N-hydroxy-pipecolic acid is a mobile metabolite thatinduces systemic disease resistance in ArabidopsisYun-Chu Chena,1, Eric C. Holmesb,1, Jakub Rajniakb, Jung-Gun Kima, Sandy Tangb, Curt R. Fischerc, Mary Beth Mudgetta,2,and Elizabeth S. Sattelyb,2

aDepartment of Biology, Stanford University, Stanford, CA 94305-5020; bDepartment of Chemical Engineering, Stanford University, Stanford,CA 94305-5020; and cChemistry, Engineering & Medicine for Human Health, Stanford University, Stanford, CA 94305-5020

Edited by Brian J. Staskawicz, University of California, Berkeley, CA, and approved April 16, 2018 (received for review March 28, 2018)

Systemic acquired resistance (SAR) is a global response in plantsinduced at the site of infection that leads to long-lasting andbroad-spectrum disease resistance at distal, uninfected tissues.Despite the importance of this priming mechanism, the identityand complexity of defense signals that are required to initiate SARsignaling is not well understood. In this paper, we describe a me-tabolite, N-hydroxy-pipecolic acid (N-OH-Pip) and provide evidencethat this mobile molecule plays a role in initiating SAR signal trans-duction in Arabidopsis thaliana. We demonstrate that FLAVIN-DEPENDENT MONOOXYGENASE 1 (FMO1), a key regulator ofSAR-associated defense priming, can synthesize N-OH-Pip frompipecolic acid in planta, and exogenously applied N-OH-Pip movessystemically in Arabidopsis and can rescue the SAR-deficiency offmo1 mutants. We also demonstrate that N-OH-Pip treatmentcauses systemic changes in the expression of pathogenesis-related genes and metabolic pathways throughout the plant andenhances resistance to a bacterial pathogen. This work providesinsight into the chemical nature of a signal for SAR and also sug-gests that the N-OH-Pip pathway is a promising target for meta-bolic engineering to enhance disease resistance.

systemic acquired resistance | signaling | plant natural products |Arabidopsis thaliana | N-hydroxy-pipecolic acid

Plants have developed a complex and dynamic innate immunesystem that relies on sensing and signaling using small mol-

ecules for defense against pathogens (1, 2). At a primary site ofinfection, plants respond to common molecular features of mi-crobes (e.g., bacterial flagellin or fungal chitin, collectively knownas microbial-associated molecular patterns or MAMPs) andpathogen-derived proteins, termed effectors, that enter the plantcell (3). Immune responses to both classes of molecules (effector-triggered or pattern-triggered; refs. 3 and 4) activate signal trans-duction networks through the action of hormones such as salicylicacid (SA), ethylene, and jasmonic acid, which cause changes indefense gene expression and production of antimicrobial metabo-lites at the site of infection (5). Interactions between these hor-mones function synergistically and antagonistically to tailora specific immune response to different pathogens (2, 6). Whilelocal defense is critical for limiting pathogen growth, plants alsopossess the ability to prime and amplify immune responses at distalsites. This global response is termed systemic acquired resistance(SAR) (2, 7). SAR is critical for preventing the spread of pathogensand protecting against new infections (1, 2). Although much isunderstood about how the immune response is activated locally,the chemical nature of the plant-derived molecules that mediatelong-distance communication between the site of infection (pri-mary) and distal uninfected (secondary, systemic) sites remainelusive.Genetic and mechanistic studies of the model plant Arabidopsis

thaliana have led to the identification of key components of theSAR pathway, providing critical insight into the mechanismscontrolling long-lasting and broad-spectrum disease resistance (7).Several plant-derived small molecules [e.g., SA (8), methyl salic-ylate (MeSA) (9), azelaic acid (AzA) (10), glycerol-3-phosphate

(G3P) (11), dehydroabietinal (DA) (12), and pipecolic acid (Pip)(13)] are associated with long-distance communication and signalamplification during SAR (8, 14, 15). However, the onset of SARsignaling also requires the uncharacterized enzyme FLAVIN-DEPENDENT MONOOXYGENASE 1 (FMO1) (16–18).Remarkably, treatment of fmo1 mutants with AzA, DA, or Pipdoes not elicit systemic resistance, suggesting that a metaboliteproduced by FMO1 plays a key role in the establishment of SARsignaling (10, 12, 13). The biochemical function of FMO1 hasremained unknown.Several forward genetic screens searching for SAR-deficient

mutants identified multiple alleles of fmo1, ald1 (AGD2-LIKEDEFENSE RESPONSE PROTEIN 1), and sard4 (SAR-DEFICIENT 4), highlighting the importance of metabolites pro-duced by these enzymes (16, 17, 19, 20). ALD1 and SARD4 areinvolved in the biosynthesis of Pip (13, 21, 22). Irrigation of wild-type Arabidopsis plants with Pip induces SAR (13), which suggestedthat Pip might be a mobile SAR metabolite. Návarová et al. (13)reported, however, that Pip could not trigger SAR in fmo1 mu-tants. In addition, fmo1 plants accumulate high levels of Pip duringa late stage of infection compared with wild-type plants (13). Thesefindings feature FMO1 as a key missing link in the mechanism ofPip-associated SAR.

Significance

Plants lack circulating immune cells and instead rely on smallmolecule chemistry for local and long-distance defense sig-naling. Following pathogen attack, plants activate innate im-mune pathways at the site of infection to limit pathogengrowth. Plants also possess the ability to prime similar immuneresponses in uninfected tissues to prevent the spread ofpathogens or protect against new infections. Despite the im-portance of systemic immunity, the mechanism for signaling isnot clear. In this study, we show that N-hydroxy-pipecolic acidmetabolites are mobile defense signals produced at the site ofbacterial infection and establish and amplify defense in un-infected, distal tissues. Our study illuminates the chemical na-ture of a mobile bioactive metabolite that confers pathogenresistance throughout the plant.

Author contributions: Y.-C.C., E.C.H., J.R., C.R.F., M.B.M., and E.S.S. designed research;Y.-C.C., E.C.H., J.R., J.-G.K., S.T., and C.R.F. performed research; Y.-C.C., E.C.H., J.R.,J.-G.K., C.R.F., S.T., M.B.M., and E.S.S. analyzed data; and Y.-C.C., E.C.H., M.B.M., andE.S.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1Y.-C.C. and E.C.H. contributed equally to this work.2To whom correspondence may be addressed. Email: mudgett@stanford.edu or sattely@stanford.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1805291115/-/DCSupplemental.

Published online May 7, 2018.

E4920–E4929 | PNAS | vol. 115 | no. 21 www.pnas.org/cgi/doi/10.1073/pnas.1805291115

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

We and others have found untargeted metabolite analysis ofArabidopsis genetic mutants to be a powerful approach for theidentification of small molecules associated with fitness pheno-types (either previously characterized or suggested by tran-scriptome analysis) (23). Examples include the identification andcharacterization of cytochromes P450 involved in phytoalexinproduction (24) and iron acquisition (25). Given that FMO1 isone of the genes most responsive to biotic stress (as indicated byanalysis of previously reported Arabidopsis microarray datasummarized in SI Appendix, Fig. S9) and that genetic data sug-gest molecules generated by FMO1 are required for initiatingSAR (16–18), we sought to apply an untargeted metabolomicsapproach to determine the products of FMO1 and their function.Here, we report the discovery of glycosylatedN-hydroxypipecolic

acid (N-OGlc-Pip) in Arabidopsis. We provide evidence that theaglycone, N-OH-Pip, is the direct product of FMO1 and has acentral role in SAR signal transduction (26). We also demon-strate that exogenously applied N-OH-Pip can rescue the SAR-deficient response of fmo1 mutants, cause systemic changes inexpression of pathogenesis-related genes and metabolic path-ways throughout the plant (and distal to the site of application),and enhance resistance to the bacterial pathogen, Pseudomonassyringae. In addition, we provide biochemical evidence that Ara-bidopsis FMO1 catalyzes N-hydroxylation of the nonproteinogenicamino acid Pip to N-OH-Pip using transient expression assays inNicotiana benthamiana. Taken together, our data indicate that N-OH-Pip is a key signaling molecule that is required to initiate SARin Arabidopsis.

