Almost all aerobic cellulolytic bacteria secrete sets ofindividual cellulases, and most ofthese cellulases contain one or more CBMs. Two such organisms, whose cellulases have been well characterized, are Cellulomonas fimi and T. fusca. The sets of six cellulases produced by these two organisms are similar in their activity, CD families and CBMs, but differ significantly in their sequences and domain order, suggesting that these organisms did not obtain their cellulase genes from a common ancestor, but rather each set is the result of convergent evolution. In five of the six corresponding cellulase pairs from each species, the family 2 CBM is at the opposite end of the enzyme (23). Both these organisms are actinomycetes and they are found in soils, but C. fimi is mesophillic with an optimum growth temperature near 30°C, while T. fusca is moderately thermophilic with an optimum growth temperature of 50°C. An interdisciplinary group at the University of British Columbia has studied C. fimi cellulases, and this group has made many important contributions to cellulase research (74).

T. fusca is often found in compost piles, rotting hay, and manure piles. The genome sequence of T. fusca was determined in 2000 by the DOE Joint Genome Institute and the finished 3.7 mb sequence is available at: http://genome. jgi-psf. org/finished_microbes/thefu/ thefu. home. html (75). These two organisms are not closely related, as C. fimi is in the suborder Micrococcineae while T. fusca is in the suborder Streptosporangineae. It is interesting that Streptomyces lividans, which is a relative of T. fusca, contains five cellulase genes that are similar to those present in T. fusca and, in every one, the family 2 CBM is in the same location as it is in T. fusca, suggesting that these sets of genes did come from a common ancestor.

In addition to the six cellulases, T. fusca secretes a number of other proteins when it is grown on cellulose. Among them are a xyloglucanase, a xylanase, a family 81 [3-1,3 endoglucanase, and two proteins that bind to cellulose but do not appear to have any catalytic activity (76-79). The xyloglucanase and the xylanase both contain a family 2 CBM and bind tightly to cellulose. The CBM on the xylanase also binds to xylan, which is unusual, as most family 2 CBMs bind only to one of these polymers (80). It seems likely that the role of the xyloglucanase is to degrade xyloglucan, which is bound to cellulose, allowing the cellulases to access the cellulose rather than to allow T. fusca to utilize xyloglucan, since T. fusca does not grow on xyloglucan, even though it completely degrades it to oligomers (76). Further evidence for this conclusion is that a xyloglucan-cellulose composite, which was not degraded by a mixture of pure cellulases, was degraded when pure T. fusca xyloglucanase was added to the cellulase mixture (76). The role of the two cellulose-binding proteins is not known, but they do stimulate the activity of several T. fusca cellulases in the initial phase of the reaction, when the cellulases are assayed at low levels (79).

Cellulase synthesis in T. fusca and many cellulolytic bacteria is regulated by at least two mechanisms: induction by cellobiose and laminaribiose (p-1,3-glucose disaccharide), and repression by any good carbon source including cellobiose (78, 80, 81). This makes sense, as the extracellular enzymes secreted by T. fusca grown on cellulose make up about 50% of the total protein in the culture. If there is sufficient sugar for growth or no cellulose, there is no reason to synthesize cellulases. The cellobiose required for induction is produced from cellulose by the uninduced level of secreted cellulase, which is mainly Cel6A. CelR is a regulatory protein that is a member of the Lac I gene family (82), and it binds to a 14 base inverted repeat sequence: TGGGAGCGCTCCCA. This sequence is upstream of the start site of all six T. fusca cellulase genes coding for secreted cellulases, as well as the genes for a number of other secreted proteins induced by growth on cellulose, and an operon that codes for a cytoplasmic 3 glucosidase and a putative cellobiose transport system (83). The CelR gene is just upstream of this operon. The binding of CelR to the regulatory sequence is inhibited by cellobiose, as expected. Since laminaribiose also induces cellulase synthesis it also probably binds to CelR but this has not been tested. At this time it is not known whether induction by laminaribiose is useful for T. fusca, or if it is an accidental result of the low specificity of the CelR sugar-binding site. A puzzling finding is that the Cel6B and Cel48A genes, which each encode an exocellulase, have a second CelR-binding site that is about 200 bases upstream of the start site of the gene and this site is not present in the other cellulase gene upstream sequences. These two cellulases are made in equal amounts and together they make up over 70% of the secreted cellulase protein. At this time there is no information about the mechanism by which good carbon sources inhibit cellulase synthesis in T. fusca.

Saccharophagus degradans is an aerobic marine plant cell wall degrading organism, whose genome was recently sequenced. Preliminary analysis of its 13 cellulase genes shows that it contains ten family GH-5 enzymes, which are most similar to endoglucanases and most of them also contain cellulose-binding domains. It also contains two family GH-9 endoglu­canase genes, one of which encodes both a family 10 and a family 2 CBM (84).

