Together, these in vitro and in vivo results
suggest that CPH inN. attenuatais responsible
for the plant’sEmpoascanonhost resistance.
NaBBL2 is required to engineer CPH
biosynthesis in crop species
The discovery of CPH and its biosynthetic
pathways underlying nonhost resistance offers
a framework for the engineering of CPH bio-
synthesisincropplantsasameansofopti-
mizing a plant’s endogenous metabolism for
defense against the attack of devastating
leafhopper pests, the diseases they vector,
and other nonhost pests. We investigated
whether CPH is widely found inSolanaceae
and other plant taxa. Metabolic profiling of
N. attenuata’s close relatives revealed that
six of sevenNicotianaspecies induced CPH in
a coordinated fashion with CP when elicited
by MeJA (fig. S26A). We selected 13 different
taxa from different plant families, including
several crop species, and compared amino acid
identities of NaAT1, NaPPO1, NaPPO2, and
NaBBL2 with their closest homologs using
Basic Local Alignment Search Tool (BLAST)
searches of the National Center for Biotech-
nology Information (NCBI) sequence database
(fig. S26B). FiveSolanaceaetaxa of the 13 spe-
cies examined contain all orthologs of the four
protein sequences. Moreover, we detected
MeJA-induced CP in eight species, including
sevenSolanaceaespecies and wheat, where-
as only two species other thanN. attenuata,
Capsicum annuumandNicotiana benthamiana,
produced CPH (fig. S26B). These results sug-
gest that CPH production may be restricted to
theSolanaceae.
Synthetic biology has enabled the transfer
of metabolic pathways among taxa because
of shared cofactors and metabolism ( 30 – 33 ).
We attempted to reconstitute the CPH pathway
in vivo (Fig. 4D). We selectedVicia fabaand
Solanum chilenseforAgrobacterium-mediated
transient expression of the CPH pathway for
several reasons. Neither species accumulated
CPH in untreated and MeJA-treated tissues.
V. fabais an ideal host plant forEmpoasca
rearing. CoP, CP, and FP do not accumulate in
V. faba, whereas CP levels are induced by
MeJA treatment ofS. chilense, which provides
an internal precursor for CPH production (Fig.
4E). Moreover, both are readily transformed
and likely to produce correctly folded active
proteins with which to test biochemical ac-
tivities of CPH biosynthetic genes in planta.
We transiently coexpressed NaPPO1 or
NaPPO2 together with (Z)-3-hexenal and CP
leaf infiltrations inV. fabaor without CP infil-
trations inS. chilense. However, we failed to
detect any CPH in either species (Fig. 4E).
PPOs are generally localized to plastids, phys-
ically separating them from their phenolic
substrates, which are known to be localized
in vacuoles ( 34 ). InN. attenuata, a thylakoid
transfer domain was identified in both NaPPO1
and NaPPO2 N-terminal sequences (fig. S27A).
Transient expression of green fluorescent pro-
tein (GFP)–tagged NaPPO1 and NaPPO2 in
N. attenuataleaves confirmed that both NaPPO1
and NaPPO2 are plastid localized (fig. S27B).
Our three-pronged pathway proposal for
CPH biosynthesis is therefore challenged by
the separate enzymatic localizations of the
different components—CP (likely vacuolar or
cytosolic) ( 34 , 35 ), GLVs, JAs, and PPOs (plas-
tidial) ( 34 , 36 ). This challenge was reminiscent
of nicotine biosynthesis, which requires a
BBL gene to join the mitochondria-localized
pyridine ring, derived from nicotinic acid,
with the pyrrolidine ring, derived from the
peroxisome-localizedN-methylpyrrolinium
cation, to produce nicotine ( 37 ). We hypothe-
sized that NaBBL2 is required for the produc-
tion of CPH in vivo, and to test this hypothesis,
we expressed NaBBL2 along with NaPPO1 or
NaPPO2 inS. chilenseplants. One day after
Agrobacteriuminfiltration,S. chilenseplants
were treated with MeJA to induce CP produc-
tion, and 3 days later, we infiltrated leaves with
(Z)-3-hexenal. After 6 hours, we harvested leaves
for liquid chromatography–mass spectrometry
(LC-MS) analysis and found that the leaves
had accumulated substantial quantities of CPH
(Fig. 4E). ForV. fabaplants, we expressed
NaBBL2 along with NaPPO1 or NaPPO2; 3 days
afterAgrobacteriuminfiltration, we infiltrated
leaves with both CP and (Z)-3-hexenal and har-
vested leaves for LC-MS analysis after 6 hours.