ResultsUntargeted Metabolomics of Arabidopsis fmo1 Seedlings. To dis-cover metabolites whose production is dependent on FMO1, weused liquid chromatography-mass spectrometry (LC-MS)-baseduntargeted metabolomics to compare the composition of meth-anolic extracts from two sets of 12-d-old seedlings grown hy-droponically: A. thaliana Col-0 WT and a T-DNA insertionmutant of FMO1, fmo1-1 (16), herein referred to as fmo1. Thebacterial pathogen P. syringae pathovar tomato DC3000 (Pst) was

added to the seedling media 24 h before metabolite analysis toelicit expression of enzymes associated with defense and SAR(including FMO1). This analysis revealed a major mass signalthat is present in WT plants in response to Pst treatment but notMock treatment (10 mM MgCl2) and was absent from all fmo1plants (Fig. 1A and SI Appendix, Fig. S1 A and B). We proposethis signal corresponds to the O-glycosylated form of the me-tabolite N-OH-Pip, based on comparison of the MS/MSspectrum to an authentic synthetic standard of the N-OH-Pipaglycone (SI Appendix, Figs. S2 and S3). Although N-OH-Piphas not previously been observed as a naturally occurring me-tabolite in plants or other organisms, pipecolic acid has beenobserved in Arabidopsis and tomato (13, 27) and was shown to beassociated with the SAR response (13). Therefore, we hypoth-esized that FMO1 catalyzes the N-hydroxylation of Pip to pro-duce N-OH-Pip. Glycosylated N-OH-Pip (N-OGlc-Pip) detect-able in Arabidopsis plant extracts is likely the product ofunknown UDP glycosyltransferases that further processesN-OH-Pip (Fig. 1C). We did not detect free N-OH-Pip in WTseedlings elicited with Pst by LC-MS analysis (SI Appendix, Fig.S1B). These data suggest that the N-OH-Pip aglycone does notaccumulate in cells and/or is an unstable metabolite.

Measurement of N-OH-Pip Derivatives in Adult Plants. We alsomeasured the level of N-OGlc-Pip in adult WT and fmo1 plants(both in lower and upper leaves), after treatment of lower leaveswith 10 mM MgCl2 (Mock) or Pst expressing the type III se-cretion (T3S) effector gene avrRpt2 (Pst avrRpt2). Pst avrRpt2 isan avirulent strain that induces effector-triggered immunity(ETI) and SAR signaling in resistant RPS2 Arabidopsis plants(28–30). N-OGlc-Pip was detected in the lower and upper leavesof WT only after infection of lower leaves with Pst avrRpt2, butnot in fmo1 plants (Fig. 1B). As in seedlings, we were not able todetect free N-OH-Pip by LC-MS. Collectively, these data in-dicate that FMO1 is required for the production of N-OGlc-Pipin Arabidopsis.

N

O OH

OH N

O OH

OGlcNH

OHOFMO1

(At1g19250)

Pip N-OH-Pip

ALD1(At2g13810)

SARD4(At5g52810)Lys

N-OGlc-Pip

N

O OH

WT fmo1 WT fmo1100

101

102

103

104

105

106

Ion

abun

danc

e

MgCl2 PstL U L U L U L U

100

101

102

103

104

105

106

Ion

abun

danc

e

MgCl2 Pst avrRpt2 MgCl2 Pst avrRpt2Col-0 WT Col-0 fmo1

A B

C

Fig. 1. Untargeted metabolomics of Arabidopsis seedlings elicited with Pst implicates FMO1 in the production of pipecolic acid derivatives. (A) Ion abun-dances for N-OGlc-Pip (gray bars) detected in extracts isolated from Arabidopsis Col-0 WT and fmo1 seedlings grown hydroponically and elicited with P.syringae pv. tomato DC3000 (Pst). Levels represent the mean ± STD of six biological replicates. Levels reported as zero indicate no detection of metabolites.(B) Ion abundances for N-OGlc-Pip (gray bars) in lower (L) and upper (U) leaves of adult Arabidopsis Col-0 plants 48 hpi with a 5 × 106 cfu/mL suspension of Pstexpressing the T3S effector avrRpt2 (Pst avrRpt2). Levels represent the mean ± STD of six biological replicates. Levels reported as zero indicate no detection ofmetabolites. (C) Proposed biosynthetic activity for Arabidopsis FMO1. The Arabidopsis enzymes ALD1 and SARD4 convert lysine to pipecolic acid (Pip) (22).FMO1 is proposed to hydroxylate Pip to N-OH-Pip. Unknown enzymes are proposed to glycosylate N-OH-Pip to produce N-OGlc-Pip, the molecule identified inthe untargeted metabolite analysis.

Chen et al. PNAS | vol. 115 | no. 21 | E4921

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

Biochemical Activity of Arabidopsis FMO1. Arabidopsis FMO1expressed and purified as a 6×-His fusion protein (FMO1-6×-His) from Escherichia coli lacked the cofactor FAD and wasnot catalytically active in any of our in vitro biochemical assays.However, we were able to detect FMO1-dependent activity whenArabidopsis FMO1 was transiently expressed in N. benthamianaonly after feeding pipecolic acid and then analyzing productspresent in leaf methanolic extracts (Fig. 2 and SI Appendix, Fig.S1 C–E). As shown in Fig. 2, FMO1-expressing leaves accumu-lated free N-OH-Pip. They also contained a second metabolitewith an m/z of 100 and a putative structure of N-hydroxypiperideine (Fig. 2 and SI Appendix, Fig. S4), which we proposeis the result of oxidative decarboxylation of N-OH-Pip. We notedthat the mass signal m/z 100 can be produced in samples ofsynthetic N-OH-Pip after heating in buffer. It is also present inCol-0 seedlings after Pst treatment and absent from fmo1 seed-lings (SI Appendix, Fig. S1B), suggesting that the m/z 100 me-tabolite is likely FMO1-derived. In addition, N-OH-Pip levels inN. benthamiana extracts decreased between 28 and 48 h afterinfiltration (hpi) while the abundance of m/z 100 increased (SIAppendix, Fig. S1C), further suggesting that N-OH-Pip is un-stable in planta and may convert to m/z 100 over time. N. ben-thamiana leaves expressing two FMO1 active-site mutants,FMO1 G17A/G19A or FMO1 G215A, did not produce N-OH-Pip or m/z 100 at 28 hpi (Fig. 2 and SI Appendix, Fig. S1 D andE). The abundance of the mutant proteins was less than WTFMO1 at this time point (SI Appendix, Fig. S1F). At 48 hpi,protein abundance was similar (SI Appendix, Fig. S1F); however,N-OH-Pip was not detected in extracts expressing FMO1 G17A/G19A or FMO1 G215A (SI Appendix, Fig. S1E). Taken together,these data demonstrate that FMO1 can catalyze the hydroxylationof Pip in planta (Fig. 2) and support the requirement of the pu-

tative FAD and NADP+ domains for FMO1 catalytic activity (18).N-OGlc-Pip was not detected in N. benthamiana leaf extractswhen Arabidopsis FMO1 was expressed, suggesting that N. ben-thamiana either does not have the necessary glycosyl transferasesor that they are not expressed under our experimental conditions.

N-OH-Pip Treatment of Arabidopsis Leaves Is Sufficient to Induce SAR.To test if N-OH-Pip treatment alone is sufficient to induce SARand rescue the SAR deficiency of fmo1 mutants, we infiltratedthree lower leaves of Arabidopsis WT and fmo1 plants with10 mM MgCl2 (Mock) or 10 mM MgCl2 containing 1 mM Pip or1 mM N-OH-Pip. We used 1 mM of each metabolite because itwas previously established that watering plants with 1 mM Pipelicits SAR in the leaves of WT but not fmo1 mutants (13). Afterchemical incubation for 24 h, one upper leaf of each plant wasinoculated with a 1 × 105 cfu/mL suspension of P. syringaepathovar maculicola strain ES4326 (Psm ES4326), a virulentbacterium (Fig. 3A). At 3 d post inoculation (dpi), bacterialgrowth was quantified (Fig. 3B), and leaf symptoms were pho-tographed (Fig. 3C). Both WT and fmo1 plants treated withN-OH-Pip contained significantly less Psm ES4326 in infectedupper leaves compared with plants treated with Mock or Pip(Fig. 3B). N-OH-Pip treatment also reduced symptom develop-ment (i.e., leaf yellowing and tissue collapse) (Fig. 3C). Theseresults suggest that treatment of leaves with N-OH-Pip, but notPip, is sufficient to induce SAR and this does not require FMO1.