Cytophaga hutchinsonii is an aerobic cellulolytic bacterium and the DOE Joint Genome Institute determined the DNA sequence of its genome (http://genome. jgi- psf. org/finished_microbes/cythu/cythu. home. html) (85). While it codes for a number of cellulase genes, most of these genes lack a CBM and all the genes appear to code for endoglucanases. These results clearly distinguish C. hutchinsonii from all other aerobic cellulolytic bacteria, whose cellulases are known, and suggest that it uses a different mech­anism for degrading cellulose than the secretion of a synergistic set of free cellulases used by the other well-studied aerobic cellulolytic microorganisms. None of its cellulase genes encode dockerin sequences, so that it does not appear to produce cellulosomes, like anaer­obic cellulolytic bacteria. Thus, it appears that C. hutchinsonii has a third mechanism for degrading cellulose. It is interesting that the anaerobic rumen bacterium, Fibrobacter suc- cinogenes, whose genome sequence was determined by TIGR funded by a USDA grant to the North American Consortium for Genomics of Fibrolytic Ruminal Bacteria, also does not code for any known processive cellulases and only one of the many endocellulases that have been cloned and sequenced from it appears to bind to cellulose (86,87). F. succinogenes does not appear to encode dockerin domains and a scaffoldin gene has not been identified. Thus, F. succinogenes also probably uses a novel mechanism for degrading cellulose. F. suc­cinogenes grows very rapidly on cellulose, so that its cellulose degrading mechanism is very efficient (88). One possible mechanism for these organisms is the one proposed for starch degradation by Bacteroides thetaiotaomicron (89). In this mechanism, starch is bound to a complex present in the outer membrane and individual molecules are transported into the periplasmic space, where they are degraded by starch degrading enzymes. This mechanism would not require processive cellulases, as individual cellulose molecules would be readily degraded by endoglucanases. If this is the process by which cellulose is degraded, it will be very interesting to determine the mechanism by which the outer membrane proteins are able to bind and transport individual cellulose molecules. It is possible that this information would allow the design of new cellulases or cellulose modifying proteins, which would be able to increase the rate of cellulose degradation by free cellulases.

Site-directed mutagenesis of cellulase genes has been used to identify residues required for activity and substrate binding. Extensive studies of T. fusca Cel6A have been carried out (90-92) and a number of Cel6A residues have been identified that are essential for activity or binding. From these studies, a detailed mechanism has been proposed for Cel6A, which differs from the standard cellulase mechanism in that there is no evidence for an essential catalytic base, although there is a conserved Asp residue (Asp 79), which is important for activity and it could be functioning as a nonessential catalytic base (91). Because it is in a flexible loop, it is not clear where this residue is positioned during bond cleavage, but it is unlikely to be in the position seen for the catalytic base in other cellulase families, which is on the other side of the sessile bond from the essential catalytic acid, which is Asp 117 in Cel6A (91). Loop movement is important in catalysis by Cel6A, as mutation of either of two adjacent Gly residues to Ala, which are at one end of the loop, reduce activity on CMC to 6% or 16% of WT activity (90).

In all WT Cel6Acd structures that contain bound sugars, the glucose molecule in subsite -1 is distorted, as is seen with most cellulases. Tyr73 is a conserved residue in family GH-6 and, when it is mutated to Phe, Cel6A activity is reduced to 10% of WT but, when it is mutated to Ser, activity is abolished (93). In a structure of the Cel6Acd Tyr73Ser mutant enzyme, the glucose in the -1 is not distorted (94). This result is consistent with molecular modeling of Cel6A, showing that the normal glucose conformation would overlap the position of Tyr73 (95). It also suggests that distortion of the glucose is essential for activity, since the Ser mutation eliminates both distortion and activity, while the Phe mutation, which does not eliminate the distortion only partially reduces activity (93). Tyr73 was reported to be the essential hydrophobic platform residue in family GH-6, and the mutagenesis data are consistent with this proposal (96). Cel6A hydrolyzes CMC and amorphous cellulose (SC) 100 times faster than filter paper and 50 times faster than bacterial cellulose. Furthermore, mutations that alter key active site residues cause much greater reductions in activity on CMC and SC than on the crystalline substrates (91). These results show that there are different rate-limiting steps for the two types of substrates. It seems likely that the rate — limiting step for CMC and SC is bond cleavage, while for crystalline substrates it maybe the binding of the substrate into the active site. In order to design new cellulases with higher activity on crystalline substrates, it is necessary to increase the rate of the rate-limiting step. Unfortunately, at this time the detailed mechanism by which a segment of a cellulose molecule is removed from its neighbors and bound into the active site is not known for any cellulase, so that it is not possible to rationally design mutant enzymes to increase the rate of this step.

Similar, but less extensive, studies have been carried out on the exocellulase, T. fusca Cel6B, and they showed that inserting a disulfide bond, which joined the two loops that cover the Cel6B active site to form a tunnel, reduced activity only 50%, so clearly loop opening, if it occurs, is not required for activity (97). These studies also found two mutations, which increased activity on crystalline cellulose, but only when the enzyme was assayed alone, not in synergistic mixtures. Several mutations were found, which reduced inhibition by cellobiose, with only a small loss of activity on crystalline cellulose. Finally, two mutations reduced activity on bacterial cellulose without reducing activity on CMC or SC, suggesting that these residues participated in a step that was only required for crystalline cellulose hydrolysis (97).

