We initially considered it instructive to determine how control over monolignol radical coupling might be effectuated, particularly given that this was a factor largely not considered as being possible during the 1950s/1960s. On the other hand, this was a relevant issue with the

large numbers of structurally related lignans now known. In this regard, many different forms of specific coupling have been reported in isolated metabolites (e. g., containing specifically — linked 8-1′, 8-5′, 8- O-4′, 5-5′, 3- O-4′, 7-1′, 8-7′, 1-5′, and 2- O-3′ interunit linkages) depending upon the metabolite and/or plant species in question (321, 322). This indicated, at the very least, that a mechanism of regiospecific coupling control had evolved in planta for phenolic coupling. Additionally, large numbers of lignan metabolites are optically active suggesting, in turn, that stereoselective coupling might also be occurring in many instances. For the purposes of this discussion, however, there are two pertinent examples of potential control over phenoxy radical-radical coupling. These include formation of the 8-8′-linked (+)-pinoresinol (69a, Figure 7.19A) in Forsythia species (323-327), and that of (meso)

8-8r-linked nor dihydroguaiaretic acid (84, Figure 7.20C) in the creosote bush (Larrea tridentata) (328, 329), respectively.

Work was thus undertaken to initially establish how the lignan (+)-pinoresinol (69a) was formed, this having been an earlier unresolved scientific interest of Holgar Erdtman. Using Forsythia suspensa as a biological partner, we were able to demonstrate that preferential stereoselective coupling of two Е-coniferyl alcohol (3) moieties occurred to afford (+)- pinoresinol (69a) when incubated with crude “cell wall’Vinsoluble preparations (323). Sub­sequent purification of the various proteinaceous components in this crude preparation ul­timately afforded (as estimated by SDS-PAGE) an ~ 110 kDa one-electron oxidase (a laccase) and an ~26 kDa protein, with the latter protein lacking monolignol oxidizing capacity (324). The laccase alone in vitro generated coniferyl alcohol radicals which underwent the well — known nonspecific coupling to afford the corresponding racemic (±)-dihydrodiconiferyl alcohols (68a/b), (±)-pinoresinols (69a/b), and (±)-threo/erythro guaiacylglycerol 8-0-4

coniferyl alcohol ethers (71a/b) (Figure 7.19A). On the other hand, when the 26 kDa pro­tein, which existed as an ~50-52 kDa dimer, was added to the assay mixture, stereoselective coupling occurred instead to afford (+)-pinoresinol (69a) (Figure 7.19A).

Interestingly, in addition to the laccase, other one-electron oxidases (peroxidase) and one-electron oxidizing agents were able to effectuate stereoselective coupling in the pres­ence of this protein (324). Given this striking ability to dictate the outcome of stereoselective coupling with the (+)-pinoresinol-forming protein, we coined the term dirigent protein (DP) from the Latin: dirigere, to guide or to align (324). The corresponding gene was next obtained, with this encoding a protein of ~18 kDa. The discrepancy in the molecular size (~26 versus ~18 kDa) was due to posttranslational (glycosylation) modification (325); functionally competent recombinant protein was also obtained when expressed in a “glyco­sylated” form using insect (Spodoptera frugiperda) cell cultures (325).

The biochemical mode of action of the (+)-pinoresinol forming DP has since been the subject of our recent work (326, 327), where it is considered to function by trapping monolignol (radicals). Figure 7.19B thus depicts our current understanding of the kinetic and “rate-limiting” processes presumed to be involved in stereoselective coupling, versus that of nonspecific coupling leading to racemic dimers (327). Importantly, it was demonstrated that proteins had indeed evolved monolignol (radical) binding capacity, and the ability to engender specific coupling modes, in vascular plants.

However, we considered the (+)-pinoresinol forming DP discovery as a special example of control over monolignol radical-radical coupling, i. e., whereby the monolignol (radical) was bound to the active site of each monomer in the DP dimer. In this way, the (+)- pinoresinol-forming DP dimer was able to orientate both coniferyl alcohol radicals in such a way as to where only (+)-pinoresinol (69a) formation could occur (Figure 7.19A).

More recent work has established that dirigent proteins and their homologues are found throughout the vascular plant kingdom (330-332). However, they appear to be restricted to land plants, suggesting they obtained their function(s) during the transition of plants to land. Dirigent proteins are generally also present in large multigene families with varying levels of homology (e. g., from 99.5 to 12.5% identity) (331, 332), with most biochem- ical/physiological functions of their homologues currently unknown. As a beginning to define their functions, we have investigated their expression profiles, using a GUS-reporter system linked to each putative DP promoter of two dirigent multigene families and/or ho­mologues in both western red cedar (Thuja plicata) (333) and Arabidopsis (334) (Kim etal., manuscript in preparation). [Western red cedar was of particular interest since it accumu­lates the (+)-pinoresinol-derived metabolite, plicatic acid (74, Figure 7.19C), and several of its DPs have the capacity for (+)-pinoresinol (69a) formation (330).] Determination of their expression profiles was carried out for both families; this appears to be a useful approach to begin to establish the biochemical/physiological functions of the various homologues.

