Recent work in Tadashi Ishi’s laboratory established a new enzyme activity that was never re­ported before, a UDP-arabinopyranose mutase (UAM). The enzyme is capable of converting UDP-Ara pyranose to UDP-Ara furanose (UDP-Ara /) and it was reported to be a reversible reaction (393). This activity was biochemically purified from rice and the corresponding gene was cloned. In rice, two homologous proteins (UAM1; AK098933, UAM2; AK071012) were identified. The recombinant enzyme at thermodynamic equilibrium produces UDP-Ara in a pyranose:furanose ratio of 90:10. The Arabidopsis homologous proteins were initially named as reversibly glycosylated protein, RGP (396, 484). Several isoforms of RGPs exist in the plant kingdom. The Arabidopsis RGP1 (At3g02230) and RGP2 (At5g15650) proteins were found to localize in the Golgi apparatus (396, 483). However, Sagi and coworkers (484) observed that a chimeric RGP-tagged to green fluorescence protein is localized to both Golgi and plasmodesmata. The specific in vivo role of RGP/UAM and the various isoforms in the (i) synthesis of the furanose form of UDP-Ara or as (ii) a reversibly glycosylated protein remains unclear.

The monolignol pathway [i. e., fromPhe (6) formation to themonolignols 1-5 and ultimately lignin] has received recent growing interest as regards the extent of gene families/metabolic networks associated with lignification and other monolignol 1-5 related metabolism. This could thus now be fully investigated given that the Arabidopsis genome was sequenced in its entirety in 2000 (79). Furthermore, since several of the enzymatic steps [e. g., to cinnamic acid (8) (via PAL)] can also involve other unrelated phenylpropanoid-forming processes (such as to flavonoids, suberins, and other non-monolignol metabolites), it was therefore useful to determine to what extent there were functionally redundant metabolic networks and/or nodes within same that were monolignol/lignin-specific. This was nec­essary to eventually understand how these processes were (constitutively) occurring in a coordinated manner with other biochemical systems (e. g., to the cell wall biopolymers, cellulose, hemicelluloses, etc.), including identification of the various transcription factors involved.

In this regard, although possible gene families have been provisionally annotated in the TAIR database (e. g., as COMTs, CADs, etc.) (93), it became clear early on in our studies (56, 94, 122) that unambiguous biochemical/physiological functional determinations for each candidate gene (mutant) were urgently needed. Moreover, given, for example, the substrate versatility of various dehydrogenases, comparative kinetic data was also required. Thus, each of the biochemical steps discussed below had to be reconsidered in terms of the above, with particular attention being given to database (mis)annotations.

In apparent agreement with previously deduced regulatory roles for carbon allocation to the monolignol-forming pathway for C4H and pC3H (together with their associated reductases and HCT) described earlier, each exists in Arabidopsis as single gene. Moreover, while as discussed earlier, F5H does not appear to affect overall carbon allocation (77), it nevertheless serves as a “switch” between G and S lignin formation, and it is also part of a very small gene family (~2). Interestingly, until recently, enzymes such as C4H were considered to be of very limited substrate versatility. On the other hand, a recent study using recombinant C4H protein from Arabidopsis described a broader substrate versatility than previously recognized — albeit with unnatural substrates (146).

By contrast, the two preceding biosynthetic steps — leading to arogenate (37) and Phe (6) formation, are encoded by multi-gene families (e. g., 6 genes for the ADT and 4 for the PAL families in Arabidopsis). Furthermore, at least for the PAL gene family, the correspond­ing genes have quite complex patterns of expression [as demonstrated by GUS-reporter promoter fusion analyses (Kim etal., manuscript in preparation)]. That is, the level of co­expression in various tissues and organs is indicative of a partially overlapping, functionally redundant, metabolic network. This is anticipated to be the case for the ADT gene family as well. Both enzymatic steps also have quite strict substrate specificities, at least for the known/tested substrates employed to date.

The remaining downstream enzymatic steps (i. e., 4CLs, CCOMTs, COMTs, CCRs, and CADs) in the monolignol/lignin-forming pathway are — as originally/currently annotated in the TAIR database — members of rather large gene families, i. e., original annotations were 4CL (14 genes), CCR (12), CAD (17), CCOMT (5), and COMT (17), respectively (93). However, this is, in fact, a considerable overestimation as discussed below.

As indicated earlier, 4CL corresponds (at most) to a four-member gene family (122), and as for PAL, the GUS-promoter fusion studies for each display a complex, partially overlapping, metabolic network (Kim et al., manuscript in preparation), i. e., in agreement with the different phenylpropanoid pathway products (from flavonoids to lignins) being formed. Additionally, there appear to be only 2-3 CCRs (130, 131) (Cochrane et al., unpublished results) and predominantly 2 CADs (56,57,71) in Arabidopsis — at least as far aslignificationis concerned. Interestingly, just one of the 17 putative COMTs has the requisite catalytic activity in vitro (Zhang et al., manuscript in preparation) for the monolignol-forming pathway, and the CCOMTs also appear to only have at most 2 genes associated with monolignol/lignin deposition (Takahashi et al., manuscript in preparation). As for PAL and other genes, the constitutive patterns of gene expression have also been established, thus providing the basis for probing/investigating the nature of the metabolic networks operative at all stages of plant growth and development.