Extensive site-directed mutagenesis studies have been carried out on the processive endo — cellulase, T. fusca Cel9A (98, 99). These studies identified key residues that are important for activity, including a Glu residue that functions as the catalytic acid, and two conserved Asp residues, which are hydrogen bonded to the catalytic water molecule in the structure of the enzyme lacking bound sugars. Mutation of either residue drastically reduces activity, even though only one functions as the catalytic base (99). In the Cel9A structure, the catalytic base forms a hydrogen-bonded network with a conserved His residue and a conserved Tyr residue. Mutation of either residue drastically reduces activity, showing that this network is important for catalysis. The conserved Tyr residue is probably the hydrophobic platform residue in Cel9A, even though a different Tyr was proposed to have this function (96). When the Y429, proposed as the platform residue, was mutated, the mutant enzyme still retained about 10-30% of WT activity on various substrates, so it clearly does not have that role in catalysis (99). Several mutations increase activity on CMC, while reducing activity on bac­terial cellulose. In most cases, these mutations increase the size of the active site cleft, which may allow binding of modified sugars. A number of CD mutations reduced processivity and all these mutations were in the —2 to —4 subsites and would be expected to weaken binding to these sites, suggesting that the loss of processivity was due to decreased affinity of these subsites for the cellulose chain bound to the family 3 CBM, which is part of the active site of this enzyme. This result is consistent with the results of a docking study of Cel7A, which calculated the binding energy of each of the glucose-binding subsites in this enzyme for glucose and found that the energies increased in a way to draw the cellulose chain into the active site consistent with its processive activity (100).

There have been extensive mechanistic and structural studies of GH-5 endocellulases, which are retaining hydrolases (101,102). The GH-5 enzymes are extremely diverse in their sequences, in their detailed structures, and in their substrate specificities. A structural study of Acidothermus cellulolyticus endocellulase, Cel5A (E1), showed that the eight totally con­served residues in this family were all in the active site and were close to the bound substrate (101). There are other residues, which are not conserved in this family, that also interact with the substrate, providing further evidence of the diversity in this family. Structural studies of Bacillus agaradhaerens endocellulase Cel5A have determined structures for every step during bond cleavage and have identified the hydrogen-bonding network to the substrate at each step (102, 103). As was found for E1, there are both conserved and nonconserved residues present in these networks. Site-directed mutagenesis of Pyrococcus horikoshii endocellulase Cel5A showed that the conserved nucleophile was essential for activity, but that the con­served acid/base residue was not important for activity (104). This is a very unusual finding but, since there are several other conserved residues adjacent to the acid/base residue, this result is not due to an incorrect alignment, so that it provides even more evidence for the extensive diversity of GH-5 enzymes. All of the studies described here used low molecular weight substrates and studied events during catalysis, but not the placement of a cellulose molecule into the active site, which appears to be a key step for crystalline cellulose hydrolysis.

11.9 Outlook

In the last decade, there has been a large increase in our understanding of cellulase structure functional relationships but there is still a considerable amount that needs to be learned. This is particularly true for crystalline cellulose hydrolysis, where the exact role of CBMs in hydrolysis, how cellulose molecules are bound into cellulase active sites, and how cellulose structure influences this process, need to be determined. Only when these processes are understood, will it be possible to engineer cellulase active sites to achieve more efficient hydrolysis of specific biomass substrates. This ability should improve the economics of con — vertingbiomass to liquid fuels, possibly leading to sustainable production of non-greenhouse gas producing liquid fuels.

To further decrease the cost of the pretreatment part of any biomass conversion process it is essential that sugar yields are increased (sugar losses are minimized), solids concentrations are as high as possible, and the capital cost of the pretreatment reactors and associated equipment are kept as low as possible. It is also desirable that the pretreatment process increases the ease with which the cellulose in the pretreated biomass can be saccharified so that it can be accomplished with lower cellulase loadings or in a shorter digestion time.

Development and improvement of existing pretreatment processes is the subject of in­tensive research at several institutions in the USA and overseas. One approach that is re­ceiving more attention is the investigation of the effects of pretreatment at a more funda­mental level; increasingly research is looking at the effects of biomass pretreatment at the cellular, ultrastructural and even molecular level of the plant cell wall. The plant cell wall is highly complex at all length scales and especially chemically heterogeneous at the molec­ular level. For example, in nature dozens of glycosyl hydrolases are involved in plant cell wall deconstruction. These glycosyl hydrolases act on cell wall polysaccharides specifically and synergistically; however, the biochemistry of these unique catalytic events, caused by
individual enzymes acting on biomass as well as the consequences of thermal chemical pretreatment on these reactions, remains poorly understood. There is an increasing focus on characterization of biomass’s molecular structure and ultrastructure in order to gain sufficient understanding of the relationships between pretreatment chemistry and enzyme digestibility. There is also a desire to determine the effects that cellulases and other enzymes have on biomass ultrastructure.

