In lignified secondary walls of grasses and in hardwoods and softwoods, hydrophobic lignins overlie and encrust the cellulose microfibrils and matrix polysaccharides and proteins, and are variously covalently complexed, to a greater or lesser extent, with these wall polymers. Lignins can be regarded as hydrophobic fillers that replace the water in the developing wall (133). Because water is displaced during lignification, increased hydrogen bonding is favored both between non-cellulosic polysaccharides and between these and cellulose microfibrils. Moreover, the chemical bonds between lignin and the non-cellulosic matrix components cross-link the matrix phase and the cellulose microfibrils ensuring coherence (134). North — cote (133) likened lignified walls to a synthetic glass-fiber composite, which is rigid because of the lignified matrix and has great strength because of the cellulose microfibrils. This composite has a porosity [20-50 nm in wheat straw and pericarp (135,136)], that limits the approach of polysaccharide hydrolases to their substrates contributing to the recalcitrance of these wall types to enzymatic digestion.

Covalent cross-linking of wall polymers, in particular the lignin-polysaccharide associ­ations, prevents extraction of matrix polysaccharides from cell walls by neutral aqueous solvents and hydrogen bond breaking reagents. Alkaline reagents, which cleave most of the lignin-polysaccharide linkages and also solubilize some lignins (93,137), are required. Thus, extraction of heteroxylans from lignified secondary walls of grasses can be achieved by alkali at room temperature (83), alkaline H2O2 (138) or oxidative degradation with chlorite(139), which however leads to polysaccharide breakdown (140). Ammonia fiber/freeze explosion effective in pretreatment of e. g., maize stover, prior to enzymatic digestion, cleaves alkali la­bile LCC complexes (99, 141). (See also Chapter 14, Pretreatments for Enhanced Digestibility of Feedstocks)

RG-I contains L-arabinose in multiple linkages (see Table 5.2). Most of the arabinose is in the furanose ring form, although a terminal arabinose exists in the pyranose form in some RG-I side chains (268). AraT activity was originally identified in microsomes from mung bean (Phaseolus aureus) (388) and bean (Phaseolus vulgaris) (389) although definitive evidence that those AraT activities were involved in pectin synthesis was not demonstrated [see Ref. (205) for review]. The bean AraT activity was primarily associated with enriched Golgi, and to a lesser extent enriched endoplasmic reticulum (390).

The study of AraTs specifically involved in pectin synthesis has been problematic for several reasons. Multiple wall polysaccharides and proteoglycans contain arabinose, in­cluding pectin, hemicelluloses (e. g., glucuronoarabinoxylan), and arabinogalactan proteins. Thus, experiments aimed at studying pectin biosynthetic AraTs by incubating microsomal membranes with radiolabeled UDP-Ara have not been very successful. The nucleotide — sugar donor, UDP-Ara, was not available and had to be synthesized (391), although more recently the UDP-p-L-arabinopyranose form has become available through CarbSource (http://www. ccrc. uga. edu/~carbosource/CSS_home. html). However, while most Ara in pectin is in the furanose form, the nucleotide-sugar synthesized by the 4-epimerization of UDP-a-D-Xyl is UDP-p-L-arabinopyranose. Thus, this has been the nucleotide-sugar form most readily available for experimental use. However, there is uncertainty as to the nature of the nucleotide-sugar substrate used for pectin synthesis. Is it UDP-p-L- arabinopyranose (UDP-Arap) or UDP-p-L-arabinofuranose (UDP-Ara/)? IfitisUDP-p-L — arabinofuranose, how is this synthesized by the plant and, experimentally, what is the most facile way to produce it? Is it synthesized by enzyme-catalyzed ring contraction of UDP-l- arabinopyranose by a mutase (392)? Indeed, recently, Ishii and colleagues (393) identified a UDP-arabinopyranose mutase (UAM) from rice that catalyzes the reversible formation of UDP-Ara/ from UDP-Arap. Interestingly, UAMs are the same proteins that previously were identified as reversibly glycosylated polypeptides (RGPs), proteins that are reversibly glycosylated in the presence of UDP-Glc and several other nucleotide-sugars (394-396). The significance of the reverse glycosylation detected in vitro, in regards to the role(s) of UAM in wall synthesis, remains to be determined.

