The database lists X-ray structures (PDB codes) and GenBank® accession information for the enzymes discussed in the present review. Endoglucanases catalyse the endohydrolysis of (1,4)-β-D-glucosidic linkages in cellulose. The enzymes belong to families GH 5–10, 12, 16, 18, 19, 26, 44, 45, 48, 51, 74 and 124. GH61 was initially described as an endoglucanase, but more recently has been shown to lack GH activity [66]. Instead, members of the family are copper mono-oxygenases that catalyse cleavage of cellulose oxidatively, releasing cellodextrins [67–69]. The mono-oxygenases can accept reducing equivalents from ascorbate or reduced dyes [67,69]. In situ, cellobiose dehydrogenase is probably the electron donor [68,69]. In the presence of cellulose, cellobiohydrolase and endoglucanase, the mono-oxygenase and cellobiose dehydrogenase act synergistically to enhance the cellulase activity by approximately 2-fold [68]. In the presence of cellulose and β-glucoside, the mono-oxygenase enhances the β-glucosidase manyfold [68]. β-Glucosidases act on soluble cello-oligosaccharides, including cellobiose. The enzymes catalyse the hydrolysis of terminal non-reducing β-D-glucosyl residues through a double displacement mechanism, with release of β-D-glucose. The enzymes belong to families GH 1, 3, 5, 9, 30 and 116. A GH3 β-glucosidase from Aspergillus oryzae expresses typical catalytic parameters: kcat of 1000 s−1 and Km of 2.0 mM (pH 5.0, 50°C) [70]. Crude industrial mixtures of enzymes that act on cellulose and soluble cello-oligosaccharides have been reviewed recently [5].
(A) Hypothetical active site of a cellobiohydrolase that cleaves cellobiose from the non-reducing end. The scissile site is the glycosidic bond held between subsite −1 and subsite +1. Numbering proceeds to the left of the scissile bond towards the non-reducing end of the oligoglucoside for negatively numbered subsites, and to the right towards the reducing end of the oligoglucoside for positively numbered subsites. (B) Substrates of Selenomonas ruminantium β-xylosidase/α-arabinofuranosidase. (C) Overlay of D-xylopyranose and L-arabinofuranose showing that the two-ring systems can occupy similar space.
Enzymes that catalyse deconstruction of hemicellulose
Below the focus is on the enzymes that operate on glucuronoarabinoxylan. Descriptions of the enzymes that act on the other forms of hemicellulose, as well as other plant polysaccharides, can be found elsewhere [71]. Heterogeneity of glucuronoarabinoxylan requires six distinct enzyme activities for complete saccharification: endoxylanase (EC 3.2.1.8), β-xylosidase (EC 3.2.1.37), α-arabinofuranosidase (EC 3.2.1.55), α-glucuronidase (EC 3.2.1.131), acetylxylan esterase (EC 3.1.1.72) and ferulic acid esterase (EC 3.1.1.73). Economic success of a biorefinery requires efficient utilization of the hemicellulose carbohydrates that can be released by enzymatic hydrolysis. In addition, xylan removal leads to greater enzymatic hydrolysis of cellulose, resulting in increased D-glucose yield. The mechanisms by which xylan removal leads to increased cellulose breakdown include increased accessibility to the cellulose fibrils [72,73] and removal of xylo-oligosaccharides [39], which inhibit cellulase enzyme activity. Similarly synergism is a common theme for the enzymes that act in the deconstruction of xylans, including glucuronoarabinoxylan. Inclusion of the activities of the following individual enzymes enhances the rate of the endoxylanase acting on glucuronoarabinoxylan: β-xylosidase [74], α-glucuronidase [75], acetylxylan esterase [76] and ferulic acid esterase [77,78].
Many industrially relevant endoxylanase genes have been cloned from a wide array of bacteria and fungi [79]. The majority of these enzymes are classified into the GH10 and GH11 families. Both categories of enzymes use a double-displacement mechanism that retains the anomeric configuration. The GH10 endoxylanases have lower substrate specificity and, therefore, a higher capacity to hydrolyse substituted xylan polymers. The recombinant endoxylanase with the highest reported specific activity is Xyn10B from Cellvibrio mixtus [80]. Acting on oat spelt xylan, Xyn10B has a Km of 6.17 mg·ml−1and a kcat of 330 s−1 at 37°C. There are also many endoxylanases that function optimally at extreme conditions: pH 2 [81–83], pH 9–10 [84–86], 85–100°C [87–89] and 4°C [90,91].