There have also been a number of other studies which provisionally suggest DP control over other coupling modes. For example, using Eucommia ulmoides “insoluble residues” Lourith et al. (335, 336) reported diastereoselective homo — and hetero-coupling of sinapyl (5) and coniferyl (3) alcohol moieties in vitro, without addition of any cofactor, to afford the 8-0-4′ lignans 75-78 (Figure 7.20A) of differing levels of erythro/threo ratios and optical activities (335, 336). The proteins involved in such coupling now need to be purified, and their encoding genes cloned, in order to more fully characterize the basis of such transformations.

Another intriguing report ofpresumed dirigent protein involvement is in the stereoselec­tive coupling of two molecules of the terpenoid, hemigossypol (79), to afford (+)-gossypol (80a) in cotton (Gossypium hirsutum L. var. marie galante) flower petals (Figure 7.20B) (337). Interestingly, different Gossypium varieties accumulate varying levels of (+)- and ( — )-gossypols (80a and 80b), with both having equal toxicity toward insects and pathogens. However, cotton seeds generally cannot be used in animal feed because of the toxicity of (-)-gossypol (80b). Breeding has thus been used to select varieties accumulating the (+)-, but not the ( —)-, antipode of gossypol (80). Research has also recently been conducted to

understand the biochemical basis of (+)-gossypol (80a) formation in cotton flowers (337). In the presence of a one-electron oxidase (peroxidase, laccase), coupling of hemigossypol (79) only afforded racemic gossypol (80a/b), whereas when a presumed dirigent protein — again lacking oxidative capacity was added — stereoselective coupling occurred to essen­tially only give (+)-gossypol (80a). Thus, it would provisionally appear that DP control over radical-radical coupling can involve metabolically quite distinct plant product classes.

Other forms of control over regiospecific radical-radical coupling have also been noted for formation of the 8-8r-linked lignans, such as in the creosote bush (Larrea tridentata). The latter accumulates (meso)-nordihydroguaiaretic acid (NDGA, 84) and other 8-8r-linked lignans. Interestingly, while the presumed precursor of NDGA, p-anol (81) can potentially undergo various forms of coupling at different sites on the molecules, such as at the 4- O, C-5, C-1, and C-8 positions, only regiospecific coupling at 8-8r occurs (see Figure 7.20C). This is again indicative of, at the minimum, regiospecific control over phenolic radical — radical coupling (328, 329). Other examples of regiospecific coupling are also apparently found in ellagitannin metabolism, such as in the intermolecular coupling of pentagalloyl glucose (85) moieties to afford tellimagrandin II (86) and subsequent intramolecular cou­pling of the latter to generate cornusiin E (87) (Figure 7.21) (338-340). While these are presumably considered to be “laccase-like” protein mediated (338-340), it will be impor­tant to establish how their biochemical mechanisms differ — if they do — from the dirigent

protein-mediated coupling. Work is currently underway to explore this possibility. In any event, biochemical mechanisms have been preliminarily described for proteinaceous control over both stereoselective and regiospecific coupling.

7.7.3.3.1 PROTEIN VERSUS NON-PROTEIN DIRECTED NATIVE LIGNIN

MACROMOLECULAR CONFIGURATION AND THE QUESTION OF RACEMISATION IN LIGNIN STRUCTURE The preceding sections have emphasized some of the difficulties in establishing native lignin macromolecular configuration — beginning with the technological limitations experienced in lignin analyses more than five decades ago — to the present date. In addition, verification of the long-known racemic nature of lignins and of lignin subunit fragments apparently con­vinced some researchers more recently that lignin formation must indeed occur randomly in vivo (175,341). However, the presence of racemic substructures does not eliminate proteina­ceous control over lignin macromolecular configuration as described below. Additionally, the notion of random coupling has in turn led to other suggestions — but again without rigorous proof — that various cell wall constituents, such as hemicelluloses, cellulose, have important roles in determining and/or establishing lignin configuration in vivo. Several of the hemicelluloses may, however, indeed be involved in forming lignin-carbohydrate bonds, e. g., via reaction with intermediate quinone methides.

However, such considerations fully ignore both the presence and the roles of cell wall pro­teins, the vast majority of whose functions remain currently unknown (342). In this regard, all of the advances made to the current day in the study of phenylpropanoid metabolism (and the effects of its modulation) resulted solely from the study of the proteins, enzymes, and genes involved. Thus, given the paucity in our knowledge of cell wall biochemistry (and of the proteins involved), we consider it inopportune not to systematically examine their roles in controlling native lignin macromolecular configuration. Indeed, this is a more likely and presumably more productive direction than the study of non-proteinaceous components, such as the effects of cellulose, hemicelluloses, etc.

In this regard, various proteins have been considered for their potential roles in establish­ing lignin macromolecular configuration in vivo. For example, using polyclonal antibodies raised against the (+)-pinoresinol-forming dirigent protein, our preliminary analyses in­dicated that their epitopes could be detected in the cell wall areas (e. g., S1 sublayers and cell corners) where lignification was initiated (284). This was considered as being due to recognition of the monolignol (radical) binding motif(s). This, in turn, led to our proposal that — in contrast to stereoselective coupling — there were arrays of dirigent (monolignol) radical binding sites in those subcellular regions, thereby providing the basis for forming a predetermined — albeit racemic — lignin structure (or structures) (29, 31, 284, 285).

