Genetic Control and Evolution of Anthocyanin Methylation

Abstract

Anthocyanins are a chemically diverse class of secondary metabolites that color most flowers and fruits. They consist of three aromatic rings that can be substituted with hydroxyl, sugar, acyl, and methyl groups in a variety of patterns depending on the plant species. To understand how such chemical diversity evolved, we isolated and characterized METHYLATION AT THREE2 (MT2) and the two METHYLATION AT FIVE (MF) loci from Petunia spp., which direct anthocyanin methylation in petals. The proteins encoded by MT2 and the duplicated MF1 and MF2 genes and a putative grape (Vitis vinifera) homolog Anthocyanin O-Methyltransferase1 (VvAOMT1) are highly similar to and apparently evolved from caffeoyl-Coenzyme A O-methyltransferases by relatively small alterations in the active site. Transgenic experiments showed that the Petunia spp. and grape enzymes have remarkably different substrate specificities, which explains part of the structural anthocyanin diversity in both species. Most strikingly, VvAOMT1 expression resulted in the accumulation of novel anthocyanins that are normally not found in Petunia spp., revealing how alterations in the last reaction can reshuffle the pathway and affect (normally) preceding decoration steps in an unanticipated way. Our data show how variations in gene expression patterns, loss-of-function mutations, and alterations in substrate specificities all contributed to the anthocyanins' structural diversity.

Figure 1.

Role of loci and enzymes required for methylation of anthocyanins in Petunia spp. A, Diagram depicting the modification of simple anthocyanin 3-glucosides by subsequent rhamnosylation, 5-glucosylation, acylation, and methylation. Enzymes involved in each reaction are indicated on the right of the arrows and genetic loci (italics) controlling the reaction on the left. B, RNA gel-blot analysis of DIFe1 expression in petals of four inbred Petunia spp. lines that are homozygous for functional (+) or mutant alleles (–) of MT2, MF1, and MF2. As a control for RNA loading and integrity, the filter was stripped and rehybridized with DIFi.

Figure 2.

Methylation of anthocyanins by DIFe1 and DIFe2 in vitro and in vivo. A, Anthocyanidins (percentage of total) after incubation of SAM and the 3-glucoside (3G), 3-rutinoside (3R), or 3,5 diglucoside (3,5G) of delphinidin with crude extracts of E. coli expressing DIFe1, the empty vector, or no extract at all (none). Mal, Malvidin; Pet, petunidin; Del, delphinidin; Peo, peonidin; Cya, cyanidin. B, Phenotype of a flower from a control plant and MAC:asDIFe1 transformant A. C, Composition of anthocyanidins in flowers of independent MAC:asDIFe1 and MAC:asDIFe2 lines and a control plant without transgene.

Figure 3.

DNA gel-blot analysis of DIFe genes/paralogs in P. hybrida. DNA from P. hybrida lines M1, V30, and R78 was digested with EcoRI (E) or BamHI (B), size separated, hybridized to DIFe2^OGB cDNA, washed at low stringency (2× SSC, 25°C; left) and subsequently at high stringency (0.1× SSC 68°C; middle), or hybridized with the DIFe1 cDNA and washed using high-stringency conditions (right). Fragments hybridizing to DIFe1 and DIFe2 at high stringency are marked with white and black arrowheads, respectively.

Figure 4.

Phylogram of A1-type plant O-methyltransferases. The tree (maximum likelihood) is based on a cured sequenced alignment (Supplemental Table S3). Numbers on branches indicate percentage bootstrap support (500 replicates). For each protein, the name, accession number, and substrates accepted in vitro is given. CafCoA, Caffeoyl-CoA; CafAcid, caffeic acids; CafEster, caffeic acid esters; At, Arabidopsis; Ckm, Cyclamen persicum × purpurascens; Fv, F. vesca; Mc, Mesembryanthemum crystallinum; Nt, Nicotiana tabacum; Os, rice; Pc, Petroselinum crispum; Ph, P. hybrida; Pp, P. persica; Pt, Populus trichocarpa; Sl, S. lycopersicum; St, S. tuberosum; Stl, Stellaria longipes; Vv, V. vinifera; Zm, maize; Zv, Zinnia violacea.

Figure 5.

