The effects of downregulating and/or mutating PAL, C4H, pC3H, HCT, and 4CL, as far as the monolignol/lignin pathway was concerned, gave predictable results. PAL, C4H, and 4CL resulted in lower lignin levels overall, with generally deleterious effects (weakening) of the vasculature being noted, albeit not quantified in any way. pC3H, a regulatory branch point to the G/S segments of the lignin-forming pathway, also gave quite predictable results, i. e., significant reductions (up to 64-68% of estimated wild-type lignin levels and/or altered vascular anatomy). Interestingly, for alfalfa, reaction (tension) wood tissue provisionally appeared to be part of a compensatory mechanism to help offset reductions in vascular in­tegrity due to lower lignin contents. No evidence was obtained, though, for “combinatorial” biochemistrybeing in effect to any measurable or significant extent in any of these manipula — tions/mutations. That is, there was no evidence of any shift to formation of non-monolignol phenolic moieties to compensate for reductions in overall lignin amounts. There was, how­ever, a relatively small increase in the level of H-lignin being formed which is considered due to limited substrate degeneracy during a proposed template-assisted polymerization, but only for the pC3H line.

Given the lack, even today, of available technologies/methodologies in adequately probing lignin structure, there were nevertheless a number of other attempts to provide depictions of what lignin structure(s) might look like. As described below, while each contains numerous discrepancies relative to recent experimental observations, progress was made though as regards identification of additional structural subunits within lignins. However, the limita­tions again further underscored the urgent need to now develop methodologies to accurately probe native lignin macromolecular configuration.

When the large computers of the future, combined with the advances in the computational tools arrive, we can expect to see all-atom models studying millions of atom-sized structures and producing statistical-thermodynamic results based on a sampling scale in the microsec­ond range. Modeling techniques will grow into the multi-scale modeling arena such that large-scale structures can be modeled reliably including the cell wall and the large structures associated with it. Figure 8.5 depicts one potential approach to developing a programmatic solution for modeling the plant cell wall, starting with the simplest problems addressable today (e. g., still a challenge today) and building toward computational interpretations of the

wall system using codes and processors not yet available. Computational approaches to the cellulose hydrolysis problem will only become more vital to the full understanding of the processes involved as the computational architectures and methods evolve, working harmo­niously with experimental and theoretical approaches to solve what are now unapproachable problems.

Acknowledgments

This work was supported by the DOE Office of the Biomass Program and by the NSF via the Strategic Applications Collaboration program at the San Diego Supercomputer Center.

Harry J. Flint

12.1 Introduction

Herbivorous mammals do not secrete digestive enzymes that are able to degrade the major structural polysaccharides of plant cell walls, but depend on the activities of symbiotic gut microorganisms to obtain energy from the plant material that makes up the bulk of their diet. Breakdown ofplant cell wall material is mediated by anaerobic microbial communities that develop in the large intestine (caecum and colon) in the case of hind-gut fermentors such as horses and rabbits, while in ruminants this breakdown occurs largely in the foregut, in the reticulo-rumen. In both cases, the short chain fatty acid products of fermentation are absorbed and used as energy sources by the animal, but the foregut location of the rumen also allows the animal to take advantage of microbial protein, by digesting the microbial cells that pass into the acidic stomach.

The rumen is the site of highly efficient breakdown of the wide variety of plant material that forms the diet of grazing animals, and harbors a complex consortium of anaerobic mi­croorganisms comprising bacteria, archaea, fungi, and protozoa (1). Plant polysaccharide­degrading enzymes are produced by a high proportion of these microorganisms, both eu­karyotes and prokaryotes. Many rumen microorganisms may, however, be considered as secondary utilizers that exist by cross-feeding, and are largely dependent on other organ­isms for primary attack upon the more recalcitrant plant structures (2-5). The role of their enzyme systems and transport machinery is to scavenge soluble polysaccharides and oligosaccharides as they are released by other microorganisms from plant material. The primary degraders of plant cell wall material on the other hand are assumed to be those that are capable of tight attachment and that possess the enzymatic machinery to access this complex, insoluble substrate. The number ofgenuinely cellulolytic species identified among the rumen microbiota is relatively small, and their populations may often be underestimated because of the difficulty of recovering them in the substrate. Interest has centered on these primary plant cell-wall-degrading species because of their key roles in initiating substrate breakdown.

