The native cellulolytic organisms, e. g., C. thermocellum, are a starting point for developing anaerobic microbes capable of one-step processing of cellulosic biomass to ethanol or other desired products in the absence of added saccharolytic enzymes (12, 15, 28). However, it is necessary to perform metabolic engineering to improve product yield and titer relative to performance obtained to date with traditional ethanologenic strains. Substantial advances have been made in developing the genetic transformation tools for cellulolytic bacteria (58, 62, 70, 71). Recently, metabolic engineering has been applied to a mesophilic, cellulolytic C. cellulolyticum to achieve high ethanol yields. This was achieved by heterologous expres­sion of Zymomonas mobilis pyruvate decarboxylase and alcohol dehydrogenase (70). The fermentation pattern was shifted significantly in that ethanol production increased by 53%, acetate increased by 93%, and lactate decreased by 48%.

From a fundamental viewpoint, it is also interesting to study the mechanisms of hydrolysis product uptake by natural cellulolytic microorganisms, determine hydrolysis product con­centrations on the cell surfaces, and study the relationships between sugar transportation, energy level, and cellulase regulation.

From a practical viewpoint, another concern about naturally cellulolytic microorganisms is their sensitivity to the final product accumulation. For example, the wild C. thermo­cellum strain cannot grow in the presence of ethanol concentrations above 1% (w/v), yet industrial yeasts can tolerate ethanol titers up to 10% (w/v). Currently, an ethanol-adapted C. thermocellum strain that can tolerate up to 5.0-8.0% (w/v) ethanol was selected by step­wise increase in ethanol concentration (72). Another new method is global transcription machinery engineering (gTME), which re-programs gene transcription to tolerate substrate and product inhibition (73). Using gTME, a recombinant S. cerevisiae increased volumetric ethanol production by 69%, specific productivity by 41%, and ethanol yield by 14% (73). We expect that this technology could be applied to natural cellulolytic microorganisms as well.

The key objective for effective, recombinant cellulolytic ethanologens, e. g., S. cerevisiae, is the introduction of an efficient cellulase enzyme system. Key objectives to enable CBP microorganisms that we learned from natural cellulolytic microorganisms are to: 1) in­crease active recombinant cellulase expression levels up to 2-20% of cellular proteins (28), 2) optimize the expressed cellulase ratio (endo-/exo-/BG) by gene regulation, depending on their turnover number and substrate characterization (74), 3) generate more ATP from intracellular substrate phosphorylation by introduction of recombinant cellobiose and cel- lulodextrin phosphorylases, especially for anaerobic fermentations (44, 49), and 4) form an enzyme-microbe complex to reduce the enzyme synthesis burden by taking advantage of enzyme-microbe synergy (55).

Acknowledgment

Y. H.P. Zhang wishes to thank the Biological Systems Engineering Department of Virginia Polytechnic Institute and State University. Funding for this work also came from the USDA — CSREES 2006-38909-03484 grant and the DOE BSE and NIST.

[1] A genetic screen led by Reiter et al. (125) identified the mur4 mutant in Arabidopsis that has a 50% reduction of arabinose in the wall. The encoded recombinant protein (Uxe1,

Homogalacturonan (HG) is the most abundant pectic polysaccharide. It is a homopolymer of 1,4-linked a-D-galactosyluronic acid that is partially methylesterified and may be O — acetylated at O-2 or O-3. As mentioned above, the distribution of methylesters in HG during synthesis or in the wall is not known (157), although there is evidence that the distribution of non-esterified galacturonosyl residues in the wall is not random (195). The degree of polymerization (DP) of HG, as well as the question of whether it is linear, or may be branched or cross-linked (204) remains a matter of debate. However, the DP has been estimated to range from 72 to 100 (304) or more (214). The synthesis of HG requires at least one HG:a1,4-galacturonosyltransferase (HG-GalAT) (Table 5.2), at least one HG- methyltransferase (HG-MT) (also referred to as pectin methyltransferase), and at least one HG:O-acetyltransferase (HG-AT).

5.4.6.1 HG:galacturonosyltransferase (HG:GalAT)

The only HG biosynthetic glycosyltransferase for which enzymatic function of the encoded gene has been established is the HG:GalAT called GAUT1 (Galacturonosyltransferase 1;

At3g61130) (137). The identification of GAUT1 followed an extensive study of HG:a1,4GalAT activity in multiple plant species including in mungbean (305-308), tomato (306), turnip (306), sycamore (309), tobacco suspension (310-312), radish roots (205), pea (230, 313), Azuki bean (Vigna angularis) (314), petunia (315), and Arabidopsis (137, 316) (see Table 5.4). This GalAT activity was shown to be particulate (i. e., membrane bound) in all species studied and was measured as the transfer of GalA from UDP-GalA onto endoge­nous acceptors. The GalAT activity in pea was localized to the Golgi (230) with its catalytic site facing the lumenal side of the Golgi (230), providing the first direct enzymatic evidence that the synthesis of HG occurs in the Golgi.

HG:GalAT activity in microsomal membranes, measured using UDP-[14C]GalA (317, 318), incorporates the GalA moiety onto endogenous acceptors to yield relatively large molecular mass labeled products of ~ 105 kDa in tobacco microsomal membranes (310) and >500 kDa in pea Golgi (230). Cleavage of the radiolabeled product into GalA, di — galacturonic acid (diGalA), and trigalacturonic acid (triGalA) by a purified endopolygalac — turonase demonstrated that the product was HG (310). In tobacco, the product produced in vitro in microsomes was ~50% esterified (310) while the product produced in pea Golgi was less esterified (230), suggesting that the degree of methyl esterification of newly synthesized HG may be species-specific and that methylesterification may occur after the synthesis of at least a short stretch of HG. GalAT activity can also be studied in detergent — permeabilized microsomes to allow access of the enzymes to exogenous pectic acceptors. For example, detergent-permeabilized microsomes from etiolated azuki bean seedlings trans­fer [14C]GalAfrom UDP-[14C]GalA onto acid-soluble polygalacturonate (PGA) exogenous acceptors (314). The azuki bean enzyme exhibited a broad pH range of 6.8-7.8 and a surprisingly high-specific activity of 1300-2000 pmol mg-1 min-1, considering the large amount (3.1-4.1 nmol mg-1 min-1) of polygalacturonase activity that was also present in the microsomal preparations.

