Convergent evolution of berberine biosynthesis

Abstract

Berberine is an effective antimicrobial and antidiabetic alkaloid, primarily extracted from divergent botanical lineages, specifically Coptis (Ranunculales, early-diverging eudicot) and Phellodendron (Sapindales, core eudicot). In comparison with its known pathway in Coptis species, its biosynthesis in Phellodendron species remains elusive. Using chromosome-level genome assembly, coexpression matrix, and biochemical assays, we identified six key steps in berberine biosynthesis from Phellodendron amurense, including methylation, hydroxylation, and berberine bridge formation. Notably, we discovered a specific class of O-methyltransferases (NOMT) responsible for N-methylation. Structural analysis and mutagenesis of PaNOMT9 revealed its unique substrate-binding conformation. In addition, unlike the classical FAD-dependent berberine bridge formation in Ranunculales, Phellodendron uses a NAD(P)H-dependent monooxygenase (PaCYP71BG29) for berberine bridge formation, originating from the neofunctionalization of tryptamine 5-hydroxylase. Together, these findings reveal the convergence of berberine biosynthesis between Coptis and Phellodendron and signify the role of the convergent evolution in plant specialized metabolisms.

Fig. 1.. Protoberberine and aporphine-type BIAs produced in P. amurense.

(A) The biosynthetic pathway of berberine (11). The black arrows represent the enzymes reported from Coptis, and the dashed arrows mean that these steps in P. amurense remain elusive. 1, dopamine; 2, 4HPAA; 3, (S)-norcoclaurine; 4, (S)-coclaurine; 5, (S)-N-methylcoclaurine; 6, (S)-3′-hydroxy-N-methylcoclaurine; 7, (S)-reticuline; 8, (S)-scoulerine; 9, (S)-tetrahydrocolumbamine; 10, (S)-canadine; and 11, berberine. Rt, retention time. (B) Phylogenetic tree and BIA metabolite profiles of tested species. The phylogenetic tree was constructed using single-copy orthologs from P. amurense and other 18 species, based on transcriptome or genome data (table S1). Metabolites were extracted from roots (yellow) and/or stems (green) of 17 species, as well as lotus plumule of N. nucifera (blue). The heatmap was developed with the BIA levels indicated by the values of log₁₀(peak area + 1). BIA types: 1-BIA, protoberberine, protopine, and aporphine. 12, (S)-magnocurarine; 13, tembetarine; 15, phellodendrine; 15, groenlandicine; 16, columbamine; 17, jatrorrhizine; 18, coptisine; 19, palmatine; 20, protopine; and 21, magnoflorine. (C) A microscopic image shows a cross section of P. amurense roots, and MALDI-MSI images generated with heatmaps show the accumulation patterns of metabolites.

Fig. 2.. Genomic features, phylogenetic tree, and WGD events of the P. amurense genome.

(A) Genome characteristics of P. amurense. The circos plot from the outmost to the inmost circle visualizes 38 pseudo-chromosomes (I), gene density represented as number of genes per Mb (II), GC content in 1 Mb windows (III), density of repeat sequences (IV), and each linking line in the center of the circos plot showing a pair of homologous genes (V). (B) A phylogenetic tree developed with single-copy orthologs from P. amurense and 12 other candidate species. Green and red colored numerals highlighted at each branch line mean the number of the expanded and contracted gene families, respectively. Pie charts show the proportions of gene families that have undergone contraction (red) or expansion (green). Divergence timings with a 95% CI were highlighted with horizontal purple bars at the internodes. WGT-γ means the gamma triplication event. (C) Collinearity between the P. amurense and C. clementina genomes. The C. clementina genome was used as reference to identify synteny blocks and orthologous gene pairs for each chromosome in the genome of P. amurense. (D) Collinearity between the Z. armatum and C. clementina chromosomes. (E) Phylogenetic trees were, respectively, constructed using the homologous genes from different chromosomes of P. amurense subgenomes and Z. armatum subgenomes according to the collinearity mapping between candidate species and C. clementina chromosomes (Chr1 to Chr9). The red, blue, green, and orange boxes in P. amurense and Z. armatum represent four types of synteny blocks with C. clementina. (F) K_S distribution for orthologous and paralogous genes among P. amurense, Z. armatum, P. amurense subgenomes, Z. armatum subgenomes, and C. clementina. (G) Predicted WGD of P. amurense and Z. armatum. WGD1 and WGD2 represent the WGD events shared between the P. amurense and Z. armatum genomes.