Recent efforts to investigate the AraT activity in mung bean confronted some of the above-mentioned problems. Incubation of mung bean microsomal membranes with UDP — p-L-[14C]arabinopyranose (UDP-[14C]Ara) resulted in the incorporation ofboth [14C]Ara and [14C]Xyl into elongated endogenous acceptors because of the epimerization of some of the UDP-[14C]Ara into UDP-[ 14C]Xylby a UDP-Xyl-4-epimerase present in the microsomal fraction (397). Furthermore, digestion of the radiolabeled product synthesized in the micro — somes with endo-arabinase yielded very little radiolabeled Ara or arabinose-containing small oligosaccharides, suggesting that the conditions used were not conducive for the synthesis of arabinans. Conversely, digestion of the product with arabinofuranosidase did release some [14C]Ara, indicating that an enzyme activity that could add at least a single Ara residue was present in the microsomes. A breakthrough in identifying pectin biosynthetic AraTs came when detergent-solubilized microsomal membranes and defined arabinooligosaccha — rides were used as acceptors for study of pectin biosynthetic AraTs (397). The incubation of detergent-solubilized microsomal membranes with (1—5)-a-L-arabinooligosaccharides of DP 5-8 and with UDP-p-L-[14C]arabinopyranose led to the identification of an AraT activity that could transfer a single arabinopyranose residue onto the non-reducing end of a1,5-arabinooligosaccharide acceptors. The enzyme had a pH optimum of 6.5 and was shown to reside predominantly in the Golgi by subcellular fractionation of organelles (397). The anomeric configuration and linkage of the arabinopyranose residue transferred by the mung bean AraT was subsequently shown to be (3-(1—>3) through the use of fluorescently labeled a-L-arabinooligosaccharide acceptors (271). Thus, mung bean contains an a1,5- arabinan:p-(1—3)arabinopyranose AraT (271).

A second mungbean arabinopyranosetransferase activity was identified (273) that could transfer an individual arabinopyranosyl residue from UDP-p-L-[14C]arabinopyranose onto the non-reducing end of fluorescently labeled 1,4-linked p-D-galactooligosaccharides of DP 3-7, with significantly better activity manifested with 1,4-linked p-D — galactooligosaccharides of DP 5 or greater. The p-1,4-galactan:AraT activity transferred the Ara residue in an a configuration on the O-4 position of the galactooligosaccharides, identi­fying the AraT as a p-1,4-galactan:a1,4AraT. The enzyme had a pH optimum of 6.0-6.5 and apparent Km(s) for UDP-p-L-[14C]arabinopyranose and fluorescently labeled galactohep — tasaccharide of 330 pM and 45 pM, respectively. Interestingly, the enzyme would not use fluorescently labeled galactooligosaccharides of DP 6-10 that had a single a-L-Arap residue at the non-reducing end as acceptors for the previously described p -1,4-galactan: p -1,4-GalT (386), indicating that the enzyme cannot use mono-arabinosylated galactooligosaccharides as acceptors. The authors propose that the presence of the a-L-arabinopyranosyl residue on the p-1,4-galactan oligosaccharides prevents further galactosylation of the galactooligosac­charides (273).

Recently, a gene encoding a putative arabinan:a-1,5-arabinosyltransferase (ARAD1; At2g35100) has been identified in Arabidopsis (270) through analysis of the CAZy GT47 family glycosyltransferase gene At2g35100 (138) (http://afmb. cnrs-mrs. fr/CAZY/) and phe­notypic biochemical and immunochemical analyses of the corresponding Arabidopsis T — DNA insert mutant ARABINAN DEFICIENT 1. ARAD1 encodes a protein with a predicted molecular mass of 52.8 kDa and a single transmembrane helix region near the N-terminus.

Although homozygous knockout mutants of ARAD1 show no visible growth differences from wild type, isolated walls from mutant leaves and stems had 25 and 54%, respectively, reduced levels of Ara compared to wild type walls (270). Transformation of the mutant plant with the ARAD1 gene complemented the mutant phenotype (i. e. restored the amount of Ara in the wall to wild type levels), thus providing evidence that ARAD1 affects wall arabinose levels. Immunocytochemical analysis of leaf, inflorescence stem, and stem revealed a reduc­tion in immunolabeling with the anti-a-1,5-arabinan antibody LM6. A lack of difference between the labeling of protein extracts from wild type and mutant with LM6 suggested that the mutant was affected in the synthesis of a-1,5-arabinans, but not in glycoprotein synthesis (270). This observation was confirmed by comparison of RG-I isolated from wild type and ARAD1 walls. Mutant RG-I had a 68% reduction in Ara content, which linkage data showed was predominantly due to a reduction in 5-linked Ara/ and also in 2,5 f — linked Ara and 2,3,5-linked Araf. These results strongly suggest that ARAD1 is a putative RG-I arabinan:a-1,5-arabinosyltransferase. Confirmation of this activity will require expression of enzymatically active enzyme expressing a-1,5-arabinosyltransferase activity.

Recently, a novel approach to identify genes involved in cell wall synthesis has been taken

(398) and offers promise in leading to the positional cloningfor a gene that affects the number of arabinan side chains in RG-I. The method takes advantage of the natural variation that occurs in cell wall synthesis in natural plant populations and of the availability of Arabidopsis recombinant inbred line (RIL) populations which facilitate the identification and cloning of quantitative trait loci (QTLs). Through the use of multiple techniques to analyze cell walls of an RIL population from a cross between Arabidopsis Bay-0 and Shahdara, including global monosaccharide composition and Fourier-transform infrared (FTIR) microspectroscopy, a major QTL was identified that accounted for 51% of the heritable variation observed for the arabinose-rhamnose ratio in the cell walls, a difference that appeared to be due to variation in the amount of RG-linked arabinan. Whether this strategy will lead to the identification of RG-I biosynthetic AraTs remains to be shown.