Success in identifying the gene encoding a GalAT required solubilizing GalAT activity from membrane preparations so as to facilitate purification of the enzyme. The first solubilization of an HG:GalAT was achieved with tobacco GalAT (311). Detergent-solubilized GalAT adds GalA onto the non-reducing end (312) of exogenous HG with a preference for HG oligosaccharides (oligogalacturonides; OGAs) of a DP of greater than 9 (311,315), although OGA acceptors as small as a trimer can be used (315, 319). Although detergent-solubilized GalAT can use polymeric pectin substrates such a polygalacturonic acid and pectin, such polymers are less favorable substrates (314).

Studies carried out under conditions that provide information regarding the mode of elongation of the OGAs by GalAT, i. e., with excess OGA acceptor to UDP-GalA ratios (320), suggest that solubilized tobacco, radish and Arabidosis enzymes, and permeabilized pea Golgi galacturonosyltransferase, have a distributive (non-processive) mode of action in vitro (230, 311, 321). Under these conditions, the bulk of the HGA elongated in vitro by solubilized GalAT from tobacco membranes (311), or detergent-permeabilized Golgi from pea (230), is elongated by a single GalA residue. As expected, as the UDP-GalA:OGA ratio is increased, the OGA products become progressively longer (137), but it is important to note that this, in itself, does not denote processivity, it simply means that the enzyme can use the product of a previous catalytic event as a substrate for a subsequent catalytic event. Interestingly, the membrane-permeabilized galacturonosyltransferase activity reported from pumpkin may represent a processive mode of elongation since reactions containing approximately equimolar amounts of UDP-GalA and acceptor yielded a population OGAs elongated by up

to five galacturonosyl residues (319). However, further information on the size distribution of the OGAs produced in reactions containing excess acceptor will be required to confirm this. Solubilized petunia galacturonosyltransferase, in reactions containing ~60-fold excess UDP-GalA to OGA acceptor, added up to 27 galacturonosyl residues onto the OGA acceptors (315), indicating that the enzyme can elongate OGA products from a previous reaction, but not specifically addressing the mode of elongation of the enzyme (320). The apparent lack of in vitro processivity of the solubilized GalAT suggests either that enzyme does not synthesize HG in a processive manner in vivo, or that the characteristics of HG-GalAT measured in vitro may be an artifact due to the dissociation of a required biosynthetic complex or loss of cofactor(s) or substrate(s) during solubilization of the enzyme. We have been unable to obtain evidence for processive in vitro solubilized GalAT activity (i. e., the production of extensively elongated acceptors under reaction conditions with excess OGA acceptor to UDP-GalA) (Quigley and Mohnen, unpublished results) [see (192)]. We also obtained no evidence that the inclusion of the methyl donor S-adenosylmethionine (308, 310, 314) and/or the acetyl donor acetylCoA promote the processivity of GalAT (192) (unpublished results). Clarification of the mode of action of GalAT and the mechanism of HG synthesis should be aidedby access to purified or recombinantly expressed enzyme(s), and may require isolation of enzyme complexes (see below).

Efforts to purify GalAT activity from tobacco proved unsuccessful due to loss of activity. Therefore, a partial purification and tandem mass spectrometry approach was used to iden­tify a gene encoding GalAT activity. The detergent-solubilized HG:GalAT from Arabidopsis suspension cells was partially purified by column chromatography to yield an enriched fraction containing approximately 20 protein bands. The proteins were trypsinized, the polypeptides analyzed for amino acid sequence by tandem mass spectrometry (137), and the sequences compared against the Arabidopsis gene database. The partially purified active

GalAT fraction contained two proteins with sequences that identified them as putative glyco — syltransferases which were eventually named GAUT1 and GAUT7 (137). Evaluation of their amino acid sequences indicated that both GAUT1 and GAUT7 had characteristics consistent with the biochemical properties of HG:GalAT: a basic PI and an apparent Type II mem­brane protein topology. The encoded proteins consisted of three domains: a short N-terminal region, a single membrane spanning region, and a larger C-terminal domain. Transient ex­pression of N-terminal truncated forms of the coding regions of GAUT1 and GAUT7 in human HEK293 cells indicated that GAUT1 had GalAT activity. Thus, GAUT1 became the first enzymatically verified HG:galacturonosyltransferase (GAlacatU ronosylT ransferase 1, GAUT1). The transiently expressed truncated from of GAUT7 did not have GalAT activity.

Sequence comparison of GAUT1 against the Arabidopsis genome identified 24 additional Arabidopsis genes with high sequence similarity to GAUT1. We named this 25-member group of related genes the GAUT1-related gene family (Table 5.5). The GAUT1-related genes represent a subclass of the Arabidopsis CAZy family GT-8 genes. Fifteen of the GAUT1- related genes have high sequence similarity to GAUT1 (37-100% identity/56-100% simi­larity) and we named these genes GAUT1-GAUT15. The 15 GAUT proteins have predicted masses of 61-78 kDa and most encode proteins with a predicted membrane anchor (GAUTs 1,6-15) or with a signal peptide (GAUTs 3-5) consistent with a Type II membrane topology or with an intramicrosomal membrane location, respectively. GAUT2 is the only GAUT not predicted to be present in, or pass through, the intracellular membrane transport system (i. e., the ER/Golgi system). The remaining 10 GAUT1-related genes have somewhat lower sequence similarity to GAUT1 (39-44% identity/43-53% similarity) and we named these the GAUT1-Like (GATL) genes. The GATL genes encode predicted 33-44 kDa proteins with predicted signal peptides. The proven location of GAUT1 and GAUTs 3, 7, 8, and 9 is in the Golgi (323, 324).

