RILs (data S2) and that seemingly minor
differences in leaf sampling and extraction
technique—potentially the simultaneous ex-
posure of leaves to aluminum foil and liquid
N 2 and unfavorable pH conditions (table S1)—
resulted in the loss of this unstable phenol-
amide in plants grown and sampled under
glasshouse conditions.
We rearedEmpoascaon irMYC2plants again
and, by optimizing extraction conditions, found
thatEmpoascafeeding strongly elicitedm/z
347.19 accumulations in empty vector (EV)–
transformed plants; these accumulations were
abolished in irMYC2lines (Fig. 3B). During
this extraction optimization effort, we real-
ized that theEmpoascaleafhopper elicitation
procedure could be replaced with the experi-
mentally more tractable elicitations by larval
oral secretions or methyl jasmonate (MeJA).
Consistently, MeJA-induced production ofm/z
347.19 was hampered in irMYC2, but also in
irMYB8and ovJAZilines (Fig. 3B). These re-
sults revealed that the unknownm/z347.19 is
regulated by the JA-JAZi-MYC2-MYB8 signaling
sector, which is likely responsible forEmpoasca
resistance.
To investigate the biosynthetic origins of
m/z347.19, we extracted OS-elicited leaves
oftheentireMAGICRILpopulationgrown
under glasshouse conditions and phenolamide-
permissive conditions and conducted an mQTL
analysis (Fig. 3C). The analysis imputed a series
of genes (withPvalues <10−^3 ) known to be
Baiet al.,Science 375 , eabm2948 (2022) 4 February 2022 5of9
A
NL similarity
NDP similarity
idMS/MS
(similarity cutoff >0.5)
Structural similarity-based MS/MS clustering^6
5 4 3 2 1 0
16293 4 5 7 8 10 11scaffold
NaPPO2
NaAT1NaPPO1 NaMYB8
Chromosome
-log (
P)
10
NaMYC2a
NaBBL2
m/z 347.19
B
C
D
0123
0
3
6
9
(Z
)-3-Hexen-1-ol (Rel. int.)
PCC=0.142
P=2.5x10-4
x10^5
MAGIC accession
x10^5
m/z 347.19 (Rel. int.)
C
W+OS
m/z 347.19
0
400
800
1200
EV PPO1 PPO2 AT1
aa
a
b
aa
a
a
b
a
C
W+OS
0
2000
4000
6000
Relative intensity
EV asHPLirLOX2irLOX2xLOX3
m/z 347.19
a aa aa
a
0
2000
4000
EV BBL2
a a
a
bm/z 347.19
Relative intensity
0
1
x10^5
2
(^3) CP
abc
a
abc
ab
c
ab
bc bc
EV PPO1 PPO2 AT1
LOX2 LOX3
Herbivory
( )
Jasmonate
HPL pathway
MYB8
AT1
CP
C6
( )GLVpathway
Polyamine
pathway
Putrescine
- Arginine
pathway
?
Unknown m/z 347.19
NaPPO1/NaPPO2
derivatives
Phenylpropanoid
NaBBL2
?
?
F
NH
NH
O
O
O
m/z 347.196
[M+H]+
2
C 19 H 27 N 2 O 4 + - C 6 H 8 O
CP+ C 6 H 8 O
H
H
NaDH29
NaAT1
NaPAL2
NaPAL1
NaBBL2
Na4CL2
Na4CL1NaC3H
NaC4H
NaHQT
0 1
Fragment-based similarity
0 1
NL-based similarity
M1
M2
M3
M4
M5
M6
M7
Module
M5
251.14iso1
251.14iso2
347.19
470.23
DCS
CP
CP
Unk.
