Flavonoid synthesis begins from phenylalanine or tyrosine via the phenylpropanoid pathway (Figure 1). Phenylalanine ammonia lyase (PAL) and tyrosine ammonia lyase (TAL) catalyze the removal of the ammonia groups from amino acids by cleaving carbon–nitrogen bonds. Phenylalanine and tyrosine are converted to trans-cinnamic acid and 4-coumarate through deamination by PAL activity and TAL activity, respectively. These enzymes form tetramers and are inhibited by their product, trans-cinnamic acid. Plants generally have monofunctional PAL activity, which shows approximately 1,000-fold higher enzyme efficiency for phenylalanine over tyrosine (Cochrane et al., 2004). However, a monofunctional TAL has not yet been identified in plant. Instead, bifunctional phenylalanine/tyrosine ammonia lyases (PTALs) have been reported in monocot grass family Poacease such as maize (Zea mays), Brachypodium, and rice (Oryza sativa) (Jun et al., 2018; Barros and Dixon, 2020). In this review, we aim to summarize only PAL activities derived from plants. In vitro PAL activities measured in several plants range from 39 to 50,000 as measured by K_cat/K_m (S⁻¹·M⁻¹) ( Supplementary Table S1 ) (Cochrane et al., 2004; Li et al., 2023). Among these, Arabidopsis (Arabidopsis thaliana) AtPAL2 has the highest enzyme efficiency, although it is difficult to compare enzyme activities exactly among different species because it depends on reaction conditions ( Table 1 ).

PAL activity measured in the same species under the same enzyme assay conditions vary greatly. For example, four different PAL activity values in Arabidopsis have been reported, with an approximately 2-fold difference in enzyme efficiencies, even when those with very low efficiency were excluded. On the other hand, enzyme engineering targeting residues suggested to play key roles in substrate selectivity between PAL and TAL successfully altered the enzyme activity and substrate specificity of PAL (Barros and Dixon, 2020). Mutating Phe144 to His in Arabidopsis PAL increased TAL activity up to 18-fold and decreased PAL activity up to 80-fold (Watts et al., 2006). The H123F mutation of Sorghum bicolor PAL1 (SbPAL1) enhanced the catalytic efficiency for Phe up to 6.2-fold and disrupted the catalytic activity for Tyr (Jun et al., 2018). Additionally, the mutation of ZmPAL2 improved the catalytic activity up to 4.5-fold (Zheng et al., 2024). Thus, optimized PAL enzymes can be selected for each reconstituted pathway and applied as parts for plant synthetic biology.

Trans-cinnamate, a compound biosynthesized by PAL, is converted into p-coumaric acid/4-hydroxy cinnamic acid in plants by the cytochrome P450 cinnamate 4-hydroxylase (C4H), a member of the CYP73A family. The C4H gene family typically has a low copy number, usually ranging from one to five members in the plant genome (Kumar et al., 2013). C4Hs localize to subcellular membranes in the endoplasmic reticulum (ER), serving as a nucleation point to form a multi-enzyme complex with PAL and 4-coumarate-CoA ligase (4CL) known as the phenylpropanoid metabolon. The electrons necessary for catalysis by C4H are provided by NADPH cytochrome P450 reductase (CPR), which is colocalized with C4H on the exterior surface of the ER membrane (Zhang et al., 2020). The K_m values of C4H derived from several plant species for trans-cinnamic acid in vitro range from 0.61 to 40.68 µM ( Supplementary Table S1 ). The formation of a complex with 4CL improved the enzyme activity more than 100-fold in Populus trichocarpa ( Table 1 ) (Chen et al., 2011). The formation of a C4H and 4CL complex might be critical for improving the production of phenylpropanoid pathway products. Indeed, expressing all genes in the phenylpropanoid pathway improved the production of downstream metabolites (Schultz et al., 2019; Lee et al., 2024).