Multiple T-DNA insert mutants of many of the GAUT and GATL genes are available (e. g., see http://signal. salk. edu/cgi-bin/tdnaexpress) and several mutants have described phenotypes. The qua1/gaut8 mutant has ~25% reduced amounts of GalA in the walls of Arabidopsis rosette leaves or total plants (135), ~30% reduced GalA levels in walls of stem (260), and modest reductions in the GalA content of suspension cultured cells (325), sugges­tive of a role of the mutated gene as a putative GalAT involved in pectin synthesis. However, the mutant walls are also reduced in Xyl (at least stem walls) and protein extracts from mu­tant stems have both reduced HG:a-1,4-GalAT activity and (3-1,4-xylosyltransferase activity (260). The lack of a confirmed enzyme activity of the recombinantly expressed or purified GAUT8 protein, along with the pleiotropic effects of the mutant, have made a conclusive identification of the function of the GAUT8 protein elusive.

The parvus/glzl/gatll mutants (136,326) also have characteristics consistent with a defect in pectin synthesis. The parvus/glzl/gatll mutants grown under low humidity are semi­sterile dwarfs that have reduced anther dehiscence. The parvus mutants also have slightly elevated Rha, Ara, and Gal and reduced Xyl compared to WT (136). These changes in neutral sugar compositions are consistent with a role of the parvus gene in the synthesis of the pectic polysaccharide RG-I. However, since the levels of GalA in the mutant walls were not determined, it is not known if the mutant walls are altered in GalA content.

We PCR-tested 57 Arabidopsis GAUT1-related gene family T-DNA insertion mutant lines from the T-DNA mutant collection (http://signal. salk. edu/cgi-bin/tdnaexpress) and identified 36 homozygous mutant lines (Caffall and Mohnen, unpublished). Glycosyl residue

composition analyses were carried out on amylase-treated walls from 82 unique tissue samples from homozygous mutants of 13 ofthe 15 GAUT family members by combined gas chromatography/mass spectrometry (GC/MS) of per-О-trimethylsilyl (TMS) derivatives of the monosaccharide methyl glycosides (327). Tissues from mutant lines of 8 of the 13 GAUT1-related family genes tested had >15% reduction in the amount of galacturonic acid in their walls, consistent with a function of the mutated genes as GalATs. However, a detailed biochemical analysis of the enzyme activity of each of the GAUTs will be required to establish gene function. Interestingly, some mutants also accumulated higher amounts of arabinose, rhamnose, fucose, and xylose and reduced amounts of mannose, galactose and/or glucose. The significance of these correlative changes with reduced levels of GalA is not yet clear; however, we speculate that the reduction of one pectic polysaccharide (e. g., HG) may be associated with, or compensated for, by a change in the amount of another polysaccharide (e. g., loss of HG may lead to an increase in RG-I which has Rha, Ara and/or Gal; alternatively, loss of an RG-I biosynthetic GalAT could lead to loss of RG-I and to an increase in HG with a resulting increase in GalA and a reduction in Rha, Ara and and/or Gal). Such a compensation of one wall polymer for another (328,329) could occur via several mechanisms including sensing of polysaccharide or nucleotide-sugar levels, or the covalent or non­covalent association of one type of polysaccharide with another (e. g., it has been reported that pectin may be covalently linked to the hemicellulose xyloglucan (107).

Although GAUT1 encodes a GalAT, its function in pectin synthesis is far from clear. For example, it is not yet understood at what stage of pectin synthesis, i. e. initiation or elongation, GAUT1 has its primary role. Also, it is not clear whether GAUT1 functions alone or in a complex. Our preliminary results suggest that, at least in vitro, GAUT1 can function in a complex with at least one other GAUT (Atmodjo and Mohnen, unpublished). The enzyme function of the other GAUT and GATL members of the GAUT1-related gene family remains to be determined. QUA1/GAUT8 is a good candidate for a GalAT, but the reduced levels of Xyl and xylosyltransferase activity of the qua1 mutants, at least in stem tissue, raises the question of what the connection is between pectin (e. g., HG) and hemicellulose (i. e., xylan) synthesis). The pectins and hemicelluloses are traditionally considered to be two different classes of wall polysaccharides, although there is some evidence that these two classes maybe tightly, and possibly covalently, linked in the wall (258, 332, 333). Thus, the characteristics of the qual mutant raise several questions. Is xylan synthesis dependent upon pectin synthesis? The characteristics of the parvus mutant and the GATL proteins are also intriguing. What is the function of the GATLs? The characteristics of the parvus mutant are consistent with a function in pectin synthesis; however, these proteins have no apparent membrane spanning domain but rather a signal peptide. If the GATLs are involved in pectin synthesis, do they interact with other lumenal pectin biosynthetic enzymes in the form of Golgi-localized complexes? The identification of GAUT1 as a functional GalAT and of the GAUT1-related gene family provides the gene/protein tools required to address some these questions.

Recently, characterization of mutants of GAUT12/IRX8 have provided evidence support­ing a possible role of GAUT12/IRX8 as a HG:GalAT involved in the synthesis of a subfraction of HG to which (3-1,4-xylan is attached (132) or as a putative a1,4-GalAT that adds a GalA onto a xylose at the reducing end of a xylan primer (129). Proof of the function of GAUT12, however, requires confirmation of enzyme activity.