N-OH-Pip Treatment of Arabidopsis Leaves Is Sufficient To InduceMetabolite Production and Gene Transcription Associated with SAR.Next, we examined the ability of N-OH-Pip to systemically in-duce the production of SAR-associated metabolites and mRNAsin Arabidopsis leaves. Three lower leaves of WT and fmo1 plantswere infiltrated with 10 mM MgCl2 (Mock) or 10 mM MgCl2containing 1 mM Pip or 1 mM N-OH-Pip. After 48 h, the threetreated lower leaves and three untreated upper leaves werepooled independently, and then analyzed by LC-MS and quan-titative real-time PCR (qRT-PCR) for measurement of metab-olites and mRNAs, respectively.For metabolite profiling, we quantified Pip, N-OH-Pip, N-

OGlc-Pip, m/z 100, and two canonical defense metabolites, thephytoalexin camalexin and SA-Glucoside (SA-Glc) (Fig. 4 A andB). After infiltration of the lower leaves of both WT and fmo1plants with N-OH-Pip, we observed accumulation of camalexinand SA-Glc in both lower and upper leaves (Fig. 4B). Ourfindings are consistent with reports showing that FMO1 is re-quired to stimulate camalexin and SA-Glc synthesis and accu-mulation in distal, uninfected leaves during SAR (31).Although we did not detect free N-OH-Pip, we observed ac-

cumulation of N-OGlc-Pip and m/z 100 in both lower and upperleaves of all plants (Fig. 4B), suggesting rapid metabolism tothese derivatives. These data show that local treatment of puri-fied N-OH-Pip alone, even in the absence of infection, can in-duce systemic changes in defense metabolite production acrossthe plant and this does not require FMO1. In contrast, the in-filtration of lower leaves of WT or fmo1 plants with Pip did notinduce changes in N-OH-Pip metabolites, camalexin, or SA-Glc.Given that fmo1 plants are unable to produce N-OH-Pip, thedetection of N-OH-Pip metabolites (N-OGlc-Pip and m/z 100) inthe upper leaves after N-OH-Pip treatment of lower leaves (Fig.4B) provides evidence that one or more of these metabolites aremobile. Taken together, these data indicate that N-OH-Pip orrapidly formed derivatives move systemically through the plantto induce metabolic changes during SAR.For transcript profiling, we quantified the mRNA abundance

for known SAR marker genes, which were reported to be in-duced in untreated upper leaves of WT plants but not fmo1plants in response to bacterial infection (i.e., Psm ES4326) oflower leaves (32). These included genes dependent on SA

FMO1

FMO1 G17AG19

A

FMO1 G215A

FMO1

FMO1 G17AG19

A

FMO1 G215A

FMO1

FMO1 G17AG19

A

FMO1 G215A

100

101

102

103

104

105

106

107

108

Ion

abun

danc

e

N

O OH

OHNH

O OH

N OH

m/z = 100.075N-OH-PipPip (fed)

FMO1

Fig. 2. Overexpression of Arabidopsis FMO1 in N. benthamiana converts Pipto N-OH-Pip. Ion abundance of Pip (green bars), N-OH-Pip (blue bars), and anunknown metabolite, m/z = 100.075 (orange bars) in leaves of N. ben-thamiana expressing Arabidopsis Col-0 WT FMO1-6×-His or two mutants,FMO1-G17A/G19A-6×-His (a FAD binding mutant) or FMO1-G215A-6×-His(a NADP+ binding mutant). One millimolar Pip was coinfiltrated with an A.tumefaciens strain carrying FMO1-6×-His, FMO1-G17A/G19A-6×-His, orFMO1-G215A-6×-His, and then leaves were harvested 28 hpi. N-OH-Pip andthe unknown metabolite were identified as FMO1-dependent signals usingan untargeted analysis. Levels represent the mean ± STD of three biologicalreplicates. Levels reported as zero indicate no detection of metabolites.Pathway represents proposed reactions occurring in N. benthamiana.

E4922 | www.pnas.org/cgi/doi/10.1073/pnas.1805291115 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

[pathogenesis-related protein (PR) 1], partially dependent on SA(PR2 and PR5), or independent of SA (ALD1; AGD2-like de-fense response protein 1; SAG13, senescence-associated gene 13;and ICS1, isochorismate synthase 1). We also monitored ALD1,SARD4 (SAR deficient 4 protein), and FMO1, three genes in theproposed N-OH-Pip metabolic pathway (Figs. 1C and 4A).WT and fmo1 plants treated with N-OH-Pip accumulated the

highest levels of ICS1 and PR1 mRNA in treated lower anduntreated upper leaves, compared with plants treated with Mockor Pip (Fig. 4C). Similar trends were observed for ALD1,SARD4, PR2, PR5, and SAG13 mRNAs (Fig. 4C and SI Ap-pendix, Fig. S5). N-OH-Pip–triggered accumulation of ALD1 andSARD4 mRNA suggests that N-OH-Pip regulates ALD1 andSARD4 transcription via a positive feedback loop. Accumulationof Pip in leaves of WT and fmo1 plants treated with N-OH-Pipimplies that N-OH-Pip induction of ALD1 and SARD4 tran-scription leads to increased synthesis of Pip in leaves. In addition,N-OH-Pip treatment increased FMO1 mRNA levels in WTleaves but not the fmo1 mutant (Fig. 4C). Our inability to detecttranscript for this fmo1 mutant is consistent with it being a nullmutant (16, 18). This implies that N-OH-Pip also positivelyregulates FMO1 transcription via a positive feedback loop. No-tably, induction of SAR gene expression was greater in N-OH-Pip–treated plants compared with Pip-treated plants (Fig. 4Cand SI Appendix, Fig. S5), demonstrating that N-OH-Pip is amore bioactive SAR molecule compared with Pip.Collectively, these data demonstrate that exogenous treat-

ment of N-OH-Pip by leaf infiltration is sufficient to activatemetabolite production and SAR-associated gene transcription.

N-OH-Pip Irrigation Induces SAR in Arabidopsis WT and fmo1 Plants.Previously, Návarová et al. (13) showed that irrigation with Pipwas sufficient to recover SAR activity in ald1 plants but not infmo1 plants. We tested if irrigation with N-OH-Pip could simi-larly induce SAR in WT and fmo1 plants. SAR assays wereperformed as described previously (13) with slight modification.WT and fmo1 plants were drenched with water, 1 mM Pip, or1 mM N-OH-Pip by root application. One day later, three lower(local) leaves per plant were inoculated with 10 mM MgCl2(Mock) or a 5 × 106 cfu/mL suspension of Pst avrRpt2, an avir-ulent strain. Two days later, the untreated upper leaves of eachplant were challenged with Mock or a 1 × 105 cfu/mL suspensionof Psm ES4326, a virulent strain. The titer of Psm ES4326 in theinfected upper leaves was determined at 3 dpi to quantify theimpact of Pip and N-OH-Pip treatment on the level of diseaseresistance. An overview of the SAR experiment is shown in Fig. 5.Before challenging upper leaves with Psm ES4326, lower

leaves infected with Pst avrRpt2 were yellow (chlorotic) andcollapsed by 2 dpi (Fig. 5B). We noticed that these symptomswere significantly delayed in WT and fmo1 plants irrigated withN-OH-Pip (Fig. 5B). Pip irrigation reduced leaf symptom de-velopment in WT plants to a lesser degree, but not in fmo1 plants(Fig. 5B). Metabolic profiling revealed that N-OGlc-Pip and m/z100 were present in Pst avrRpt2-infected leaves of WT plantsirrigated with Pip or N-OH-Pip, and fmo1 plants irrigated withN-OH-Pip (SI Appendix, Fig. S6). These results indicate that N-OH-Pip treatment reduces symptom development of Arabidopsisleaves in an FMO1-independent manner. The delay of symptomdevelopment caused by N-OH-Pip and Pip could be attributed to