Two Arabidopsis putative arabinosyltransferases, designated reduced residual arabinose — 1 and -2 (RRA1; At1g75120 and RRA2; At1g75110) were recently identified by Egelund et al.

(399) based on a 20% reduction in the arabinose content in pectin — and xyloglucan-depleted cell wall fractions from meristematic tissue of rra1 and rra2 mutant plants. However, whether these genes, which are classified in CAZy family GT77, encode functional arabinosyltrans­ferases, and if so, whether they function in pectin, arabinoxylan, wall structural protein, or other syntheses, remains to be determined.

DNA is a water-soluble molecule, aided by tens of proteins to helping in the packaging and re-packing during RNA and DNA synthesis. Proteins are also soluble and if they are too hydrophobic, either a portion will be embedded in membranes or hydrophobic domains will assemble together with other hydrophobic proteins to maintain their solubility. On the other hand, plant polysaccharides provide us with a challenge in terms of understanding how such massive amounts of insoluble or gel-like structures are made inside the Golgi apparatus without hindering other cellular processes. Bacteria, for example, form insoluble lipo-glycans in stages: UDP-sugars made in the cytosol contribute to synthesis of short side chain glycans attached to the inner membrane. These short glycans are made facing the cytosol. The side chains are then flipped and transferred through the outer membrane to the outside of the cell, where specific enzymes cut and assemble the side chains to form complete glycan structures. This process, by analogy, is very similar to the synthesis of core N-linked glycoproteins. But how are pectins and hemicelluloses made in plants? Are Golgi — synthesized polysaccharides made completely in the Golgi, or are they made in smaller fragments that are assembled together at the wall, as is the case in lipo-glycan synthesis? If the entire pectic polymers are made in the Golgi, then are they sequestered or packed temporarily with or by proteins to help with the challenge posed by their physical properties (i. e., solubility, size, etc.)? Further questions will arise. Is HG made inside the lumen (thus making a jelly-like lumen) or perhaps made in specific Golgi stacks, designated only for the wall (not glycoprotein biosynthesis)? The latter scenario could be attractive in light of the fact that unlike humans and fungi, one plant cell consists of hundreds of Golgi stacks. It is possible that an entire “designated Golgi” is moving with its packed glycan(s) to the wall to “unload” the insoluble matrix.

While many of the putative GTs are Type II membrane proteins (i. e., predicted to have their catalytic domain facing the lumen), some GTs, such as mannan synthase, have multi domains that span the membranes. Where is the catalytic domain of such Golgi-enzymes facing? Furthermore, if a glycan is fully made in the Golgi, is it made in one subcompartment of the Golgi, for example, cis-Golgi, or is it initiated in the cis — and further modified in the medial — Golgi (like glycoproteins)? Certainly, understanding where each wall biosynthetic enzyme functions and where its catalytic domain faces is essential to address the above questions.

In addition to location and topology, it is puzzling how polysaccharides are made in the small Golgi cisternae (estimated to be smaller than 20 x 200 nm). The same cisternae are temporarily packed not only with wall-glycans but also with numerous Golgi-resident proteins and large numbers of secretory proteins. If the average size of the catalytic domain of GTs and membrane-bound NDP-sugar biosynthetic enzymes is 30-40 kDa, they can “touch” each other if inserted opposite to each other. Given the low quantity of these metabolic enzymes and the potential solubility problems of some glycans, one can wonder if synthesis of a specific Golgi-glycan is done in a complex of enzymes, as is the case in the synthesis of cellulose.

7.6.1.1 PAL

This step catalyzes the entry point into phenylpropanoid metabolism, and thus can lead to a diverse range ofphenylpropanoid-derived products (i. e., lignins, lignans, hydroxycinnamic acids, suberins, flavonoids, proanthocyanidins, etc.), depending upon the species, tissue, and/or cell type in question. Reduction of overall PAL enzymatic activity (through inhibition, mutation, downregulation, etc.) would thus be expected to have negative consequences on either all aspects of phenylpropanoid metabolism or on specific elements, depending upon which PAL gene(s) is (are) affected. This was facilely demonstrated first using the PAL inhibitor, L-AOPP (57, Figure 7.11), as early as in 1977 and in subsequent studies extending through 1985 (200-205). This resulted, depending upon the plant species investigated, in reductions in formation/accumulation of anthocyanins, isoflavones, hydroxycinnamic acids, and lignins, respectively. As an example, Figure 7.12A depicts the effects of AOPP (57) inhibition on lignification in mungbeans (2004); the cross-section of the AOPP (57)-treated plant line clearly has collapsed xylem due to presumed reductions in lignin contents and thus a weakened vasculature — as would be anticipated from first principles.