UDP-GalA, a major sugar donor for pectin synthesis, is made by (i) the salvage pathway, by the phosphorylation of D-GalA to GalA-1-P in the presence of ATP and kinase activity (GalAK). The GalA-1-P pyrophosphorylase catalyzes the conversion of GalA1P and UTP to UDP-GalA. (ii) UDP-GlcA 4-epimerase (UGlcAE), a reversible 4-epimerase that converts UDP-GlcA to UDP-GalA.

1 Feeding experiments with radioactive GalA demonstrated that the label is readily incorpo­rated into pectin (459). Soluble enzyme preparations from a 1-day-old germinating seeds of mungbean have high GalAK activity (468), while no GalAK is observed in a 4-day-old etiolated seedlings (411). The collective biochemical data suggest that the soluble GalA kinase differs from the membrane-associated kinase activity that phosphorylates GlcA. A functional gene encoding GalAK activity was recently discovered in Arabidopsis [Ting and Bar-Pled, 2006, unpublished]. Proton NMR analysis confirm that pure reecombinant GalAK phosphoylates D-GalA in the presence of ATP to GalA-1-P. Currently however, it is difficult to predict the relative amount of GalA that is recycled back to UDP-GalA. Cleaarly, high activities of pectin derading enzymes during pollen germination could pro­vide suffcient substrate for AalAK. Indeed, the non-specific UDP-sugar pyrophosphorlase (fondly named SLOPPY in Arabidopsis (414) and also identified in pea (469)) that con­verts GalA-1-P with UTP to UDP-GalA is highly expressed during pollen germination. The relative amount of GalA recycled from pectins as free sugars in other plant tissues is unknown, although the “fate” of free GalA could be tissue specific. For example, during strawberry fruit maturation, an increase in GalA-reductase (GalUR) activity was shown to direct GalA, released from pectin, for the synthesis of ascorbic acid (470). Thus, the relative contribution of the GalA-salvage pathway for synthesis of wall polymers remains unclear. Hopefully, current analyses of GalAK mutants in Arabidopsis will shed light on its physiological functioon (i. e. in the wall and in ascorbic acid metabolism).

2 UDP-GlcA 4-epimerase (UGlcAE) is a membrane-bound enzyme that reversibly catalyzes the 4-epimerization of UDP-GlcA to UDP-GalA. In Arabidopsis, six distinct functional genes (UGlcAe) encode isoforms having UDP-GlcA 4-epimerase activity (450, 471, 472). The isoforms can be divided into three evolutionary clades: Type A, B, and C. Members of UGlcAE are predicted to be Type II membrane proteins, suggesting that their catalytic

domain faces the lumen of an endomembrane. The topology of the catalytic domain was not validated experimentally. Mohnen’s laboratory has shown that the activity of UGlcAE co-fractionated with the Golgi on sucrose-gradients (230) and expressing UGlcAE1-GFP in plants demonstrates that the chimeric fusion targets UGlcAEl to the Golgi as well (Gu and Bar-Peled, unpublished data). Multiple UGlcAE isoforms are common and found in other plants, for example, rice and maize. Biochemical analyses of Arabidopsis, rice, and maize UGlcAE isoforms showed that the reversible UGlcAE epimerase has a preference (2:1) to form UDP-GalA and is inhibited by both UDP-Ara and UDP-Xyl. Recent studies (Gu and coworkers, submitted) demonstrated that the maize UGlcAE is especially sensitive to, and strongly inhibited by, UDP-Xyl when compared with Arabidopsis UGlcAE2. The authors suggest that in maize, the relative low amount of pectin (when compared to dicots) could be due to inhibition of UGlcAE by UDP-Xyl. Hence, this may provide an explanation as to why maize has less pectin and more xylan when compared to dicots. The role of multiple UGlcAE isoforms is currently being addressed. One possible explanation is that different isoforms localize to distinct endomembranes, as was suggested by Pattathil and coworkers (473) for the different Uxs isoforms.

Both cinnamoyl CoA reductases (CCRs) and cinnamyl alcohol dehydrogenases (CADs) are substrate versatile NADPH-dependent enzymes. These proteins were detected and purified in the laboratories of Zenk and coworkers (118, 120, 123) in the early 1970s; CCR is a type-B reductase abstracting the 4-pro-S-hydrogen of NADPH (124), whereas CAD is a type-A reductase (4-pro-R-hydrogen abstraction) (123). As for 4CLs, all known CCRs and CADs are quite substrate versatile: CCR can catalyze the in vitro reduction of p-coumaroyl (14), caffeoyl (15), feruloyl (16), 5-hydroxyferuloyl (17), and sinapoyl (18) CoAs into the resulting aldehydes 19-23, and the first gene encoding a CCR was described in 1997 (125). CADs are able to catalyze the conversion of aldehydes 19-23 into the monolignols 1-5, with the first bona fide gene encoding a CAD reported by Knight et al. (126); earlier reports of CAD being cloned (127) later turned out to be malic enzyme (128, 129).

In the Arabidopsis Information Resource (TAIR), there were some 11 genes annotated as CCR/CCR-like, and another 17 as CAD/CAD-like (93). To date, only two CCRs have been demonstrated to have both this activity and physiological function (130-132). A similar situation exists for CADs: two of these (AtCAD4 and AtCAD5) are the most catalytically active, and another four (AtCAD2, 3, 7, and 8) have very low overall activities making their involvement in lignification/monolignol formation suspect (56); additionally, AtCAD1, 6, and 9 did not have any detectable CAD activities proper in vitro, and thus their biochemical function currently remains unknown (56).

The biochemical mechanism of CAD has also been the subject of considerable interest, and the X-ray crystal structure of AtCAD5 was recently described (133). Site-directed mu­tagenesis has implicated Glu70 in catalysis, the latter being coordinated with the catalytic Zn2+ through its side chain (Figure 7.9A). A proton shuttle mechanism for CAD has also been provisionally proposed (Figure 7.9B) but remains to be experimentally proven.