Taken together, the entire suite of genes excluding transcription factors possibly involved in monolignol/lignin deposition — at least to the monolignols — accounts for <25 genes from prephenic acid (36) onwards, rather than the more than 10-fold higher number originally annotated.

The opportunities and challenges that now remain in further understanding monolig — nol/lignin pathway metabolism are to complete the full biochemical/physiological charac­terization of these different gene family members and to establish the “rules” for overall pathway induction/deployment — particularly since the pathway genes are “scattered” over different chromosomes. This potentially appears to be a very straightforward task, given that each of the members has now been identified — at least through their in vitro biochem­ical conversions. The data also underscore the need for circumspection when considering database annotations: while useful repositories, these cannot be considered definitive in any way until the physiological and biochemical functions are unambiguously determined.

Beginning with the studies of Freudenberg and others, it was initially assumed that lignins were three-dimensional cross-linked biopolymers, albeit also in the absence of any experi­mental data in support of this assumption. Such assumptions, however, did not survive fur­ther experimental scrutiny either. In a study by Dolk etal. (316), using kraft lignin derivatives, it was estimated that at best there was less than one cross-link per 19 monomeric units. Later studies by Mlynar and Sarkanen (317) and by Smith etal. (318), using ultracentrifugation/ size exclusion chromatography and modeling of delignification with a computer program (SIMREL), respectively, also concluded that there were no detectable cross-links. Moreover, other studies probing lignin biophysical properties in situ have indicated that lignins were more akin to linear-like polymeric entities (319). Interestingly, Forss and Fremer (308) had underscored many of the experimental shortcomings in the proposed Freudenberg three­dimensional cross-linked structure which went unanswered. It is thus quite bewildering why many researchers have arbitrarily dismissed the concept of regular structure in lignins without the appropriate experiments having even been either designed and/or conducted. Yet this incorrect notion of three-dimensional cross-linked polymer for lignins, however, exists even today in many literature contributions.

Lignins also display a profound tendency to associate, as demonstrated, for example, in the study of kraft lignins (8). Specifically, it has been reported that kraft lignins — while chemically modified but which nevertheless are presumed to retain important vestiges of native lignin structure — contain discrete molecular entities. These can associate with one another to form multimodal distributions of interconvertible supramolecular complexes estimated to be comprised of 103-104 individual species. Such selectivity is considered to result from association of well-defined (regular) structures in the participating polymer chains (320). This presumably would not be expected if random (1066/100-mer) and/or combinatorial chemistry was occurring.

In presumed agreement with a concept of defined template polymerization, one-electron oxidation of coniferyl alcohol (3) moieties in open solution in the presence of a lignin template apparently engendered preferential formation of high molecular weight species in preliminary studies. In this regard, Sarkanen etal. (50, 51) reported the effects of polymer­ization of coniferyl alcohol (3) in the presence of peroxidase/H2O2 and a methylated “kraft” lignin preparation (Mw ~15 400,2.7 x 10-8 M initial concentration). In the presence of the putative template, higher molecular weight entities were reportedly preferentially formed as shown by the elution profiles (Figure 7.18A), and to a much lower extent in absence of the template (Figure 7.18B). These researchers interpreted these findings as due to template poly­merization effects, where the preformed (methylated kraft) lignin macromolecule assumed the role of a progenitorial template in vitro. That is, monolignol radicals were considered to be positioned on adjacent loci of the template with new interunit linkages determined by either the corresponding substructure in the lignin template chain and/or the chemical nature of the monolignol (radical) species aligned on the template. Work has, however, not yet been described as to the chemical (subunit) nature of these interunit linkages and how/if the fidelity of the replication process is maintained relative to that of the macromolecular template itself. Nevertheless, the presumed ability to polymerize monolignol radicals on a preexisting lignin template needs to be considered as regards possible relevance to cell wall assembly mechanisms.

Dehydration of xylose was initially investigated by modeling the reactions of protonated p- D-xylose in vacuum (17,18). The absence of solvent water molecules allows the study of the intrinsic dehydration mechanisms, without the interference ofthe solvent. Furthermore, the quantum mechanical calculations without solvent molecules are more tractable, particularly for static calculations to obtain the energetics of the reactants, intermediates, and products. In addition, the knowledge of the intrinsic reaction in vacuum is often necessary to model solvated systems.