Researchers at Michigan State University have been studying the changes that occur at the cellular level in corn stover upon AFEX pretreatment (78). After staining with a lignin — specific dye, safranin, they observed a change in the distribution of lignin in the cells by confocal laser microscopy. Scanning electron micrographs also appeared to indicate a change in distribution of lignin-like compounds and they postulated that the ammonia partially cleaved the lignin resulting in a decrease in its glass transition temperature so that the lignin could be relatively easily mobilized at temperatures close to 100oC. They also saw a 10-30% decrease in the oxygen:carbon ratio by X-ray photoelectron spectroscopy (XPS) at the surface of the cells indicating an increase in carbon-rich species such as lignin. A redistribution of lignin to the surface of plant cell walls caused by pretreatment could have a significant effect on the digestibility of the cellulose.

Similar effects have been observed in dilute acid pretreatment of corn stover. Dilute acid pretreatments are performed at much higher temperatures than AFEX pretreatments (140- 200oC versus about 90oC). The level of lignin redistribution is therefore much higher. In dilute acid pretreatment, droplets are seen to form that appear to be at least lignin derived. The droplets appear in the liquid hydrolyzate, adhere to the surface of the biomass and also collect in spaces between cells (Figure 14.1). Visually, these microscopic changes appear very significant; however, it is not known at this time how significant they are in altering the digestibility or accessibility of the cellulose in the pretreated materials.

The plant cell wall microfibril, the primary target substrate of bioconversion, is believed to consist of a cellulose-elementary-fibril core surrounded by a hemicellulose sheath forming a macromolecular composite (79). In acidic pretreatments hydrolysis and solubilization of the

1.5% H2SO4 ; 140°C

hemicellulose is thought to be the main mechanism by which accessibility of the cellulose to enzymes is increased. Changes in xylan distribution are being studied by probing pretreated materials with xylan-specific monoclonal antibodies. The distribution of xylan in cells and within cell walls can then imaged by binding the antibody with a secondary fluorescent dye (Alexa-488) for confocal laser scanning microscopy (CLSM) or a gold nanoparticle for transmission electron microscopy (TEM). In dilute acid pretreatment of corn stover rind differences in the pattern of fluorescence have been observed that are attributable to changes in the distribution of xylan in the cell wall (Figure 14.2). The cause of this is still being investigated and the effect of differences in xylan distribution on cellulose digestibility is still being interpreted.

It is anticipated that studies such as those described above will lead to significant ad­vances in our understanding of how pretreatment can overcome the natural recalcitrance of biomass and increase the digestibility of biomass in cost-effective pretreatments. An im­proved understanding should then allow for design of improved pretreatments that decrease the cost of converting biomass feedstocks into fermentable monomeric sugars.

Acknowledgment

This work was supported by the US DOE Office of the Biomass Program.

Lignin-carbohydrate complexes (LCC) can be selectively isolated from lignified secondary walls of woods, for example, bald cypress (Taxodium distichum), birch (Betula platyphylla) (89), spruce (90), beech (Fagus crenata) and pine (Pinus densiflora) (91), and grasses, for ex­ample, sugarcane (Saccharum officinarum) bagasse (92), rice (Oryza sativa) (89), and wheat (Triticum aestivum) straws (93) by fractionation, co-extraction, and co-chromatography of native or derivatized preparations (89, 94-96). Comparative chemical (97) and physico­chemical procedures, including 13C-NMR (97), 2D NMR (98), and FTIR (99) have been used in the characterization of the chemistry of the lignin-polysaccharide linkages.

The carbohydrate portion of LCC from the softwood bald cypress (Taxodium distichum) consists of galactoglucomannan, arabino(4-O-methylglucurono)xylan, and

Figure 4.4 Feruloyl units on neighboring GAX chains in cell walls may be cross-linked by radical coupling into terulate dehydrodimers (structures 1-5) or dehydrotrimers (structure 6). Dotted arrows indicate potential sites tor radical coupling with hydroxycinnamoyl alcohols or lignin oligomers in Iignitying cell walls, resulting in cross-linking of GAXs to lignin. "Ara" is an arabinoturanosyl residue on a GAX. (Reprinted with permission, from Grabber, J. H., Hatfield, R. D., Ralph, J., Zoh, J. & Amrhein, N. (1995) Ferulate cross-linking in cell walls isolated from maize cell suspensions. Phytochemistry, 40, 1077-1082, Fig. 1.)

arabinogalactan. In contrast, the LCCs of the hardwood birch (Betula platyphylla) and rice straw (Oryza sativa) are composed exclusively of 4-O-methylglucuronoxylans and arabino(4-O-methylglucurono)xylan, respectively (89). Laine and coworkers (67) found 4-linked xylan, 3,6-galactan, 4-linked galactan, and 3-linked glucan in the residual LCCs isolated from the kraft pulps of coniferous gymnosperms spruce (P. abies) and pine (Pinus sylvestris). Lawoko and coworkers (90) obtained four major LCC fractions from P abies and characterized a galactoglucomannan LCC containing ~8% of the wood lignin, a glucan LCC containing ~4% of the wood lignin, a xylan-lignin-glucomannan network LCC (xylan > glucomannan) containing ~40% of the wood lignin, and a glucomannan-lignin-xylan net­work LCC (glucomannan > xylan) containing ~48% of the wood lignin. It was concluded that carbohydrate-free lignin, i. e., lignin without covalent bonds to carbohydrates, probably does not occur in spruce wood.