Sakakibara models

The 1970s and 1980s witnessed several additional attempts to adequately depict possible lignin structures. The most complex structure resulted from studies by Glasser and Glasser (13), using a computerized model called SIMREL. In hindsight, this model suffered from various critical limitations: For example, it was unable to predict the presence of various substructures, such as dibenzodioxicin (5-5′, 8-0-4′, 7-0-4′, substructure V in Figure 7.2D) moieties and it also contained a large number of hypothesized interunit linkages that have never been observed in lignins, e. g., C7-C8, C8-C7, C7-C6, and C7-0-C9. As for the Freudenberg and Forss/Fremer lignin models, hypothetical thioacidolytic cleavage of potential monomer/dimer releasable moieties in this representation would not provide quantitative data in agreement with later experimental observations, e. g., by thioacidolysis. This computer-generated model can, therefore, be eliminated from further consideration as adequately representing lignin structure.

At more or less the same time, a proposal for a representative lignin structure in the angiosperm, beech (Fagus silvatica), was made by Nimz (309). This model was based upon the analysis, identification, and quantification, of various monomeric, dimeric, trimeric, and tetrameric fragments released when beech woodmeal was treated with thioacetic acid and catalytic amounts of boron trifluoride at 20°C for 1 week, followed by hydrolysis of the thioacetates with NaOH (2 N, 60°C, 24 h) and Raney nickel treatment (8 h) (310). The proposed beech wood lignin structure (not shown) contained 27 inter-linked monomeric units of which, on a per monomer basis, there were 7 potentially cleavable G/S monomers, as well as 8-І7 (4), 8-5′ (2), 8-8′ (1.5), 7-8′ (1), 5-5′ (1), and 4- 0-5′ (0.4)/7-6/ (0.1) dimers. In hindsight, some of these fragments may have resulted from rearrangement, e. g., of sy — ringaresinol (substructure IIIC in Figure 7.2D) during the chemical degradative procedures employed, as observed for thioacidolysis cleavage/Raney Ni treatment where syringaresinol (70) can be converted into the tetrahydronaphthalene derivative 73 (Figure 7.17B) (217, 311) (Jourdes etal., unpublished results). Furthermore, the linkage frequencies in the Nimz model are not fully consistent with data obtained for other angiosperm lignins, such as with Arabidopsis and alfalfa, this perhaps being due to low chromatographic recoveries of various products, etc. Other linkages in the Nimz model, such as 7-8′, are also not known to be present, and may thus potentially represent artifacts as well. As before for the other lignin structures discussed, various other known subunit structures, such as 5-5′, 8- O-4′, 7-0-4′-dibenzodioxocin, were absent from this model.

Two other “representative” structures for gymnosperm lignins were reported by Adler (312) and by Sakakibara (313). The first (not shown) contained a proposed spruce lignin depiction of 16 aromatic monomer (H/G) units, with 5 potential thioacidolysis releasable monomers, including one again being the unlikely sinapyl alcohol (5). Other linkages in­cluded thioacidolysis cleavable 8-5′ (1), 5-5′ (1), and 8-6′ (1) dimeric entities, with the remaining five monomers envisaged to be linked together via 8-1′,4- O-5′, 8-8′, and 5-5′ in­terunit linkages. Sakakibara (313), by contrast, proposed another “representative” softwood (gymnosperm) lignin structure (also not shown), based upon hydrolysis and hydrogenolysis analyses. This contained 35 monomeric aromatic units which could be potentially cleaved (e. g., by thioacidolysis) to afford G/S monomers (7), 8-1′ dimers (3), 8-5′ dimers (3), 8-8′ dimers (1), trimers (3) consisting of 5-574-0-5′, 8-875-5′, and 4-0-178-5′ linkages, as well as a proposed five-unit (8-5′, 8-5′, 5-5′, 8-8′) linked moiety. None of these proposed structures, however, again apparently adequately account for gymnosperm lignins, in terms of interunit type and frequency. Nor did they contain the additional substructures more recently discovered, such as the 5-5′, 8-O-4′, 7-0-4′-dibenzodioxocin (V, Figure 7.2D). Thus, while recognizing all of these to be valuable studies, they fell short of being adequate representations of native lignin structure(s).

Xianghong Qian and Mark R. Nimlos

9.1 Background

With the depletion of fossil fuels and the increase in oil prices, biofuels have become ever more attractive as an alternative source of energy. The US Department of Energy (DOE) recently released its roadmap aiming to reach the goal of supplying 30% of the US motor vehicle gasoline from cellulosic ethanol (1). In addition, the European Union has a plan to produce 25% of its transportation fuels from biofuels (2). Toward these goals, significant progress has been made in biomass conversion of cellulose to fermentable sugars. Many schemes for the utilization of biomass as a source of renewable fuels and chemicals rely upon the ability to deconstruct the polysaccharides in plant cell walls into constituent sugars.

Acid hydrolysis is commonly used during pretreatment to break down the structures of plant cell walls and prepare plant polysaccharides for hydrolysis using biological catalysts. In this step, hemicellulose is hydrolyzed to monomeric sugars, the majority of which are pentosans, such as p-D-xylose. Depending on the severity (temperature and acidity) of this pretreatment process, some xylose molecules undergo an undesirable dehydration process, thus lowering the biomass conversion efficiency. In addition, the main degradation product, furfural, is a toxin for fermentative organisms (3, 4) and can polymerize to reduce access to other polysaccharides, such as cellulose. Thus, dehydration reactions in acid pretreatment of biomass present a barrier to economical conversion of biomass. In the past, sugar yield and acidic sugar degradation products were found to be strongly dependent upon the reactor configuration, the reaction media, and the reaction temperature (5-9) during dilute acid hydrolysis.