Furthermore, although the reliability of many of the analytical techniques used in later studies by other researchers was clearly questionable, the subsequent standard biotechnolog­ical manipulation of PAL activity in tobacco (through antisense and sense co-suppression, etc.) apparently resulted in poorly formed (weakened) xylem tissue (206,207); additionally, the plant lines were also severely stunted, had curled leaves, localized lesions, less pollen, reduced viability and deformed flowers. These are presumably not desirable traits, and in­deed appear to fall under the rubric of pleiotropic “unintended” effects: these, in turn, reflect our severely limited current understanding of overall plant metabolism, growth, and development. Moreover, while the lignin levels were reportedly lower in these studies as in­dicated above, the analytical procedures used were very questionable (i. e., thioglycolic acid lignin determination and Klason lignin analyses of “neutral detergent fibers”) as discussed in Anterola and Lewis (77). Another very preliminary study also reported that generation of double PAL mutants (e. g., pall pal2) in Arabidopsis resulted in lignin levels reduced to circa 30-35% of wild-type levels (208). This study, however, lacked any systematic analysis of the effects on lignification over different phases of growth/development until maturation. It did though provisionally indicate that lignin levels were reduced.

Perhaps, most importantly, none of the above studies explored any (quantitative) mea­surement of effects on structural integrity of the vasculature; nor was there either any new or additional insight gained on lignin macromolecular configuration/assembly, and/or effects of modulation of same. As predicted from first principles by ourselves there was clearly no “instant response,” as proposed by Ralph (197) and Ralph et al. (175), to severe enzymatic shifts that would suggest that lignin’s structure was not important; nor were “perfectly vi­able” plants produced, based on the defects that had been noted as early as 1983. That is, the genetically modified tobacco PAL plant lines produced were only able to survive with a number of serious defects; additionally, these transformants were not stable, and reverted to wild type in subsequent generations (209). Accordingly, such preliminary (phenomeno­logical) studies on PAL modulation now need to be expanded in depth, in order to gain a more definitive understanding of the actual effects on, for example, lignin macromolecular configuration/assembly, vascular integrity and overall physiology.

The bond-stretching and angle-bending terms, discussed above, are often referred to as “hard” degrees of freedom since large energies are required to cause significant deviations from the equilibrium geometries. Most of the complex variations in structure and relative energies observed in biological systems are due to the “softer” torsional and non-bonded contributions.

The barriers to rotation about a bond can be modeled in one of two ways. In very early force fields it was believed that rotational barriers could be omitted. The gauche-trans energy differences would be reproduced by the non-bonded interactions. However, it was quickly realized that for organic molecules neglecting dihedrals made successful parameterization of force fields, to reproduce experimental observables, an almost impossible task and so dihedral terms were explicitly included. The AMBER force field, in common with a number of other biologically oriented force fields, uses a Fourier series expansion for the torsional potential

N у

У(Ф) = £ у [1 + cos(^ — 7)] (8.4)

n=0 2

where Vn is the relative barrier height to rotation, n is the multiplicity (number of minima in a 360° rotation), ф is the dihedral angle, and у is the phase factor which determines the location of the minima. Vn is often termed the relative barrier height since other terms in the force field equation contribute to the barrier height as the bond is rotated, especially the 1-4 non-bonded interactions discussed below. The advantage of using a Fourier series expansion for the dihedral terms centers on the fact that terms of differing multiplicity can be combined to describe complex torsional profiles (Figure 8.2).

Improper torsion angles, also known as out-of-plane bending, are defined for four atoms that are not bonded in a serial manner. They are used to maintain planarity where necessary. The AMBER force field accounts for improper torsions in the same way as regular torsion angles but using a twofold multiplicity.

The term hemicellulase is often used to refer to a mix of enzyme activities that act upon the non-cellulose, non-pectin polysaccharides in biomass. The diversity and complexity of plant cell wall structural hemicelluloses preclude a more precise definition. In general terms, the enzymes involved in hemicellulose degradation can be divided into two major categories: depolymerizing and debranching. Tables 10.3-10.6 have been compiled from several online enzyme databases, the Carbohydrate Active eZyme database (CAZy, http://www. cazy. org), the Expert Protein Analysis System (ExPASy, http://www. expasy. org), and BRENDA, the Comprehensive Enzyme Information System (http://www. brenda. uni-koeln. de) (14-16). These tables further categorize the enzymes involved in cellulose and hemicellulose degra­dation and demonstrate the complexity of the problem. It is important to understand that the classification of biomass-degrading enzymes spans several nomenclatures, with some, but nowhere near complete, cross-referencing. One common system is the IUMB system, with the enzymes being classified by four numbers, designating the enzyme type and activity, i. e., З.2.1.4. The second system, which has gained much popularity over the last decade or so, is the so-called CAZy database, which, in contrast to the IUMB system, classifies glyco — syl hydrolases (and other carbohydrate-active enzymes) into families based on structural and evolutionary similarities, GH7 for example. As there is no comprehensive and simple cross-reference available between these two systems, enzyme designations here are given in either the original connotation from the referenced material, or in the current most common usage.

