Environmental Chemistry Letters

, Volume 4, Issue 3, pp 127–136

The meta-hydroxylation step in the phenylpropanoid pathway: a new level of complexity in the pathway and its regulation


  • Guillaume A. Schoch
    • Department of Plant Metabolic Responses, CNRS-UPR 2357Université Louis Pasteur
  • Marc Morant
    • Department of Plant Metabolic Responses, CNRS-UPR 2357Université Louis Pasteur
    • Department of Plant BiologyThe Royal Veterinary and Agricultural University
  • Nawroz Abdulrazzak
    • Department of Plant Metabolic Responses, CNRS-UPR 2357Université Louis Pasteur
  • Carole Asnaghi
    • Department of Plant Metabolic Responses, CNRS-UPR 2357Université Louis Pasteur
  • Simon Goepfert
    • Department of Plant Metabolic Responses, CNRS-UPR 2357Université Louis Pasteur
  • Maike Petersen
    • Institut fur Pharmazeutische BiologiePhilipps-Universitaet Marburg
  • Pascaline Ullmann
    • Department of Plant Metabolic Responses, CNRS-UPR 2357Université Louis Pasteur
    • Department of Plant Metabolic Responses, CNRS-UPR 2357Université Louis Pasteur
Review article

DOI: 10.1007/s10311-006-0062-1

Cite this article as:
Schoch, G.A., Morant, M., Abdulrazzak, N. et al. Environ Chem Lett (2006) 4: 127. doi:10.1007/s10311-006-0062-1


Together in silico and genetic mining approaches have recently designated the CYP98 family of plant cytochromes P450 as the family of enzymes that catalyzes the meta-hydroxylation step in the phenylpropanoid pathway. This meta-hydroxylation is not catalyzed on the free p-coumaric acid as anticipated, but on its conjugates with shikimic, quinic, or phenyllactic acids. While all CYP98s have in common phenol meta-hydroxylase activity, p-coumaroylshikimate remains their preferred substrate. High expression of CYP98s is detected in lignifying tissues in stems, roots, and siliques. The CYP98A3 gene disruption in Arabidopsis thaliana leads to a drastic inhibition of lignin synthesis, cell growth, and plant development. The meta-hydroxylation of phenolic precursors is thus essential for higher plant development.

Isolation of coding sequences belonging to the CYP98 family from basil, wheat, and extensive functional analysis of the recombinant enzymes, together with CYP98s from other plant taxa, helps shedding some light on mechanisms of P450s evolution. Most importantly, the occurrence of the meta-hydroxylation on esters of shikimic or quinic acids introduces a new biochemical regulation mechanism in the phenylpropanoid pathway.


Cytochrome P450CYP98Chlorogenic acidLigninAdaptation


The mechanism of the 3-hydroxylation of the simple phenolic precursors of lignin, pigments, aromatic, and defense compounds in higher plants has long remained elusive. Meta-hydroxylation is, however, needed for the biosynthesis of guaiacyl and syringyl units of lignin, of UV absorbing pigments and seed storage compounds in crucifers such as sinapoyl malate and sinapoyl choline, but also for the formation of many antioxidant and bioactive esters or amides (such as chlorogenic acid, rosmarinic acid, and caffeoyltyramine), coumarins (such as scopoletin), flavors, and fragrances (such as vanillin, gingerol, capsaicin, safrole, and (methyl)eugenol) (Fig. 1).
Fig. 1

The meta-hydroxylation of the phenolic compounds in plants. The recent discovery of the enzyme catalyzing the meta-hydroxylation in the phenylpropanoid pathway yields a better understanding of the entire biosynthetic routes of caffeoyl esters, sinapoyl esters, and monolignols. The meta-hydroxylation is also a necessary step for the synthesis of coumarins and of a large set of plant-specific flavor and flagrance molecules. The pathways leading to these compounds are not yet fully understood. Phenylalanine ammonia-lyase (PAL); Cinnamate 4-hydroxylase (C4H); (4-hydroxy)cinnamoyl-CoA ligase (4CL); Hydroxycinnamoyl-CoA:shikimate/quinate hydroxycinna-moyltransferase (HCT); Coumaroyl ester 3′-hydroxylase (C3′H)