Xylan is a prevalent hemicellulose in many sources of biomass, and for example, makes up roughly 20% of corn stover (10). The polymer backbone of xylan is composed of xy­lose monomers joined by p-1,4 ether linkages, which hydrolyze to form xylo-oligomers and eventually xylose. Earlier studies (11) show a biphasic behavior of xylan hydrolysis, a fast breakdown of xylan to xylo-oligomer followed by a slow depolymerization of the residual xylan. The reasons for this biphasic behavior remain elusive. The hydrolysis of low degree of polymerization (DP) xylo-oligomer to xylose is typically fast (11). Reaction (9.1) shows the hydrolysis of a backbone section of xylan by an SN1 water substitution

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

mechanism. The mechanism involves addition of a proton to the oxygen atom of the ether linkage, which leads to the dissociation of the polymer to form a positively charged oxo — nium ion and a shorter xylan chain. A nearby water molecule quickly interacts with the oxonium ion forming a neutral xylan chain. The extra proton from the water molecule is recycled back into the solution by forming a hydronium ion with the surrounding water molecules. The stability of the oxonium ion suggests fast kinetics for the first half of the hydration reaction. As a result, xylo-oligomers and xylose should be released quickly during the acid pretreatment of xylan and biomass as long as protons are readily available and eas­ily transferred to the ether bond. The rate of protonation of the ether linkage is unknown. Most likely, it will depend on the macroscopic acid concentration and microscopic atomic and molecular environment, which may hinder/promote proton transport to the ether linkage.

The mechanisms of the dehydration or degradation reactions that lead to the destruction of sugar molecules have been less clearly understood. The decomposition of xylose in acidic aqueous solutions has been the focus of a number of studies dating back (12) to the 1930s. Furfural is found to be the main product from the decomposition reactions and this process is used in the industrial production of furfural from oat hulls. The mechanism for furfural formation was initially proposed (13-15) to occur via the open chain form of the sugar structure as is shown in (9.2). Even though xylose molecules have a predominant ring structure in water, the dehydration of the open chain form to furfural drives the equilibrium to the right. Moreover, the protonation of the ring oxygen opens up the ring structure. Degradation proceeds via the elimination of the two water molecules and eventual closure of the open chain to form a furan ring. No intermediate has been detected experimentally to validate this reaction mechanism.

Antal and coworkers (16) proposed another mechanism for xylose degradation to furfural via direct conversion from the six-carbon pyranose ring structure to the five-carbon furan ring structure. The existence of this mechanism was confirmed by recent ab initio molecular dynamics simulations and quantum mechanical calculations (17, 18). Two mechanisms, (9.3) and (9.4), were proposed (16) that involve direct rearrangement of the cyclic ring structure after the protonation ofthe hydroxyl groups and loss ofwater. Further elimination of the water molecules leads to the formation of furfural. The kinetics of furfural formation from xylose in acid solutions has been measured (19) and the reported activation energy is about 32 kcal mol-1. Direct dehydration or degradation of xylan has not been reported, though degradation ofxylo-oligomer, particularly xylotriose has been reported when certain inorganic salts are added to the solution (9). It is likely that these processes also have relatively high activation energies. Since hydrolysis is facile if protons are readily available for the (3-1,4 ether linkages, and dehydration reactions have high activation energies, it is theoretically possible to obtain high yields of xylose during dilute acid pretreatment of xylan and biomass.

However, the low yields of xylose (60-65%) (20-22), and high level of furfural formation (15%), suggest that more needs to be learned about the mechanisms and kinetics of hydrolysis

There is good evidence that the majority of plant cell-wall-degrading enzymes in R. flavefa­ciens are retained on the bacterial cell surface via a cellulosome-type multienzyme complex. The assembly of this complex depends on the specific interaction of dockerin and cohesin domains within the component proteins. The known structural components of this com­plex are encoded by the sca gene cluster, which was first described in R. flavefaciens 17 (10). A very similar gene cluster has now also been identified from the partial genome sequence of R. flavefaciens FD1 (11). In R. flavefaciens 17, the scaffolding protein ScaA carries three cohesins and a C-terminal dockerin that in turn binds to any one of seven cohesins present in the larger noncatalytic protein ScaB (12). Recent work has shown that the C-terminus of ScaB also contains an unusual type of dockerin domain that interacts with a single cohesin present in the small, cell-surface-anchored protein, ScaE (13). ScaE is covalently bound to the peptidoglycan of the cell surface at its C-terminus via a sortase-mediated mechanism (Figure 12.1). Meanwhile, a second small protein encoded by the sca cluster, ScaC, possesses a dockerin that binds to ScaA, but carries a single, distinctive cohesin. This has been de­scribed as an adaptor, because it recognizes a different type of dockerin from that recognized by ScaA cohesins (14). Dockerin-carrying enzymes from R. flavefaciens 17 have so far been shown to bind mainly to ScaA cohesins, although some divergent enzyme dockerins such as those from CesA (Ce3B) and XynE (Xyn11E) have distinct binding specificities for which the corresponding cohesin partners remain unknown (12, 15, 16).

The proteins encoded by the homologous sca cluster of R. flavefaciens FD1 differ in a few interesting respects from their homologues in R. flavefaciens 17. ScaAFD1 carries only two cohesins, while ScaBFDi carries nine, five homologous to those of ScaBi7, and four closer to those of ScaA17. Thus, ScaBFD1 is predicted to be able to bind four enzyme subunits directly, and 10 more via attached ScaA molecules, while ScaB17 is predicted to bind 21 enzyme subunits via ScaA molecules (11). It is not yet clear whether these differences have functional consequences. Significant strain diversity exists in R. flavefaciens both at the genetic level and at the level of plant cell wall degradation (17, 18).