The simulations started when a proton was added to each of the four hydroxyl groups or the ring oxygen on the xylose ring structure. The subsequent reactions in vacuum were followed using ab initio molecular dynamics simulations. The results of these simulations indicate that protonation at O1 and O4 does not lead to any observable xylose degradation during the course of the 2 ps simulations time. Protonation at O5 results in a reversible ring opening and closing reaction. Protonation at O2 results in irreversible dehydration and degradation reac­tion to form furfural. Protonation at O3 leads to the fragmentation of a xylose molecule to a one-carbon (formic acid) and a four-carbon product. Formic acid has been observed exper­imentally. Mechanism 3, which results from the protonation at O1, was not observed during ab initio MD simulations. Static electronic structure calculations confirmed these simula­tion results and determined the likely intermediates in the mechanism of furfural formation. Furthermore, these calculations demonstrated that mechanism (9.4) is the most likely route to furfural formation based upon the calculated activation energies of the reaction steps. Figure 9.1 shows an overview of the reactions of these five isomers of a protonated xylose molecule.

Molecular dynamics simulations starting with protonated xylose provide an indication of the reactivity of that protonated isomer. The results of molecular dynamics simulations for protonation (17) at O2 and O3 are shown in Figures 9.2 and 9.3. The colored pictures at the top of these figures show snap shots of the progress of the reaction as a function of simulation time. Lewis structures of the progress of the reactions are shown at the bottom of the figures to provide clarity. As can be seen in Figure 9.2, xylose protonated at O2 dehydrates and rearranges to a furanyl compound within 700 fs. This mechanism is identical to reaction mechanism (9.4) for the formation of furfural proposed by Shafizadeh (41). As shown in Figure 9.3, xylose protonated at O3 dehydrates and breaks the C1-C2 bond within 250 fs. The resulting ether intermediate readily breaks apart to form formic acid and a 1,4-diol. This mechanism accounts for the formation of formic acid that has been reported in the literature (42). The diol that is formed as a by-product in this reaction is likely reactive and may polymerize to form a resin that is commonly found during acid degradation of

Figure 9.1 Scheme tor reactions of protonated xylose as determined by quantum mechanical modeling (CPMD and Gaussian). Protonation at 02 leads to the formation of furfural while protonation at 03 leads to formic acid. Protonations at 01 and 04 lead to dehydration products that readily recombine with water to reform xylose. Protonation at 05 leads to ring opening, which readily re-closes.

Figure 9.2 Results of CPMD simulation of xylose degradation after protonation of the hydroxyl group on O2. After 125 fs xylose is dehydrated and at approximately 659 fs the remaining carbocation rearranges to form the dehydrated furanyl form of xylose. This product will need to undergo two additional dehydrations to form furfural. (Reproduced in color as Plate 24.) xylose. Early reports (42) suggest that the formation of resin is accompanied by formic acid formation.

Molecular dynamics simulations of xyloses protonated at the other oxygen atoms were run well beyond 1 ps and showed no net reaction. MD simulations of xylose protonated at O1 did not also lead to the formation of the furanyl product as proposed in reaction

Figure 9.3 Results of CPMD simulation of xylose degradation after protonation of the hydroxyl group on O3. After 25 fs xylose is dehydrated and at approximately 250 fs the C1-C2 bond has broken. In this product the C1-O5 will eventually break to yield formic acid as shown in the Lewis structure at the end. (Reproduced in color as Plate 25.)

mechanism (9.2). These simulations showed that protonation at O1 leads to dehydration (see Figure 9.1) to form the oxonium, which did not react further. This ion will most likely react quickly with solvent water molecules to reform xylose. Simulations show that protonation and dehydration of xylose at O4 leads to the formation of the species depicted in Figure 9.1 containing a three-atom ring. This species does not react further, but is likely to recombine with water to reform xylose. Of particular interest is the mechanism for the formation of furfural from the open form of xylose, reaction mechanism (9.2). Molecular dynamics simulations show that protonation of O5 leads to the open form of the sugar, which quickly reverts to the cyclic form.