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

12.2 Cellulolytic and hemicellulolytic bacteria from the rumen

Three rumen bacterial species, in particular, were recognized from early cultural studies to be actively cellulolytic. Two Ruminococcus species, R. flavefaciens and R. albus, are repre­sentatives of the Gram-positive Clostridial cluster IV, while Fibrobacter succinigenes belongs to a divergent group of Gram-negative bacteria. Recent molecular work has confirmed the importance of all three species in the rumen ecosystem (6-8) although there are indications from 16S rRNA diversity studies that hitherto uncultured cellulolytic bacteria remain to be recovered from the rumen (9).

William S. Adney, Daniel van der Lelie, Alison M. Berry, and Michael E. Himmel

15.1 Introduction

Traditionally, microbiologists have taken a reductionist approach to understanding micro­bial biomass decay communities focusing on the analysis of individual genes, microorgan­isms, and biochemical reactions. Recent advances in molecular biology have identified many genetic components but have provided limited information on the mechanisms of biomass decay. In the area of biochemistry, advances in proteomics and high-throughput enzyme assays are providing new theories into the mechanisms of biomass decay, but limited in­formation on individual microorganisms and their interactions. Combined, however, these technologies are providing a new “systems biology” approach to understanding the biomass decay communities. This approach will allow microbiologists to envision and model micro­bial biomass decay as a set of interacting processes that when combined effectively degrade plant biomass.

The first major step towards clarifying the fundamental principles of biomass recalci­trance is to understand the scale and complexity of natural systems involved in biomass decomposition. Heterotrophic microorganisms are major players in biomass decomposi­tion and in the global cycling of terrestrial carbon. Our terrestrial biosphere depends on heterotrophs functioning in complex and dynamic communities to breakdown the natural accumulation of biomass. Although the subject of study for several decades, we still know little about the diversity and complex interrelationships of the individual organisms. How­ever, we now understand that these communities vary in spatial and temporal dimensions to control the biochemical rate of carbon cycling, as well as the cycling of other essential elements, like nitrogen, sulfur, and phosphorus. In fact, the production of carbon dioxide by chemoorganotrophs is the single most important contribution of CO2 to the atmosphere (1).

Microbial communities are complex networks of individual organisms that include every ecological relationship ever described, ranging from coexistence to commensalism, mutu­alism, and parasitism. There are direct symbioses between individual microorganisms and indirect symbioses in which metabolic processes of one species modify the habitat and/or physiology of another species. Studying microbial communities in most environments has

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

thus presented a challenge in that it is very difficult to simulate environmental conditions and ecological relationships adequately in the laboratory in order to satisfy the physiological requirements for the reproducible cultivation of a representative community. This maybe due to our inability to reproduce the physical, chemical, and temporal conditions needed for the multiplex interactions and unknown species.

Until recently, biodiversity estimates were based only on those species that could be culti­vated by using traditional in vitro microbiology techniques. Unfortunately, these techniques only allow 1-5% of the total community members to be examined (2, 3). Traditional mi­crobiology techniques were developed to study the growth and metabolic requirements of individual organisms in pure culture. As such, they provide limited information into the ac­tions of biomass-degrading communities determined by diverse and dynamic biochemical pathways. Recently, new molecular technologies have provided valuable information about the in situ biodiversity of plant decaying microbial communities. However, while genomic approaches provide important information about the diversity of individual species within populations they do not predict the biochemical outcome on functional terms. More than a biotic inventory of microorganisms is needed to develop a complete understanding of the organization and interactions within the community. Individual microorganisms display such strong interactions that new capabilities are needed to understand recalcitrance in comprehensive and integrated way. Therefore, a “systems biology” approach is needed to envision how biomass decomposition functions as a complete set of intersecting processes. This requires that we also understand the biochemical reactions and catalysts involved in the deconstruction of biomass. Clearly, no technique alone can provide the broad range of information needed to understand community structure and system function.