Most plant cell wall hydrolases studied from R. flavefaciens 17 have a dockerin-like module located either at the C-terminus, or internally (15,19). One non-dockerin containing extra­cellular hydrolase is, however, known that carries family 11 and 10 xylanases domains at its N and C terminus, respectively, connected by an unusual NQ-rich linker (20). In addition to at least three types of enzyme dockerin, the dockerins found in the structural proteins ScaA

and ScaB are also distinct in their sequences and binding characteristics, suggesting that there are at least five different dockerin specificities in this strain (21). Meanwhile, genome sequencing of R. flavefaciens FD1 has revealed at least 180 polypeptides that carry dockerin sequences (22). The sequences of these dockerins fall into a number of branches in phy­logenetic analyses. Fewer than 50% of these carry glycoside hydrolase domains, and many carry proteinase domains or domains of unknown function. It remains unclear whether all these domains are required primarily for plant cell wall breakdown. Proteinases may well be involved in removing structural proteins in the plant cell wall, but it is also possible that dockerin-cohesin interactions mediate the assembly of protein complexes that play other roles on the bacterial cell surface.

Uniquely among known cellulosome systems, the R. flavefaciens scaffolding proteins so far identified carry no identified carbohydrate-binding modules. Carbohydrate-binding modules occur in many enzyme subunits, including a module in the abundant cellulase EndB (Cel44A) that represents a new CBM family. The gene that precedes ScaE, however, is now known to encode a protein (CttA) that is attached to the bacterial cell surface via ScaE, and binds crystalline cellulose (23). CttA is, therefore, a strong candidate for a carbohydrate-binding protein that may play a key role in positioning cellulosomal enzymes close to their substrates. The possible role played in adhesion by the yellow affinity substance of R. flavefaciens (24) remains unclear.

Microorganisms involved in the decay of plant biomass use an array of biochemical processes to mediate conversion. They can be categorized either genetically using molecular biology tools and/or biochemically, with emphasis on the types ofhydrolytic enzymes they produce. The biodegradation of complex polysaccharides is a multistep, hierarchical process. It is hierarchical in two senses. In general, removal of sugar modifications makes subsequent attack of main chain-degrading enzymes much more effective. Secondly, in nature there is a hierarchy of polymer degradation, from the least to the most recalcitrant compounds. During the early stages, the rate of decay is determined by the accessibility of the more easily decomposed carbohydrates. In general, pectins are degraded first, followed by hemicellulose, and then finally cellulose and lignin. This results in a succession of microorganisms that are able to assimilate specific substrates.

Hydrolytic enzymes capable ofbreaking down cellulose and hemicellulose (complex poly­mers of arabinose, mannose, glucose, and xylose) are widespread and are produced by a wide range of microorganism that include fungi, actinobacteria, Clostridiaceae, and mem­bers of the a-, (3-, and у -proteobacteria. The efficient conversion of plant polysaccharides is achieved by an array of enzymes with varying but synergistic specificities and activities. The diversity of these enzymes in nature is illustrated by the growing number of entries of hydrolytic enzymes in glycosyl hydrolase-specific databases such as the Carbohydrate-Active enzymes (CAZy) database (5). With the increasing number of sequenced genomes available it was recently estimated that over 12 000 glycosyltransferase and glycoside hydrolase open reading frames will have been added to the database during 2006 (6).

A second and more narrowly defined functional group includes the white rot Basidiomy — cota and xylariaceous Ascomycota, which are the primary agents of lignin degradation by secretion of nonspecific, extracellular enzymes that modify the lignin macromolecule. Lignin is a polyphenolic heteropolymer that is interconnected by stable ether and carbon-carbon bonds making it extremely resistant to microbial attack. Lignin degradation is carried out by the action of lignin modifying enzymes that are produced primarily by the white rot fungi. Lignin breakdown is an obligate aerobic reaction carried out by oxidative ligninolytic enzymes that remove various functional groups, side chains, and aromatic rings randomly from the lignin macromolecule. The initial steps often involve O2- or H2O2-dependent, lignin peroxidase, Mn peroxidase, or laccase. Frequently, flavin-dependent enzymes supply H2O2 to the lignin peroxidase or Mn peroxidase (7, 8). The lignin polymer is reduced to smaller, low molecular weight fragments that are then available for further decomposition by either fungi or bacteria.

Historically, microbiologists have defined decay communities based on individual isolated bacteria and on the basis of their physiologic and nutrition requirements. In general, classical isolation techniques require substantial knowledge of the organism’s habitat so that it can be reproduced and one population can be enriched over others. Within a small particle of biomass several different microenvironments can exist differing both chemically and physically. Physiochemical conditions, such as oxygen concentration change both temporally and spatially during decay. In many cases, individual bacteria cannot be isolated due to obligate interdependence upon other bacteria for growth that must be supplied using co­cultures of helper bacteria. Microorganisms may also be slow — growing, or obligate anaerobes requiring patience and tenacity to isolate and characterize. With the advent of new molecular tools for describing microbial communities, it has become clear that bacteria grown isolated in “pure culture” underrepresent the microbial diversity in most environments and are generally only a minor component.

Covalent associations between lignin and wall proteins including GRP and HRGP have been proposed from analysis of fractions and immunocytochemical and recombinant DNA studies on secondarily thickened cell walls of mature and developing woods and legumes [reviewed by Lam and coworkers (127)] but the linkages have not been chemically characterized.

4.3.2.2 Suberin: a polyaliphatic-polyphenolic association

Some plant cells such as cottonseed hairs and the cork cells of the periderm, which forms bark, develop a suberin layer between the primary wall and the plasma membrane prior to secondary wall formation. Suberin has a polyaliphatic domain and a unique polyphenolic domain containing hydroxycinnamates (128, 129).