Fig. 3. Infiltration of N-OH-Pip into Arabidopsis Col-0 WT and fmo1 leaves inhibits growth of virulentPsm ES4326 in distal leaves. (A) Photograph showinga representative plant and leaves used for SAR ex-periments. Typically, leaf numbers 7, 8, and 9 wereused for chemical treatments (treated lower leaf)and leaf numbers 11, 12, or 13 was challenged withbacteria (upper leaf). (B) Titer of Psm ES4326 inchallenged upper leaves of Arabidopsis WT andfmo1 plants. Three lower leaves of Col-0 WT andfmo1 were infiltrated with 10 mM MgCl2 (Mock) or10 mM MgCl2 containing 1 mM Pip or 1 mM N-OH-Pip. After chemical incubation for 24 h, one upperleaf of each WT and fmo1 plant was inoculatedwith a 1 × 105 cfu/mL suspension of Psm ES4326. Thenumber of Psm ES4326 was quantified 3 d later tomeasure resistance of leaves. Bars (black, WT; gray,fmo1) represent the mean ± STD of four biologicalreplicates. Asterisks denote the significant differ-ences between indicated samples using a one-tailedt test (**P < 0.01). The experiment was repeatedtwice with similar results. (C) Symptoms of Arabi-dopsis WT and fmo1 upper leaves infected with PsmES4326 following the chemical treatment describedin B. Leaves were photographed 3 dpi. The SAR ex-periment was repeated twice with similar results.

Chen et al. PNAS | vol. 115 | no. 21 | E4923

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

metabolite-induced defense priming in infected leaves or in-herent bactericidal activity of the metabolites.After challenging upper leaves with Psm ES4326, the titer of

Psm ES4326 in water-treated WT plants infected with PstavrRpt2 was lower than that in water-treated WT plants in-oculated with Mock (Fig. 5C). These data indicate that PstavrRpt2 induces SAR under the conditions tested. By contrast,Psm ES4326 titers were similar in water-treated fmo1 plants in-oculated with Mock or Pst avrRpt2 (Fig. 5C). These data areconsistent with previous findings that fmo1 mutants are deficientin SAR signaling (16, 18).Irrigating plants with Pip and N-OH-Pip enhanced SAR, albeit

to different extents. Pip absorption by roots provided a moderatebut significant reduction in Psm ES4326 titer in Mock and PstavrRpt2-inoculated fmo1 plants (Fig. 5C). However, the level ofprotection was less than SAR-induced protection elicited by Pst

avrRpt2 in water-treated WT plants. Interestingly, N-OH-Pipabsorption by roots provided the strongest level of protection(Fig. 5C). Psm ES4326 titers in Mock and Pst avrRpt2-inoculatedfmo1 plants were significantly less than those detected in water-or Pip-treated fmo1 plants. Moreover, N-OH-Pip-induced pro-tection in fmo1 plants was similar to that observed for PstavrRpt2-induced protection in WT plants (Fig. 5C). Collectively,these data indicate that N-OH-Pip is a potent SAR-inducingmetabolite, and irrigation of plants with N-OH-Pip is also suf-ficient to establish SAR in WT plants and complement the SARdefect of fmo1 plants.

N-OH-Pip Is Not Bactericidal. To rule out that N-OH-Pip is toxic tobacterial growth in planta, we performed a minimum inhibitoryconcentration (MIC) assay (33) to measure potential bactericidalactivity. Cultures of Psm ES4326, Pst DC3000, and Pst avrRpt2

Fig. 4. N-OH-Pip induces expression of SAR markergenes and metabolites in treated lower leaves anduntreated upper leaves of Arabidopsis Col-0 WTand fmo1 plants. (A) Proposed pathway of metaboliteand gene induction in N-OH-Pip–associated SAR. Inthis study, FMO1 is hypothesized to convert Pip toN-OH-Pip and unknown enzymes convert N-OH-Pip toN-OGlc-Pip and m/z = 100.075. Pip biosynthesis re-quires ALD1 activity and is partially dependent onSARD4 activity (13, 21, 22). N-OH-Pip treatment leadsto the accumulation of camalexin and ALD1, SARD4,FMO1, ICS1, and PR2 mRNAs. ICS1 converts chorismicacid to isochorismic acid, leading to SA production. SAis required for camalexin production in during SAR(31), and SA induces the expression of NPR1. NPR1 is atranscription factor that regulates expression of PR1and other defense-associated genes. In plants, SA-Glucoside (SA-Glc) is thought to be nontoxic storageform of SA (40). (B) Ion abundance of Pip, N-OGlc-Pip,m/z = 100.075, camalexin and SA-Glc in leaves. Threelower leaves of Col-0 WT and fmo1 were infiltratedwith 10 mM MgCl2 (Mock) or 10 mM MgCl2 contain-ing 1 mM Pip or 1 mM N-OH-Pip. After chemical in-cubation for 48 h, three treated lower (L) leaves andthree untreated upper (U) leaves of WT and fmo1were analyzed by metabolite profiling and qPCR (C).Bars (black, WT; gray, fmo1) represent the mean ±STD of three biological replicates. (C ) Relative ex-pression of SAR marker genes in leaves: ALD1(AT2G13810), AGD2-like defense response protein 1;SARD4 (AT5G52810), SAR deficient 4; FMO1(At1g19250), flavin-dependent monooxygenase 1;ICS1 (At1g74710), isochorismate synthase 1; PR1(At2g14610), pathogenesis-related protein 1 andPR2 (At3g57260), β-1,3-glucanase 2. Bars (black, WT;gray, fmo1) represent the mean ± STD of three bi-ological replicates. The experiment was repeatedtwice with similar results.

E4924 | www.pnas.org/cgi/doi/10.1073/pnas.1805291115 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

were incubated with different concentrations (0–1 mM) of Pip,N-OH-Pip, or SA and then grown at 28 °C for 36 h. SA wasincluded as a positive control because it inhibits multiplication ofPseudomonas aeruginosa and Agrobacterium tumefaciens in vitro(34, 35). Similar to previous studies, 1 mM SA inhibited themultiplication of Psm ES4326, Pst DC3000, and Pst avrRpt2 invitro (SI Appendix, Fig. S7). However, neither Pip nor N-OH-Pipinhibited Psm ES4326, Pst DC3000, and Pst avrRpt2 growth invitro. These data indicate that Pip and N-OH-Pip are notbactericidal.

N-OH-Pip–Treated Arabidopsis Plants Exhibit a Faster HypersensitiveResponse. Next, we investigated if N-OH-Pip enhances ETI inlocal, infected leaves by performing electrolyte leakage and HRassays. WT and fmo1 plants were irrigated with water, 1 mM Pip,or 1 mM N-OH-Pip. One day later, six to seven leaves of eachplant were inoculated with a 3 × 108 cells per mL suspension ofPst DC3000 carrying an empty vector (Pst vector) or expressingavrRpt2 (Pst avrRpt2). Ion leakage was quantified at 5 hpi (Fig.6A), and HR symptoms were recorded at 8 hpi (Fig. 6B).

WT and fmo1 leaves infected with Pst vector exhibited similarelectrolyte leakage (below 20%) when irrigated with water, Pip,and N-OH-Pip (Fig. 6A), and no HR phenotypes were observedfor these infected plants at 5 hpi (Fig. 6B). WT plants irrigatedwith water and infected with Pst avrRpt2 exhibited more elec-trolyte leakage at 5 hpi than similarly treated fmo1 plants (Fig.6A). HR symptoms were not observed at 8 hpi (Fig. 6B); how-ever, full leaf collapse occurred by 10 hpi. Compared with waterirrigation, Pip irrigation resulted in a significant increase inelectrolyte leakage from WT leaves infected with Pst avrRpt2(Fig. 6A), with most (7/10) of the leaves undergoing HR by 8 hpi(Fig. 6B). No significant changes were detected for similarlytreated fmo1 leaves. Notably, N-OH-Pip irrigation resulted in asignificant increase in electrolyte leakage from both WT andfmo1 leaves infected with Pst avrRpt2 (Fig. 6A). HR phenotypesfor the N-OH-Pip–irrigated plants appeared faster and weremore severe than those observed for the Pip-irrigated plants(Fig. 6B). Taken together, these results show that N-OH-Piptreatment stimulates AvrRpt2-elicited HR in plants compared

Fig. 5. Root application of N-OH-Pip elicits localdefense and SAR in Arabidopsis Col-0 WT and fmo1plants. (A) Experimental design of SAR assay. EachCol-0 WT and fmo1 plant was treated with 10 mL ofwater, 1 mM Pip, or 1 mM N-OH-Pip by root appli-cation. One day later, three lower leaves of eachplant were infiltrated with 10 mM MgCl2 or a 5 × 106

cfu/mL suspension of Pst DC3000 expressing avrRpt2(Pst avrRpt2) in 10 mM MgCl2. Two days later, oneupper leaf of each plant was inoculated with a 1 ×105 cfu/mL suspension of Psm ES4326. The number ofPsm ES4326 in upper leaves was quantified 3 d later.(B) Phenotype of lower leaves inoculated with PstavrRpt2 after chemical treatment. Phenotypes offour biological replicates were recorded 2 dpi withPst avrRpt2. Similar phenotypes were observed inthree independent experiments. (C) Titer of PsmES4326 in upper leaves of Col-0 WT (black bars) andfmo1 (gray bars). Bars represent the mean ± STD ofthree biological replicates. Asterisks denote thesignificant differences between indicated samplesusing a one-tailed t test (**P < 0.01; *P < 0.05; ns,not significant). The experiment was repeatedthree times with similar results.