Again, CPH accumulated (Fig. 4E). From these
results, we infer that NaBBL2, although not
required for in vitro synthesis, is required for
in vivo CPH biosynthesis. Additional work is
required to evaluate whether NaBBL2 plays a
role in solving the localization challenge, which
could have other possible solutions (figs. S28
and S29 and supplementary materials). Finally,
we conductedEmpoascafeeding trials on
the CPH-engineeredV. fabaandS. chilense
plants and observed that theseEmpoasca
host crop plants become lethal host plants
forEmpoasca(Fig. 4E).
This mechanistic analysis ofEmpoascanon-
host resistance provides another example of
the innovative chemical solutions that native
plants have evolved to solve their ecological
challenges ( 38 ). The natural history–driven
multi-omics framework that we used for the
discovery of CPH and its marriage with syn-
thetic biology approaches highlights how read-
ily the results of millions of years of innovation
by natural selection can be transferred to our
crop plants to catalyze the next, greener, and
ecologically more nuanced revolution in plant
protection ( 39 ) and domestication ( 40 – 42 ).
Crop plants face challenges not substantially
different from those of native plants, being
constantly tested by an herbivore community
that challenges the host-nonhost distinction.In a world of climate change and globally ho-
mogenized herbivore communities, opportun-
istic associations will dominate natural and
man-made ecosystems. Insight into how native
plants cope with opportunistic associations will
help us to design crops that are more resilient
in the face of unknown stresses as the world’s
climate changes ( 43 ).Materials and methods summary
Two replicates of the 650 RILs from a 26-parent
MAGIC population and their 26 parental lines
were planted at the WCCER field station in
Prescott, Arizona, USA. To elicit a standardized
herbivory response, leaves of all RILs, which
were in the early flowering stage, were wounded
and immediately treated with dilutedM. sexta
oral secretions (W + OS) or were left untreated
(control). Leaves were then harvested on dry
ice at 1 and 72 hours. One week after metab-
olite sampling, all plants of the field popu-
lation were screened for naturalEmpoasca
leafhopper numbers and damage. These leaf-
hoppers had opportunistically sampled the
N. attenuataplants from neighboring native
cucumber host plants. The mQTL and eQTL
mapping between SNPs and the relative abun-
dance of each compound or transcript using a
set of 646 RILs of the MAGIC population was
done with the R package software GAPIT using
general linear models (GLMs). The multi-omics
coassociation network was built from corre-
lations among metabolomes, transcriptomes,
phytohormones, and SNPs. ForEmpoasca
choice assays, transgenic lines ofN. attenuata
at the early rosette growth stage were ran-
domly placed in an open-choice glasshouse
environment containingEmpoascaleafhoppers
reared on bean plants in the MPI-CE glass-
house in Isserstedt, Germany. Y2H assays
and quantitative reverse transcription poly-
merase chain reaction (qRT-PCR) were used
for characterizingEmpoasca-induced JA sig-
naling genes. Compound-specific idMS/MS
was constructed using ultrahigh-performance
liquid chromatography–electrospray ionization
(UHPLC-ESI)–quadrupole time-of-flight mass
spectrometry (qTOF-MS) for idMS/MS acquisi-
tion and rule-based computational approaches
for idMS/MS assembly. Metabolome diversity
and specialization and metabolic specificity
were calculated using information theory by
considering the Shannon entropy of the idMS/
MS frequency distributions. In vivoEmpoasca
choice and in vitroEmpoascafeeding assays
were conducted by infiltrating synthetic CP,
CoP, or FP into irMYC2leaves or by feeding
Empoascawith the compounds diluted in
10% glucose solutions. Fifteen RILs, which in-
duced putrescine-containing phenolamides
after OS elicitation and accumulated a di-
verse set of known and unknown phenol-
amides, were used to construct idMS/MS for
MS/MS structural metabolomics analysis.Baiet al.,Science 375 , eabm2948 (2022) 4 February 2022 8of9
RESEARCH | RESEARCH ARTICLE