4.3.2.3 Cutin: an aliphatic polyester

The outer wall of the epidermal cells of stems and leaves has a multilayered cuticle, which includes a thick cuticular layer composed of cutin, a high molecular weight polyester of C16 and C18 hydroxy fatty acids which may be covalently bound to polysaccharides that form part of the underlying cell wall (130, 131). A second component of the cuticle, cutan, is a non-hydrolysable, aliphatic biopolymer believed to be composed of polyunsaturated fatty acids joined by ether bonds (132). Waxes overlie the cuticle surface.

It is not known whether RG-I is synthesized onto existing HG or rather is synthesized independent of HG. If it is synthesized onto HG, it not known whether GAUT1 is that GalAT responsible for synthesizing the HG backbone region which would serve as a primer for RG-I backbone synthesis, or whether other GAUTs or other enzymes would perform this function.

It is also not known whether the alternating [^4)-a-D-GalpA-(1^2)-a-L-Rhap-(1^] backbone repeat is synthesized by a single glycosyltransferase containing both RhaT and GalAT activity, or whether, alternatively, the RG-I backbone is synthesized by a protein complex containing both a GalAT and a RhaT. If the backbone is synthesized by a complex, it is also not known whether GAUT1 or one or more of the GAUT1-related gene family members are part of the complex. To date, no RG-I-specific GalAT or RhaT has been reported.

5.4.10.1 RG-I:galactosyltransferase (RG-I:GalT)

RG-I synthesis requires at least eight galactosyltransferase (GalT) activities (Table 5.2). Probable p-1,4-GalT and p-1,3-GalT activities were originally identified in studies of mi­crosomal preparations from mung bean (375, 376) and more recently a mung bean p-

1,4- galactosyltranferase activity with a pH optimum of 6.5 was confirmed (377). Multiple galactosyltransferase (GalTs) activities have also been reported in particulate homogenates (378, 379) and solubilized enzyme (380) from flax (Linum usitatissimum L.). Detergent- solubilized flax microsomal GalTs transferred [3H]Gal from UDP-[3H]Gal onto exogenous RG-I-enriched and pectic p-1,4-galactan acceptors (381) to yield high molecular mass radiolabeled products. Surprisingly, the pH optimum for transfer onto lupin pectic p 1,4- galactan (i. e., pH 6.5) was different than the pH optimum for transfer of Gal onto an endopolygalacturonase-treated RG-I-enriched fractions from flax (i. e., two optima: pH 6.5 and 8.0) (381). Analysis of the products using RG-I-specific enzymes confirmed that the GalTs indeed added Gal onto RG-I (381), and thus, represented RG-I:GalTs. Furthermore, fragmentation of at least part of the product with p-1,4-endogalactanase demonstrated that at least some of the GalT activity represented p-1,4-galactosyltransferase (381). At pH 8, the GalT activity had an apparent Km of 460 ^M for UDP-Gal and characteristics consistent with a function in catalyzing the addition of galactose onto short galactan side branches of RG-I.

Microsomal membranes from potato suspension cultured cells have been shown to contain RG-I:p-1,4-galactosyltransferase activities that both initiate and elongate p-1,4- galactan side chains of RG-I (382). The potato RG-I:p-1,4-GalT activity in microsomal membranes had a pH optimum of 6.0-6.5 and produced a >500 kDa product using UDP — [14C]Gal and endogenous acceptor(s) in microsomal membranes. The product was frag­mented by endo-p-1,4-galactanase into [14C]Gal and [14C]galactobiose and into radiola­beled fragments between 50 and 180 kDa in size (382) when treated with the RG-I-specific rhamnogalacturonase A, an endohydrolase that cleaves the glycosydic linkage between the GalA and Rha in the RG-I backbone (383). The GalT activities in the microsomal membranes could be solubilized from the membranes using detergent (382) and the solubilized enzyme fraction was shown to contain at least two distinct GalT activities, one with a pH optimum of 5.6 that preferentially added Gal onto an ~1.2-MDa RG-I acceptor with a mole % Gal/Rha ratio of 0.7; and the other with a pH optimum of 7.5 that preferentially added Gal onto a 21-kDa RG-I acceptor with a mole % Gal/Rha ratio of 1.2. Neither activity could use RG-I acceptor containing lower Gal/Rha ratios, RG-I backbone without side chains, or galac­tan polymers or oligomers as acceptors, suggesting that the activities identified required recognition of the RG-I backbone and some existing Gal in a side chain (384). Interest­ingly, only the product synthesized onto the 21-kDa RG-I acceptor was digestible with a

1.4- p-endogalactase, suggesting that either the Gal transferred onto the larger RG-I acceptor was of a linkage other than (3,1-4, or that the length of the galactan side chain synthesized was less than three, the minimum size recognized by the 1,4-p-endogalactanase. The RG-I:p-

1.4- GalT that elongates the (3-1,4-side chains of RG-I was shown, by subcellular organelle fractionation and protease sensitivity experiments, to be a Golgi-resident protein with its catalytic site facing the lumen of the Golgi (385), a location consistent with its role in pectin synthesis. No gene for any RG-I galactosyltransferase has been reported.

More recently, (3-1,4-GalT activity in mung bean detergent-treated microsomal mem­branes was identified that transferred up to eight galactosyl residues in a (3 -1,4-linkage onto the non-reducing end fluorescently labeled exogenous (1^4)-p-galactooligosaccharide ac­ceptors (386, 387). Of the galactooligosaccharide acceptors used, i. e., degree of polymeriza­tion (DP) of 3-7, the galactoheptaose was a most effective acceptor although acceptors of DP 4-6 also functioned. The fluorescently labeled trimer was not active. Interestingly, fluo­rescently labeled RG-I backbone oligosaccharides of DP 5-7 were also not active, suggesting that the GalT activity identified could not add Gal onto oligosaccharide RG-I backbone regions. The (3-1,4-galactan:p-1,4-GalT activity had a pH optimum of 6.5 and apparent Km of 32 pM for UDP-Gal and 20 pM for the fluorescently labeled galactoheptaose (386).