Supporting Information Legends
Table S1. Sequences of primers used in this study.
Figure S1. Position of all nucleotide changes identified by TILLING of the L. japonicus
BGD4 gene. Exons 7 and 8 are shown in blue. Point mutations in the 15 lines are highlighted
in yellow. In bold and italic the 7 lines with amino acids substitutions, and underlined the 5
lines for which seeds were available (furthermore distinguished in colors that indicate the
individual lines JI7270, JI7175, JI8554, JI8251, JI10179). The positions of the primers used
for TILLING are underlined.
Figure S2. Full amino-acid sequence alignment of BGD2, BGD4, and TrCBG.
Figure S3. Overlay of β-glucosidase structural models and the catalytic site residues. The
figure depicts the homology models of the β-glucosidases from Lotus japonicus and the GH1
modeling templates from rice (Os4bglu12, PDB code: 3PTM, chain A) and white clover
(Trifolium repens β-glucosidase 2, PDB code: 1CBG). (A) top view of the structural
alignment of β-glucosidase from rice (orange) and white clover (white) and BGD2 models
using white clover template (cyan) and rice (green), as well as BGD4 using white clover
template (blue) and rice (magenta). (B) bottom view of overlay shown in panel A. (C)
overlay of the catalytic and substrate binding residues interacting with the glucosyl moiety
bound at subsite –1, labeled according to numbering in the rice structure (3PTM, chain A).
The 2-deoxy-2-fluoro-α-D-glucopyranose stacking onto W442 comprising the aromatic
platform (not labeled for clarity) is shown also to depict the polar interactions to the bound
glucosyl (yellow dotted lines). The depicted residues overlay almost perfectly (except for
Y322 which is modeled in two conformations one of which differs from the consensus)
highlighting the high quality of the models.
Figure S4. Cyanogenesis and LC-MS analysis of hydroxynitrile glucoside content in the
selected Lotus species. (a) Cyanogenesis in the Lotus species indicated by their IPK stock
code (see Table 2). HCN release is assayed from individual leaves collected in a 96-well
microtitre plate. Duplicate samples for each Lotus accession are shown. Tissues are disrupted
by a freeze-thaw cycle. HCN release is visual using Feigl-Anger paper. (b) Extracted ion
chromatograms of LC-MS analysis of leaf extracts from the Lotus species indicated.
Extracted ion peaks are for sodium adducts of linamarin (light-blue, m/z 270),
rhodiocyanoside D and A (brown, m/z 282, rhodiocyanoside A is the major peak), and
lotaustralin (dark-blue, m/z 284). The characteristic hydroxynitrile glucoside profile of L.
japonicus MG-20 shows rhodiocyanoside A and lotaustralin as the two major compounds.
The hydroxynitrile glucoside profile of L. krylovii shows rhodiocyanoside A as the major
compound. The hydroxynitrile glucoside profile of L. burttii shows rhodiocyanoside A as the
major compound. Lotaustralin and rhodiocyanoside A are the major hydroxynitrile
glucosides in L. filicaulis. Out of three L. tenuis accessions tested, only LOT44 contains
rhodiocyanosides. None of the tested accessions of L. corniculatus contains rhodiocyanosides.
Linamarin and lotaustralin typically occur at similar levels in this species. L. pendunculatus
only contains a trace amount of lotaustralin, lacks hydroxynitrile glucoside hydrolyzing
activity and is acyanogenic. Linamarin is the dominant hydroxynitrile glucoside compound in
L. ornithopodioides.
Figure S5. Amino acid alignment of the BGDs amplified from Lotus spp. Amino acids
alignment of the partial BGD cDNA fragments amplified from the various Lotus species
indicated. The two regions containing the conserved catalytic motifs are indicated in orange.
The position of the G, V, or S residue at position 211 is also indicated in color. The
sequences of BGD2 and BGD4 from L. japonicus are included, as is the sequence of TrCBG
from T. repens. Sequences are aligned with ClustalW2.
Figure S6. Phylogenetic tree with dN/dS ratios. The DNA based phylogenetic tree presented
in Figure 5b was used to calculate the dN/dS ratios for the individual branches of the tree
using the codeml program of the PAML package (Yang, 2007). A dN/dS < 1 suggests
purifying selection, dN/dS = 1 neutral selection, and dN/dS > 1 positive selection. For some
branches the dN/dS ratio could not be calculated due to the absence of silent or replacement
substitutions for that branch, and this is indicated with --. Symbols in front of the species
name indicate the type of β-glucosidase sequence isolated as indicated in Table 2: ● BGD2type sequence, ○ BGD4-type sequence, ♦ BGD with G211.