Cellulose is reported to be covalently linked to lignin in both hardwoods and softwoods (100) as judged by the identification of the polysaccharides associated with lignin in the water insoluble fraction after carboxymethyl etherification. In the softwood spruce (Picea jezoensis), more than half the cellulose was linked to lignin, but in the hardwood beech (Fagus crenata), only one-sixth. The major non-cellulosic polysaccharides of spruce wood were also covalently linked to lignin, but in beech wood the extent of their linkage to lignin was low.

43.2.2.1 TYPES OF COVALENT POLYSACCHARIDE-LIGNIN CROSS-LINKS The major types of covalent linkages involved in lignin-carbohydrate associations are shown in Figure 4.5. One of these is indirect, through a bridging hydroxycinnamate molecule, the others are direct covalent linkages. These linkage types are described below.

Ester-ether cross-links. Hydroxycinnamate esters on heteroxylans and dimeric hydroxycin­namate esters bridging heteroxylan (GAX) chains, are etherified through hydroxyl(s) on lignin monomers (Figure 4.3a) and are quantitatively important in secondary walls of grasses (101), wheat (Triticum aestivum) (102-105) and Phalaris acquatica internodes (106), and maize (Zea mays) stover (83). The ether bond maybe at the (З-position of the lignin unit (102, 107-109) or possibly at the a-(benzylic)-position (110). A number of isomeric bridging homo — and hetero-dimers involved in the ester-ether bridging have been characterized (Figure 4.4) (83). The dimers involve mostly FA and dehydrodiferulic acid (DDFA) and to a lesser extent SA (32, 83). pCA does not appear in dimeric form (32, 83). In addition, there is evidence from NMR experiments with 13C-labelled Italian ryegrass (Lolium multiflorum) for other linkages between hydroxycinnamates and lignin monomers that are not easily broken by solvolytic analyses e. g., alkali at high temperatures (4 M NaOH, 170°C, 2 h) (103). Thus C-C, 8- O-4 styryl ether, and biphenyl ether cou­pling of ferulate and diferulates to lignins are not determined and only etherified ferulates are quantified (111-113). These may account for as little as 15% of total cross-linking (83).

Direct ester linkages. Carboxylic acid groups of uronic acids on matrix polysaccharides e. g., GAX and HGA may esterify alcoholic hydroxyls on lignin monomers (Figure 4.6). These alkali labile linkages have been reported in LCC from the woods of spruce (Picea sitchenis), pine (Pinus resinosa) and aspen (Populus tremuloides) and fibers from the

(^): p-coumaroyl feruloyl

Figure 4.5 Schematic diagram showing possible covalent crosslinks between polysaccharides and lignins in secondarily thickened cell walls of commelinid monocotyledons, including grasses (o)p-coumaryl, (•) feruloyl, (•-•) dehydrodiferuloyl residues. (a) direct ester linkage;(b) direct ether linkage;(c) ferulic acid esterified to polysaccharide;(d) p-coumaric acid esterified to lignin; (e) hydroxycinnamic acid etherified to lignin; (f) ferulic acid ester-ether bridge;(g) dehydrodiferulic acid ester-ether bridge. (Reprinted with permission, from liyama, K., Lam, T. B-T. & Stone, B. A. (1994b) Covalent cross-links in the cell wall. Plant Physiology, 104, 315-320, Fig. 2.)

eudicotyledons jute (Corchorus capsularis) and mesta (Hibiscus cannabinus), and the commelinid monocotyledon pineapple (Ananas comosus) (2). Borohydride reduction of aspen LCC leads to the loss of lignin and the formation of 4- O-methyl glucose from 4-O-methylglucuronic acid residues presumably from a glucuronoxylan (114). In one case, the LCC from beech (Fagus crenata) wood, the entity on the lignin involved in the ester linkage has been identified as a benzyl unit using 2,3-dichloro-5,6-dicyano — 1,4-benzoquinone (DDQ) and conjugate acid oxidation (91, 115). There is also direct evidence from 2D heteronuclear 3H-13C NMR for the presence of direct ester linkages in an acetylated LCC preparation from Eucalyptus globulus wood (98). In this LCC no benzyl ester was detected but the у -position of a lignin unit is esterified by a carboxyl group

of a uronic acid which was either GalA or GlcA or both. The extent of esterification was estimated as 9 per 100 lignin monomeric units. Pectin lyase treatment of birch (Betula maximowiczii) wood suggested that lignin maybe esterified to HGA (116).

Benzyl ether linkages. Linkages between hydroxyls on matrix polysaccharides and lignin monomers (Figure 4.7) have been described on the basis ofstudies with model compounds and observations on LCC fractions (2, 83, 85). These ether linkages have been character­ized by DDQ oxidation. In normal and compression woods of Japanese red pine (Pinus densiflora) (117), galactoglucomannan and (1^4)-p-galactan were bound to the lignin

preferentially through the C(O)6 position of the hexoses and the 4- O-methylglucurono — arabinoxylan was bound to the lignin through C(O)2 and C(O)3 positions of xylose units (Figure 4.7). Benzyl ether linkages in P. densiflora LCC have also been detected by ozonol — ysis (118). DDQ oxidation of beech (Fagus crenata) wood LCC showed that xylan was ether-linked through C(O)2 and C(O)3 (117). Phenolic benzyl ether linkages, in contrast to their non-phenolic counterparts, are susceptible to alkaline hydrolysis (1 M NaOH, 100°C). They are also sensitive to acid hydrolysis (0.3 M H2S04, 120°C). The lignins in the beechwood LCC are 100 times larger than those in the pinewood LCC but are less frequent (119).