Xylan 1,4-β-D-xylosidase (EC 3.2.1.37) catalyses the hydrolysis of single D-xylose units from the non-reducing end of xylo-oligosaccharides. It is classified in the CAZy database under GH families 1, 3, 30, 39, 43, 51, 52, 54, 116 and 120 [65]. As well as being required for the complete saccharification of xylan [92], β-xylosidase has been demonstrated to act synergistically with other hemicellulases in the degradation of xylan [74]. At present, the preponderance of characterized β-xylosidases belong to GH family 43. GH43 is targeted in part because this family catalyses hydrolysis using an inverting single-displacement mechanism. This precludes transxylosylation [93], which could otherwise impinge on efficiency, as in the case of high concentrations of a competing nucleophile, such as xylobiose.
GH43 β-xylosidase SXA, isolated from the organism Selenomonas ruminantium, a resident of bovine rumen [94,95], has the highest reported kcat value (185s−1 at pH 5.3, 25°C) for hydrolysis of xylobiose (X2, Figure 2B), with a Km of 2.1 mM [96]. GH43 β-xylosidase, XylBH43, from Bacillus halodurans acting on xylobiose exhibits a kcat of 117 s−1 and a Km of 3.02 mM at pH 6.5 and 25°C [97]. Owing presumably to the spatial similarity of β-D-xylose and α-L-arabinofuranose (Figure 2C), β-xylosidases ordinarily exhibit secondary α-arabinofuranosidase activity [98,99]. Many β-xylosidases have been studied exclusively with colorimetric model substrates. However, to evaluate the enzymes' potential performance in a saccharification reactor, it is necessary to determine their activity on relevant natural substrates, as demonstrated by β-xylosidase GbtXyl43A. This enzyme exhibits greater catalytic efficiency in hydrolysing 4NPA (4-nitrophenyl-α-L-arabinofuranose) (Figure 2B) than for 4NPX (4-nitrophenyl-β-D-xylopyranoside) (Figure 2B), but displays insignificant α-arabinofuranosidase activity and is limited to β-xylosidase activity on natural substrates [100]. Also, it has been demonstrated that there is only a modest correlation between the activity exhibited on a natural substrate, X2, and the model substrate, 4NPX (Figure 3) (D.B. Jordan and J.D.Braker, unpublished work) [92].
The β-xylosidases with literature values are from Selenomonas ruminantium (1), Bacillus pumilus 12 (2), Bacillus pumilus IPO (3) and Bacillus halodurans C-125 (4) [92,97]. The remaining β-xylosidases without literature values are from Bacillus subtilis subsp. subtilis str. 168 (5), Alkaliphilus metalliredigens QYMF (6), Bacillus sp. (7) and Bacillus sp. (8) (D.B. Jordan and J.D. Braker, unpublished work).
β-Xylosidases can be inhibited by D-xylose and D-glucose at low millimolar levels, both of which could be present at high concentrations in an industrial saccharification process. This precludes the use of β-xylosidase in an SHF process because the concentrations of D-glucose and D-xylose would reach high concentrations (>1 M) in excess of 100-fold that of Ki. Increased attention has been directed towards characterizing and engineering decreased monosaccharide inhibition of β-xylosidases [101]. Also, there are several reports of D-xylose-tolerant β-xylosidases from both fungal sources [102–104] and bacterial sources [105,106].