Chen et al. PNAS | vol. 115 | no. 21 | E4925

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

with water and Pip treatment, demonstrating that N-OH-Piptreatment also affects ETI.

DiscussionFMO1 Produced Metabolite N-OH-Pip Activates SAR. FMO1-dependent regulation of defense signaling has emerged as a crit-ical determinant of the establishment and amplification of SAR inArabidopsis (16, 31, 32). The SAR deficiency of fmo1 mutantscould be due to a failure to produce a priming signal, to recognizedefense signal(s), or to transmit defense signal(s) from infectedtissue to uninfected tissue. Our discovery that FMO1 produces N-OH-Pip, a bioactive metabolite that is mobile in plants andconfers pathogen resistance in systemic tissues, reveals that thekey biological function of FMO1 is to produce the primingsignal for SAR. The accumulation of N-OH-Pip–derived me-tabolites (N-OGlc-Pip and m/z 100) in untreated, distal leavesof fmo1 plants following infiltration of N-OH-Pip in local leavesindicates that one or more of these metabolites are transportedduring SAR.N-OH-Pip establishment of SAR in Arabidopsis is character-

ized by the induction of transcriptional and metabolic programsin systemic plant tissue. Application of N-OH-Pip to local leavesresulted in the robust activation of several SA-dependent andSA-independent SAR marker genes in distal leaves (Fig. 4C andSI Appendix, Fig. S5). This is consistent with reports showing thatFMO1 is required for SA-dependent and SA-independent de-fenses (31, 32). N-OH-Pip also activated transcription of the N-OH-Pip biosynthetic genes (ALD1, SARD4, and FMO1) in bothlocal and systemic leaves (Fig. 4C), highlighting the importanceof transcriptional feedback regulation for amplifying N-OH-Pip–dependent signaling during SAR.Remarkably, disease symptoms in Arabidopsis leaves were

dramatically reduced in leaves of bacterially infected plantstreated with N-OH-Pip (Figs. 3C and 5B). We speculated that N-OH-Pip might directly affect bacterial growth and/or the abilityof bacteria to deliver T3S effectors to plant cells. However,N-OH-Pip was not toxic to pathogen growth in culture (SI Ap-pendix, Fig. S7) and did not inhibit the activation of ETI in re-sistant Arabidopsis plants (Fig. 6). These data suggest thatN-OH-Pip stimulates or enhances plant-specific processes ininfected leaves to prevent their collapse and yellowing. Whatthese physiological processes are remains to be determined.We also found that N-OH-Pip–treated leaves displayed

enhanced resistance in SAR assays (Fig. 5C) and are highlysensitized for ETI (Fig. 6). This suggests that the level of N-OH-Pip in tissues plays a critical role in establishing the mag-nitude and timing of the disease resistance response. Consistentwith our findings, overexpression of FMO1 in Arabidopsisresulted in enhanced basal resistance to the oomycete Hyalo-peronospora parasitica and P. syringae pv. DC3000 (Pst), as wellas enhanced resistance to Pst avrRpt2 (17). In contrast, loss ofFMO1 function resulted in enhanced susceptibility to Pst andH. parasitica (17, 18).

Biochemistry of FMO1 and Chemistry of Pip Hydroxylation. Flavo-protein monooxygenases (FMOs) catalyze incorporation of oneatom of molecular oxygen onto a nucleophilic substrate (36).Similar to plant cytochromes P450 that activate and transferoxygen, the FMO family of genes has greatly expanded in plants,suggesting they play critical roles in metabolism and fitness. InArabidopsis, 29 genes encode FMO-like proteins, compared withonly 5 in animals (37). These proteins have been divided intothree clades: Clade 1 includes FMO1 and a pseudogene. Clade2 includes the YUCCA group of 11 enzymes, which have beenwell-studied and are associated with auxin biogenesis. Clade3 has 16 enzymes, including an enzyme that S-oxygenates glu-cosinolates, molecules critical for pathogen defense (37). Ofthese, FMO1 is notably one of the most overexpressed metabolic

genes after pathogen challenge (SI Appendix, Fig. S9). Our dis-covery that the FMO1 product N-OH-Pip is a bioactive SARmetabolite is further evidence that FMOs have a privileged rolein plant defense metabolism. Moreover, it predicts that directproducts of FMO1 homologs in other plants may be novel nat-ural products involved in defense priming.Numerous nonproteinogenic amino acids have been described

in biological systems (38), yet N-OH-Pip has not previouslybeen reported. It remains a mystery why N-oxidation of pipecolicacid, a metabolite produced in both plants and bacteria, hasevolved as a key step in generating an active signaling moleculefor SAR. The hydroxylamine functionality bears resemblanceto the hydroxamic acids required for binding iron in bacterial

Fig. 6. AvrRpt2-elicited HR phenotype occurs earlier in Arabidopsis Col-0 WT and fmo1 plants treated with N-OH-Pip compared with water or Pip.Col-0 WT and fmo1 plant pots were treated with 10 mL of water, 1 mM Pip,or 1 mM N-OH-Pip by root application. Twenty-four hours later, leaves ofCol-0 WT and fmo1 plants were infiltrated with a 3 × 108 cells per mL sus-pension of Pst DC3000 with an empty vector (Pst vector) or avrRpt2 (PstavrRpt2). (A) Percent ion leakage of inoculated leaves at 5 hpi. Bars (black,WT; gray, fmo1) represent the mean ± STD of nine randomly inoculatedleaves of three WT or fmo1 plants. Asterisks denote the significant differ-ences between indicated samples using a one-tailed t test (**P < 0.01; *P <0.05). (B) HR phenotypes of inoculated leaves at 8 hpi. Fraction refers tonumber of leaves showing HR of 10 randomly inoculated leaves. Experi-ments were repeated three times with similar results.

E4926 | www.pnas.org/cgi/doi/10.1073/pnas.1805291115 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

siderophores, and intriguingly, some bacterial siderophores arenow thought to be involved in signaling (39). Further studies arerequired to determine if the hydroxylamine substituent of N-OH-Pip has a functional role in its transport, perception, and/or metabolism.Results from our transient expression experiments in N.

benthamiana indicate that N-OH-Pip may be unstable in plantaas FMO1-dependent levels of N-OH-Pip decreased from 28 to48 h while levels of m/z 100 increased (SI Appendix, Fig. S1D).We speculate that O-glycosylation of N-OH-Pip could serve as astabilization mechanism and that N-OGlc-Pip may be the pri-mary storage form of the molecule in Arabidopsis. This proposedprocess is reminiscent of SA glycosylation and storage of SA-Glcin response to pathogen infection (40). While we cannot de-finitively rule out that m/z 100 is a product of extraction, itspresence could indicate a passive or active attenuation strategy.For example, the level of N-OH-Pip could be controlled viaconversion to N-OGlc-Pip for storage or m/z 100 for de-activation. Future work should investigate the precise roles of N-OH-Pip and its derivatives in SAR.