Phenyl glycoside linkages. Glycosylation of lignin phenolic monomer hydroxyls or sidechain hydroxyls on lignins by monosaccharides, oligosaccharides, or polysaccharides (Figure 4.8) is a form of carbohydrate-lignin association that has experimental sup­port (95, 120, 121). The presence of a phenyl glycoside linkage was reported in the galactoglucomannan-rich LCC from spruce (P. abies) wood (90). There is direct evidence from 2D heteronuclear :H-13C NMR for the presence of phenyl glycosides in an acetylated LCC preparation from Eucalyptus globulus wood (98). The content of phenyl glycoside moieties in the acetylated LCC was ~0.08 per monomeric lignin unit. Phenyl glycosidic linkages are acid labile but stable to alkalis.

RG-II contains 3-O-acetylaceric acid and acetylated methyl fucose. No gene for RG-II acetyltransferase has been identified. However, O-acetyltransferase activity in microsomes from suspension-cultured potato cells (339) has been shown to transfer [14C]acetate from [14C]acetyl-CoA onto endogenous acceptors in the microsomes to yield a salt/ethanol pre- cipitable product from which approximately 8% of the radioactivity could be solubilized by treatment with endopolygalacturonase and pectin methylesterase. Thus, it is possible that the radiolabeled acetate solubilized by the glycanase treatments represented acetylated RG-II and that the activity identified was RG-II acetyltransferase.

5.4.9.3 Other RG-II transferases

The glycosyltransferases that insert fucose, KDO, DHA, and aceric acid into RG-II have not been identified. A 10-member Arabidopsis gene family (118) that has 35-73.8% amino acid sequence identify to an Arabidopsis a 1,2-fucosyltranferase that fucosylates a side branch in the hemicellulose xyloglucan has been described (116). Whether one or more of these genes encodes fucosyltransferase(s) involved in RG-II synthesis remains to be investigated.

Feingold (403) summarized the NDP-sugars identified in plants. Those that were not described above are ADP-L-Ara, GDP-L-Ara; ADP-ribose; GDP-Xyl, ADP-Gal, GDP-d — Gal; ADP-D-Man; UDP-Fructose, ADP-D-Fructose; UDP-D-digitoxose (2,6-dideoxy-D — ribohexose); TDP-GalA; UDP-2-deoxy-2-acetamido-D-Glc, UDP-2-deoxy-2-acetamido-D — Gal; UDP-cellobiose. In addition to these NDP-sugars, pectin consists of other sugar residues: aceric acid and DHA (deoxylyxoheptulopyranosilaric acid). The formation of these NDP — sugars is not well studied.

It is not clear if modifications of sugar residues on pectin (such as methylation or acety­lation) occur after transferring the sugar from the respective NDP-sugar. In chloroplasts for example, sulfolipid biosynthesis requires the activated sugar UDP-Glc-6-sulfonate, UDP 5′- diphospho-sulfoquinovose. In this case, the sugar-linked to NDP is modified with a sulfate group prior to the transfer of the sulfoquinovose. Whether such NDP-sugar modifications occur with NDP-sugars required for wall synthesis is unknown. While more is known about synthesis of activated sugars less is known about “catabolism” of NDP-sugars. Recently, work on ascorbic acid metabolism in plants revealed two mutants, vtc2 and vtc4, involved in the degradation of GDP-L-Gal. First, a GDP-L-Gal phosphorylase (vtc2, At4g26850) con­verts GDP-L-Gal and Pi into L-Gal-l-P and GDP (502) and subsequently a dephosphorylase

activity on L-Gal-1-P yields L-Gal (458) (vtc4, At3g02870). Interestingly, the phosphorylase converts GDP-d-G1c to d-G1c-1-P and GDP as well.

Alkaline nitrobenzene oxidation (NBO), thioacidolysis degradation, and permanganate ox­idation procedures are also currently routinely applied to the analysis of plant materials, due to their abilities to cleave various linkages in lignins, as well as that of related phenolics. For example, NBO oxidation of lignin “model” compounds 42-44 results in formation of the various lignin-derived products (45-50) (Figure 7.10B) (189, 190), and is thus employed to probe both lignin compositions and contents. With lignins, this results in homolytic oxidative fission of their 7-8 linkages, and ultimately cleavage of the 8- O-4′ bonds (Figure 7.10C) (189, 190). Moreover, if there are significant amounts of other non-lignin cell wall bound p-hydroxycinnamic acids/aldehydes present, this method gives overestimations of lignin contents/compositions (77). Thioacidolysis is also now perhaps even more routinely employed, again to probe both lignin contents and composition; with this method, the over­all main monomeric, monolignol-derived, cleavable degradation products, obtained from lignins proper are compounds 54-56 (Figure 7.10D). These conversions have also been well studied with model compounds 51-53 that contain 8- O-4′ interunit linkages. Other work is currently under way to fully identify dimeric and other oligomeric entities released by this method. Both NBO and thioacidolysis procedures are generally also employed to estimate H:G:S ratios, as well as amounts of releasable products relative to lignin contents and/or cell wall residues (CWR). Overall, the recoveries of monomeric and dimeric components, rela­tive to original lignin contents, are very low. For example, alkaline NBO and thioacidolysis generally only account for circa 15-25% and 20-40% by weight, respectively, of the lignin presumed present (71, 132) (Figure 7.10E). Thus the bulk of the lignin polymer(s) remains unaccounted for using either procedure. Lastly, permanganate oxidation has also been quite extensively employed with various carboxylic acid methyl esters ultimately being formed (191). The yields of these products are also low (192) (~ 10-30% or so, Figure 7.10E), with the bulk of the lignin polymer again being unaccounted for.