Depolymerizing enzymes act on the backbone sugar chain and are usually classified as either endo-acting, which cut the chain in the midst of a long polymer, or exo-acting, which work from the end of the chain. Several, however, have been reported to be associated with both types of activity. In addition, there are a series of enzymes that act on the oligomers

Table 10.3 Cell wall polysaccharide depolymerizing |3-glucanases IUPAC Name Families Dominant substrates Dominant products Dominant linkages 3.2.1.4 Cellulase 5, 6, 7, 8, 9, 10, 12, 26, 44, 45, 48, 51,61, 74 Cellulose, |3-glucans, mixed linkage |3-glucans Glucan oligomers (1 —>4)-P-d glucoside 3.2.1.6 Endo-1,3(4)-|3-glucanase 16 Laminarin, lichenin, mixed |3-glucans Glucan oligomers (1 —>3,4)-P-d glucoside 3.2.1.21 |3-glucosidase 1, 3, 9 Cellobiose, other oligoglycans Glucose (1 —>4)-P-d glucoside 3.2.1.39 Glucan endo-1,3- |3-D-glucosidase 15, 17, 55, 64, 81 Laminarin Glucan oligomers (1 —>3)-P-d glucoside 3.2.1.58 Glucan 1,3-fi-glucosidase 3, 5, 17, 55 Laminarin Glucose Non-reducing end (1 v3)-|3-D-glucoside 3.2.1.73 Licheninase 5, 8, 11, 12, 16, 17 Lichenin, mixed P-(1 —>3,4) glucans Mixed |3-(1 -^3,4)-glucans (1 —>4)-P-d glucoside 3.2.1.74 Glucan 1,4-|3-glucosidase 3 Cello-oligomers Glucose exo (1 —>4)-P-d glucoside 3.2.1.91 Cellulose 1,4-|3-cellobiosidase 5, 6, 7, 9 Cellulose Cellobiose Non-reducing end |3-(1 v4) glucoside 3.2.1.120 Oligoxyloglucan |3-glycosidase Xyloglucan oligomers Isoprimeverose (xyl-p-a- (1 ^ 6)-|3-D-glc-p) (1 —>4)-P-d glucoside 3.2.1.150 Oligoxyloglucan reducing-end-specific cellobiohydrolase 74 Xyloglucan oligomers Cellobiose — non-reducing glc may be substituted xyl-p-a-(1 ^6) Reducing end (1 —>4)-P-d glucoside 3.2.1.151 Xyloglucan-specific endo-|3-1,4-glucanase 5, 12, 16, 26, 44, 74 Xyloglucan Xyloglucan oligomers (1 -^4)-|3-D-glucoside 3.2.1.155 Xyloglucan-specific exo-|3-1,4-glucanase Xyloglucan Xyloglucan oligomers (1 -^4)-|3-D-glucoside, mixed endo/exo modes

generated by the combination of endo — and exo-activity. These include p-glucosidase (3.2.1.21), p-xylosidase (3.2.1.37), and p-mannosidases (3.2.1.25) among others. Due to the low degree of polymerization of their substrates, distinction between endo — and exo­hydrolysis modes of action is difficult.

Debranching enzymes, often referred to as accessory enzymes, can be further subdivided into those acting on glycosidic linkages and those acting on ester-linkages. The dominant enzymes of the former include a-L-arabinofuranosidases (3.2.1.55) and a-glucuronidases (3.2.1.139), both of which remove glycosidic side chains from xylan (2, 13). The dominant esterases include acetyl xylan esterase (3.1.1.72) and feruloyl esterase (3.1.1.73), both of which also act on xylan (11). Enzymes acting on the major hemicelluloses are described in Figure 10.1. There are also reports of other esterases acting on other acetylated polysaccha­rides, including glucomannan, galactomannan, and even cellulose (12).

Some hemicellulases and accessory enzymes exhibit cross reactivity across the different hemicelluloses, while others may be very specific for a particular oligomers sequence or conformation. p-xylosidase, for example, may preferentially hydrolyze xylobiose, but may also act on xylotriose and higher xylo-oligomers, gentibiose, and cellobiose. Feruloyl esterase, though normally most active on ferulic acid ester-linked to arabinose, may also be active on coumaroyl esters. p-glucosidases have a wide range of specificities, with certain enzymes acting across a broad range of p-(1—>4), p-(1 —>3), and mixed p-(1—>3,4) linkages and other versions of the same enzymes requiring a specific linkage sequence or side chain pattern. Due to the high degree of heterogeneity in hemicellulose structure, the number of permutations involved in the conformation of sugars, non-sugars, and their linkages

precludes a comprehensive, or even in-depth, discussion of the specific enzymes and their activities. Instead, we will outline the basic activities of a variety of hemicellulase enzymes and indicate their potential use in the depolymerization of the key bioenergy feedstock hemicelluloses.

The binding of the cellulosome to microcrystalline cellulose has been suggested in the past to be mediated directly via the scaffoldin-based, cellulose-binding CBM (37, 38). In the original description of the cellulosome from C. thermocellum, a mutant was isolated that lacked cell-surface cellulosomes and consequently failed to bind to the substrate (27, 108). Since the scaffoldin is anchored to the cell surface via the various anchoring scaffoldins, the same CBM appears to be responsible for binding the entire bacterial cell to cellulosic substrates, as will be described below (109). Nevertheless, the C. thermocellum cellulosome also includes numerous enzymes that also bear family-3 CBMs (93, 95) that can potentially supplement the cellulose-binding role of the scaffoldin-based CBM.