Static electronic structure calculations (CBS-QB3) were used to determine the energetics of the reaction pathways (18) shown in Figure 9.1. The proton affinities, or the gas phase enthalpies to add a proton, provide an indication of the likelihood of the proton to attack the oxygen atoms on xylose. The proton affinities are found to have the following order: O2 > O5 > O3 > O4 > O1 with values shown in Table 9.1. These values suggest that O2 has the highest proton affinity and is more susceptible to protonation, while O1 is least susceptible to protonation. These results show that the proton has a preference for O2, which leads to the formation of furfural. This is consistent with the experimental obser­vation that acid treatment of xylose leads primarily to furfural. Static electronic structure calculations of the activation energies for the steps shown in Figure 9.1 are also consistent with the observations from the MD simulations. Table 9.1 also lists the calculated acti­vation energies for the steps shown in this figure. Protonation at O1 readily leads to the oxonium ion as confirmed by the low barrier. However, the subsequent reaction to form the furanyl compound shown in Reaction (9.3) has a high-calculated barrier (32.0 kcal mol-1). Protonation at O2 has a low barrier for the first step and all subsequent steps. This is con­sistent with experimental studies that have failed to observe any intermediates. Protonation at O3 has low barriers to form formic acid, consistent with the experimental observation of this product. No reactions were found from the protonation at O4. Protonation at O5 readily leads to ring opening, but subsequent reactions that lead to the formation of fur­fural have high barriers (18) (27-29 kcal mol-1). Thus, the original mechanism proposed for furfural formation, Reaction mechanism (9.2), appears to be less likely than Reaction mechanism (9.4), which has barriers of 10-17 kcal mol-1. These results also suggest that protonation at O2 leads to furfural formation and protonation at O3 leads to formic acid, both of which have been observed experimentally. NMR measurements (18) also confirm these two products and experiments with 13C labeled are consistent with the proposed mechanisms.

Members of the CFB phylum, which includes Prevotella and Bacteroides, account for a very significant proportion (>30%) of total rumen bacteria (43, 50-51). Many of these species appear to be unique to the rumen (52). None of the available isolates are known to be cellulolytic, but a number of species, notably P. bryantii and P. ruminicola, possess carboxymethylcellulases, hemicellulases, and pectinases (53). This suggests a role in utiliz­ing products of plant cell wall breakdown released by primary degraders, as suggested by coculture studies that demonstrate cross-feeding of oligosaccharides (2, 3, 54).

Gene clusters concerned with polysaccharide utilization have been studied in P. bryantii B14. One six-gene operon encodes two glucanases and a mannanase (55). The main role of at least one of the P. bryantii glucanases maybe in degrading p (1,3-1,4) glucans (56, 57). Another operon encodes a family 43 xylosidase/exoxylanase and a family 10 endoglucanase (58) and shows homology with an operon from Bacteroides ovatus that is essential for xylan utilization (59). Expression of this gene cluster in P. bryantii is upregulated in response to substrate availability (60) with xylo-oligosaccharides providing the induction signal (61). Most of the xylanase activity found in P. bryantii cells is released only upon cell disruption, suggesting that it is located in the periplasm or membranes (62). The situation resembles that for starch-degrading enzymes in the human gut bacterium, Bacteroides theotaiotaomicron, where limited hydrolysis is thought to occur at the outer membrane, followed by extensive hydrolysis in the periplasm (63). An unusual family 10 xylanase encoded by an unlinked gene XynC (64) that has a preference for large xylo-oligosaccharides is a possible candidate for the limited extracellular xylanase activity in P. bryantii Bi4.

Genome sequences are available for some non-rumen representatives of the CFB divi­sion. The human colonic strain, Bacteroides thetaiotaomicron 5482, has a large genome (6.26 Mb) that exhibits considerable redundancy with respect to polysaccharide-degrading en­zymes (65). This species is not a primary degrader of plant cell walls, and does not possess cellulases or xylanases belonging to the two major GH families 10 or 11, but possesses multiple xylosidase and glucosidase genes. Genome sequencing of rumen Prevotella species is likely to reveal greater complexity of xylanases and related enzymes than is known at present.

Microorganisms associated with upper soil and decaying biomass are comprised of a rel­atively small group of microorganisms that can be grown in vitro and a larger majority that cannot be grown. The most extensive zone of microbial growth occurs on surfaces of biomass, usually within the rhizosphere region of soil. The most studied biomass degrad­ing microbial communities are those involved composting, wet wood, estuaries, and forest floor (60-63). Historically, the ecology of these sites has been investigated by examining those organisms that could be isolated and grown in the laboratory or by measuring the enzymes they produce. Most reports describing the population dynamics have been done by traditional culture and phenotyping methods (64), through the use of systems such as the BIOLOG method (65, 66), by the measurement of phospholipid fatty acid patterns in soil or litter samples (67, 68), and by extracting and monitoring enzyme activities (69). It has also been reported that augmentation of a biomass with specific enzymes, primarily cellulases, stimulates decomposition (70), although the mechanisms at the community or molecular level are not known.