It was long believed that the 3-hydroxylation occurred on the free acids, the side chains of which were subsequently conjugated and/or reduced. The enzymes catalyzing this reaction, however, could not be characterized. They were alternatively described as an ascorbate-dependent, NADPH-dependent, or FAD-dependent oxygenase (Vaughan and Butt 1970; Stafford and Dresler 1972; Boniwell and Butt 1986; Kojima and Takeuchi 1989), a plastidic enzyme using plastoquinone or ferredoxin as the electron donor (Bartlett et al. 1972) or a nonspecific phenolase that also oxidizes dihydroxyphenols to orthoquinones (Vaughan and Butt 1970). More recently, involvement of ester intermediates was suggested, with three enzymes being described to catalyze the 3-hydroxylation of coumaroyl-CoA. First is a nonspecific polyphenol oxidase (Wang et al. 1997), the second a soluble FAD-dependent hydroxylase (Kamsteeg et al. 1981), the third a Zn2+-dependent dioxygenase that was described to be inactive at a normal cytoplasmic pH (Kneusel et al. 1989). Involvement of P450 enzymes was also suggested for the meta-hydroxylation of p-coumaric acid esters of shikimate and quinate, leading to the synthesis of chlorogenic acid or caffeoyl shikimate (Heller and Kühnl 1985; Kühnl et al. 1987). But these enzymes were not characterized at the molecular level, nor was any of these enzymes confirmed to contribute to the 3-hydroxylation of lignin monomers or to the 3-hydroxylation of simple phenolic compounds.
Fig. 2

Phylogram of the cytochrome P450 superfamily in A. thaliana. The phylogenic representation was built as follows: The first sequences of each subfamily were taken from http://www.p450.kvl.dk/Arab_cyps/table.shtml. Sequences were aligned with ClustalX (v 1.8), and the membrane regions excluded up to the hinge motif for further analysis. The tree was bootstraped 1,000 times and rooted on CYP51. Cytochromes P450 in group A form a plant specific clade. P450s involved in the phenylpropanoid pathway are shown in grey box. The branch of CYP98A3 has been highlighted

Fig. 3

Tissue-specific expression of CYP98A3.A RNA blot hybridization: sixteen microgram of total RNA were loaded on each line for different Arabidopsis tissues and separated on a 1.2 % denaturating gel. The CYP98A3 transcripts were hydridized with a 32P cDNA probe corresponding to the whole CYP98A3 cDNA sequence (upper panel). The lower panel is a loading control, staining with methylene blue of the 28S rRNA band. B Lignin staining and immunolocalization of CYP98A3 and of CYP98A3 expression in leaves (A and B) and roots (C and D). Hand transversal sections of stems and roots were either stained with phloroglucinol-HCl (A and C) for lignin coloration, or printed onto nitrocellulose for hybridization with polyclonal anti-CYP98A3 antibodies (B and D) revealed with alkaline phosphatase (purple). Ep: epiderm; c: cortex; ph: phloem; mx: metaxylem; px: protoxylem; pd: periderm; sph: secondary phloem; vc: vascular cambium; sx: secondary xylem

Fig. 4

Coumaroyl-ester metabolism by recombinant CYP98A3. Products of metabolism by CYP98A3 were separated by HPLC as described in Schoch et al. (2001). Absorbance was monitored at 320 nm. A and B Conversion is shown (right panels) after a 5-min incubation of 2.5 pmol of recombinant CYP98A3 in a 200 μl assay. Controls (left panels) in the absence of NADPH or incubated with microsomes of yeast transformed with a void plasmid gave identical results. A Conversion of CQ, trans-5-O-(p-coumaroyl)-d-quinate (2 nmol in the assay). The product peak (P1) has been identified as chlorogenic acid. B Conversion of CS, trans-5-O-(p-coumaroyl)shikimate (4 nmol in the assay). Product peak (P2) has been identified as trans-5-O-(caffeoyl)shikimate. C and D Substrate specificity of CYP98A3 for coumaroylshikimate isomers. C The mix of cis and trans forms was generated by UV irradiation. D The mix of 3-isomer, 4-isomer, and 5-isomer was generated by incubation of the 5-isomer for 1 h at 90°C. Isomers were used as substrates or incubated without NADPH (left panels). Right panels: Metabolites obtained after 5 min (dashed line) or 70 min (solid line) incubation at 28°C with 1.2 pmol of recombinant CYP98A3. 3-O-CS, trans-3-O-(p-coumaroyl)shikimate; 4-O-CS, trans-4-O-(p-coumaroyl)shikimate, P3, trans-3-O-caffeoylshikimate; P4, trans-4-O-caffeoylshikimate. Estimated turnovers are around 600 min−1 for CS in both experiments, which shows that isomers are not strong competitors of the metabolism of trans-5-O-(4-coumaroyl)shikimate