4CL catalyzes the formation of p-coumaroyl-CoA by attaching coenzyme A (CoA) to p-coumaric acid. The 4CL enzymes exist in several isoforms and are encoded by multiple genes within a species (Lavhale et al., 2018). These enzymes have been identified and functionally characterized in various plant species. Plant 4CLs are classified into four groups: I, II, III, and IV. Dicot plants contain group I and II 4CLs, while monocot plants contain group III and IV 4CLs. Group II and IV 4CLs are mainly involved in flavonoid biosynthesis, whereas group I and III 4CLs are involved in lignin biosynthesis. 4CLs have a broad substrate spectrum and function at a crucial branchpoint that determines the biosynthesis of lignin, flavonoids, or other phenylpropanoid derivatives. Arabidopsis 4CLs use diverse substrates such as cinnamic acid, p-coumaric acid, caffeic acid, and ferulic acid, and the enzyme kinetics for diverse substrates vary. Among the five Arabidopsis 4CLs, At4CL3 shows much higher enzyme efficiencies for 4-coumaric acid than for other substrates (Costa et al., 2005). In rice, Os4CL3 has the highest enzyme efficiency for 4-coumaric acid ( Supplementary Table S1 ) (Gui et al., 2011). Thus, Os4CL3 and At4CL3 might be good candidate molecular parts for flavonoid production ( Table 1 ). These three enzymes (PAL, C4H, and 4CL) are key components of the phenylpropanoid biosynthesis pathway to generate flavonoids. The formation of complexes by these enzymes can lead to different enzyme performances and can be engineered to increase biomass production via the regulation of lignin production. Therefore, their co-expression might improve the production of phenylpropanoid pathway products (Schultz et al., 2019; Lee et al., 2024).

CHS, a class of Type III polyketide synthases (PKSs), catalyzes the first step of the flavonoid biosynthesis pathway. One molecule of p-coumaroyl-CoA and three molecules of malonyl-CoA are condensed by CHS to synthesize one molecule of naringenin chalcone (4,2′,4′,6′-tetrahydroxychalcone). The first plant CHS gene to be isolated and identified was from parsley (Petroselinum crispum). Subsequently, CHS genes have been functionally identified in more than 20 plant species (Abe and Morita, 2010; Dao et al., 2011). The plant-type III PKSs share 30–95% amino acid sequence identity. CHS and its homologs catalyze the conversion of a variety of CoA-linked starter substrates, ranging from aliphatic-CoA to aromatic-CoA and from small acetyl-CoA and polar malonyl-CoA to nonpolar n-hexanoyl-CoA substrates. For example, Jez et al. (2001) reported that CHS2 from alfalfa (Medicago sativa) has quite high and similar enzyme efficiencies for p-coumaroyl-CoA and several substrates with CoA functional groups, such as malonyl-CoA and feruloyl-CoA. CHS is a promiscuous enzyme that catalyzes the formation of other polyketides, at least in vitro. This catalytic promiscuity provides an important basis for the adaptive evolution of plant specialized metabolism. CHS can be used as a key molecular part in the reconstitution of metabolic pathways to synthesize diverse and novel chemicals, depending on the substrates. Waki et al. (2020) compared the activities of CHS isolated from five species: PpCHS from Physcomitrella patens, SmCHS from Selaginella moellendorffii, GbCHS from Gingko biloba, OsCHS-1 from Oryza sativa, and AmCHS from Antirrhinum majus. The efficiencies of these enzymes for p-coumaroyl-CoA differed by up to 17.8-fold ( Supplementary Table S2 ). Among these, GbCHS and SmCHS exhibited high catalytic efficiencies for naringenin chalcone production from p-coumaroyl-CoA ( Table 2 ). Thus, selecting a CHS with high enzyme efficiencies is crucial for improving naringenin chalcone production, and GbCHS and SmCHS could be used as molecular parts for synthetic biology.

CHI catalyzes the stereospecific cyclization of chalcones to (2S)-flavanones, such as naringenin (2S-5,7,4′-trihydroxyflavanone) (Lewis et al., 2024). CHI performs this reaction much more efficiently than the spontaneous cyclization reaction, which produces both 2S- and 2R-isomers. Only 2S-isomers, which act as biological precursors for flavonoid biosynthesis in nature, are produced through catalysis by CHI (Bednar and Hadcock, 1988). The CHI superfamily comprises four types: I, II, III, and IV (Park et al., 2018; Ralston et al., 2005). Type I CHI enzymes, which are ubiquitous in vascular plants, exhibit typical CHI catalytic activity, converting naringenin chalcone to (2S)-naringenin. Type II CHI enzymes (which are specific to legumes) have broader substrate specificity, catalyzing the conversion of isoliquiritigenin (6′-deoxychalcone) to 2S-liquiritigenin (5-deoxyflavanone) as well as naringenin chalcone (6′-hydroxychalcone) to 2S-naringenin (5-hydroxyflavanone) (Cheng et al., 2018). Type III CHI enzymes lack chalcone cyclization activity but possess fatty acid binding properties, influencing fatty acid biosynthesis and storage in developing Arabidopsis embryos (Ngaki et al., 2014). Type IV CHI enzymes lack catalytic activity but can enhance flavonoid production and flower pigmentation (Morita et al., 2014). CHI-like protein (CHIL) binds to CHS and enhances CHS-catalyzed THC production, resulting in improved metabolic flux from the general phenylpropanoid pathway to the flavonoid pathway (Waki et al., 2020).