In spite of these serious shortcomings, as discussed below, little has been done until now to attempt to establish overall trends in interunit linkage placement (as a function of lignin deposition); in accurately determining response factors, etc., for calibration and quantifica­tion purposes; in developing new/improved methodologies to study both lignins and their degradation products. For these reasons, and before discussing the results so obtained, it was thus first necessary to consider the advantages and limitations of the procedures currently employed.

It is difficult to accept that the behavior of atomistic systems, which behave according to quantum rather than classical laws, could be accurately described by the application of classical Newtonian mechanics. This approach can be justified, however, by considering the de Broglie expression for the thermal wavelength-Л,

where T is the temperature and M is the atomic mass. The approximation of classical behavior holds if Л = a, where a is the mean nearest neighbor separation. This holds for “heavy” liquid systems at all but the lowest temperatures at which quantum effects become important.

To describe the molecular dynamics of a system classically a function representing the potential energy of the system, together with the related parameters is required. Typically, the energy is calculated from the sum of bonded and non-bonded interactions,

E total Ebond + E angle + E dihedral + E non-bond

The exact form of the terms in the above potential function and the associated parameters varies across different molecular mechanical force fields. Some force fields also include a cross term representing coupling between the first three terms in Equation (8.2). Examples of commonly used force fields include the Allinger MM2 and MM3 series (3,4), CHARMM (5), AMBER (6), and GROMOS (7). Each force field has a slightly different ethos and is typically suited to the study of one class of molecules. Of those mentioned above CHARMM, AMBER, and GROMOS would be considered protein force fields suited to the simulation of proteins, nucleic acids, and with suitable parameterization, carbohydrates. The underlying equations for these three force fields are similar, although there are subtle differences. For the purposes of discussion here we will restrict ourselves to the AMBER force field equation, which is

V(rn) =J2 Kr (Г — req )2 +J2 Kв (Є — ®eq )2

bonds angles

+ ^ V [1+cos(^- 7)]+^ Rj

dihedrals i < j |_ i

where the potential energy V is a function of the positions r of n atoms. Kr, req, Kв, Qeq, Vn, n, ф, 7, Aij, Bij, £r, qi, and qj are all empirically defined parameters. The first three terms in the above equation correspond to the bond, angle, and dihedral terms, respectively, while the last term describes the non-bonded van der Waals and electrostatic interactions.

In order to investigate the formation and destruction of xylo-oligomers further, experi­ments were conducted using xylan. Both Birchwood xylan and Beechwood xylan were used for these studies. These materials were supplied as fine powders from Sigma and used as received. Because this biopolymer is not soluble, stock solutions of the samples in acid could not be prepared. Instead, approximately 20 mg of sample was added to 2 mL of a 1.2% w/w aqueous solution of sulfuric acid and the measured products were normalized to the actual amount of xylan used. In the absence of heating, HPLC measurements of the liquid portion of these slurries detected no xylose or furfural, while these species were observed when the samples were heated in the microwave reactor. Figure 9.10 shows plots of xylose and furfural

concentrations (normalized to 10 mg/mL of added xylan) as a function of residence time of the xylan slurry in the reactor. Figure 9.10a shows the plot of xylose formation at 115° C, where it is apparent that increasing the residence time results in increased formation of xylose. This result suggests that, at this temperature, xylan has not been fully converted to xylose. At 125, 135, and 145°C, the amount of xylose formed remains constant at all measured residence times, which suggests that the xylose residues in the xylan have been completely released. We note that approximately 73% of the mass in the xylan has been converted to xylose in these experiments. This value is close to the known concentration of xylose in this xylan [Johnson, D. K. (2006) unpublished work] (69%). At higher temperatures, the concentration of xylose decreases with increasing residence time. This observation suggests that at these tempera­tures, the xylose is being dehydrated to form furfural, which is also confirmed by the furfural plot in Figure 9.10b. Note that at temperatures below 145°C, independent measurements with pure xylose in acid show little formation of furfural, whereas at these temperatures large amounts of xylose are formed from xylan. Xylobiose was only observed at 115° C and only at a normalized level of 0.002 mg mL-1. These observations again show that hydrolysis reactions of xylo-oligomers and xylan are much faster than dehydration reactions of xylose.