Xylem fibers and tracheids are the main cell types in hardwoods and softwoods, respec­tively, and have thick, lignified secondary walls that have been extensively studied. In these secondary walls, the cellulose microfibrils are more densely packed and highly ordered than in primary walls, and usually occur in three layers (Si, S2, and S3) (156). In the thinner outer (S1) and inner (S3) layers, the microfibrils have a cross-helical organization, whereas in the thicker middle layer (S2), the helical organization is uniform and steeper (Figure 4.10) (157). In the S2 wall layer, there is evidence from microscopy that aggregates of cellulose microfibrils and matrix material form alternating, concentric lamellae (158-160). In the walls of the softwood spruce (P. abies), the galactoglucomannans are associated more with the cellulose than the lignin, whereas with the heteroxylans the reverse is true (161, 162). These results led to a model (Figure 4.11) in which much of the galactoglucomannan is asso­ciated with the aggregates of cellulose microfibrils, and the matrix material contains lignin with associated heteroxylans and the remaining galactoglucomannans (10). This is consis­tent with the observations that heteromannans (glucomannans and galactoglucomannans) form oriented associations with surfaces of cellulose microfibrils since the conformation of their backbone chains is similar to cellulose (163) and are dissociated from the surfaces of cellulosic microfibrils only in strongly alkaline borate solutions [17.5% NaOH-4% borate (6 M NaOH-0.81 M H3BO3)] (164).

In the xylem fibers of hardwoods, immunogold labeling indicates that heteroxylans (4- O — methylglucuronoxylans) with low degrees of substitution are associated with the aggregates of cellulose microfibrils, whereas heteroxylans with higher degrees of substitution are in the matrix (165). Alow degree of acetylation of polysaccharides enhances their water solubility (23) and may prevent their association with one another and other polysaccharides in walls.

Although the cellulose microfibrils in the secondary walls of sclerenchyma fibers of some species of grasses are organized in a similar way to those in the xylem fibers and tracheids of hardwoods and softwoods (166), the walls of bamboo fibers have many layers and are described as polylamellate (167). As in the walls of hardwood fibers, in the lignified secondary walls of the grasses there maybe populations of heteroxylans with low degrees of substitution

associated with the aggregates of cellulose microfibrils and heteroxylans with higher degrees of substitution in the matrix (8, 168).

Studies of lignified secondary walls of hardwoods, softwoods, and grasses by transmission electron microscopy after preparation by fast freeze, deep etch, and rotary shadowing re­vealed bridges between cellulose microfibrils (147, 148, 169). Unlike the bridges in primary walls, these occur across slit-shaped pores, 8-40 nm wide, in the cell walls (148). The bridges were visible only in walls before lignification had occurred, or in lignified walls that had been delignified with acid sodium chlorite solution. Although the model proposed for the secondary wall of spruce tracheids, showed no bridges between the aggregates of cellulose microfibrils (10), similar principles used in modeling primary walls can be applied to mod­eling lignified secondary walls (8,168). Thus, polysaccharides associated with the aggregates of cellulose microfibrils may also form the bridges between the aggregates.

The building blocks for polysaccharide synthesis are nucleotide sugars (NDP-sugars). The sugar moieties in NDP-sugars are incorporated into growing polysaccharide polymers by glycosyltransferases (GTs). A major contributor to glycan diversity is the number of NDP — sugars that an organism produces. In the plant kingdom, 30 different NDP-sugars have been identified (403). It is estimated that over 50 enzymes are directly involved in the synthesis of NDP-sugars in plants. To date only 22 NDP-sugar biosynthetic genes have been functionally identified [see (404) and text below] (Table 5.6). While it is widely accepted that NDP-sugars are the precursors for cell wall polysaccharides, glycoproteins, and glycolipids, it should be kept in mind that in addition to NDP-sugars, lipid-bound sugars are also sugar donors for the synthesis of glycans (for example, synthesis of the N-glycan core of glycoproteins in eukaryotes and addition of galacturonic acid to Rhizobium lipopolysaccharides via a prenyl phosphate-galacturonic acid donor substrate) (405, 406). Whether the initiation of plant cell wall polysaccharide synthesis is mediated by a core glycan that requires lipid-bound sugars at the ER, remains to be established.

Relatively few NDP-sugars are made inside the ER and Golgi apparatus where glycans are made. Most NDP-sugars are produced in the cytosol. Thus, specific NDP-sugar trans­porters exist to facilitate the import of NDP-sugars from the cytosol into the correct lumen of the endomembrane where GTs reside. It is predicted that ~20 NDP-sugar transporters exist in plants of which functionally only six have been characterized (A. Orellana, personal communication). These transporters are localized to the ER and Golgi apparatus (93, 407). While some of the NDP-sugar transporters are specific, the relatively low number of trans­porters would suggest that some NDP-sugar transporters may accept several NDP-sugars. In addition to the need for GTs, NDP-sugars, and their transporters for wall polysaccharide synthesis, some wall polysaccharides (i. e. pectins and hemicelluloses) are also modified by acetyl and methyl groups. Thus, diverse acetyltransferases and methyltransferases are also required. Little is known about their substrate specificity, as none has been biochemically purified. Basic questions such as what controls the degree and number of methyl modifi­cations on a specific glycan remain elusive. Hence, no wall-related acetyl — or methyltrans — ferases have been functionally cloned. The methyltransferases (MetTs) and acetyltransferases (AceTs) generally utilize S-adenosyl-L-methionine (SAM) and acetyl-CoA as methyl and acetyl donors, respectively (229, 285, 335, 337). A recent study in A. Orellana’s laboratory led to the biochemical identification of a SAM transporter activity in the Golgi apparatus of pea (A. Orellana, personal communication). Beyond synthesis, plant glycans undergo further modification, including degradation and remodeling by specific glycosidases and esterases. Due to space limitations, transporters and glycan modifying enzymes will not be summarized in this review, rather, the reader is referred to a recent review (246).