α-Arabinofuranosidases (EC 3.2.1.55), classified in the CAZy database under GH families 3, 43, 51, 54 and 62 [65], catalyse the hydrolysis of terminal non-reducing L-arabinose side chains from the xylan backbone, where they can be found both singly- and doubly-substituted at C-2 and/or C-3 of the xylopyranose backbone via α-1,2- and α-1,3-linkages [107]. α-Arabinofuranosidases that cleave L-arabinose exclusively from glucuronoarabinoxylan polymers are termed glucuronoarabinoxylan arabinofuranohydrolases [108]. Similar to the β-xylosidases, many of the α-arabinofuranosidases display secondary β-xylosidase activity [109,110]. α-Arabinofuranosidases have generally been characterized kinetically on artificial substrates, which does not correlate with activity on natural substrates. This is illustrated in the recent study of two α-arabinofuranosidases, AF30 and AF47, isolated from a fungal pathogen of preharvest corn, where the relative kcat (AF47/AF30) is ~2 for 4NPA hydrolysis, whereas the specific activity ratio for corn fibre glucuronoarabinoxylan hydrolysis (AF47/AF30) is ~0.33 [111]. α-Arabinofuranosidase kcat values for hydrolysis of natural substrates are rarely reported, a recent notable exception being the modular GH43 glucuronoarabinoxylan arabinofuranohydrolase from the ruminal bacterium Fibrobacter succinogenes. This enzyme has reported kinetic parameters of a kcat of 240 s−1 and Km of 4.1 mg·ml−1 acting on natural glucuronoarabinoxylan substrate for the wild-type enzyme, and moreover, a truncated site-directed mutant resulted in an increased kcat of 630 s−1 and Km of 24 mg·ml−1 [112].
The α-glucuronidases (EC 3.2.1.131) catalyse hydrolysis of the 1,2-linked glucuronosyl side chains from xylan [113]. The glucuronosyl substitutions inhibit both enzymatic and acidic hydrolysis of the xylan polymer [114]. In addition, the glucuronosyl group can form covalent cross-links to lignin [115–117]. The majority of α-glucuronidases are categorized as GH67 and remove only the glucuronosyl group that is attached to the terminal residue at the non-reducing end of xylo-oligosaccharides. The α-glucuronidase with the highest reported catalytic activity (a kcat of 202s−1 at 40°C and pH 4.8) on a native substrate (aldotetraouronic acid) is from Aureobasidium pullulans [118]. Most α-glucuronidases have an optimal pH of 4.5–6.5 and an optimal temperature of 40–65°C. There are several examples of α-glucuronidases that have acidic (pH 3–3.5) optima, but none with an alkaline optimum [119–121]. The α-glucuronidase from Thermotoga maritima has the highest reported temperature optimum (85°C) [122]. Previously, a new category of α-glucuronidases was discovered that acts on a polymeric substrate, and these enzymes are classified in the GH115 family [123].
CEs (carbohydrate esterases)
CEs are organized into 16 families in the CAZy database. The CE enzyme activities targeted for glucuronoarabinoxylan saccharification are acetylxylan esterases (EC 3.1.1.72; CE families 1–7, 12 and 15) and ferulic acid esterases (EC 3.1.1.73; CE family 1). CEs belong mainly to the α/β hydrolase fold superfamily, which includes hydrolases, dehalogenases, lipases and peroxidases [124]. The active sites generally contain a serine-histidine-carboxylate catalytic triad, and exhibit modest catalytic rates and much smaller rate enhancements than the GH enzymes. Alkaline pretreatment conditions could effectively saponify the ester bonds, rendering the esterases redundant.
Acetylxylan esterases remove acetyl groups from the xylan backbone. The acetyl groups can attach to the 2-O, 3-O or both positions of the xylosyl monomer. Most acetylxylan esterases are in families CE 1, 4 or 5. The enzymes were previously demonstrated to preferentially remove acetyl groups from the 2-O position [125]. However, in an aqueous environment, the acetyl groups can easily migrate between the 2-O and 3-O positions, thus it is not critical to use two esterases to target the acetyl groups at both sites. Acetylxylan esterases have long been recognized as being important for the enzymic saccharification of acetylated biomass [126]. A kcat of 24 s−1 at 50°C has been determined for the enzyme from T. reesei acting on D-xylose tetraacetate [127].
Ferulic acid esterases (EC 3.1.1.73) catalyse hydrolysis of L-arabinose-ferulate ester bonds [128–130]. L-Arabinose side chains of glucuronoarabinoxylan can be substituted with ferulic acid (4-hydroxy-3-methoxycinnamic acid) at the C-5 and C-2 hydroxy groups of L-arabinose. There it can form ferulate bridges consisting of dehydrodimers and dehydrotrimers of various linkage configurations between carbohydrate chains, and between carbohydrate chains and lignin (Figure 1). Cross-linking increases plant cell wall structural rigidity, which simultaneously hinders access to the hemicellulosic substrate by GH enzymes [131], and has been likened to “the molecular equivalent of spot-welding a steel-mesh frame” [132]. Most activity studies with ferulic acid esterases have been conducted on methyl ester model substrates, which display a maximum kcat of 200 s−1 [133,134]. The turnover number for the release of ferulic acid from hemicellulose derived from rye grass is approximately 4 s−1 at 37°C [135].