Cheng et al. (2018) compared the activities of CHIs isolated from five different plants. The efficiencies of these enzymes for naringenin chalcone ranged from 49.8 × 10⁶ to 122 × 10⁶, representing a 2.4-fold difference ( Supplementary Table S2 ). Among these enzymes, MpCHI isolated from Marchantia paleacea exhibited the highest enzyme efficiency. Ralston et al. (2005) reported the enzyme kinetics of three soybean (Glycine max) CHIs (CHI1A, CHI1B2, and CHI2) for seven chalcone substrates with different degrees of hydroxylation. GmCHI2, a type I CHI, uses only naringenin chalcone (4,2′,4′,6′-tetrahydroxychalcone) as a substrate. Two CHI type II enzymes, GmCHI1A and GmCHI1B2, have similar enzyme efficiencies for seven different chalcone substrates. In addition, Deschampsia antarctica CHI (DaCHI) exhibits strong substrate preference for naringenin chalcone (type I substrate) but weak substrate preference for isoliquiritigenin (type II substrate) (Park et al., 2018). Of the four OsCHIs examined in rice (OsCHI1, OsCHI3, OsCHI6, and OsCHI7), only type I OsCHI3 exhibited CHI activity for naringenin chalcone. However, OsCHI1 activity for isoliquiritigenin was not detected (Park et al., 2021). Thus, CHI enzymes, which show promiscuity for diverse substrates, play important roles in increasing the diversity of flavonoids together with CHS. Among CHI enzymes mentioned above, GmCHI2 showed the highest enzyme efficiency for naringenin chalcone, followed by DaCHI1. Thus, these two enzymes could be selected to improve naringenin production ( Table 2 ).

The evolution of land plants was accompanied by the emergence of different classes of flavonoids. Liverworts, the most primitive land plants, utilize FNSI/flavanone 2-hydroxylase (F2H) to produce flavones and 2-hydroxyflavones. FNSI/F2H evolved into FNSI/F3H in moss. Such a transition further developed in gymnosperms, and the enzyme completely shifted to the bona fide F3H in angiosperms. FNSI has been lost in most angiosperms, as its role in flavone production was replaced by the cytochrome P450 monooxygenase FNSII enzymes that have emerged in angiosperms (Li et al., 2020a; Du et al., 2016). However, distinct FNSIs have emerged through duplication and substitution events of F3H in Apiaceae plants (Gebhardt et al., 2005, 2007). Besides Apiaceae FNSIs, additional FNSIs have arisen independently from different lineages in O. sativa, Z. mays, and Arabidopsis. FNS converts flavanones to flavones by forming a double bond between C2 and C3 of flavanones. FNSI, a member of the 2-oxoglutarate-dependent dioxygenase (2-ODD) superfamily, requires a non-heme ferrous iron, 2-oxoglutarate, and O₂ for the catalytic reaction, while FNSII is a member of the CYP93 subfamily (Martens and Mithöfer, 2005). Most FNSIIs in monocots and dicots belong to the CYP93G and CYP93B subfamilies, respectively (Akashi et al., 1999). Among FNSIs, although the primitive liverwort (Conocephalum japonicum) CjFNSI/F2Hs show promiscuous characteristics, exhibiting FNS, F2H, and FLS activities and relatively high substrate-binding affinity, FNSI (PcFNSI) from the Apiaceae family member parsley shows the highest catalytic efficiency owing to its overwhelmingly high binding affinity for naringenin. Among FNSIIs, Lonicera (honeysuckle) FNSIIs show the highest V_max values, and they prefer eriodictyol to naringenin as a substrate for their FNS activity ( Table 2 ), suggesting that they are suitable for 3′,4′-dihydroxyflavone production.

F3H converts flavanones such as naringenin, eriodictyol, and pentahydroxyflavanone to the dihydroflavonols dihydrokaempferol (DHK), dihydroquercetin (DHQ), and dihydromyricetin (DHM), respectively. Dihydroflavonols serve as common precursors for three major classes of end products: flavonols, anthocyanins, and proanthocyanidins (Park et al., 2019; Lei et al., 2023). Thus, F3H plays a crucial role as a key branchpoint in the flavonoid biosynthesis pathway. F3H belongs to the 2-ODD superfamily, which also includes FLS, anthocyanidin synthase (ANS), FNSI, and flavonol 6-hydroxylase (Wang et al., 2021). The 2-ODDs catalyze a variety of oxidation reactions (such as hydroxylation, desaturation, and oxidative ring closure) in plants, animals, and microorganisms, participating in a diverse array of primary and specialized metabolic pathways (Cheng et al., 2014). Among the three rice F3Hs, OsF3H1 has the highest enzyme efficiency for eriodictyol, as it is 1,000 times more efficient than OsF3H2 and ~100 times more efficient than OsF3H3 (Kim et al., 2008). Although enzyme activities were not characterized, several F3H gene and promoter combinations were tested to produce DHQ in yeast, with Citrus sinensis F3H (CsF3H) showing the highest activity (Yu et al., 2022). Studies of F3H activity from three different species (AtF3H, Arabidopsis thaliana; CtF3H, Carthamus tinctorius; GmF3H, Glycine max) for naringenin or eriodictyol substrate (Owens et al., 2008; Tu et al., 2016; Kim et al., 2008) revealed that AtF3H had the lowest K_m value of the three ( Supplementary Table S2 ). Although enzyme activity data are lacking, OsF3H1 can be used to convert eriodictyol to DHQ ( Table 2 ).

