Biosynthesis of Allyl and Propenyl Phenols and Related Phenylpropanoid Moieties

Besides lignin, specialized plant metabolism can utilize monolignols in the formation of lignans (phenylpropanoid dimers) and, as recently elucidated, allyl and propenyl phenols. Allylphenols differ from propenylphenols in their side-chain double bond position, with the former having terminal (C-8–C-9) desaturation and the latter having the chemically more stable internal (C-7–C-8) double bond. Several biochemical hypotheses had been created to explain their distinctive lack of a C-9 oxygenated functionality (Canonica et al., 1971; Klischies et al., 1975; Manitto et al., 1974a,b), but experimental support for any was lacking until recently.

Our interest in the allyl/propenyl phenols first began with studies directed at elucidating the biochemical pathway(s) to the lignan nordihydroguaiaretic acid (43, NDGA, Fig. 13.10) and its congeners (Cho et al., 2003; Moinuddin et al., 2003), these being abundant metabolites in the creosote bush (Larrea tridentata). These substances are of increasing interest due to their potent biological and (potential) medicinal applications. For example, the NDGA derivative 44 is apparently proceeding smoothly through National Institutes of Health trials as an effective chemotherapeutic treatment for the usually hard to treat (refractory) brain and central nervous system tumors (Chang et al., 2004) (see also http://www. clinicaltrials.gov/ct/show/NCT00404248, http://www.clinicaltrials.gov/ct/show/ NCT00259818 and http://www.cancer.gov/search/viewclinicaltrials.aspx?cdrid= 455645). Interestingly, the creosote bush has long been part of traditional Native American Indian medicine, being used to treat more than 50 diseases, most commonly those of renal and gynecologic origins (Arteaga et al., 2005). These lignans, however, lack oxygenated carbon 9,90 functionalities that are present in most lignan classes [e.g., podophyllotoxin (45), secoisolariciresinol (46), Fig. 13.10], as well as in the polymeric lignins and monomeric phenylpropanoids (e.g., monolignols 19–23) in the vast majority of plant species. Based on previous radiolabeling/stable isotope labeling studies (Moinuddin et al., 2003), it was presumed that the unusual ‘‘loss’’ of the oxygenated functionality occurred at the monomer stage; that is, allyl and/or propenyl phenols could be serving as the precursors/substrates for

FIGURE 13.10 Lignans commonly used in medicinal applications, particularly for cancer treatment/prevention.

dimerization to form these less common lignans. In this regard, the biosynthetic pathways to the allyl/propenyl phenols had not been elucidated in any organism and, in particular, the precise precursor (substrate) undergoing deoxygenation represented both a long-standing question and a biochemical mystery (Canonica et al., 1971; Klischies et al., 1975; Manitto et al., 1974a,b; Senanayake et al., 1977).

FIGURE 13.11 Hypothetical biosynthetic pathways from p-coumaryl alcohol (19) to chavicol (31) and methylchavicol (32), of which pathway B was demonstrated to occur. Source: Vassão et al. (2006b).

Basil (Ocimum basilicum, Thai variety) was used as a suitable study system since it accumulates the simplest allylphenol, methylchavicol (32); based on various radiolabeling studies, it was shown that the latter was derived from the corresponding monolignol, p-coumaryl alcohol (19) (Vassão et al., 2006b). Three potential mechanisms for the conversion of 19 into 32 included reduction of the monolignol side-chain (i.e., saturation) followed by dehydration (Fig. 13.11A); methylation of the phenolic moiety preceding further side-chain modification (Fig. 13.11C); and/or activation of the terminal (C-9) oxygenated functionality prior to side-chain double bond reduction (Fig. 13.11B). Pathways A and C (Fig. 13.11) were eliminated since no experimental evidence in support of either route was obtained.

Interestingly, however, a double-bond reductase was discovered and characterized, which utilized p-coumaryl aldehyde (14) as the preferred substrate to afford the corresponding side-chain reduced aldehyde (48, Fig. 13.11) (Kasahara et al., 2004, 2006; Youn et al., 2006b). This alkenal reductase activity was the first to be reported in the phenylpropanoid pathway, with the corresponding enzymes isolated from A. thaliana (AtDBR) and Pinus taeda (PtPPDBR) also being homologous to a terpenoid double-bond reductase (pulegone reductase, PulR) from Mentha piperita and mammalian alkenal reductases as well (Fig. 13.12). AtDBR and PtPPDBR catalyze the NADPH-