Only recently, genomic analysis and genetic approaches led to the CYP98 family of cytochrome P450 enzymes to demonstrate that it was responsible for the 3-hydroxylation of the phenol ring in different plant species. This work also revealed an unexpected derivation and puzzling complexity in the phenylpropanoid pathway.

Identification of the candidate gene in A. thaliana

A bioinformatics approach led to the identification of one of the 272 cytochrome P450 genes from A. thaliana, as a candidate gene for encoding the meta-hydroxylase of phenolic compounds (Schoch et al. 2001). Our attention was first drawn to the CYP98 family by its phylogeny and expression. An analysis of P450 phylogeny in A. thaliana (Fig. 2) showed that the CYP98 family was most closely related to CYP73A5, coding for the cinnamic acid 4-hydroxylase, the second enzyme and first P450 in the phenylpropanoid pathway. From the three genes belonging to this family, CYP98A3 was not duplicated like most other genes belonging to the phenylpropanoid pathway in A. thaliana. Such a situation is unusual in other P450 families, most of them showing multiple duplications. Phylogenetic analysis thus pointed to CYP98A3 as an enzyme likely to be involved in the phenylpropanoid pathway. This hypothesis was comforted by the high frequency of CYP98A3 ESTs reported from many Arabidopsis tissues (roots, rosette, inflorescence, siliques, seeds), and by the high frequency of related CYP98 ESTs in a variety of plants species and tissues (poplar and pine xylem, soybean hypocotyl, cotton fibers).

Message frequency, wide distribution, and location thus pointed to probable involvement of CYP98A3 and orthologues in a pathway highly expressed in many tissues including stems and cotton fibers. Expression profile of the CYP98A3 gene was first confirmed by Northern blotting (Fig. 3A) to show that expression was ubiquitous, but highest in lignifying tissues. It was further confirmed via tissue-print hybridization with specific antibodies that clearly indicated high expression associated with lignifying tissues both in stems and roots (Fig. 3B). The CYP98A3 gene was therefore expressed in yeast to investigate its catalytic activity.

While this work was carried out, a parallel study also initiated in silico and based on comparative gene expression also led to CYP98A3 as a gene with an expression pattern similar to the 4-hydroxylases and 5-hydroxylases of phenolic compounds and highest expression in stems, roots, and siliques (Nair et al. 2002). A more detailed analysis of the expression profile during Arabidopsis development using promoter-GUS fusions, however, revealed a more restricted tissue specific expression of CYP98A3 compared to the cinnamate 4-hydroxylase (CYP73A5): e.g., CYP98A3 is not expressed in the seeds and does not show an overall expression in flower tissues as CYP73A5, but promoter activity is restricted to vascular tissues; in the root, the promoter does not seem active in the meristem, cortex, and epidermis. As discussed by Nair et al. (2002), this restricted expression pattern, however, needs to be reconsidered in light of a possible other cis-acting element.

A third track also led to the CYP98A3 gene as a candidate 3-hydroxylase. A screening of a collection of Arabidopsis mutants for reduced epidermal fluorescence (ref) phenotype due to the absence of sinapoyl malate in the leaf tissues led to the ref8 line that was unable to synthesize caffeic acid. Positional cloning of the ref8 gene revealed that it was CYP98A3 (Franke et al. 2002a). Protein inactivation resulted from a point mutation just downstream of the region coding for the heme-binding site of the P450 protein.

The mutation not only affected the chemical composition of the plant, but strongly reduced plant growth and fertility, induced a collapse of the vascular bundles and enhanced vulnerability to fungal attack.