Plant biomass is a chemically diverse substrate varying in composition, but predominated by complex and interactive polymers of cellulose, hemicellulose, and lignin. Evolution has developed plants that are naturally recalcitrant to degradation by microbial communities. Major contributing biochemical features to the recalcitrance of terrestrial plants not shared by bryophytes and earlier plants are the lipid materials of the cuticle and its wax, and the diversity of phenolics in lignin and flavinoid compounds (4). The arrangement and density of the vascular bundles, the relative amount of sclerenchymatous (thick wall cell) tissue, the structural variation, and complexity of cell wall constituents also contribute to recalcitrance. The result is that the natural plant biomass decay involves diverse groups of heterotrophic bacterial and fungal communities that degrade and metabolize plant material. In short, Nature’s answer to the innate chemical and structural complexity found in plants is greater diversity and synergistic interactions. The full extent of this diversity is still up to debate.

Carbon turnover in the terrestrial biosphere occurs chiefly in the soil from complex and yet unclassified microbial communities. Turnover rates vary dramatically depending on environmental conditions such as temperature, water availability, inorganic nutrients, pH, organic carbon input such as plant exudates, biomass composition, and the presence of microorganisms producing hydrolytic enzymes such as cellulases and hemicellulases. The exact mechanisms of degradation carried out by complex interactions of members of biomass-degrading community remain elusive. The arrival of new biotechnology tools such as ribosomal rRNA gene sequencing, comparative metagenomics, transcriptome, and se — cretome analysis are allowing for habitat-specific fingerprinting of microbial communities. Genomic-based studies are providing vast amounts of sequence data from various environ­mental samples and have led to new insights into microbial populations. The improvement of sequencing technologies has made metagenome shotgun sequencing of an environmental sample feasible; however, most environmental communities are far too complex to be fully sequenced in this manner.

Metagenomic analysis of representative bacterial assemblages is beginning to provide a knowledge base and a source of genetic material for further studies on the ecology of ligno — cellulose degradation and biotechnology applications. One obvious limit of environmental metagenomic sequencing is the sheer diversity of the microbial communities that populate rich environments, which calls for large sequencing efforts, to assemble long DNA sequence data. This complexity can lead to bias through overrepresentation of only a few common genomes, and because of the difficulty of assembling long sequences for analysis. Also of sig­nificance is the increasing number of published genomes of biomass-degrading organisms. Genomics research is not only providing phylogenic information about microbial popula­tions but in conjunction with new biochemical tools is generating a broader understanding of the complete biophysical activity of the community. Together these new tools are rapidly advancing our understanding ofthe principles that underlie microbial biomass degradation and carbon cycling in nature. The knowledge gained from these studies is a stepping-stone to the development of optimized enzymes and microorganisms for the production of com­modities such as ethanol and hydrogen from biomass.

In cell walls, proteins and polysaccharides may interact both non-covalently and covalently. Thus non-covalent associations between positively charged lysine and histidine residues on HRGP and negatively charged non-cellulosic polysaccharides, e. g., pectins and GAX have been postulated (122). Covalent intermolecular isotyrosine (phenyl ether) bridges between HRGPs have been proposed to account for the ability of acidified chlorite (123) to allow solubilization of the HRGP by cleavage of these linkages, but have not been identified, however, intramolecular isotyrosine linkages are found in HRGPs.

GAX in secondary walls of grasses may also be covalently linked to protein though dimer­ization of FA on GAX to tyrosine on proteins as proposed by Geissmann and Neukom (124) (Figure 4.3c). Such a mixed dimer cross-link has been isolated from an endosperm prolamin-AX complex formed during bread making (125) and has been proposed to occur in walls of lignified pericarp cells in wheat bran (126). In aleurone walls from wheat bran, a fraction that resists digestion by hydrolases for AX and (1—>3, 1—4)-p-D-glucan, con­tains a highly branched AX that appears to be linked to a protein, supposedly through an FA-tyrosine bridge (75).