Other studies have also implicated various other proteins (e. g., proline-rich proteins, PRP) as potential “lignin scaffolds” in the cell wall, based on co-localization ofPRP epitopes and lignins in developing cell walls of maize coleoptiles (343) and in secondary cell walls of soybean (Glycine max) differentiating protoxylem elements (344), albeit without any precise indication as to what this meant either biochemically or in terms of how they influ­ence lignin macromolecular configuration. Before investigating the involvement of any one of these possibilities, we considered it useful to begin to develop more robust approaches to: probe native lignin structure in vivo (type and frequency of interunit linkages); identify conditions for obtaining native lignin facsimiles through in vitro assays, as well as to identify the biochemical (structural motifs) in dirigent proteins that are required for monolignol (radical) substrate binding (work in progress).

Proponents of the random coupling/combinatorial biochemistry model — affording 1066 (per 100-mer) isomers (196) alone for monolignols 1-, 3-, and 5-derived structures — have given additional speculation as to why it is their consideration that no proteinaceous control over lignification is in effect. These include: (i) the presumed absence of optical activity in lignins; (ii) that for proteinaceous control, there would be a requirement for complementary chains of proteins in both d and l configurations (of their amino acids) for monolignol (radical) binding leading to racemic lignins; and (iii) that presumed lignin subunits, such as the ~1% or so of secoisolariciresinol components considered part of gymnosperm lignin, are formed nonenzymatically.

In this regard, in 1999, Ralph et al. (341) endeavored to demonstrate that the pres­ence of racemic lignin fragments eliminated the possibility of proteinaceous control over macromolecular lignin configuration. This was an unexpected interpretation, given that provisional reasoning had been provided beforehand (31) and later (52) to rationalize the hitherto well-known absence of lignin optical activity. Nevertheless, presumed 8-5′, 8-1′, 8­8′, and 8- O-4′ lignin fragments were isolated from pine sapwood, using the reductive DFRC method (341), which converts benzylic hydroxyl groups of various lignin/lignan-derived entities to their corresponding methylenic functionalities. Analyses of these products estab­lished that two (8-5′ and 8-1′-derived) reduction products were, as expected, racemic; on the other hand, the reduced derivatives obtained from the presumed 8-8′ pinoresinol (69) and the 8-O-4′ (71) substructures could not be resolved into specific enantiomeric forms. Interestingly, the 8-8′-linked product was not pinoresinol (69) per se, but was instead sec­oisolariciresinol (89) (tetraacetate): as indicated above, the latter 89 is currently considered by some researchers as being a minor (~1%) component of gymnosperm (spruce) lignin (345). [Indeed, the abundance of this presumed structure was estimated to be ~1.0 unit per 100 C9 units in spruce lignin, and 1.0-1.5 units per 100 C9 units in kraft and kraft pulp residual lignin based on quantitative 13C NMR and HSQC NMR analyses (345).]

Formation of secoisolariciresinol (89) from pinoresinol (69) (Figure 7.22A), however, has been extensively investigated in this laboratory, this resulting from the action of ei­ther enantiospecific pinoresinol-lariciresinol reductases (PLRs) or with PLRs that do not display strict enantiomeric preferences (346-348). Moreover, the X-ray crystal structure of PLR has been obtained (348), a catalytic mechanism proposed (348), and work has been completed to identify the changes in PLR substrate-binding pockets for differing enan- tiospecificities/preferences (Kim et al., manuscript in preparation). It is now quite well established that enantiospecific differences and/or slight enantiomeric preferences involve, as anticipated, changes in the nature of the substrate-binding pocket(s). That is, this does not require an interconversion of the d — and L-configurations of the amino acids present in the protein/enzyme structure, as proposed by Ralph and Brunow (349). In a somewhat analogous manner, the 7-O-4′ reduction of (±)-dehydrodiconiferyl alcohols (68a/b) by phenylcoumaran benzylic ether reductase (PCBER) results from binding either enantiomer in its large substrate-binding pocket (348), i. e., again without any necessity to have d and l forms of the amino acids in the enzymes involved.

Interestingly, Holmgren etal. (350) also attempted to demonstrate the reductive conver­sion of pinoresinol (69) into secoisolariciresinol (89) upon incubation of coniferyl alcohol (3) with horseradish peroxidase/H2O2 in presence of NADH, i. e., in the absence of any

(PLR) protein. As expected, no such reduction occurred in the absence of functional PLR; indeed, such experiments are generally carried out as controls in our enzyme assays. Thus, given that biochemical mechanisms for the reduction of both (+)- and ( — )-pinoresinols (69a and 69b) have been described (347), it may be more instructive to examine whether such processes are involved — even to a small amount — in either the lignification process, or whether they result from an infusion of heartwood-forming components, including small amounts of (±)-secoisolariciresinols (89).