FERMENTATION
Economically competitive ethanol production from lignocellulosic materials requires efficient use of both the hexose and pentose monosaccharides. Although there are other hexose sugars, in addition to D-glucose, galactose and mannose (e.g. glucuronic acid, galacturonic acid and rhamnose), depending on the feedstock, they are present in small amounts relative to the hemicellulose fraction consisting mainly of D-xylose. L-Arabinose is also present in the hemicellulose fraction, but again, it is a small portion of the hemicellulose relative to D-xylose. The recognition that utilization of the D-xylose fraction will be required to make lignocellulosic ethanol cost-competitive has resulted in a great deal of work to engineer pathways and organisms to convert D-xylose.
Fermentative production of biofuels from lignocellulosic feedstocks provides unique challenges for micro-organisms. In addition to efficient pentose metabolism and inhibitor tolerance, the ideal fermenting micro-organism would also maintain the productivity measures of the current starch- and cane sugar-based ethanol systems, being tolerant to ethanol, low pH, high osmolarity and high temperature (to lower the cooling cost of removing heat generated by the fermentation). No micro-organism has been discovered that is capable of fermenting D-xylose at high titres and high productivity using a homo-ethanol pathway [136]. As a result, several groups have undertaken research to isolate and/or engineer, through directed and random methods, organisms with well-defined properties and reliable genetic transformation systems, such as Zymomonas mobilis [137] and Escherichia coli [138,139], along with additional yeast species such as Scheffersomyces stipitis [40]. Despite some success in engineering these organisms for potential use in lignocellulosic biofuel production, most are not as tolerant as Saccharomyces cerevisiae towards ethanol or some of the inhibitors found in lignocellulosic hydrolysates [140]. Additionally, while numerous organisms naturally metabolize D-xylose, and some will ferment D-xylose to ethanol, D-glucose fermentation rates are typically orders of magnitudes lower by these organisms compared with Saccharomyces. Thus, owing to its high ethanol yield, high productivity under anaerobic conditions and wide use in industrial fermentation processes, S. cerevisiae remains the preferred organism for converting biomass-derived monosaccharides into bioethanol.
Unfortunately, S. cerevisiae does not naturally ferment D-xylose. Several pathways that exist in nature for metabolizing D-xylose (Figure 4) have been engineered into S. cerevisiae. Most efforts towards engineering Saccharomyces yeasts for D-xylose fermentation have focused heavily on reconstitution of the two xylulose 5-phosphate-producing pathways (i.e. D-xylose reductase/xylitol dehydrogenase or D-xylose isomerase) (reviewed in [141–144]). Native D-xylose-metabolizing fungi typically use a two-step reduction/oxidation, whereas bacteria usually employ a single isomerization of D-xylose. Both pathways convert D-xylose into xylulose, which is phosphorylated by xylulokinase to produce the PPP (pentose phosphate pathway) intermediate xylulose 5-phosphate. A series of carbon-transfer reactions in the non-oxidative branch of the PPP results in the production of fructose 6-phosphate and glyceraldehyde 3-phosphate, which can be metabolized by S. cerevisiae.
Xylulose 5-phosphate route: XR, D-xylose reductase (E.C. 1.1.1.21); XOHDH, xylitol dehydrogenase (E.C. 1.1.1.9); XI, D-xylose isomerase (E.C. 5.3.1.5); XK, xylulokinase (E.C. 2.7.1.17); RPE, ribulose 5-phosphate-3-epimerase (E.C 5.1.3.1); TKL, transketolase (E.C. 2.2.1.1); AP, acylphosphatase (E.C. 3.6.1.7); AK, acetate kinase (E.C. 2.7.2.1); PTA, phosphotrans-acetylase (E.C. 2.3.1.8); and ACS, acetyl-CoA synthetase (E.C. 6.2.1.1). Non-phosphorylated intermediate route: XDH, D-xylose dehydrogenase (E.C. 1.1.1.175, E.C. 1.1.1.179); XL, xylonolactonase (E.C. 3.1.1.68); XAD, xylonate dehydratase (E.C. 4.2.1.82); KDXA, 2-oxo-3-deoxy xylonate aldolase (E.C. 4.1.2.28); KDXD, 2-oxo-3-deoxy xylonate dehydratase (E.C. 4.2.1.-); and αKGSADH, α-oxoglutaric semialdehyde dehydrogenase (E.C. 1.2.1.3).