Functional characterization of CYP98A3 as a phenol ring 3-hydroxylase

One of the approaches used to demonstrate CYP98 biochemical function was recombinant expression in yeast. This approach was used independently in three laboratories (Franke et al. 2002a; Nair et al. 2002; Schoch et al. 2001). It showed that p-coumaric acid and p-coumaroyl-CoA were very poorly or not metabolized by recombinant CYP98A3. However, further exploration of CYP98A3 substrate specificity, partially driven by the previous reports of Heller and Kühnl (1985) and of Kühnl et al. (1987), led to the demonstration that while glucose esters are not substrates, the shikimate and quinate esters of p-coumaric acid are very actively hydroxylated in the meta position on the phenolic ring (Schoch et al. 2001). The turnover and catalytic efficiency of this reaction was higher than that of the para-hydroxylation of cinnamic acid by CYP73A5, with the shikimate ester being the best substrate. CYP98A3 converts exclusively the trans isomers, and metabolizes the naturally abundant 5-O-p-coumaroyl isomers with significantly higher efficiency than that of the 4-O-isomers and 3-O-isomers of shikimate and quinate (Fig. 4).
Fig. 5

Characterization of the cyp98A3 TDNA insertion mutant. A Phenotype of the cyp98A3 plants grown on soil, showing a dramatic reduction of the size and inhibition of development of the homozygous plants. B HPLC analysis of lignin composition: comparison of the composition in lignin monomers obtained after thioacidolysis of WT plants and two cyp98A3 lines, showing an accumulation of H units in the mutant instead of G and S in the WT plants. C Genetic complementation of the cyp98A3 mutant restoring a WT phenotype

CYP98A3 was thus confirmed as a meta-hydroxylase of the phenolic ring, also catalyzing the last step of the biosynthesis of chlorogenic, but with a totally unexpected substrate specificity regarding the currently accepted scheme of the phenylpropanoid pathway.

Further confirmation of the essential role of CYP98A3 in the phenylpropanoid pathway came from the analysis of the phenolic composition of the ref8 mutant (Franke et al. 2002b). This analysis demonstrated that both soluble (sinapoyl malate in leaves and sinapoyl choline in seeds) and lignin-associated meta-hydroxylated G and S units were drastically reduced in the mutant.

A major impact of the CYP98A3 gene on the plant development was recently further confirmed by the analysis of a TDNA insertion KO mutant (Abdulrazzak et al. 2006). The cyp98A3 mutant in which the gene is totally inactivated shows very strongly impaired growth and never develops a fertile inflorescence stem (Fig. 5).

Inhibition of cyp98A3 plant development is associated with reduced cell growth. Its lignin composition is deeply modified and consists mainly of H units, with very reduced contents of G and S units. Interestingly, a higher residual content of S units than G units is detected in the aerial parts of the plant. The reverse is observed in roots. Decreased accumulation of soluble meta-hydroxylated simple phenolics and their esters is paralleled by a strong accumulation of flavonoids.

All data thus concur to demonstrate the involvement of CYP98A3 in the synthesis of meta-hydroxylated phenolics in vitro and in vivo, and point to a major role of CYP98A3 in plant growth and development.

CYP98s in other plant species

Simultaneously to this work in Arabidopsis, a search for candidate genes potentially involved in the biosynthesis of eugenol in sweet basil (Ocimum basilicum) led to the isolation of two variant cDNAs CYP98A13 from peltate glands (Gang et al. 2002). The search for candidates was initially based on the screening of a collection of ESTs in a basil that produced methyl chavicol instead of eugenol, but the isolation of a full-length cDNA was only possible from a line that produced eugenol. The difference between the two lines was at the level of transcript processing of the cDNA. This suggests that regulation of the CYP98 gene might in part occur at the level of splicing efficiency. When expressed in yeast, CYP98A13s were shown to catalyze the meta-hydroxylation of shikimate and quinate in a similar way to CYP98A3. Like other Lamiaceae, sweet basil is reported to synthesize rosmarinic acid. The phenyllactic ester of p-coumaric acid was thus assayed as a substrate of CYP98A13s. The meta-hydroxylation of the phenyl moiety of this ester proceeded at a rate that was estimated to be about 15% of the rate of metabolism of the shikimate ester.