Wall biogenesis is a complex cellular event similar to an assembly line. It requires the supply of a wide range of precursors targeted to different subcellular locations for a process that begins in one subcompartment and continues in other subcompartments as the glycans are synthesized and modified. During this process the concerted action of a range of cytosolic,

Table 5.6 NDP-sugar biosynthetic genes
Mutant, isoforms (putative?), locus (aa) Cell location Sloppy, At5g52560 614 UGIcPP#1 470 At5g17310 469 UGIcPP#2 At3g03250 SuSyl, At5g20830 808 Mito SuSy2, At3g43190 807 Cyto SuSy3, At5g49190 808 Golgi At4g02280 809 ADG1, Aps1, 520 Chlo At5g48300 476 Chlo Aps2, At1 g05610? 521 Chlo Арі 1, ADG2 523 At5g19220 521 Apl2, At1g27680 520 Apl3, At4g39210 Apl4, At2g21 590 GalK, At3g06580 496 UGE1, At1g12780 351 Cyto UGE2,At4g23920 350 Cyto UGE3, At1 g63180 351 Cyto Rhd1,UGE4, At1g64440 348 Cyto-Golgi UGE5, At1g10960 351 associated Urs1, At1 g78570 669? Cyto At3g14790 667 Rhm2, mum4, At1g53500 NRS/er, At1 g63000 301 Ugd1, At5g39320 480 Ugd2, At3g29360? 480 Ugd3, At5g1 5490 480 Ugd4, At1g26570 481
Enzyme U DP-sugar PPase UDP-GIc PPase Sucrose synthase ADP-GIc PPase	Activity Sugar-1 — P + UTP U DP-sugar + РРІ Glc-1 — P + UTP+* UDP-GIc+ РРІ	Syn Sloppy UGlc PPase
Sucrose + UDP UDP-GIc + Frc
SuSy
Glc-1 — P + ATP ADP-GIc + РРІ	AGPase small sub AGPase large sub LS
Gal + ATP Gal-1 — P + ADP UDP-GIc UDP-Gal	Gall, GalK UGE
Gal К UDP-GIc 4-epimerase UDP-Rha synthase UDP-Rha epimerase/reductase UDP-GIc dehydrogenase
UDP-GIc + NAD(P)H ^UDP-Rha
URS Rhm NRSer UGlcDH UGD
UDP-4keto 6deoxyGlc + NAD(P)H -^UDP-Rha UDP-GIc + 2NAD —)>UDP-GlcA + 2NADH

UGlcAEI, At2g45310 UGIcAE2, Gae6, At3g23820 UGIcAE3, Gael, At4g30440 UGIcAE4, At4g00110 UGIcAE5, Gae2 At1g02000? UGIcAE6, Gae5, At4g12250?

Uxs1, At3g53520 Uxs2, At3g62830 Uxs4, At2g47650 Uxs3, At5g59290 Uxs5, At3g46440 Uxs6, At2g28760

AXS1, At1 g08200 At2g27860

Aral, At4g16130

Mur4, UXE1, At1g30620 UXE2, At2g34850?

Uxe3, At3g34850 Uxe4, At5g44480?

Cyt1, At2g39770 At4g30570?

GMD1, At5g66280?

GMD2, mur1, At3g51160

Ger1, At1g73250 Ger2, At1 g1 7890

Gme1, At5g28840

kdsA1, At1 g79500 kdsA2, At5g09730

kdsB, At1 g53000

ER, and Golgi enzymes, as well as ER, Golgi, vesicular and plasma membrane proteins is required to facilitate the production of one type of glycan. Therefore, knowledge regarding the catalytic topology of each membrane enzyme and the cellular machinery that partitions, regulates, and traffics each protein and the corresponding glycan-intermediates, to their correct subcompartments must be understood to truly comprehend wall assembly and synthesis.

The prevailing paradigm regarding the structures of native celluloses during most of the last century held that cellulose is inherently crystalline in the native state and that removal of other cell wall constituents results in exposure of the cellulose in its native state. This is best represented by the models of native celluloses presented by Preston (1) and by Frey-Wyssling (2) in their respective classic treatises. The paradigm has been the basis of all crystallographic models of the structures of native celluloses. This view was based on early observations of the birefringence of cellulose, and after X-ray diffraction was discovered, by observations of X-ray diffraction by cellulosic substances. As new instrumental methods have been devel­oped and as progress has been made in computational modeling ofcelluloses, it has become clear that questions of structure are more complex.