Fig. 3.. Elucidation of the berberine biosynthetic pathway in Phellodendron.

(A) Pearson correlation coefficient between the P. amurense transcriptome data and berberine content (r > 0.80). The heat map displays the z-score calculated from log₁₀ (FPKM + 1). CYP450s, highlighted in gray, are unsuccessfully cloned, partial genes, or the homologous genes of reported limonin biosynthesis (fig. S21). FPKM, Fragments Per Kilobase of exon model per Million mapped fragments. (B) Extracted ion chromatograms of compounds catalyzed by purified PaOMT1 and PaOMT2 using 3, 4, 6, and 8 as substrates. (C) Extracted ion chromatograms of compounds catalyzed by PaCYP71BG29 microsomes using 5 and 7 as substrates. (D) Pearson correlation coefficient between the P. amurense transcriptome data and PaOMT1 gene. Gray-highlighted genes are unsuccessfully cloned or the homologous genes of reported limonin biosynthesis. (E) Extracted ion chromatograms of compounds catalyzed by purified PaOMT4/5/7/8/9 using 3, 4, 6, and 8 as substrates. (F) This scheme summarizes these catalytic steps that have been verified with enzymatic data provided in this report. The catalytic activities of PaCYP71BG29 were doubled confirmed using yeast and tobacco expression systems. N-methylated OMT members were renamed as NOMTs. T_min, time in minutes.

Fig. 4.. Crystal structure and catalytic mechanism of PaNOMT9.

(A) An overall view of the crystal structure of PaNOMT9/4/SAM. (B) Protein-ligand interaction diagram between PaNOMT9 and the ligand 4. Red marks signify the hydrophobic residues. (C) The sequence alignment of amino acids among N-methylation active OMT proteins PaNOMT9/2/7/8 and O-methylation OMT proteins PaOMT1/3/4/5. (D) Catalytic activities of PaNOMT9 and 20 variants resulted from targeted mutations are shown with stacked bars. (E) Crystal structure of ternary complex formed by PaNOMT9/4/SAM, in which the mutations of those crucial residues reduce the catalytic activities. The residues highlighted with cyan color indicate the proximity to the substrate within 5 Å, while the residues colored with orange are outside this range. The dotted gray line means the distance between the N atom of 4 and the methylation group of SAM is 4.5 Å. (F) Overlay of the crystal structures of two ternary complexes: Tf6OMT/norlaudanosoline/SAH and PaNOMT9/4/SAM. The protein-ligand binding conformation differences of these two ternary complexes lead to the function divergence of O-/N-methylation.

Fig. 5.. Determination of the neofunctionalization of PaCYP71BG29 and its catalytic mechanism.

(A) A maximum-likelihood phylogenetic tree was built with the PaCYP71BG subfamily members from P. amurense, Z. armatum, and C. clementina. OsCYP71P1 was chosen as the outgroup. K_a/K_s, nonsynonymous (K_a) with synonymous (K_s) substitution rates. (B and C) Extracted ion chromatograms (EIC) characterized compounds formed in the enzymatic reactions consisting of candidate CYP microsomes and two substrates-tryptamine 27 and (S)-reticuline 7. (D) Sequence alignment of PaCYP71BG29, ZaCYP71BG29, and other CYP71BG subfamily members. Twenty-four amino acids were presented, and six variation sites colored with red are within 5 Å from 7. (E) Stacked bars show catalytic preference of PaCYP71BG29 and its 24 variants toward two substrate: (S)-reticuline 7 and (S)-scoulerine 8. (F) Molecular docking model of PaCYP71BG29/HEME/7 indicates those crucial residues, which enhance (red) and reduce (blue) the catalytic activities. The distances among between substrate 7 and crucial residue E213 and HEME are 3.4 and 2.6 Å. (G) A mechanism is proposed for the formation of the berberine bridge catalyzed by PaCYP71BG29.