Rosmarinic acid is the dominant hydroxycinnamic ester in Lamiaceae, but also in Boraginaceae. A differential display approach, aimed at the identification of genes induced upon elicitation of rosmarinic acid accumulation by yeast extract in Lithospermum erythrorhizon cell cultures, led to the isolation of the CYP98A6 gene (Matsuno et al. 2002). When expressed in yeast, the CYP98A6 protein was shown to catalyze the 3-hydroxylation of 4-coumaroyl-4′-hydroxyphenyllactic acid. Metabolism of other p-coumaroyl esters was however not assayed. Expression of the CYP98A6 gene was strongly induced by yeast extract, and even more by jasmonic acid. This expression closely paralleled the accumulation of rosmarinic acid and was much stronger than phenylalanine ammonia-lyase induction in elicited cultured cells. It seems thus likely that CYP98A6 constitutes a rate-limiting step for rosmarinic acid accumulation in L. erythrorhizon.

More recently, eight CYP98 coding sequences were isolated from wheat based on a new PCR approach using very short conserved sequences (Morant et al. 2002). Seven of them showed open reading frames allowing successful expression in yeast of polypeptides with visible spectrum characteristics typical of P450 proteins. Four, with significant sequence changes compared to canonical CYP98s, including a frameshift at the C-terminus, turned out being totally inactive for metabolism of p-coumaroyl esters. The activity of three others was compared to some other available CYP98 recombinant proteins from A. thaliana and sweet basil (Morant et al. 2006). This comparative study shows that (1) the p-coumaroyl ester of shikimic acid is the preferred substrate of all the tested CYP98s, (2) highest turnover is obtained with wheat CYP98A11, (3) different CYP98s, within a same plant and in different plant species, show very different substrate preferences and substrate specificities, (4) only basil CYP98A13s significantly metabolize the phenyllactate ester of p-coumaric acid, (5) wheat CYP98A11 and CYP98A10 also catalyse meta-hydroxylation of p-coumaroyl tyramine. Cytochromes P450 in the CYP98 family, while remaining primarily trans-5-O-(p-coumaroyl)shikimate 3′-hydroxylase thus seem to have developed different substrate preferences and subsidiary activities in different plant species, driven by a specific biochemical context.
Fig. 6

Revised model of the phenylpropanoid and monolignol biosynthetic pathways. The pathway that seems to be active in lignification is shown in black. The core phenylpropanoid pathway is catalyzed by phenylalanine ammonia-lyase (PAL), cinnamate 4-hydroxylase (C4H) and (4-hydroxy)cinnamoyl-CoA ligase (4CL). The meta-hydroxylation of the phenyl ring involves esterified phenolics, and hydroxycinnamoyl-CoA:shikimate/quinate hydroxycinnamoyl transferase (HCT) and coumaroyl ester 3′-hydroxylase (C3′H). More specific to monolignol pathway are p-coumaroyl-CoA 3-hydroxylase (CCoA3H), cinnamoyl-CoA reductase (CCR), caffeic-acid/5-hydroxyferulic-acid O-methyltransferase (COMT), coniferaldehyde 5-hydroxylase (F5H), and cinnamoyl alcohol dehydrogenase (CAD)

Only the CYP98A8 and CYP98A9 from A. thaliana, show no activity for the conversion of phenolic esters. While still belonging to the CYP98 family, these two P450s, however, have strongly diverged sequences (less than 49% identity) compared to CYP98A3. They probably correspond to an old duplication of the CYP98A3 gene that led the evolution of a new function, and should thus be classified in a distinct CYP98 subfamily.

The revised phenylpropanoid pathway

Figure 6 shows the updated grid of the phenylpropanoid pathway. If the methylation of the caffeoyl unit occurs on the CoA ester and the 5-hydroxylation on the aldehyde form of the C6-C3 unit, it turns out that the meta(3)-hydroxylation is catalyzed on the shikimate, quinate, or phenyllactate esterified form of the phenylpropane skeleton.

This was largely unexpected, and it gives a new importance to and sheds a new light on the biological function of the hydroxycinnamoyl-CoA shikimate/quinate: hydroxycinnamoyl transferases (HCTs) that convert the CoA esters into shikimate or quinate esters. HCTs also catalyse the reverse reaction, with a rather broad substrate specificity with regard to the esterified phenolic moiety.