Along with FNSI, F3H, and ANS, FLS belongs to the 2-ODD superfamily. Phylogenetic analysis suggested that the divergence of F3H, ANS, and FLS preceded the split of gymnosperms and angiosperms and that FLS emerged most recently (Choudhary and Pucker, 2024). FLS was first identified in parsley suspension cell cultures and was subsequently characterized in various species, such as tea plant (Camellia sinensis), Zea mays, and Gingko biloba ( Table 2 ). FLS catalyzes the oxidation of the C-ring of dihydroflavonols (DHK, DHQ, and DHM) to generate flavonols (kaempferol, quercetin, and myricetin, respectively). Several FLSs are bifunctional, exhibiting both F3H and FLS activities (Chua et al., 2008; Park et al., 2023). Based on studies of Arabidopsis FLS1, five residues were shown to be involved in the binding of dihydroflavonol substrates. Among them, H132, which interacts with the B-ring hydroxyl group of the substrate, is thought to determine substrate preference. H132 allows FLS to accept all three types of dihydroflavonol substrates. This residue has been substituted with Y in gymnosperms and some monocots, and F in most other monocots, which allows FLSs to preferentially accept DHK and DHQ, respectively (Choudhary and Pucker, 2024). According to the available enzyme kinetics data, Zea mays FLS (ZmFLS) with an F132 residue shows strong catalytic efficiency among FLSs, with a superior V_max value for dihydroflavonol substrates. Its most preferred substrate is DHK, whereas FLS from the monocot onion (Allium cepa; AcFLS-HRB), which harbors Y132, exhibits a preference for DHQ over DHK ( Supplementary Table S2 ). Therefore, the position corresponding to H132 might be an important factor in determining the flavonol product. The kinetics data show the F3H activities of Ornithogalum caudatum FLS1 (OcFLS1) and Rubus chingii FLS1 (RcFLS1), which are available to convert flavanones to dihydroflavonols ( Table 2 ), indicating that the underlying genes could be employed for metabolic engineering to improve the efficiency of flavonol production.

DFR catalyzes reduction of dihydroflavonols (DHK, DHQ, and DHM) to form leucoanthocyanidins (leucopelargonidin, leucocyanidin, and leucodelphinidin, respectively), which are then converted to colored anthocyanidins by leucoanthocyanidin dioxygenase/anthocyanin synthase (LDOX/ANS). Proanthocyanidins first appeared in lycophytes, a group of seedless vascular plants. However, phylogenetic analysis showed that the ancient ancestor of DFR can be traced back to moss (Campanella et al., 2014). DFRs, which belong to the short-chain dehydrogenase family, require NAD(H) or NADP(H) as a cofactor. To date, most DFRs have been isolated from flowering plants, and their characteristics have been elucidated. DFR sequences harbor a conserved NADPH-binding domain and a substrate-binding domain. The third residue within the 26-amino-acid substrate-binding domain determines the substrate specificity of DFR. Monocots commonly contain an N residue at that position, which allows DFR to accept all three types of dihydroflavonols as substrates, whereas most dicots have D or A as well as an N residue. The D residue confers preferences for DHQ and DHM, while the A residue confers a strong preference for DHK. Camellia sinensis DFRa (CsDFRa), with a substrate specificity–determining N residue, strongly prefers DHQ and DHM over DHK. Strawberry (Fragaria × ananassa) ‘Elsanta’ DFR1 and DFR2 (DFR1 ES and DFR2 ES), which harbor an A and N residue, respectively, and show strikingly low K_m values for DHK and DHQ, respectively, compared to other DFRs ( Table 2 ). To maximize the efficiency of producing anthocyanin-derived end products in plants using DFR, an elaborate strategy is required to avoid overlapping between the substrate preferences of FLS in the host plant and the DFR utilized, which could mitigate the competition between these enzymes for dihydroflavonol substrates.