Although some success has been reported on the aforementioned pathways in S. cerevisiae, a number of shortcomings remain and warrant exploration of additional pathways. Another potential pathway to D-xylose metabolism could be initiated by the enzyme D-xylose dehydrogenase (Figure 4). This pathway splits at the intermediate 2-oxo-3-deoxyxylonate where it is (i) cleaved by an aldolase to generate pyruvate and glycolaldehyde, or (ii) acted on by two additional enzymes to form α-oxoglutarate. This pathway has the potential to avoid limitations imposed by low PPP flux in S. cerevisiae because xylulose 5-phosphate is not formed. The initial enzyme of the pathway, D-xylose dehydrogenase, has a higher affinity (Km<4 mM) for D-xylose [145–148] than D-xylose reductases or isomerases which have Km values of approximately 20–60 mM [149–151]. D-Xylose isomerases with higher affinity for xylose are common in bacteria; however, many of these bacterial D-xylose isomerases do not work in S. cerevisiae, and D-xylose isomerases that have been shown to function in this yeast show poor affinity for D-xylose. Considering the inefficient transport of D-xylose into S. cerevisiae (see below), a pathway with increased affinity for D-xylose could improve D-xylose utilization. Archaea and some bacteria are the only organisms to date that have been shown to contain complete pathways to metabolize D-xylose via an initial oxidation [146–148]. D-Xylose dehydrogenase has recently been expressed in Kluyveromyces lactis for the production of xylonic acid [152], but engineering of the entire pathway has not been reported.
Another possible route to explore is a branch of the xylulose 5-phosphate route which includes cleavage of xylulose 5-phosphate to acetyl phosphate, and glyceraldehyde 3-phosphate by the enzyme PK (phosphoketolase) (Figure 4). Acetyl phosphate can be converted into acetyl-CoA, which can be metabolized via the TCA (tricarboxylic acid) cycle. Alternatively, it can be further converted into ethanol by acetaldehyde dehydrogenase (acylating) and alcohol dehydrogenase (not shown). The role of PK in D-xylose (and D-glucose) metabolism in certain bacteria has long been established [153,154]. Later, PK activity was shown to be induced in yeasts up to 70-fold by D-xylose [155], suggesting that this branch plays an important role in D-xylose metabolism, at least in certain yeasts. A PK pathway has been engineered into S. cerevisiae, resulting in an increased ethanol yield [156]. Aside from this work, little has been reported pursuing this approach.
In addition to exploring the D-xylose dehydrogenase and PK routes to D-xylose metabolism, a number of additional problems associated with the two main D-xylose metabolism pathways are also in need of resolution. Additional concerns related to D-xylose transport, redox imbalance, flux limitations through the PPP and the role of gluconeogenesis must be resolved to further increase D-xylose fermentation rates.
D-Xylose transport
Saccharomyces yeasts do not possess D-xylose-specific transporters. D-Xylose gains entry to the cell through the HXT family of D-glucose transporters [157]. To overcome this bottleneck, a number of groups have expressed D-xylose transporters from bacterial, yeast and plant species into yeast [158–163]. This approach has improved D-xylose uptake for some of the strains. Unfortunately, progress in this area has been slow because the number of known D-xylose transporters available for expression and study in S. cerevisiae is limited. While kinetic data for D-xylose transport in native D-xylose-metabolizing yeasts have been available for decades [164–170], most of the D-xylose-specific transporters from these organisms have not been isolated. Work with additional transporters, preferably passive transporters that do not require the expenditure of energy, will shed additional light on the efficiency of D-xylose transport and its role in improving D-xylose fermentation.
Redox imbalance
Under anaerobic conditions, cofactor differences between the first two enzymes in the fungal pathway result in NADPH depletion and NADH accumulation [171]. NADH accumulation favours the production of xylitol and glycerol at the expense of ethanol. NADPH depletion results in decreased D-xylose reduction and limits reducing power for generating cell biomass and for inhibitor tolerance [172]. Multiple strategies have been investigated to alter the redox balance. Protein engineering of the S. stipitis xylose reductase to increase the use of NADH over NADPH by the enzyme has shown some benefit [173–176]. However, many of the strategies have failed to show significant improvement, and others have intensified the problem [177,178]. Theoretically, expression of a D-xylose isomerase would alleviate this imbalance of cofactors [179]. However, most D-xylose isomerases expressed in S. cerevisiae function poorly and the equilibrium does favour D-xylose. NADPH may still be limiting in S. cerevisiae expressing a D-xylose isomerase pathway due to its inability to recycle D-xylose back to D-glucose 6-phosphate to regenerate NADPH through the oxidative branch of the PPP [180].
