A flavin-dependent monooxygenase catalyzes the initial step in cyanogenic glycoside synthesis in ferns

Abstract

Cyanogenic glycosides form part of a binary plant defense system that, upon catabolism, detonates a toxic hydrogen cyanide bomb. In seed plants, the initial step of cyanogenic glycoside biosynthesis-the conversion of an amino acid to the corresponding aldoxime-is catalyzed by a cytochrome P450 from the CYP79 family. An evolutionary conundrum arises, as no CYP79s have been identified in ferns, despite cyanogenic glycoside occurrence in several fern species. Here, we report that a flavin-dependent monooxygenase (fern oxime synthase; FOS1), catalyzes the first step of cyanogenic glycoside biosynthesis in two fern species (Phlebodium aureum and Pteridium aquilinum), demonstrating convergent evolution of biosynthesis across the plant kingdom. The FOS1 sequence from the two species is near identical (98%), despite diversifying 140 MYA. Recombinant FOS1 was isolated as a catalytic active dimer, and in planta, catalyzes formation of an N-hydroxylated primary amino acid; a class of metabolite not previously observed in plants.

Fig. 1. A schematic overview of the biosynthetic pathway of cyanogenic glycosides in plants.

In ferns, the conversion of the parent amino acid into an oxime is catalyzed by a multifunctional FMO, whereas in all higher seed plant species analyzed, the reaction is catalyzed by a cytochrome P450 from the CYP79 family^,,,,,,–. *Uncharacterized pathway partners.

Fig. 2. Cyanogenic glycoside content and FMO transcript abundances in tissues from the two modern fern species Pteridium aquilinum and Phlebodium aureum.

a Two phenylalanine-derived cyanogenic glycosides have been reported from ferns: the monoglucoside prunasin (D-mandelonitrile-β-D-glucopyranoside) and the diglycoside vicianin (6-O-arabinopyranosylglucopyranoside). b Phylogenetic relationship between Pteridium aquilinum (Dennstaedtiaceae) and Phlebodium aureum (Polypodiaceae) showing that these two modern ferns species diversified 140 million years ago (tree adapted from). c Content of vicianin across different tissue types of P. aureum (see also Supplementary Fig. 1), with the blue bars indicating the tissue selected for downstream transcriptomic analysis. d Content of prunasin in the pinnae of a population of 25 field-collected P. aquilinum, with the green bars indicating individuals selected for transcriptomic analysis. e Transcript abundance of predicted flavin monooxygenases (FMOs) in tissues of P. aquilinum and P. aureum containing high (gray bars) or low (black bars) cyanogenic glycoside levels. Arrows indicate candidate genes. TPM transcripts per million mapped reads. f Schematic illustration of identity and motifs in the transcriptome-deduced amino acid sequences of Paq18302 (PaqFOS1) and Pa22578. Differences between Pa22578 and the isolated PaFOS1 are indicated by blue lines. The position of putative binding motifs for FAD and NADPH, the FATGY and FMO-identifier motifs conserved across plant FMOs are highlighted. Supporting alignment is shown in Supplementary Fig. 1.

Fig. 3. Schematic diagram illustrating the occurrence of CYP families across plant taxa, based on the known CYP families present in eudicots.

The diagram is based on analysis of the OneKP database, which includes transcriptomes from 74 ferns. The presence of 8930 cytochromes P450 fern sequences were predicted and sorted into families in accordance with nomenclature. The analyzed fern transcriptomes harbor at least 81 different P450 families of which 49 (60%) are novel fern-specific families. Approximately half of the CYP families present in higher plants are also found in ferns. The CYP79 family is present from gymnosperm to eudicots, but based on the >40% sequence identity, the CYP79 is absent in ferns. This also applies to the other known CYP families involved in cyanogenic glucoside biosynthesis in plants: the CYP71, CYP706, and CYP736 families.

Fig. 4. LC–MS based metabolite analyses of Nicotiana benthamiana leaves transiently expressing PaqFOS1.

a Extracted ion chromatograms (EICs) for m/z 136 corresponding to the [M+H]⁺ adduct of authentic phenylacetaldoxime (upper panel), metabolite extracts from N. benthamiana leaves expressing PaqFOS1 (middle panel) and empty vector control (lower panel). b m/z 318 EICs corresponding to the [M+Na]⁺ adduct of an authentic prunasin standard (upper panel), metabolite extracts from N. benthamiana transiently expressing PaqFOS1 in combination with PdCYP71AN24 and PdUGT85A19^, (middle panel) and the control expressing PdCYP71AN24 and PdUGT85A19 (lower panel). c Base peak chromatograms (BPCs) of the metabolite extracts from N. benthamiana leaves expressing PaqFOS1 using expression of p19 as an empty vector control show the formation of additional products: m/z 320 at 6.7 min corresponds to the [M+Na]⁺ adduct of glucosylated phenylacetaldoxime; m/z 406 at 7.6 and 7.9 min correspond to the [M+Na]⁺ adduct of a glycosylated, phenylacetaldoxime-malonic acid conjugate; and m/z 393.11 at 8.1 min correspond to the [M+Na]⁺ adduct of phenylethanol glucoside malonate ester^,. For MS/MS of additional products, see Supplementary Fig. 3.

Fig. 5. Size exclusion chromatography elution profile of PaqFOS1 monitored at 280, 254, and 450 nm corresponding to absorbance (mAU) of the polypeptide, NADPH, and FAD, respectively.

Based on the elution volume in comparison to a set of reference proteins with known molecular masses, the PaqFOS1 protein eluted as a dimer with a mass of ~120 kDa. SDS-PAGE analysis of the PaqFOS1 containing fractions obtained by size exclusion chromatography demonstrating that PaqFOS1 migrated with an apparent molecular mass of 60 kDa in agreement with a calculated molecular mass of 62.5 kDa for the monomeric protein.

Fig. 6. The formation of N-hydroxyphenylalanine (NOH-phe).

LC–MS chromatograms from targeted analysis of samples compared with the authentic standard, derived from a transient expression of FOS1 in N. benthamiana showing the p19 as a negative control, b heterologous expression of FOS1 in E. coli using the absence of NADPH as a negative control, and c the presence of N-hydroxyphenylalanine in P. aureum tissue. The chromatographic trace represents the abundance of the fragment ion of N-hydroxyphenylalanine (precursor ion → fragment ion of 182.1 → 136.0; see Supplementary Table 4). d The hypothesized biosynthetic route from phenylalanine to phenylacetaldoxime, as catalyzed by FOS1. e In vitro activity assay of recombinant PaqFOS1 using l-phenylalanine (Phe) as a substrate, with different combinations of the necessary cofactors FAD and NADPH. Bars represent mean ± SE (n ≥ 3). f The presence of N-hydroxyphenylalanine in P. aureum fiddlehead and young pinnae metabolite extracts (n = 3).

Fig. 7. Phylogenetic tree of the flavin monooxygenase (FMO) superfamily containing all predicted full-length FMOs from ferns (P. aureum and P. aquilinum) together with FMOs from eight higher plant species.

As all species are represented in each of the tree clades, the phylogenetic analysis suggests an early diversification of the groups prior to species differentiation. Employed sequence IDs are compiled in Supplementary Data 2. Characterized FMOs (or clusters of all characterized FMOs such as the Arabidopsis YUCCAs) are indicated by a star.