FIGURE 13.12 Alignment of double-bond reductases from Pinus taeda (PtPPDBR), Arabidopsis thaliana (AtDBR1), and homologues from Mentha piperita (pulegone reductase, PulR), rat (Rattus norvegicus, AOR), guinea pig (Cavia porcellus, 12-HD/PGR), and mouse (Mus musculus, 1VJ1). The nucleotide-binding domain is indicated by the dotted line, with conserved AXXGXXG motif in red. Conserved catalytic Tyr residues (Y260 for AtDBR1) are highlighted in light blue, and secondary structural elements are indicated in colored bars (blue for α-helices and orange for β-strands). Source: Redrawn from Youn et al. (2006b).

dependent reduction of p-coumaryl (14) and coniferyl (16) aldehydes to the corresponding dihydroaldehydes, (Fig. 13.13) and AtDBR has also been shown to catalyze the reduction of 4-hydroxynonenal (4- HNE, 51), a pro-apoptotic lipid peroxidation product, to 4-hydroxynonanal (Kasahara et al., 2006; Youn et al., 2006b). Based on substrate versatility studies and an X-ray crystal structure for AtDBR, a concerted mechanism involving an enol intermediate was proposed for these zinc-independent alkenal reductases (Youn et al., 2006b). While the corresponding dihydroalcohol product 49 is a wellknown plant defense metabolite (Kraus and Spiteller, 1997), it was not, however, converted in basil into either chavicol (31) and/or p-anol (37) (Vassão et al., 2006b). Accordingly, it was not considered as being involved in allyl/propenyl phenol biosynthesis.

Instead, a quite novel metabolic process converting monolignols [such as p-coumaryl (19) and coniferyl (21) alcohols] into allyl/propenyl phenols [chavicol (31) and eugenol (33), respectively] was discovered (Fig. 13.14), with two enzymes being implicated in their formation in planta. The first step is activation of the monolignol side-chain alcohol by conjugation to an activated acid (acyl-CoA), resulting in formation of a monolignol ester. This modification results, in

FIGURE 13.13 Possible enzymatic mechanism for AtDBR-mediated conversion of p-coumaryl (14) and coniferyl (16) aldehydes and 4-HNE (51) into their corresponding dihydroaldehyde derivatives. Source: Redrawn from Youn et al. (2006b).

energetic terms, in formation of a more facile leaving group (carboxylate ester), which is more readily displaced by an incoming reducing hydride, for example, in the form of NAD(P)H. Indeed, such coniferyl alcohol acyl transferases have been recently characterized in basil (O. basilicum) (Harrison and Gang, 2006) and petunia (Petunia hybrida) (Dexter et al., 2007), utilizing acetyl-CoA and coniferyl alcohol (21) to afford coniferyl acetate (53), and it is anticipated that substrate-versatile acyltransferases may be able to utilize different monolignols and acyl/aroyl-CoA cofactors to generate different esters. One such ester, p-coumaryl coumarate (54), had been previously shown to serve as substrate for enzyme preparations from Asparagus officinalis (Suzuki et al., 2002) and Cryptomeria japonica (Suzuki et al., 2004), generating the nor-lignans (Z)- and

FIGURE 13.14 Biosynthetic pathway to chavicol (31) and eugenol (33) from the corresponding monolignols, p-coumaryl (19) and coniferyl (21) alcohols. CS, chavicol synthase; ES, eugenol synthase.

(E)-hinokiresinol (56/57; Fig. 13.15A) respectively. Although the proteins responsible for the latter conversions remain to be fully characterized and/or described, potential mechanisms whereby the departing carboxylate (as CO₂) facilitates formation of the final C8–C70 bond, without any additional cofactors, can be envisaged (Fig. 13.15A). It is also possible to propose potential mechanisms where p-coumaryl coumarate (54) [or other pcoumaryl alcohol esters, e.g., p-coumaryl acetate (52)] generates, through the addition of an incoming hydride, chavicol (31) and/or its regioisomer p-anol (37) (Fig. 13.15B). Indeed, the second step in monolignol reduction was shown to be the action of regiospecific reductases that transfer a hydride from NADH or NADPH into either the C-7 or the C-9 of the corresponding monolignol ester (or a quinone methide derivative thereof), thus forming either an allyl or propenyl phenol, respectively (Figs. 13.14 and 13.15B).

FIGURE 13.15 Possible mechanisms for conversion of p-coumaryl alcohol esters into (A) hinokiresinol (56/57) and (B) chavicol (31) and p-anol (37). (A) (a) Concerted or (b) through intermediacy of a quinone methide and (B) (c) and (d) formation of a quinone methide intermediate through displacement of the (interchangeable) ester leaving group, with subsequent reduction by hydride [from NAD(P)H] and rearomatization to form either (c) chavicol (31) and/or (d) p-anol (37). The reactions in (B) may also proceed through direct displacement, without intermediacy of the quinone methide, by an incoming hydride at carbons 7 or 9 to form chavicol (31) or p-anol (37), respectively (not shown). Source: Modified from Vassão et al. (2006b).