163.04
502.26
308.20CS
501.24
355.10CGA
202.12
163.04
147.06
252.07
425.21
131.05
403.23 163.04
219.15
179.12
553.30
459.22
225.11
535.29
CPD
3
Relative intensity (10 counts)
0
3
6
EV irMYC2
Control
Empoasca feeding
b
a
aa
m/z 347.19
Lan (C)
15
10
5
0
EV irMYC2irMYB8ovJAZi
m/z 347.19
Lan+MeJA
a aa aa
a
a
b
E
NH
NH
O
O
O
m/z 251.139
[M+H]+
2
C 13 H 19 N 2 O 3 +
CP
H
H
NaPPO1/NaPPO2? NaBBL2
JAZi
MYC2
MYB8
AT1
Jasmonate signalling
GLV pathway
Polyamine
pathway
?
Fig. 3. Elucidating an herbivory-elicited GLV-caffeoylputrescine metabolite
and its three-pronged biosynthetic pathways by combining MS/MS structural
metabolomics with forward and reverse genetics.(A) (Left) Biclustering
of 518 idMS/MS spectra constructed from 15 RILs of the field-planted MAGIC
population based on shared fragments (NDP-based similarity) and shared
neutral losses (NL-based similarity) reveals seven distinct modules (M1 to M7) in
the molecular network. (Right) Close-up of module 5 in which putrescine- or
caffeoyl-derived phenolamides are enriched harboring an unknown metabolite
m/z347.19 that is directly linked to two isomers of CP (circled in green).
CS,N-caffeoylspermidine; CoCS,N′,N′′-coumaroyl, caffeoylspermidine; CFS,
N′,N′′-caffeoyl, feruloylspermidine; DCS,N′,N′′-dicaffeoylspermidine; CPD,
caffeoylputrescine dimer; CGA, chlorogenic acid; Unk., unknown. (B) Accumulations
ofm/z347.19 inEmpoasca-elicited EV and irMYC2lines (top) and MeJA-induced
EV, irMYC2, irMYB8, and ovJAZilines (bottom). (C) (Left) Manhattan plot for
herbivory-induced unknownm/z347.19 from an mQTL analysis of W + OS–elicited
leaves from the MAGIC RIL population grown in the glasshouse and extracted
with procedures that minimize losses of the phenolamide sector. Core JA signaling
gene,NaMYC2a; phenolamide regulator,NaMYB8; CP biosynthetic gene,NaAT1;
and two unknown biosynthetic candidate genes,NaPPO1andNaPPO2, were
imputed in the mQTL analysis (Pvalue cutoffs = 10−^3 ) as well as an uncharacterized
candidate gene,NaBBL2(P= 0.0013). (Right) Gene coexpression network
constructed using a previously published microarray dataset of irMYB8plants
harvested 1 and 5 hours after W + OS elicitation. The phenolamide biosynthetic
genes,NaAT1andNaDH29, were used as baits (diamonds). Yellow dots depict
genes coexpressed with both baits, whereas the green (NaAT1) and blue (NaDH29)
dots depict genes coexpressed with a single bait. (D) VIGS of biosynthetic gene
candidates involved inm/z347.19 production. SilencingNaPPO1,NaPPO2,
NaAT1, andNaBBL2expression abolished the elicitation ofm/z347.19 by W + OS
treatment observed in EV control plants (C indicates untreated controls).
(E) Proposed three-pronged biosynthetic pathway for theEmpoasca-elicited
m/z347.19 production, which requires the LOX2-HPL–dependent C6 GLV
metabolism, LOX3-dependent and JA-regulated phenylpropanoid metabolism,
and polyamine metabolism, the outputs of which are putatively conjugated
in NaBBL2- and NaPPO1/2-dependent reactions. (F) Scatterplots of metabolite
abundance ofm/z347.19 against (Z)-3-hexen-1-ol volatile emissions in MAGIC
accessions from the glasshouse (top) andm/z347.19 accumulation after
W + OS treatment in leaves of stably transformed EV, asHPL, irLOX2, and irLOX2
or irLOX3lines (bottom).
RESEARCH | RESEARCH ARTICLE