PPP flux limitations
Although low flux through the PPP in S. cerevisiae is beneficial for D-glucose fermentation, low flux through the non-oxidative part of the PPP limits D-xylose fermentation. Elevated expression of the non-oxidative branch enzymes has been used to increase D-xylose fermentation [181]. It was also previously discovered that strains used for production of ethanol from cane sugar have duplicate genes involved in thiamine (vitamin B1) and vitamin B6 synthesis [182]. Thiamine pyrophosphate is a cofactor of the non-oxidative PPP pathway enzyme transketolase, and this adaptive duplication may be helpful for increasing flux through the PPP. NADPH produced by the oxidative branch of the PPP pathway is also used in detoxifying furfural and hydroxymethylfurfural, and low flux through this pathway potentially further limits the ability of Saccharomyces to efficiently ferment D-xylose from lignocellulosic hydrolysates [172].
Requirement for gluconeogenesis
NADPH regeneration in yeast occurs mainly through the oxidative branch of the PPP. Native D-xylose-utilizing yeasts appear to recycle D-xylose back to D-glucose 6-phosphate to be used for glucan synthesis of cell wall components and in the oxidative branch of the PPP for NADPH production required for anabolic reductive reactions. S. cerevisiae expressing the S. stipitis D-xylose reductase/xylitol dehydrogenase pathway was recently shown to be unable to induce the genes required to regenerate NADPH in this manner, thus limiting D-xylose utilization [180]. Resolving the redox imbalance by use of a D-xylose isomerase pathway or introducing alternative NADPH regeneration mechanisms will alleviate the need for recycling D-xylose to D-glucose, but may not eliminate it. Regardless of the pathway used to metabolize D-xylose, D-glucose will still be required for the glucan component of the cell wall and NADPH (via D-glucose 6-phosphate and the oxidative PPP) for cell growth and inhibitor tolerance. Whereas native D-xylose-utilizing yeasts are able to induce enzymatic activity to produce D-glucose 6-phosphate from D-xylose, the transcriptional response of S. cerevisiae to D-xylose is not optimized for this pathway, and may actually induce genes that negatively affect D-xylose utilization. A recent paper describes expression of a cellodextrin transporter and an intracellular β-glucosidase in S. cerevisiae grown in a medium containing cellobiose and D-xylose [183]. This strategy partially overcomes the problem by providing low levels of D-glucose during D-xylose fermentation. SSCF can also provide low levels of D-glucose during D-xylose fermentation, allowing more efficient D-xylose uptake and fermentation.
D-Xylose-regulated promoters
A wide variety of promoters are available for constitutive and regulated expression of foreign genes in S. cerevisiae. In many cases, however, constitutive expression is a waste of cellular resources during the D-glucose phase, when D-xylose is not metabolized. For example, the expression of genes for D-xylose transport, or for improving redox imbalance during the D-xylose consumption phase, could induce an imbalance during D-glucose fermentation. The ability to fine-tune the expression of the multiple genes required for D-xylose fermentation, to be expressed only when needed, will allow better control of the genetically engineered pathways. Unfortunately, D-xylose-regulated promoters are not yet available for control of gene expression in S. cerevisiae.