These regiospecific reductases (e.g., chavicol and eugenol synthase, CS/ES, and isoeugenol synthase, IES) have been studied to a larger extent than the monolignol-specific acyltransferases. Computational analyses of CS/ES isolated from basil and IES isolated from petunia indicate greatest homology (~40–45% identity) (Koeduka et al., 2006) to members of the PIP family of reductases (pinoresinol-lariciresinol, isoflavone, and phenylcoumaran-benzylic ether reductases) we have either discovered and/or extensively characterized (Dinkova- Kostova et al., 1996; Fujita et al., 1999; Gang et al., 1999), and for which crystal structures have been determined (Min et al., 2003). Significantly, based on sequence homology, one such PLR/CS/ES homologue from L. tridentata (LtCES1) was recently isolated and characterized (Fig. 13.16) (Vassão et al., 2007).

While PLRs are involved in formation of other medicinally important plant metabolites [e.g., podophyllotoxin (45) and secoisolariciresinol (46)] and various plant defense compounds [e.g., plicatic acid (30) in western red cedar heartwood], the biochemical mechanisms of PLR, PCBER, and CS/ES share common properties, including (1) a necessity for a free phenolic functionality in the substrate, indicative of a common quinone methide intermediate (Figs. 13.14 and 13.17) (Kim et al., 2007; Koeduka et al., 2006; Min et al., 2003), and (2) a highly conserved Lys residue (K138 in PLR from T. plicata, K133 in its homologue in L. tridentata, and K132 in CS/ES from basil) required for catalysis (Min et al., 2003). Figure 13.18 depicts the X-ray crystal structure of one member of this class of reductases, PLR from T. plicata (Min et al., 2003). Based on the proposed catalytic mechanisms of CS/ES (Koeduka et al., 2006) and PLR (Min et al., 2003), it is hardly surprising that a high level of similarity was observed between these proteins. All of the PIP reductases, as well as CS/ES, utilize NAD(P)H as the source of a hydride that is regiospecifically (or stereospecifically) added to a carbon that originated from a phenylpropanoid side chain (i.e., either a monolignol derivative or dimer). In fact, a brief phylogenetic analysis indicates that these homologues cluster together with PLR (e.g., from T. plicata), PCBER (e.g., from P. taeda), IFR (e.g., from Medicago), and leucoanthocyanidin reductases (LACR, e.g., from Vitis vinifera), with the L. tridentata homologue clustering closer to more distant PCBER and IFR homologues (Fig. 13.19).

The biochemical characteristics of these enzymes have been studied, with basil CS/ES and petunia IES reported to have substrate affinities [K_m, coniferyl acetate (53)] of 1.6–5.1 mM and V_max of 7–20 pkat/mg protein. These are indicative of relatively low substrate affinity, although not far from the range of other enzymes involved in volatile oil biosynthesis (Koeduka et al., 2006). Additionally, the corresponding PLR homologue in the creosote bush (L. tridentata) catalyzes similar conversions, but interestingly with apparently higher catalytic efficacy [K_m values of a few hundred micromolar and V_max values of a few hundred pkat/µg protein for coniferyl acetate (53), p-coumaryl acetate (52), and p-coumaryl coumarate (54) (Vassão et al., 2007); see Table 13.5]. We are currently examining the properties of other PIP reductases regarding their abilities to form allyl and propenyl phenols and have thus far seen evidence of some substrate versatility in PLRs acting on monolignol esters (unpublished observations).

FIGURE 13.16 Amino acid alignment of basil (Ocimum basilicum) chavicol/eugenol synthase (ObEGS1), Petunia hybrida isoeugenol synthase (PhIGS1), and PIP reductases from Medicago sativa (MsIFR), Thuja plicata (TpPLR), Pinus taeda (PtPCBER), Forsythia intermedia (FiPLR), and Larrea tridentata (LtCES1).

In effect, the long-standing question regarding the biochemical formation of these widely used compounds, allyl and propenyl phenols, has now been elucidated and shown to utilize the same pathway precursors as lignin biosynthesis. Proteins (and their corresponding genes) involved in this process have been isolated and characterized, thus presenting a new approach with which to study and alter the lignification program of woody plants, as well as enabling the production of these compounds in more commonly cultivated plants.