The recombinant, scaffoldin-derived, cellulose-binding family-3 CBM was also shown to bind chitin but not xylan (110). The CBM binds to celluloses of very high crystallinity — even to cellulose of the highest known crystallinity, namely, from the cell walls of the al­gae, Valonia ventricosa. It also binds strongly to amorphous forms of cellulose, including phosphoric acid swollen cellulose of minimal crystallinity. The very high capacity to bind to the amorphous cellulose (111) likely reflects increased accessibility to the binding sites, rather than a preference for amorphous regions. In fact, the comparative crystal structures of different families of cellulose-binding CBMs suggest that, in each case, an aromatic strip of amino acids on the flat surface of the CBM molecule mediates strong binding with the glucose chain of the hydrophobic face of the cellulose surface (112). Additional hydrogen­bonding interactions between hydrophilic amino acids and polar groups on neighboring glucose chains are thought to provide additional protein-cellulose contacts that stabilize the strong interaction. Docking analyses thus suggest that a patch along three adjacent cellulose chains interacts with the flat surface of the CBM molecule, which indicates why cellobiose and other cellodextrins fail to inhibit CBM binding to the insoluble substrate (110).

The C. thermocellum genome includes several other, enzyme-borne family-3 CBMs that exhibit most or all of the latter proposed binding residues (113-115), all of which would presumably bind strongly to microcrystalline cellulose. However, the latter five or six en­zymes, lack dockerins and are thus free, non-cellulosomal enzymes. In addition, another sub-class of family-3 CBMs, termed CBM3c, is fused to a GH9 catalytic module. This type of family-3 CBM is modified in the same surface residues that are considered to bind crystalline cellulose, and the CBM3c is believed to assist the adjacent catalytic module in binding to structural intermediates of the substrate or to alter the mode of activity (113,116,117). The CBM3c fails to bind strongly to crystalline cellulose substrates, and thus plays an ancillary role in breakdown of the cellulose chain by the catalytic module. The balance of the other dockerin-containing, enzyme-borne CBMs lack several but not all of the cellulose-binding residues. In some cases (118,119), cellulose-binding properties have been reported, but their contribution to the primary strong recognition of crystalline cellulose by the cellulosome, as observed for the scaffoldin-based family-3 CBM, remains unclear.

The binding of the cellulosome to cellulose is inhibited by low ionic strength, and water can be used to release at least part of a cellulosome or CBM preparation from the cellu­lose matrix (111). Increasing the salt content of the medium increases the binding of the cellulosome to the substrate. Interestingly, maximal enzymatic activity of the cellulosome complex was observed at suboptimal conditions of adherence to the substrate. At low salt concentrations, the low enzymatic activity most likely reflects the lack of sufficient adsorp­tion, whereas above-optimum concentrations of salt, the reduced activities may be due to restricted mobility of the cellulosome under such conditions. Once attached, the cellulosome seems to remain static, as shown by laser bleaching experiments (unpublished results). A micrograph of cellulosomes bound to the surface of the super-crystalline Valonia cellulose is shown in Figure 13.3. Note the apparent cellulosome-induced stripping of the microfibrils from the surface of the cellulose crystal.

The organization of enzymes into cellulosomes ensures their multiple, concerted at­tachment to the substrate surface. This arrangement affords a distinct advantage over the distribution of free cellulases, since it allows multiple hydrolysis and facilitates access of additional appropriate cellulases and other enzymes (e. g., hemicellulases, pectate lyases, carbohydrate esterases) at or near the cleavage sites for enhanced processing of the sub­strate. The proximity ofcomplementary enzymes at the cellulose surface provides a suitable remedy for counteracting substrate recalcitrance, as will be further discussed below.

DNA microarrays are powerful and versatile tools for monitoring the expression of tens of thousands of genes simultaneously. This technology has successfully been applied to mon­itor transcriptome regulation in cancer studies (114, 115), the discovery of drug targets (116), and importantly for studying microbial gene expression and regulation under differ­ent growth conditions (117). Similar to the situation in which microprocessors have sped up computation, microarray-based genomic technologies have revolutionized the genetic analysis of biological systems, by moving from assaying gene expression, one gene at a time,

Isolation of RNA and DNA

Hybridization

Whole genome microarray

Data analysis

Figure 15.3 Example of a microarray hybridization experiment using a DNA reference. Depending upon the label used an expressed gene can be differentiated from silent genes based on the spot color.

to the ability to visualize the dynamics of the entire transcriptome of an organism in one hybridization step.

The analysis of gene expression using microarrays involves the steps outlined in Figure 15.3. The first step is the acquisition and purification of RNA of samples grown under differ­ent experimental conditions. Depending on the experimental design this may also involve the isolation of DNA. Methods for simultaneous isolation and purification of both RNA and DNA (118) from the same sample have been developed. After the isolation of nucleic acids from the sample, a fluorescent label must be incorporated into each. In the case of RNA,
labeled dUTPs are incorporated into the sample, as it is reverse transcribed into cDNA. DNA labeling is accomplished using the Klenow polymerase fragment to incorporate the labeled nucleotide. Following this simultaneous hybridization and subsequent laser excita­tion and imaging of each of the dye-labeled samples simultaneously, allows one condition to be compared to the other. Transcribed RNA can be compared to baseline DNA from the same sample allowing the gene expression level to be calculated, relative to the initial sample composition.