Edward A. Bayer, Bernard Henrissat, and Raphael Lamed

13.1 Introduction

Plant cell wall polysaccharides provide an exceptional source of carbon and energy that can be potentially utilized as a low-cost renewable source of mixed sugars for fermentation to biofuels like ethanol (1-3). Perhaps the major bottleneck for conversion of biomass to ethanol is the combined high cost and low efficiency of the enzymes — the cellulases and other glycoside hydrolases — that are capable of degrading cellulose and myriads of complex plant cell wall polysaccharides to simple sugars. Efficient hydrolysis is impeded by limited accessibility of the enzymes and the recalcitrance of cellulose, owing to the extremely stable “paracrystalline” arrangement of the cellulose chains in the microfibrils (4).

The rate-limiting step in the hydrolysis of cellulose is not the catalytic cleavage of the (3 -1,4- glucosidic bond, but the disruption of a single chain of the substrate from its native crystalline matrix, thereby rendering it accessible to the active site of the enzyme. Thus, the processes and interactions that are most significant are those that facilitate the disruption of strong interchain hydrogen-bonding network that characterizes the microcrystalline arrangement of the insoluble cellulose structure (5). Moreover, single cellulolytic enzymes alone are generally incapable of efficient cellulose hydrolysis (6). The mode of action of the various cellulases is different, and they act synergistically, such that the combined extent of hydrolysis is much more than the sum of the individual parts (6-8). A simplistic example of this phenomenon is apparent when comparing the action of endoglucanases versus that of the exoglucanases: Endoglucanases cleave at random points along the cellulose chain, whereas exoglucanases cleave successive units (e. g., cellobiose or cellotetraose) from the chain ends in a “processive” (sequential) manner (9). Processive endoglucanases have also been described (10,11), whereby an initial endo cleavage is followed by more systematic, successive cleavage along the chain, owing to either an appropriate change in the active-site architecture or the effects of an associated ancillary module(s) that might confer processivity on an otherwise endo-acting enzyme. In any case, the secret to potent enzymatic degradation of a recalcitrant cellulosic substrate is how these different types of enzymes work together, which is difficult to approach experimentally but is a key conceptual component for improving the overall degradation of cellulosic biomass.

Biomass Recalcitrance: Deconstructing the Plant Cell Wall for Bioenergy. Edited by Michael. E. Himmel © 2008 Blackwell Publishing Ltd. ISBN: 978-1-405-16360-6

In nature, degradation of cellulosic substrates is accomplished by various microorgan­isms, thus contributing a central component to the carbon cycle. In some cases, free-living microorganisms exploit such polysaccharides from decaying plant matter, found, for ex­ample, in compost piles and sewage sludge; in others, the microbes assist higher animals (e. g., ruminants, termites, etc.) in converting the polysaccharides to digestible components. Microbial degradation of cellulosic materials is one of the most important processes in na­ture, and different bacteria and fungi approach the task in different ways (12-16). Whereas aerobic microbes commonly produce copious quantities of the relevant enzymes (e. g., cel — lulases and hemicellulases), the biosynthetic apparatuses of anaerobes are much more frugal in their output of such enzymes. In this context, it is believed that the anaerobic environ­ment presents a greater selective pressure for the evolution of highly efficient machinery for the extracellular degradation of polymeric substrates, such as the recalcitrant crystalline components of the plant cell wall. The energy yield of aerobes per hydrolyzed sugar unit is much greater than that of anaerobes, which have evolved extensive energy-conserving mechanisms for physiological adaptation to environmental stresses such as novel enzyme activities (17). Consequently, the anaerobes tend to adopt alternative strategies for degrad­ing plant matter, and of these the organization of the enzymes into cellulosomes appears to be the most remarkable.

Much of the pioneering work on cellulases was first performed on “free” enzyme systems, commonly produced by fungi and some, aerobic bacterial species. The free cellulases are relatively simple macromolecules consisting generally of a catalytic module and cellulose­binding module (CBM) on a single polypeptide chain. The CBM targets the catalytic module to the cellulose surface, whereupon it begins to disrupt and degrade the cellulose chains. The different types of cellulase enzymes are thus distributed and interact freely in a synergistic manner (8).

An emerging alternative approach is to use oligonucleotide or cDNA microarrays designed to detect specific rDNA or gene sequences known to be present in important lineages of microorganisms that are thought to be in a specific microbial community (104-106). Loy and coworkers successfully developed and used a microarray consisting of 132 16S rRNA gene-targeted oligonucleotide probes covering all recognized lineages of sulfate reducing prokaryotes (SRP) for high-resolution screening of clinical and environmental samples (107, 108). The microarray, named SRP — PhyloChip, has great potential for rapid screening of SRP diversity in complex samples and microarray SRP diversity fingerprints allow identification of relevant probes for further characterization of a sample by PCR or quantitative hybridiza­tion. This is a valuable option if large numbers of samples are to be analyzed for temporal or spatial variations in SRP diversity. While this approach can provide a comprehensive survey for the presence or absence of a particular sequence, the technique has a closed architecture and cannot identify novel sequences nor can microarrays easily distinguish between two or more closely related sequences in mixed samples.