The same HCT enzymes thus seem to shuttle the coumaroyl and caffeoyl units between their CoA- and shikimate/quinate-esterified forms. The in vivo function and importance of these enzymes has been confirmed both for the synthesis of lignin (Hoffmann et al. 2003, 2004) and the accumulation of chlorogenic acid (Niggeweg et al. 2004). The p-coumaryol shikimate ester is by far the best substrate for all of the recombinant CYP98 proteins tested so far. Shikimate is the main intermediate in the so-called “shikimate pathway” that produces aromatic amino acids in the plastids from higher plants. It is thus very likely that the shikimate esters are the natural intermediates in the high turnover lignification pathway in planta. Shikimate esters of p-coumaric or caffeic acids were never reported to accumulate and are not detected in cyp98 mutant plants. The activity of the CYP98s and of the HCTs thus seems to be highly coupled, possibly via a physical interaction for optimal substrate channelling.

It is interesting to note that, while CYP98s described so far display some functional versatility but keep a same principal function, specialized acyl and amine transferases, transferring preferentially quinate (Niggeweg et al. 2004), or even more selectively phenyllactate (Gang et al. 2002) or tyramine (Negrel and Javelle 1997) have been described. It is thus possible that association with a specific transferase directs the CYP98s substrate specificity in vivo.

A question remains open, looking at the data available so far. It is: Why the shikimate esters seem to flow straight through the pathway, while very high concentrations of chlorogenic acid can accumulate in some plant species? This could be due to a transferase bottleneck for the quinate to CoA ester conversion, or, more likely, to a fast partitioning and storage in the vacuole of chlorogenic acid. The reasons for the accumulation remain to be elucidated, as is the re-mobilization mechanism of chlorogenic acid in accumulating plants.
Fig. 7

Hypothetical model of direct modulation of lignin biosynthesis by plastid energetic level

Role of CYP98s in the regulation of the phenolic pathway

Now, why has such a complicated solution been selected by evolution to achieve the 3-hydroxylation of the lignin monomers? The derivation via shikimate/quinate esters must represent a selective advantage over hydroxylation that would occur on simple p-coumaric acid or its CoA-ester. One possible advantage is stabilization versus spontaneous oxidation of the caffeoyl ortho-diphenol. This is probably achieved in two different ways: conjugation, and protein channelling of the reactive catechol for methylation to the more stable ferulic acid. But such a protection might be achieved via just conjugation with coenzyme A.

Another potential advantage is the introduction of a new and versatile level of regulation in the pathway. This hypothetical regulation is illustrated in Fig. 7. It places lignification under the control of chloroplast activity, in particular with respect to the shikimate pathway. Shikimate and quinate are intermediates of the plastid located shikimate pathway that produces phenylalanine and the C6-C3 backbone. Conjugation with shikimate/quinate of the C6-C3 unit is needed for 3-hydroxylation to occur. The KM for shikimate or quinate, reported for HCTs in many plant species, is always high, in the 500 μM to several mM range (Lotfy et al. 1994; Hoffmann et al. 2002; Niggeweg et al. 2004), which implies that cytoplasmic concentrations of these metabolites are likely to exert a control on the rates of the 3-hydroxylation step if CYP98 is present in limiting concentration. The reaction does not consume shikimate or quinate, at least as long as no chlorogenic acid is accumulated, but conjugation provides a way of “sensing” activity of the upstream shikimate pathway.

When shikimate/quinate are present at high concentration in the cytoplasm (optimal light and temperature), production of lignin monomers, and end-chain phenolics is accelerated. When they are present in limiting amounts, priority is given to the synthesis of aromatic amino acids for protein, and to the synthesis of compounds directly deriving from p-coumaryol-CoA (e.g., flavonoids). Such a sensing biochemical regulation is expected to allow fast and versatile adaptation to environmental conditions (from night/day regime to biotic stress) without requiring costly protein synthesis and regulation of gene activity. It might constitute a major adaptative advantage for higher plants grown under challenging conditions.

Copyright information

© Springer-Verlag 2006