15.3 Conclusions

The use of lignocellulosic material as a feedstock for producing fuels and chemicals has enormous potential considering its relative abundance. However, its use continues to be cost prohibitive because of pretreatment and enzyme costs (119). What lacks is a fun­damental understanding of how natural systems efficiently degrade and utilize decaying biomass. Microorganisms of various types are intimately involved in biomass degradation and occasionally are the only biological agents capable of doing so. Gaining a fundamen­tal understanding of biological mechanisms of biomass decay by microbial communities will help us develop cost-effective ways to convert these abundant substrates to fuels and chemicals.

Microorganisms are not alone in nature. Each microorganism involved in biomass degra­dation interacts with both its surroundings and with other organisms. These interactions result in chemical and physical changes that in turn lead to microbial communities that are complex and dynamic but ultimately effective in transforming biomass. Conventional microbiological methods used to evaluate them fail to capture their diversity or biochemical complexity. This is due in part because cultivating microorganisms in the laboratory depends on supplying the right nutrients and growth conditions. Because biomass conversion is a dynamic process there are countless microenvionments that would need to be considered.

New methods for examining the diversity and biochemistry of biomass degradation are emerging. It is now possible to characterize microbial populations using large-scale inte­grative approaches. Rapid advances in the fields of genomics and proteomics have spawned multiple new “omic” subdisciplines as evidenced by the number of recent papers detailing the evolution and molecular basis of microbial communities. Large-scale proteomics-level examination of natural microbial communities is now possible allowing one for not only the functional analysis of enzymes directly involved in biomass conversion but also for the analysis of the processes and interactions involved in community formation. These new technologies will revolutionize the way microbial communities are defined in the future and allow microbiologists to envision and model microbial biomass decay as a set of interacting processes that when combined effectively degrade plant biomass.

Acknowledgment

This work was supported by the US DOE Office of the Biomass Program. The research was also financed as part of the BioEnergy Science Center, a U. S. Department of Energy Bioenergy Research Center supported by the Office of Biological and Environmental Research in the DOE Office of Science.

Polysaccharides with (3-1,4-mannan and (3-1,4-glucomannanbackbones are abundant con­stituents of the wood of gymnosperms (78, 79) the cell walls of certain algae (80), and are present in lower amounts in many other species (81). Mannans also serve as carbohydrate reserves in a variety of plant species (78,82). Several groups have biochemically characterized glucomannan synthase activities from a variety of plant species (78,79,82,83). The enzymes were shown to use GDP-mannose and GDP-glucose as substrates and to produce polymers with varying ratios of the two sugars, depending on the ratio of the sugar nucleotides (83) Recently, transglycosylase enzymes that modify the architecture of mannan polysaccharides in plant cell walls have also been discovered (84).

Mannan synthase is encoded by genes of the CSLA gene family (85). Expression of a cDNA from guar in soybean cells led to the synthesis of a (3-1,4-mannan. Similarly, expression of three Arabidopsis CSLA proteins in Drosophila cells resulted in proteins that catalyzed synthesis of mannan from GDP-mannose (86). Similarly, several poplar orthologs expressed in Drosophila cells exhibited glucomannan synthase activity (87). Functional analysis of

CSLA genes from diverse species is consistent with the hypothesis that the function of the CSLA genes is conserved in all plants (81).

As noted previously in studies of impure enzymes (83), when provided with both GDP — glucose and GDP-mannose, the enzymes produce mixed linkage mannans. One of the proteins produced glucan when provided only with GDP-glucose. These studies suggest that no primer is required to initiate mannan synthesis. The ability of the CSLA enzymes to accept GDP-glucose or GDP-mannose is compatible with earlier suggestions that the ratio of mannose:glucose in glucomannans may be controlled by regulating the availabil­ity of the two sugar nucleotides. The observation that GDP-glucose is a substrate also may explain old observations from in vitro polysaccharide synthesis experiments that had initially been interpreted as evidence that GDP-glucose was the substrate for cellulose synthesis.

Genes encoding the galactomannan galactosyltransferase responsible for attaching galac — tan side chains have also been identified following purification of the enzyme from fenugreek (88). The a-1,6GalT galactosyltransferase cDNA encodes a 51282 Da protein, with a single transmembrane alpha helix near the N terminus. The protein has been functionally ex­pressed in the yeast Pichia pastoris and is active when the membrane-spanning domain is removed. Thus, presumably the membrane-spanning domain is required only to localize the protein to the Golgi apparatus where mannan is thought to be synthesized. The degree of substitution when UDP-galactose is available is variable and appears to be a stochastic process controlled both by enzyme specificity and the levels of a1,6GalT activity (89). Eight Arabidopsis gene sequences are very similar to the a-GalT from fenugreek (88).

