Handbook of Vanilla Science and Technology

Similar complexities have been encountered in studies on related molecules. For example, salicylic acid (SA, 2-hydroxy-benzoic acid) was long thought to be synthesized through the phenylpropanoid pathway via L- phenylalanine, followed by chain shortening to a benzoic acid followed by subsequent 2-hydroxylation (Yalpani et al. 1993; Leon et al. 1995), and this was supported by genetic studies in which modification of expression of L-phenylalanine ammonia-lyase (the first enzyme of the phenylpropanoid pathway) gave disease response phenotypes predictably associated with modification of SA levels (Pallas et al. 1996). The subsequent demonstration that, at least in Arabidopsis, defense-associated SA formation occurs directly from the shikimate pathway via isochorismate (Wildermuth etal. 2001) came as a total surprise. Similarly, recent labeling and genetic studies have demonstrated that the formation of benzoic acids in Petunia flowers occurs by multiple pathways involving both oxidative and non-oxidative chain shortening (Boatright et al. 2004; Orlova et al. 2006). This complexity makes it difficult to interpret labeling studies, particularly if (as in the case of most studies on vanillin to date) multiple tissue types are being labeled and the labeling is only carried out over a short period relative to the period of biosynthesis and accumulation. It has been argued that the existence of multiple pathways to benzenoid natural products within one plant might reflect a biological need for flexible responses to different environmental conditions (Wildermuth 2006). This is quite plausible, but it seems to the present author that constitutive vanillin biosynthesis during the development of the vanilla pod is more likely to occur via a single major pathway. The question is how to elucidate that pathway when enzyme promiscuity can mislead in vitro studies.

Early labeling experiments suggested that vanillin biosynthesis in plants occurs via ferulic acid, a molecule known to be synthesized via the phenylpropanoid/monolignol pathway (Zenk 1965). Although subsequent studies have suggested other alternatives (Table 18.1), it is instructive to consider this model for the formation of vanillin because it allows discussion of the types of enzymes that may be involved in the ring modification reactions, and their identification through functional genomics approaches.

At least in Arabidopsis, ferulate is formed from 4-coumarate by six enzymatic steps in a pathway, shared with monolignol biosynthesis, that is considerably more complex than envisaged at the time that the first labeling studies on vanillin biosynthesis were performed. The first step is the formation of a Coenzyme A ester through the action of 4-coumarate: CoA ligase (4CL), an enzyme generally encoded by multiple genes in plants (Ehlting et al. 1999) (Figure 18.1). The subsequent coumaroyl CoA ester is potentially a substrate for β-oxidative chain shortening (Figure 18.1) but, in the monolignol pathway, is directly converted to the corresponding shikimate ester by the action of hydroxycinnamoyl CoA: hydroxycinnamoyl transferase (HCT) (Hoffmann et al. 2003); it is this shikimate ester that undergoes hydroxylation of the aromatic ring at the 3-position by a second cytochrome P450 monooxygenase (Schoch et al. 2001). However, the subsequent 3-O-methylation does not happen at the shikimate ester stage; rather, the shikimate ester is converted back to the CoA ester through HCT acting in the reverse direction, and the resulting caffeoyl CoA is then methylated via caffeoyl CoA 3-O-methyltransferase (CCoAOMT) to yield feruloyl CoA. This compound is reduced to coniferaldehyde by the action of a cinnamoyl CoA reductase (CCR), another enzyme that is encoded by multiple genes in plants (Escamilla-Trevino et al. 2009). Finally, coniferaldehyde is converted to ferulic acid by the action of an aldehyde dehydrogenase (Nair et al. 2004) (Figure 18.1). It is important to note that detailed biochemical and genetic studies support the operation of this complex pathway over the simple mechanism whereby coumarate is converted to ferulate in two steps by 3-hydroxyl-ation followed by 3-O-methylation, at least in dicotyledonous plants. However, early enzymatic work with crude and partially purified plant extracts did indeed suggest that this simpler pathway might operate.

The alternative and much simpler pathway to vanillin involves non-oxidative chain shortening. At least in vitro, 4-coumarate can be converted to 4-hydroxybenzaldehyde through a non-oxidative process requiring the presence of a thiol reagent but no other cofactor (Podstolski et al. 2002) (Figure 18.1), although no gene has yet been identified to encode this type of enzyme. Conversion to vanillin then simply requires 3-hydroxylation and O-methylation. Classical COMT enzymes are able to catalyze this methylation at the level of the benzaldehyde (Kota et al. 2004).