Presence of N-OH-Pip Biosynthesis Pathway in Other Plant Species.Because SAR is a broad-spectrum defense strategy shared amongmany plants, we mined available plant genomes to determine theprevalence of enzymes involved in N-OH-Pip biosynthesis. Wetabulated the best BLAST hit between the Arabidopsis SARD4,ALD1, and FMO1 proteins and those of sequenced plantgenomes, and found that >94% of species analyzed hadFMO1 homologs with greater than 50% amino acid identity (SIAppendix, Fig. S8). All plants within the Brassicaceae that weevaluated have enzymes with >88% amino acid identity toArabidopsis SARD4, ALD1, and FMO1, and these plants con-tained the top five homologs to the Arabidopsis FMO1. InArabidopsis, the closest homolog to FMO1 has only a 74% aminoacid identity, and this protein originates from a likely pseudogeneof FMO1 that is missing a large internal coding region. The nextbest BLAST hit has only a 26% amino acid identity to FMO1. Thescarcity of FMO1 homologs in Arabidopsis suggests that the ho-mologs in other plants (and especially those of the Brassicaceae)may have the same function. While theN-OH-Pip pathway may bebroadly conserved, it is possible that this pathway may have arisenindependently in the Brassicaceae and other metabolites maymediate defense signaling outside this plant family.

Is N-OH-Pip a Long-Distance Mobile Signal for SAR? The moleculethat initiates systemic defense signaling is still up for debate (7,14). It is not clear if there is one molecule or a variety of mol-ecules that orchestrate broad-spectrum plant disease resistance.Several mobile metabolites (MeSA, AzA, G3P, and DA) areknown to prime resistance in uninfected tissues (7, 14) and somerequire DIR1 (a lipid transport protein) (10–12, 41), suggestingthat DIR1 is required for the accumulation and/or transport ofmolecules that induce SAR. Notably, AzA, DA, and Pip requireFMO1 for the initiation of SAR (10, 12, 13), highlighting thepossibility that FMO1 may produce one of the elusive long-distance mobile signals critical for SAR. Our work demon-strates that the product of FMO1, N-OH-Pip, is necessary andsufficient to initiate SAR in local and systemic tissue, making it aprime candidate for a long-distance SAR signal. It remains to bedetermined how and where in the plant the N-OH-Pip signal isperceived to initiate SAR.In conclusion, the discovery of N-OH-Pip as a mobile signal

provides critical insight into how plants use small molecules toresist the spread of infection. This work provides an opportunityto address fundamental open questions in SAR biology, in-cluding the mechanism of transport, signal perception, and signalattenuation once a heightened defense response is no longerrequired. The sensitivity of plants to N-OH-Pip treatment also

highlights the possibility for translating a chemical or metabolicengineering approach to prime or enhance disease resistance inplants under pathogen pressure.

Materials and MethodsBacterial Strains and Growth Conditions. P. syringae pv. maculicola strainES4326 (Psm ES4326), P. syringae pv. tomato DC3000 (Pst DC3000), P. syrin-gae pv. tomato carrying pVSP61 + avrRpt2 (Pst avrRpt2), P. syringae pv.tomato carrying pVSP61 vector (Pst vector), E. coli strain DH5 alpha, and A.tumefaciens strain C58C1 pCH32 were used in this study. Growth conditionsfor all bacterial strains are described in SI Appendix, Table S1.

Plant Materials and Growth Conditions. A. thaliana ecotype Col-0 (WT) andfmo1 (SALK_026163) plants were grown in a controlled growth chamber(22 °C, 80% relative humidity, 103 μmol/m2 per s of light intensity) on a 10-hlight/14-h dark cycle. Genotypes were confirmed by PCR using FMO1 gene-specific primers and a T-DNA–specific primer (SI Appendix, Table S2). Four- to5-wk-old adult plants were used for SAR assays, qRT-PCR, and metaboliteprofiling. For seedling hydroponics experiments, Arabidopsis seeds were ster-ilized by suspending seeds in 50% ethanol for 1 min, washing three times insterile water, suspending in 50% bleach for 10 min, and washing three moretimes in sterile water. Seeds were resuspended in 1× Murashige–Skoog (MS)medium with vitamins (PhytoTechnology Laboratories) (pH 5.7) and vernalizedfor 48 h at 4 °C. Seeds (15 ± 1) were placed into 3 mL of MS medium + 5 g/Lsucrose in wells of a six-well microtiter plate. Plates were sealed with Micro-pore tape (3M) and grown in a chamber at 50%humidity, 22 °C, and 100 μmol/m2

per s photon flux under a 16-h light/8-h dark cycle. After 1 wk of growth, spentmedium was replaced with 3 mL of fresh MS medium + 5 g/L sucrose and plantswere grown for one additional week. N. benthamiana plants were grown in soilon a growth shelf with a 16-h light cycle for 5 wk.

Elicitation Methods for Seedling Hydroponics Experiments. For bacterial elici-tation, Pst was grown on LB agar plates at 30 °C. A single colony was pickedand grown to exponential phase in liquid LB at 30 °C. Cells were thencentrifuged at >13,000 × g and resuspended in sterile 1× MS media to anOD600 of 0.2. Seventy microliters of the Pst suspension was added to eachseedling-containing well for elicitation. Plants were elicited 24 h beforesample harvest.

Plant Extraction for Metabolite Profiling. Arabidopsis seedlings were lyophi-lized to dryness and ground using a ball mill (Retsch MM 400). A single 5-mmsteel ball was added to each sample and samples were shaken at 25 Hz for2 min. Samples were resuspended in 10 mL of 80:20 MeOH:H2O per gramwet tissue and heated at 65 °C for 10 min. Samples were filtered through0.45-μm polytetrafluoroethylene (PTFE) filters and analyzed via LC-MS. Adultleaf tissue was flash frozen and ground in liquid nitrogen using a mortar andpestle. Samples were resuspended in 10 mL of 80:20 MeOH:H2O per gramwet tissue and heated at 65 °C for 10 min. Extracts were filtered through0.45-μm PTFE filters and analyzed via LC-MS.

LC-MS Methods. N-OGlc-Pip, SA-Glc, and camalexin were measured usingreverse-phase chromatography on an Agilent 1260 HPLC coupled to anAgilent 6520 Q-TOF ESI mass spectrometer as previously described (24). A5-μm, 2 × 100 mm Gemini NX-C18 column (Phenomenex) was used for sep-aration. The mobile phases were A [water with 0.1% formic acid (FA)] and B[acetonitrile (ACN) with 0.1% FA] and the following gradient was imple-mented at a flow rate of 0.4 mL/min (percentages indicate percent B): 0–30 min (3–50%), 30–45 min (50–97%), 45–50 min (97%), 50–51 min (97–3%),51–55 min (3%). The MS was run in positive ion mode with conditions asused previously (24). For MS/MS analysis, a fragmentor voltage of 10 V wasused with an m/z window of 1.3.

Pip, N-OH-Pip, and hypothesized N-hydroxy piperideine were measuredusing hydrophilic interaction chromatography on an Agilent 1290 UHPLCcoupled to an Agilent 6545 Q-TOF ESI mass spectrometer. A 130-Å, 1.7-μm,2.1 mm × 50 mm Acquity UPLC BEH Amide column (Waters) was used forseparation. The mobile phases were A (water, 10 mM ammonium formate,0.125% FA) and B (95% ACN, 10 mM ammonium formate, 0.125% FA), andthe following gradient was implemented at a flow rate of 0.6 mL/min(percentages indicate percent B): 0–1.5 min (100%), 1.5–6 min (100–40%), 6–8 min (40%), 8–8.5 min (40–30%), 8.5–9.5 min (30–100%), 9.5–12 min(100%). The MS was run in positive ion mode with the following parameters:mass range: 30–1,700 m/z; drying gas: 250 °C, 12 L/min; nebulizer: 10 psig;capillary: 3,500 V; fragmentor: 140 V; skimmer: 65 V; octupole 1 RF Vpp:750 V; 333.3 ms/spectrum. For MS/MS analysis, a fragmentor voltage of 10 V

Chen et al. PNAS | vol. 115 | no. 21 | E4927

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

was used with an m/z window of 1.3 for N-OH-Pip and a fragmentor voltageof 20 V with an m/z window of 1.3 was used for hypothesized N-hydroxypiperideine. The initial 0.5 min of each run was sent to waste to avoid saltcontamination of the MS.