In Arabidopsis, mannans have been localized not only in thickened secondary cell walls of xylem elements, xylem parenchyma, and interfascicular fibers, but also in the thickened walls of the epidermal cell of leaves and stems and, to a lesser extent, in most other cell types (90). Analyses of Arabidopsis mutants containing a transposon insertion in exon seven of the CslA7 gene showed that disruption of this gene results in defective pollen tube growth and disruption of embryonic development (91). Mutants (called rat4) containing a T-DNA insertion in the 3′ untranslated region of the AtCslA9 gene display resistance to Agrobacterium tumefaciens transformation, apparently caused by decreased binding of bacterial cells to roots. AtCslA9 promotor-GUS fusions indicated that this gene is expressed in a variety of Arabidopsis tissues, including lateral roots and the elongation zone, where the root is most susceptible to Agrobacterium transformation (92). In both mutant studies, the authors suggested that the mutant phenotypes resulted from alterations in polysaccharide content; however, in neither case was such a defect demonstrated.

A GDP-mannose transporter that is localized to the Golgi has been characterized (93).

ADP-Glc is the major precursor for starch synthesis. Starch is a polymer of a-D-Glc consisting of two types of molecules: amylose (linear a-1,4-linked glucose) and amylopectin in which one Glc in every 20 or so residues on an amylose-like structure is branched by an a(1- 6)-linkage connected to an a-1,4-linked chain. Although starch synthesis is not related to pectin synthesis, the information available about the regulation of the synthesis of the starch precursor, ADP-Glc, can be used as a paradigm when considering the synthesis and regulation of other NDP-sugars.

Starch, the main carbon storage form in plants, is made in plastids of photosyn­thetic and non-photosynthetic tissues. Adenosine 5′-diphosphate:glucose pyrophospho — rylase (ADPGlc:PPase) catalyzes the first and rate limiting step in starch biosynthesis (i. e., the conversion of Glc-1-P and ATP to ADP-Glc and pyrophosphate (436). In cereal en­dosperms, two distinct ADPGlc PPases exist, one is found in the cytosol and the other in plastids (437-439). By contrast, ADPGlc PPase is exclusively located in plastids of leaves of both mono — and dicotyledonous plants, as well as in heterotrophic organs of dicotyledonous plants. Plant ADPGlc PPase is composed of two types of subunits (small and large) and is allosterically regulated by 3-phosphoglycerate and phosphate. The Arabidopsis genome con­sists of six ADPGlc PPase-encoding genes (two small subunits, ApS1 and ApS2; and four large subunits, ApL1-ApL4). Based on recombinant enzyme activities, mRNA expression and the fact that recombinant Aps2 has no ADPGlc PPase activity, it has been proposed (440, 441) that ApS1 is the main catalytic isoform responsible for ADPGlc PPase activity in all tissues of Arabidopsis. The authors suggested that each isoform of the large subunits plays a regulatory role. The large subunit, ApL1 is expressed in source tissues, whereas ApL3 and ApL4 are the main isoforms expressed in sink tissues. Thus, in source tissues, ADPGlc PPase could be regulated by the 3-phosphoglycerate/phosphate ratios, while in sink tissues; the enzyme would be dependent on the availability of substrates for starch synthesis.

In cereal endosperm, on the other hand, a different regulation of starch synthesis may operate. It appears that the transport of ADP-Glc from the cytosol into the plastid is the limiting factor. This became clear during the characterization of a plastidial ADPGlc trans­porter (HvNstl) barley mutant with low-starch content (442). The mutant accumulates high levels of ADP-Glc in the developing endosperm indicating that cytosolic pool of ADP­Glc is not under metabolic control in this tissue. Lastly, leaves overexpressing SuSy showed a large increase in the levels of both ADP-Glc and starch, compared with WT leaves, while leaves overexpressing antisense SuSy accumulated low amounts of both ADP-Glc and starch (438). The above findings, which originated in the Pozueta-Romero’s laboratory, show that in source leaves ADP-Glc produced by SuSy (outside the chloroplast) is directly linked to,

and appears to control starch biosynthesis. This implies that SuSy, but not ADPGlc:PPase, controls the level of ADP-Glc in the cytosol in source leaves (438, 439).

More recently, a new enzyme activitywas identified in Arabidopsis thatmustbe considered to better evaluate the metabolic fate of ADP-Glc in the cytosol. Recombinant Arabidopsis At5g18200 has ADP-Glc phosphorylase activity (please note it is not a PPase). The enzyme is capable of transferring AMP from ADP-Glc onto either Pi or Gal-1-P (443) as shown in the scheme below:

ADP-Glc + Pi ^ Glc-1-P + ADP ADP-Glc + Gal-1-P ^ Glc-1-P + ADP-Gal

Unlike the human and fungal GalT enzyme, which transfers UMP from UDP-Glc onto Gal — 1-P forming Glc-1-P and UDP-Gal, the ADP-Glc phosphorylase cannot utilize UDP-Glc as a donor substrate (443).

We put forward that the above-described distinct regulatory role of ADPGlc PPase and SuSy are examples that highlight the possibility that different isoforms of nucleotide-sugar biosynthetic enzymes may have distinct roles in plants and that different plant species may regulate the same metabolic pathway in different ways.