Expression pattern and regulation of DIFe genes. A, Quantitative RT-PCR analysis of MT2, MF1, and MF2 mRNAs in different tissues of the P. hybrida F1 hybrid M1 × V30. Floral tissues were from buds of three different stages. Stage 1/2, buds up to 20 mm; stage 3/4, 30- 40-mm buds; and stage 5/6, open flowers. GLYCERALDEHYDE 3-PHOSPHATE DEHYDROGENASE (GAPDH) and ACTIN were used as internal controls. B, Real-time PCR analysis of mRNAs expressed in stage 3/4 petals of regulatory mutants. Lines R27 (AN1 AN2 AN11), W225 (an1), and W134 (an11) are isogenic. Line W242 is in a distinct genetic background and harbors a mutable an2 allele (an2^mut). A germinal AN2^REV revertant in which the transposon had excised from AN2 was used as an isogenic wild type. mRNA levels were normalized using ACTIN, and expression in R27 was set to 1. Primers used for real-time PCR could not distinguish between DIFe2a/MF1 and DIFe2b/MF2.

Figure 6.

Structure of wild-type and mutant alleles of MT2 (DIFe1), MF1 (DIFe2b), and MF2 (DIFe2a). A, Quantitative RT-PCR analysis of DIFe mRNAs in genotypes with mt, mf1, and/or mf2 alleles. B, Structure of DIFe1/MT2 locus in P. hybrida (Ph), P. inflata (Pi), and P. axillaris (Pa) lines with different MT2 MF1 MF1 genotypes (+, homozygous for dominant allele; –, homozygous for recessive nonfunctional allele; and ×, homozygous dominant for MF1 and/or MF2). C, Structure of DIFe2a/MF1 and DIFe2b/MF2 locus in P. hybrida (Ph), P. inflata (Pi), and P. axillaris (Pa) lines with different genotypes. Coding and noncoding sequences (introns are not drawn to scale) are indicated by rectangles and lines, respectively, and start and stop codons by white and black circles. Numbers denote the number of nucleotides in (functional) exons and introns. Lesions found in mutant alleles are indicated above the gene maps.

Figure 7.

Complementation of mf and mt mutants by 35S:MT2, 35S:MF2, and 35S:VvAOMT1. A, Flowers of untransformed R78 and R78 × V32 plants and transgenic siblings that express 35S:MT2, 35S:MF2, or 35S:VvAOMT1. B, Composition of anthocyanidins obtained by acid hydrolysis of anthocyanins in petals of transgenic and control plants, as determined by HPLC. Number signs indicate specific transformants. C, LC-MS/MS analysis of anthocyanins in R78 petals (control) and transgenic siblings expressing 35S:MT2, 35S:MF2, or 35S:VvAOMT1. D, Quantitative RT-PCR analysis of transgene expression in distinct transformants. Number signs denote individual transformants. E, LC-MS/MS analysis of anthocyanins in R78 × V32 petals (control) and transgenic siblings expressing 35S:MT2, 35S:MF2, or 35S:VvAOMT1. Diagrams in C and E show LC profiles and, for the major peaks, m/z values of the corresponding ion and, in brackets, subfragments observed in MS/MS(+) spectra (see Supplemental Figs. S4–S9).

Figure 8.

Two possible binding modes of delphinidin 3-(p-coumaroyl) rutinoside 5-glucoside. A, Alignment of CCoA-MTs and AMTs. Amino acids and motifs of Ms CCoA-MT involved in dimerization, binding of SAM/S-adenosyl-L-homo-Cys (SAH), the caffeoyl or CoA moiety, and divalent cation are indicated. B, Substrate binding with the 3-(p-coumaroyl)rutinoside chain pointing into the active site and clashing with bulky residues (Tyr-208 and Tyr-212), which are, in the Petunia spp. AMTs, replaced by smaller ones (Gly and Leu). For clarity, the 5-Glc group of the anthocyanin is not shown. C, Side view of B. D, Substrate binding with the 3-(p-coumaroyl)rutinoside chain oriented toward the active site and clashing with bulky residues (Trp-58, Asn-116, and Tyr 208), which are, in the Petunia app. AMTs, replaced by Ala/Gly, Ala, and Gly, respectively. For clarity, the 5-Glc group of the anthocyanin is not shown. E, Side view of D.