N-OH-Pip Chemical Synthesis. N-OH-Pip was synthesized from L-pipecolic acid(98% purity; Oakwood Chemical) using a modified protocol (42). To begin,10.1 g of L-pipecolic acid was added to a cooled solution of 4.93 g of 88%pure KOH (1 equivalent). Acrylonitrile (5.58 mL; 1.1 equivalents) was thenadded drop wise to the solution over 5 min. The solution was stirred for 1.5 hin an ice bath and another 1.5 h at room temperature. Then, the pH of thesolution was adjusted to 6.6 with 12 M HCl to quench the KOH. A rotaryevaporator was used to evaporate the solvent. Four hundred milliliters ofacetone was added to the residue, and the solution was brought to a boil.After several minutes of boiling, the solution was filtered, and a rotaryevaporator was used to remove approximately half of the solvent. Then, theresidue from the filtration and fresh acetone was added back into theremaining filtrate, and the solution was brought to a boil, filtered, and seton the rotary evaporator. The cycle of heating the solution and filtering wasrepeated five times. Two hundred milliliters of filtrate from the fifth cyclewas then stored at −20 °C overnight to recrystallize.

Crystallized product (2-cyanoethyl-pipecolic acid; 3.78 g) was mixed with60 mL of MeOH and 4.80 g of 70% metachloroperoxybenzoic acid (mCPBA)(1 equivalent) was added to 20 mL of MeOH. The mCPBA solution was addeddrop wise to the cooled 2-cyanoethyl-pipecolic acid slurry over 30 min. Theslurry was stirred for an additional hour in an ice bath, and as the reactionprogressed, the slurry dissolved into solution. Then, 300 mL of precooleddiethyl ether was added to the reaction and the mixture was stored at−20 °C overnight to recrystallize.

Crystallized product (2-cyanoethyl-pipecolic acid oxide; 0.5 g) was dis-solved in 150 mL of acetone in a 250-mL flask with a short path distillationhead. Acetone was slowly distilled drop by drop for 3 h, and fresh acetonewas periodically added to keep the original volume. Then, a rotary evapo-rator was used to evaporate the majority of the solvent and the remainderwas evaporated to dryness under reduced pressure.

1H and 13C NMR spectra were taken of a 25mM solution ofN-OH-Pip in D2Owith a Varian Inova 500 NMR spectrometer. The following parameters wereused for 1H NMR spectra: temperature: ambient; probe: 5-mm PFG switchable;scan number: 16; receiver gain: 40; relaxation delay: 0; pulse width: 8; fre-quency: 499.75 Hz. The following parameters were used for 13C NMR spectra:temperature: ambient; probe: 5-mm PFG switchable; scan number: 120; re-ceiver gain: 54; relaxation delay: 0.5; pulse width: 7; frequency: 125.67 Hz.

Chemical Treatment of Leaves for Bacterial Growth Assay. Leaf numbering wasperformed according to refs. 43 and 44. Three lower leaves (leaf nos. 7–9) ofWT and fmo1 Arabidopsis plants (4- to 5-wk-old) were infiltrated with10 mM MgCl2, 1 mM Pip in 10 mM MgCl2, or 1 mM N-OH-Pip in 10 mMMgCl2. After 24 h, one untreated upper leaf (leaf nos. 11, 12, or 13) of eachplant was inoculated with a 1 × 105 cfu/mL suspension of Psm ES4326. Theinoculated plants were then covered with a dome to maintain humidity. Thetiter of Psm ES4326 in the upper leaves was quantified at 3 dpi by homog-enizing leaves discs in 1 mL of 10 mM MgCl2, plating appropriate dilutionson nutrient yeast glycerol medium supplemented with 1.5% wt/vol agar(NYGA) with rifampicin (100 μg/mL), incubating plates at 28 °C for 2 d, andthen counting bacterial colonies. Four biological repeats were performedper treatment in two independent experiments.

Chemical Treatment of Leaves for qRT-PCR and Metabolite Profiling. Threelower leaves (leaf nos. 7–9) of WT and fmo1 Arabidopsis plants (4- to 5-wk-old) were infiltrated with 10 mM MgCl2, 1 mM Pip in 10 mM MgCl2, or 1 mMN-OH-Pip in 10 mM MgCl2. After 48 h, the three treated lower leaves andthree untreated upper leaves (leaf nos. 11–13) were harvested, pooled, re-spectively, and then frozen in liquid nitrogen. Frozen tissue was pulverizedand divided into two aliquots: one for qRT-PCR and the other for metabolicprofiling. Three biological repeats were performed per treatment in twoindependent experiments.

RNA Isolation and qRT-PCR. Total RNA was isolated from leaves using TRIzolreagent (Invitrogen) according to the manufacturer’s instructions. Two micro-grams of RNA were used to synthesize cDNA by oligo dT and reverse tran-scriptase. For qRT-PCR, each cDNA sample was amplified with gene-specificprimers (SI Appendix, Table S2) using Green Taq DNA polymerase (GenScript)with EvaGreen Dye (Biotium) and theMJ Opticon 2 (Bio-Rad).UBC21 (Ubiquitin-Conjugating Enzyme 21; At5g25760) mRNA abundance was used to normalizethe expression value in each sample. The comparative Ct method (2−ΔΔCt) was

performed to determine the relative expression. The fold change of each valuewas normalized to the value of MgCl2-treated local leaves of WT.

SAR Assay. SAR bacterial growth assays were performed as described (13) withslight modification. Each plant pot was drenched with 10 mL of water, 1 mML-(-)-Pipecolinic acid (Pip) (Oakwood) or 1 mM N-OH-Pip. After 24 h, threelower leaves of each plant were infiltrated with 10 mM MgCl2 or a 5 ×106 cfu/mL suspension of Pst avrRpt2 in 10 mM MgCl2. Two days later, oneupper leaf of each plant was inoculated with a 1 × 105 cfu/mL suspension ofPsm ES4326, and then plants were covered with a dome to maintain hu-midity. The titer of Psm ES4326 in the upper leaves was quantified at 3 dpi byhomogenizing leaves discs in 1 mL of 10 mM MgCl2, plating appropriatedilutions on NYGA medium with rifampicin (100 μg/mL), incubating plates at28 °C for 2 d, and then counting bacterial colonies. Three plants were usedper condition, and the experiment was repeated more than three times.

Construction of FMO1 Mutants. The ORF of FMO1 lacking the stop codon wasamplified from Arabidopsis Col-0 WT cDNA by PCR using FMO1-specificprimers (SI Appendix, Table S2) and cloned into the pCR8/GW/TOPO vector(Life Technologies). Two alanine substitution mutants, FMO1(G17A/G19A)and FMO1(G215A), were generated using pCR8/GW/TOPO-FMO1 as tem-plate and fmo1 mutant primers (SI Appendix, Table S2). All constructs wereconfirmed by DNA sequence analysis. WT and mutant FMO1 cDNAs weresubcloned into pEAQ-HT-DEST3 (45) to create C-terminal 6× His-tagged fu-sion proteins. Plasmids were introduced into E. coli DH5 alpha and A.tumefaciens C58C1 by heat shock transformation.

Transient Expression in N. benthamiana. Agrobacterium strains harboring thepEAQ-gene constructs were grown on LB agar plates with the appropriateantibiotics. After 48 h of growth, cells were removed from plates using aninoculation loop and resuspended in 1 mL of LB. Cells were centrifuged at4,000 × g for 5 min, the supernatant was removed, and cells were resus-pended in 1 mL of Agrobacterium induction medium (10 mM MES buffer,10 mM MgCl2, 150 μM acetosyringone, pH 5.7) and incubated at roomtemperature with shaking for 2 h. Cells were then diluted to a final OD600 of0.3 in induction medium. In tests with supplemented Pip, cells were dilutedto a final OD600 of 0.3 in induction medium + 1 mM Pip. These solutionswere then infiltrated into the underside of N. benthamiana leaves (threeleaves per plant) using a needleless 1-mL syringe. Plants were grown on agrowth shelf with a 16-h light/8-h dark cycle for 28 or 48 h before sampleharvest for metabolic analysis and immunoblot. Total protein of each samplewas extracted from two leaf discs (1-cm diameter per disk) by using ureabuffer (8 M urea, 15% β-mercaptoethanol, 3× Laemmli buffer). Proteinswere separated by 12% SDS/PAGE analysis and transferred to a PVDFmembrane and visualized by Ponceau S red staining before immunoblotanalysis. FMO1-6×-His, FMO1(G17A/G19A)-6×-His, FMO1(G215A)-6×-Hisproteins were visualized by chemiluminescence using anti-His (Qiagen),peroxidase-conjugated secondary antibodies (Bio-Rad), and ECL reagent(GE Biosciences).