The evolution of crystallographic models of cellulose has been reviewed elsewhere (3). There had been little consensus regarding the crystallographic structures of the native form well into the early 1980s. The reason for the uncertainty was that different investigators had data sets from different tissues and species, and the different investigators used dif­ferent constraints on their solutions of the structures. The constraints are necessary be­cause the data sets are not adequate for achieving a definitive solution of the structural problem.

In 1984, Atalla and VanderHart (4, 5) reported that native celluloses are composites of two forms, Ia and Ip, which coexist in all native forms. Two new instrumental methods for that time, Raman spectroscopy and solid state 13CNMR, contributed to the finding. In the initial reports, the two forms were described as “two distinct crystalline forms.” In retrospect, use of the term “crystalline” was unfortunate. Raman spectroscopy also led to the conclusion that the two forms have the same conformation but different hydrogen bonding patterns (6).

The crystallographic studies have been limited historically by the inadequacy of the num­ber of reflections observed for a definitive solution. Diffraction patterns include at most about 300 reflections from celluloses, whereas a definitive solution requires many more re­flections. The approximately 300 reflections achieved in the most recent studies of cellulose structures (7, 8) are in contrast to 1257 reflections observed in the study of cellobiose by Chu and Jeffrey (9) and 1724 reflections in the study of methyl p-cellobioside by Ham and Williams (10). In the crystallographic studies of cellulose cited above, the reflections are complemented with constraints imposed on the solutions of the structural problem. The key constraints are assumptions regarding the symmetry of the crystal structures that have been controversial since Honjo and Watanabe first reported that such symmetry is not consistent with the electron diffraction patterns (11).

Regarding the organization of celluloses in their native states in higher plants, the central flaw in crystallographic studies is that the constraint represented by the assumption of translational symmetry, implicitly imposed in the mathematical analysis of diffractometric data, is not consistent with the curvature of the microfibrils nor does it accommodate the naturally occurring long-period helical twist. Both the curvature and the twist are distinctive of the morphology of the cell walls and of the tissues and species where they occur. To understand the genomic encoding of species and tissue specificity, it is necessary to have a structural paradigm that can be related to the variability in morphology.

Structures derived from diffractometric measurements require that the forms Ia and Ip belong to different space groups. Ia must have one chain per unit cell with a distinctive conformation, and Ip has two nonequivalent chains per unit cell distinct from the chain in the Ia form. These findings cannot be reconciled with the two forms of native cellulose coexisting in a nanofibril 3-5 nm in diameter. These findings also contradict those by Sugiyama and coworkers in the classic study reporting the lattice images of Valonia ventricosa fibrils (12), which shows that the fibril is a single crystal though it is approximately 65% Ia and 35% Ip. Finally, recent observations of the Raman spectra of celluloses Ia and Ip leave little question that the conformations in both forms are nearly identical. These matters have been discussed in detail elsewhere (13); they will be considered in overview here.

From a more conceptual perspective, the fundamental flaw of the crystallographic models is that they do not provide a basis for rationalizing the species and tissue specificity of the blends of the two forms that occur in native tissues. It is clear that the distinctive patterns of different native celluloses are related to the aggregation of the most elementary fibrils at the levels immediately above the aggregation of the individual nanofibrils emerging from individual cellulose synthase complexes (see Chapter 5 for more discussion).

To clarify terminology used in the following discussions, we define nanofibrils as the aggregates of cellulose chains emerging from individual synthase complexes, whereas a microfibril is used to define the next level of aggregation. A microfibril is thus an aggregate of nanofibrils distinctive of a particular tissue in a particular species, and it is usually the first level observable through high magnification microscopy, whether electron microscopy, or during the last decade or so, atomic force microscopy.

The point of departure for our discussion will be the observations by electron microscopy that microfibrils of native celluloses are first and foremost biological structures spatially orga­nized in a periodic helical form. Furthermore, their period varies with the lateral dimensions of the fibrils and is species — and tissue-specific. Recent molecular modeling studies of ag­gregates of cellulose chains in a hydrated environment have shown that the periodicity is inherent in the nature of cellulose molecules (14).

Modulation of 4CL activity can also, depending upon the gene(s), be targeted to affect monolignol, flavonoid, suberin, and related metabolism. Standard biotechnological ma­nipulations of overall 4CL activity have been described thus far using tobacco (221-223), Arabidopsis (185), and aspen (Populus tremuloides) (224). Very preliminary descriptions of each ofthe corresponding phenotypes so obtained were also reported to be dramatically dif­ferent, even though lignin levels were apparently diminished to a similar extent by circa 50% or so, but only when 4CL activity levels were greatly reduced (down to ~7% of wild-type levels) (77). The reports on the transformants ranged from having no visible phenotype in Arabidopsis, to dwarfing in tobacco, and to a claim of spectacularly enhanced growth in aspen.