PATH FORWARD
Improving the commercial feasibility of second generation bioethanol production requires integrating goals for feedstock development, pretreatment, enzymatic saccharification, fermentation, waste treatment and process water recycling. In the area of feedstock development, better analytical techniques will allow for better understanding of the interrelationship among the major cell components, hemicellulose, lignin, pectin and cellulose. Immediate progress in feedstock development is likely to be dominated by lignin modification because decreased lignin content is highly correlated with improved enzymatic saccharification yields, and considerable progress has been made in determining lignin synthetic pathways [25,26]. Several pretreatments are able to affect alteration of cellulose lattice structure, of which room temperature ionic liquids are the newest candidates; however, doubts remain regarding cost and the ability to recycle these solvents. Most pretreatment strategies being pursued [44,184] do not fully hydrolyse xylan in order to avoid generation of high levels of simple sugars in the pretreatment step; ultimately, such procedures maximize yield of monosaccharides and minimize formation of furans [41,42]. Greater understanding of the structure of residual fermentation oligosaccharides and their impact on complete hydrolysis and/or enzyme activity may lead to knowledge of additional enzyme activities needed to complete saccharification. Newly discovered enzymes (formerly GH61) that oxidatively cleave cellulose offer an exciting possibility for obtaining large improvements in cellulose deconstruction rates [66–68]. Natural enzyme sources, metagenomic DNA libraries and genetically engineered libraries should be searched for enzymes with improved turnover numbers on natural substrates, increased tolerance to soluble inhibitors (e.g. D-glucose, D-xylose and furans), and lower non-specific binding to lignin.
Improving the efficiency of D-xylose fermentation will remain a major research focus. Although yield has been improved considerably (to >0.4 g of ethanol/g of D-xylose) [185], specific ethanol productivity (g of ethanol/g of cells per h) from D-xylose still lags D-glucose fermentation [143] by an order of magnitude. In the future, genetic engineering of strains will continue to be a powerful tool to improve yeast strains, as work with native D-xylose-utilizing yeasts has uncovered genes that may assist with D-xylose metabolism that are either not present or regulated improperly in D-xylose-grown S. cerevisiae. Additionally, a number of microarray and proteomic studies have also shown that multiple genes are regulated in strains that have been engineered to ferment D-xylose, suggesting that D-xylose utilization will depend on genes and/or pathways beyond what have already been engineered into the strains [186–188] and will probably need to be evolved/adapted simultaneously. Increased understanding of how native D-xylose-utilizing yeasts efficiently metabolize D-xylose and re-engineering these additional pathways, along with strategies for improved inhibitor tolerance, into robust industrial S. cerevisiae strains will lead to further improvements in ethanol productivity, yield and cost competitiveness.
Abbreviations: CBP, consolidated bioprocessing; CE, carbohydrate esterase; DP, degree of polymerization; GH, glycoside hydrolase; 4NPA, 4-nitrophenyl-α-L-arabinofuranose; 4NPX, 4-nitrophenyl-β-D-xylopyranoside; PK, phosphoketolase; PPP, pentose phosphate pathway; SHF, separate hydrolysis and fermentation; SSCF, simultaneous saccharification and co-fermentation; SSF, simultaneous saccharification and fermentation
- © The Authors Journal compilation © 2012 Biochemical Society
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.
- 35.
- 36.
- 37.
- 38.
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.
- 53.
- 54.↵
- 54a.
- 54b.
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.
- 83.↵
- 84.↵
- 85.
- 86.↵
- 87.↵
- 88.
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.
- 104.↵
- 105.↵
- 106.↵
- 107.↵
- 108.↵
- 109.↵
- 110.↵
- 111.↵
- 112.↵
- 113.↵
- 114.↵
- 115.↵
- 116.
- 117.↵
- 118.↵
- 119.↵
- 120.
- 121.↵
- 122.↵
- 123.↵
- 124.↵
- 125.↵
- 126.↵
- 127.↵
- 128.↵
- 129.
- 130.↵
- 131.↵
- 132.↵
- 133.↵
- 134.↵
- 135.↵
- 136.↵
- 137.↵
- 138.↵
- 139.↵
- 140.↵
- 141.↵
- 142.
- 143.↵
- 144.↵
- 145.↵
- 146.↵
- 147.
- 148.↵
- 149.↵
- 150.
- 151.↵
- 152.↵
- 153.↵
- 154.↵
- 155.↵
- 156.↵
- 157.↵
- 158.↵
- 159.
- 160.
- 161.
- 162.
- 163.↵
- 164.↵
- 165.
- 166.
- 167.
- 168.
- 169.
- 170.↵
- 171.↵
- 172.↵
- 173.↵
- 174.
- 175.
- 176.↵
- 177.↵
- 178.↵
- 179.↵
- 180.↵
- 181.↵
- 182.↵
- 183.↵
- 184.↵
- 185.↵
- 186.↵
- 187.
- 188.↵
- 189.↵
- 190.
- 191.
- 192.↵
March 2012
Volume: 442 Issue: 2
Open in Utopia Docs
