The causal mutation leading to sweetness in modern white lupin cultivars

Abstract

Lupins are high-protein crops that are rapidly gaining interest as hardy alternatives to soybean; however, they accumulate antinutritional alkaloids of the quinolizidine type (QAs). Lupin domestication was enabled by the discovery of genetic loci conferring low QA levels (sweetness), but the precise identity of the underlying genes remains uncertain. We show that pauper, the most common sweet locus in white lupin, encodes an acetyltransferase (AT) unexpectedly involved in the early QA pathway. In pauper plants, a single-nucleotide polymorphism (SNP) strongly impairs AT activity, causing pathway blockage. We corroborate our hypothesis by replicating the pauper chemotype in narrow-leafed lupin via mutagenesis. Our work adds a new dimension to QA biosynthesis and establishes the identity of a lupin sweet gene for the first time, thus facilitating lupin breeding and enabling domestication of other QA-containing legumes.

Fig. 1.. The distinctive chemotype of sweet white lupins of the pauper type.

(A) Early QA biosynthesis pathway in bitter lupins according to the prevalent pathway hypothesis. spont.: spontaneous reaction. (B) Proposed QA pathway blockage accounting for sweetness in pauper white lupin. Distinctive metabolites accumulate in pauper plants (pauper metabolites), suggesting that the blockage occurs just downstream of the step catalyzed by CAO. Dashed arrows show the formation of pauper metabolites upon accumulation of the immediate product of CAO (Δ¹-piperideine, 3) and its spontaneously formed dimer (tetrahydroanabasine, 5). The origin of the pauper metabolites was inferred from their MS² spectra (fig. S1), particularly from the presence of fragments of m/z 84.08 (corresponding to Δ¹-piperideine, 3) and m/z 167.15 (corresponding to tetrahydroanabasine, 5). (C) Representative LC-MS chromatograms of leaf extracts from two bitter (P27174, Graecus) and one pauper sweet (Amiga) white lupin lines (top three chromatograms) as well as an extract of N. benthamiana leaves transiently expressing LDC and CAO (bottom chromatogram). Accumulation of pauper metabolites is observed in the leaves of N. benthamiana upon transient coexpression of LDC and CAO. Traces are extracted ion chromatograms (EICs) of the most representative pauper metabolites (mean m/z ± 0.01). Signal intensities were adjusted to aid visualization (see Materials and Methods for scaling factors).

Fig. 2.. A SNP variant in AT is uniquely associated with the pauper chemotype.

(A) Position of four SNPs in the coding sequence of AT and their associated variants in Amiga (pauper sweet), P27174 (bitter), and Graecus (bitter). The SNP variants found in Amiga (pauper sweet) are highlighted with colored backgrounds. (B) Leaf chemotype and AT SNP variants of a diversity panel derived from 24 white lupin accessions (table S1), as determined by LC-MS and Sanger sequencing. SNP variants identical to those found in Amiga (pauper sweet) are highlighted with colored backgrounds [as in (A)], with heterozygous SNPs displaying diagonally split colored backgrounds. The homozygous SNP_2 variant “A” is exclusively associated with the pauper sweet chemotype.

Fig. 3.. A single SNP in AT leads to strongly impaired AT enzyme activity.

(A) Representative LC-MS chromatograms of extracts from leaves of N. benthamiana expressing LDC, CAO, and AT (P27174 bitter version) in different combinations, showing the accumulation of ammodendrine (6) and its precursor, tetrahydroanabasine (5). MS² spectrum (22.9 eV) of ammodendrine (6) is also shown (light gray box). Traces are combined EICs of m/z 167.15 ± 0.01 and m/z 209.16 ± 0.01. (B) In vitro acetylating activity of AT (P27174 bitter version) against the Δ¹-piperideine monomer (3) and its dimer (tetrahydroanabasine, 5), showing strong preference for the latter. Monomer and dimer solutions were incubated at pH 7.8 for up to 24 hours before initiating the assay to probe the equilibrium between (3) and (5) at near-physiological pH. Lines connect the mean relative activities at each time point (n = 4). (C) In vitro acetylating activity of different versions of AT against tetrahydroanabasine (5) showing the effect of the pauper variant at SNP_2. In the box plots, the center line represents the median, the box limits represent the upper and lower quartiles, and the whiskers represent maximum and minimum values. Data labels are mean activities relative to P27174 (bitter version) ± 95% confidence interval (n = 4). Letters indicate significant differences [one-way analysis of variance (ANOVA) with Tukey’s post hoc test, P < 0.05].

Fig. 4.. Inactivation of the AT gene in narrow-leafed lupin leads to pauper-like sweetness.

(A) Genotyping of mutagenized narrow-leafed lupin plants carrying the wild-type AT allele (AT^WT) or an early stop codon allele (AT^KO). (B) Distribution and abundance of 11 major QAs (7 to 17) in the leaves of narrow-leafed lupin carrying the AT^WT and AT^KO alleles. Traces are combined EICs of m/z 235.18 ± 0.01, m/z 249.20 ± 0.01, m/z 265.19 ± 0.01, m/z 347.23 ± 0.01, m/z 369.22 ± 0.01, m/z 395.23 ± 0.01, m/z 411.23 ± 0.01, and m/z 429.24 ± 0.01. In the box plots, the center line represents the median, the box limits represent the upper and lower quartiles, and the whiskers represent maximum and minimum values. Data labels represent the relative abundance (%) in homozygous AT^KO plants relative to homozygous AT^WT plants (n = 6). n.d., not detected at the working dilution. (C) Representative LC-MS chromatograms showing the accumulation of pauper metabolites and the absence of ammodendrine (6) in homozygous AT^KO plants. EIC traces correspond to m/z 167.15 ± 0.01 (blue), m/z 199.14 ± 0.01 (cyan), m/z 209.16 ± 0.01 (orange), and m/z 343.19 ± 0.01 (red).