RG-I accounts for 7-14% of the primary wall (157) and 20-33% of pectin (194). Unlike HG and RG-II, RG-I has a backbone of up to 100 repeats of the disaccharide [^4)-a — D-GalpA-(1^2)-a-L-Rhap-(1^] (2, 157, 261, 262, 361, 362). The GalA residues in RG-I maybe O-acetylated at C-3 or C-2 (157, 199, 288-290). The average molecular weight of sycamore RG-I has been reported to be 105-106 Da (157). Between 20 and 80% of the rhamnosyl residues are substituted at C-4, and sometimes at C-3, with side chains composed mostly of arabinosyl and/or galactosyl residues (2, 157, 264), referred to as galactans (157, 264, 278), arabinans (214, 264, 290), and arabinogalactans (2, 157, 190, 214, 264). These side chains may range in size from 1 to 50 or more glycosyl residues (2, 157, 290). A large amount of immunocytochemical evidence based on antibodies against specific RG-I carbohydrate epitopes (160) indicates that the precise structure of the side chains of RG-I varies in a cell type and development-specific manner (214, 363). Representative side chains or portions of side chains that have been identified in RG-I have been previously summarized in (192).

The RG-I galactans may contain only galactosyl residues or may contain other neutral glycosyl residues (157) or acidic residues such as GalA (190), GlcA (157,190, 278), or 4- O — methylGlcA (278). Some galactans also have p-1,6-branching (190). As mentioned above, the size and linkages in the galactan side branches of RG-I vary depending upon the species (157).

RG-I also contains side chains of individual or multiple L-arabinofuranosyl (Araf) residues or chains of 1,5-linked-a-L-Ara f substituted at O-3 and occasionally at O-2 with additional Ara f residues (190, 290, 364). Such side chains are referred to as arabinans.

Some RG-I side chains contain both arabinosyl and galactosyl residues. These side chains are referred to as arabinogalactans that have been divided into Type I and Type II arabino­galactans. Type I arabinogalactans contain a 1^4-linked p-D-galactan backbone while the Type II arabinogalactans contain a 1^3-linked p-D-galactan backbone and are often highly branched (2, 157, 190, 214). Type II arabinogalactans maybe associated with glucurono- mannoglycans (190). Some studies suggest that mannose may be a component in some pectins, probably as a side branch to RG-I (190), however, the structural role of mannose in pectin has not been clearly demonstrated and therefore mannose is not discussed as a component in pectin here. Some of the Type II arabinogalactan is associated with arabino — galactan proteins (AGPs) (365-368), hydroxyproline-rich proteins located at the plasma membrane, cell wall, or in media surrounding suspension-cultured cells (366, 367, 369, 370). It is not always clear whether specific Type II arabinogalactan structures isolated from wall extracts are associated with RG-I, AGPs, or both. However, multiple lines of evidence show that some Type II arabinogalactan is associated with RG-I. This includes the cross reactivity of the antibody CCRC-M7 with both RG-I and arabinogalactan proteins (371). CCRC-M7 recognizes a trimer or larger of p-(1,6)-Gal carrying one or more Ara residues (372). Pectic polysaccharides from the Chenopodiaceae family including spinach (Spinacia oleracea) and sugar beet (Beta vulgaris) are esterified with phenolics such as ferulic acid (157, 291, 373), on galactose and arabinose residues that are likely substituents in the side branches of RG-I (157, 291, 292, 374).

Currently, little is known about how the synthesis of nucleotide-sugars is controlled in time or space, and how it relates to the glycosyltransferases that actually make the diverse glycan polymers. What is the limiting factor in wall synthesis? Is it supply of NDP-sugars (as is the case for starch) or glycosyltransferases?

We will divide this section into three topics: sugar flux, role of isoforms, topology and protein complexes.

2.5.2.1 Sugar flux

Although a considerable proportion of cellular sugar ends up in wall polysaccharides, some sugar-derivatives are required for glycoprotein, glycolipid, and glycoside synthesis. In addi­tion, significant amounts of sugars are stored either as large glycans such as starch, small-sized glycans (e. g., raffinose, fructan), or as the disaccharide sucrose. We would like to point out two issues related to flux: 1) growth potential of a cell; 2) whether some wall components compensate for the lack, or reduced amount, of other glycans.