In the case of 4CL downregulation in tobacco, the data obtained were the most compre­hensive and generally appeared the most reliable (77, 221-223), albeit where several of the transformants were dwarfed at the flowering stage. In terms of the 13C NMR spectra of the lignin-enriched preparations isolated from the wild type and 4CL downregulated lines, both were very similar except for a few slightly enhanced resonances which were subsequently interpreted by Anterola and Lewis (77) as being mis-assigned by Kajita etal. (222). We con­sidered these to correspond instead to p-coumaroyl (59)/feruloyl (60) tyramine moieties, which if correct, were then in accordance with our previous work on this molecular species in tobacco (225). There was no evidence that the overall amounts of these well-known tyramine-derived moieties in the Solanaceae had either increased, or were covalently linked to the lignin in the lignin-enriched isolates. Interestingly, and as expected, the 4CL downreg — ulated line displayed a significant reduction in vascular integrity as evidenced by collapsed xylem in the tobacco stem cross-sections (223) (see Figure 7.12D). Again, though, there were no comprehensive studies on tobacco growth/development and structural integrity of the vasculature of these transformants carried out to gain a more full understanding of the overall effects engendered by this manipulation.

For 4CL downregulation of Arabidopsis (185), the phenotype was apparently similar to wild type, although no data were provided describing either growth and/or developmental processes. Additionally, the results obtained were not explicable based on known lignin chemistry, even though ostensibly there were reductions in lignin contents. For example, the approach taken by these researchers to estimate lignin amounts used the unreliable thio — glycolic acid method. Nitrobenzene oxidation (discussed earlier) was also carried out, this being a degradative method which generally accounts at best for circa 15-25% of the lignin present in plant tissue (Figure 7.10E). Recalculation and comparison of the thioglycolic acid and nitrobenzene oxidation data by Anterola and Lewis (77) determined, however, that the latter data accounted for >115% of the thioglycolic acid lignin perhaps indicat­ing that either the true lignin contents were grossly underestimated, or the alkaline NBO analyses were incorrect. This is an important matter as it again underscores the serious discrepancies in which many lignin “analyses” are often being carried out in this field. For example, currently many researchers typically do not either compare or contrast results be­tween the different methods used for consistency, accuracy and/or quantification reliability. Accordingly, meaningful trends are potentially being missed in terms of overall effects of manipulation of lignin contents and compositions. Hence, taken together, the effects of Arabidopsis 4CL downregulation provided little insight yet other than perhaps (and quite unexceptionally) possibly reducing lignin levels. Again, these studies simply represent very preliminary analyses and now need to be extended to probe both effects on lignification proper and on vascular apparatus assembly.

4CL downregulation in aspen resulted in quite unusual assertions based on the data actu­ally obtained (224). In that study, it was reported that the downregulated lines grew ~50% taller and that cellulose synthesis was markedly increased. Our reanalysis (77) gave a very different interpretation. In terms of accelerated growth, the lignin contents of the controls were 21.62%, whereas the “growth acceleration” reportedly occurred in the various trans­genic lines having between 20.6 and 11.8% lignin, respectively (224). That is, a very small reduction in lignin content from 21.62 to 20.6% was considered sufficient to result in acceler­ated growth. This effect has not been observed in any other study of lignin downregulation/ mutation, and requires scientific explanation, if correct. Moreover, to date, there have been no further reports of effects on poplar growth by 4CL downregulation. These investigators, or others, should thus clarify at the earliest opportunity whether growth enhancements are still observed for both greenhouse and field-grown 4CL downregulated lines, and what ef­fects on lignification, cell wall architecture, and physiology in general result. One possibility to be considered is that this plant line is capable of overproducing reaction (tension wood) tissue as a mechanism of partially “compensating” for reduced lignin levels, i. e., as noted for pC3H downregulation in alfalfa (72).

If so, this could further explain the reportedly enhanced cellulose contents, which had also only “increased” proportionally in large part because of reductions in lignin levels. Furthermore, although the full NMR spectroscopic data was not provided for that study (224), it was stated that the lignin in the 4CL downregulated line (reduced by ~45%) was apparently the same as that of wild type, as were also the G/S ratios by thioacidolysis. This would again provisionally be expected from first principles (77): namely that there were simply reductions in lignin content when 4CL activity was greatly suppressed without any “compensation” by other non-monolignol phenolic entities into the lignin core structure as suggested by others (173-175, 226). Finally, these preliminary studies again emphasize the need to conduct more extensive analyses of the lignins and to determine the effects on the vascular apparatus integrity in 4CL downregulated lines.