1 New meristematic cells need to expand and grow to their prospective developmental tissue (e. g., leaf cells). What determines the growth potential and the cell’s final size is unclear. Logically, with limited wall precursors the potential for growth is restricted since wall polymers are not made. The underlying mechanism that controls this complex develop­mental process is still unknown and poses a fascinating scientific quest. For example, do transcription factors regulate coordinately the expression of “tissue-fate genes” as well as NDP-sugar biosynthetic genes and genes involved in the supply of carbon? If carbon flux is not limited and all NDP-sugar biosynthetic genes are highly expressed — would the cell be larger? Do young, old, or stressed cells sense sugar availability or sugar status for growth and/or for storage in different ways? What are the ultimate determinates for growth; sugars or sugar-phosphates? Several sugar-sensing (signaling) proteins (and cor­responding genes) have been isolated. It is assumed that sugar sensing (i. e., the interaction between a sugar molecule and a sensor protein) mediates a signal which initiates signal transduction cascades that result in cellular responses such as altered gene expression and enzymatic activities. Sugars as signaling molecules affect the plants at all stages of growth starting from seed germination to seed development. Sugars, like hormones, can act as primary messengers and regulate signals that control the expression of various genes in­volved in sugar-phosphates and wall metabolism. But do NDP-sugars function, in part, as signal molecules? In human cells, UDP-GlcNAc serves as a glucose sensor and moves between the cytosol and the nucleus. A cytosolic and nuclear-localized soluble enzyme, known as OGT, catalyzes the O-linked transfer of GlcNAc from UDP-GlcNAc directly to Ser/Thr of target proteins (503). For example, the O-GlcNAcylation, of the transcription factor Sp1 promotes nuclear localization of Sp1 and its ability to transactivate calmod­ulin (CaM) gene transcription (504). Whether plant cells consist of analogous signals to suppress or activate wall-biosynthetic genes by monitoring levels of NDP-sugars is unknown.

A major task for future research will be to investigate the relationships between isoforms that produce the same nucleotide-sugar, GTs, and sugar-sensing genes. Once the function of wall-related genes becomes known, bioinformatics will be useful in identifying a common set of genes that are coordinately expressed or suppressed to form a specific glycan.

Synthetic dehydropolymerizates (DHPs, so called “lignin-like”) can be obtained by H2O2/peroxidase catalyzed one-electron oxidation of monolignols. While generating some of the substructures found in native lignins, there is, of course, little opportunity for control over primary chain interunit linkage (sequence) designation with in vitro preparations. Thus, numerous important differences between synthetic DHPs [which, for example, re­lease dimeric pinoresinol-derived substructures (see Figure 7.2D, substructure Illb) upon cleavage] and native lignin polymers [which do not(193)] have long been known, and which diminish greatly their utility (29). While it is possible to synthesize 8- O-4′ linked oligomers (194), and also to modulate the amounts of 8-O-4′ interunit linkages through varying peroxidase activity in vitro (195), such preparations are not lignins proper. Indeed, in many respects, the comparison of synthetic DHPs to native lignins might be likened to that of randomly linking amino acids through amide linkages in the same relative proportions as found in insulin, and describing the resulting products as an insulin mimic.

8.2.2.1 The stretch energy

In the AMBER force field there are three terms describing the interactions between atoms that are either directly bonded or separated by two or three bonds. These three terms are known as the bonded terms and correspond to bonds, angles, and dihedrals, respectively. The first term (Kr(r — req)2) describes a simple harmonic potential obeying Hooke’s law and is used to represent the energy involved in the stretching of a bond between two directly linked atoms. 2Kr is the force constant for the bond and (r — req) the distortion from the equilibrium bond length. It should be immediately obvious from this expression that such an approach prohibits the study of bond breaking since the expression tends to infinity as the bond length is increased significantly beyond the equilibrium value.

The harmonic representation of the potential is the simplest possible but provides a fair description of the energetics of bond stretching and compression when the bond length remains close to the equilibrium value (Figure 8.1). While more complex descriptions are available they come at the price of increased computational complexity and increased pa­rameterization requirements and are rarely used for modeling biological systems.