Electrolyte Leakage and Hypersensitive Reaction Assays. Electrolyte leakageand hypersensitive reaction (HR) assays were performed according to ref. 46.Thirty- to 32-d-old WT and fmo1 plants were irrigated with 10 mL of water,1 mM Pip, or 1 mM N-OH-Pip. One day later, six to seven random leaves ofeach plant were inoculated with a 3 × 108 cells per mL suspension of PstDC3000 (vector) or Pst DC3000 (avrRpt2) and then incubated at room tem-perature under lights. For electrolyte leakage assay, at 5 hpi, three leaf discs(7 mm diameter) of each plant were pooled and floated in 20 mL of water inPetri dishes. Five minutes later, the leaf discs were transferred to a 15-mLtube containing 3 mL of water and incubated at room temperature for 1 hwith shaking. Conductivity of each sample before and after boiling wasmeasured using an electrical conductivity meter (Spectrum Technologies).The percentage of electrolyte leakage was calculated as conductivity beforeboiling/conductivity after boiling. For HR assays, leaf phenotypes of WT andfmo1 for each condition were record at 8 hpi. Three plants were used percondition, and the experiment was repeated three times.

ACKNOWLEDGMENTS. We thank Russ Li and Russ Stabler for assistance pre-paring synthetic N-OH-Pip and George Lomonossoff (John Innes Centre) forproviding pEAQ plasmid. This work was supported by an HHMI and SimonsFoundation Grant 55108565 (to E.S.S.), NIH DP2 Grant AT008321 (to E.S.S.),National Science Foundation Graduate Research Fellowship DGE-1656518 (toE.C.H.), National Science Foundation Grant IOS-1555957 (to M.B.M.), Bina-tional Science Foundation Grant 2011069 (to M.B.M.), and Ministry of Sci-ence and Technology of Taiwan Grant 105-2917-I-564-093 (to Y.-C.C.).

E4928 | www.pnas.org/cgi/doi/10.1073/pnas.1805291115 Chen et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

1. Spoel SH, Dong X (2012) How do plants achieve immunity? Defence without spe-cialized immune cells. Nat Rev Immunol 12:89–100.

2. Pieterse CMJ, Leon-Reyes A, Van der Ent S, Van Wees SCM (2009) Networking bysmall-molecule hormones in plant immunity. Nat Chem Biol 5:308–316.

3. Cui H, Tsuda K, Parker JE (2015) Effector-triggered immunity: From pathogen per-ception to robust defense. Annu Rev Plant Biol 66:487–511.

4. Couto D, Zipfel C (2016) Regulation of pattern recognition receptor signalling inplants. Nat Rev Immunol 16:537–552.

5. Glazebrook J (2005) Contrasting mechanisms of defense against biotrophic and ne-crotrophic pathogens. Annu Rev Phytopathol 43:205–227.

6. Robert-Seilaniantz A, Grant M, Jones JDG (2011) Hormone crosstalk in plant diseaseand defense: More than just jasmonate-salicylate antagonism. Annu Rev Phytopathol49:317–343.

7. Fu ZQ, Dong X (2013) Systemic acquired resistance: Turning local infection into globaldefense. Annu Rev Plant Biol 64:839–863.

8. Shah J, Chaturvedi R, Chowdhury Z, Venables B, Petros RA (2014) Signaling by smallmetabolites in systemic acquired resistance. Plant J 79:645–658.

9. Park S-W, Kaimoyo E, Kumar D, Mosher S, Klessig DF (2007) Methyl salicylate is acritical mobile signal for plant systemic acquired resistance. Science 318:113–116.

10. Jung HW, Tschaplinski TJ, Wang L, Glazebrook J, Greenberg JT (2009) Priming insystemic plant immunity. Science 324:89–91.

11. Chanda B, et al. (2011) Glycerol-3-phosphate is a critical mobile inducer of systemicimmunity in plants. Nat Genet 43:421–427.

12. Chaturvedi R, et al. (2012) An abietane diterpenoid is a potent activator of systemicacquired resistance. Plant J 71:161–172.

13. Návarová H, Bernsdorff F, Döring AC, Zeier J (2012) Pipecolic acid, an endogenousmediator of defense amplification and priming, is a critical regulator of inducibleplant immunity. Plant Cell 24:5123–5141.

14. Shah J, Zeier J (2013) Long-distance communication and signal amplification in sys-temic acquired resistance. Front Plant Sci 4:30.

15. Dempsey DA, Klessig DF (2012) SOS - too many signals for systemic acquired re-sistance? Trends Plant Sci 17:538–545.

16. Mishina TE, Zeier J (2006) The Arabidopsis flavin-dependent monooxygenase FMO1 isan essential component of biologically induced systemic acquired resistance. PlantPhysiol 141:1666–1675.

17. Koch M, et al. (2006) A role for a flavin-containing mono-oxygenase in resistanceagainst microbial pathogens in Arabidopsis. Plant J 47:629–639.

18. Bartsch M, et al. (2006) Salicylic acid-independent ENHANCED DISEASESUSCEPTIBILITY1 signaling in Arabidopsis immunity and cell death is regulated by themonooxygenase FMO1 and the Nudix hydrolase NUDT7. Plant Cell 18:1038–1051.

19. Jing B, et al. (2011) Brush and spray: A high-throughput systemic acquired resistanceassay suitable for large-scale genetic screening. Plant Physiol 157:973–980.

20. Cecchini NM, Jung HW, Engle NL, Tschaplinski TJ, Greenberg JT (2015) ALD1 regulatesbasal immune components and early inducible defense responses in Arabidopsis. MolPlant Microbe Interact 28:455–466.

21. Hartmann M, et al. (2017) Biochemical principles and functional aspects of pipecolicacid biosynthesis in plant immunity. Plant Physiol 174:124–153.

22. Ding P, et al. (2016) Characterization of a pipecolic acid biosynthesis pathway re-quired for systemic acquired resistance. Plant Cell 28:2603–2615.

23. Patti GJ, Yanes O, Siuzdak G (2012) Innovation: Metabolomics: The apogee of theomics trilogy. Nat Rev Mol Cell Biol 13:263–269.

24. Rajniak J, Barco B, Clay NK, Sattely ES (2015) A new cyanogenic metabolite inArabidopsis required for inducible pathogen defence. Nature 525:376–379.

25. Rajniak J, et al. (2018) Biosynthesis of redox-active metabolites in response to irondeficiency in plants. Nat Chem Biol 14:442–450.

26. Hartmann M, et al. (2018) Flavin monooxygenase-generated N-hydroxypipecolic acidis a critical element of plant systemic immunity. Cell 173:456–469.e16.

27. Camañes G, Scalschi L, Vicedo B, González-Bosch C, García-Agustín P (2015) An un-targeted global metabolomic analysis reveals the biochemical changes underlyingbasal resistance and priming in Solanum lycopersicum, and identifies 1-methyl-tryptophan as a metabolite involved in plant responses to Botrytis cinerea andPseudomonas syringae. Plant J 84:125–139.

28. Kunkel BN, Bent AF, Dahlbeck D, Innes RW, Staskawicz BJ (1993) RPS2, an Arabidopsisdisease resistance locus specifying recognition of Pseudomonas syringae strains ex-pressing the avirulence gene avrRpt2. Plant Cell 5:865–875.

29. Mindrinos M, Katagiri F, Yu G-L, Ausubel FM (1994) The A. thaliana disease resistancegene RPS2 encodes a protein containing a nucleotide-binding site and leucine-richrepeats. Cell 78:1089–1099.

30. Cameron RK, Dixon RA, Lamb CJ (1994) Biologically induced systemic acquired-resistance in Arabidopsis thaliana. Plant J 5:715–725.

31. Bernsdorff F, et al. (2016) Pipecolic acid orchestrates plant systemic acquired re-sistance and defense priming via salicylic acid-dependent and -independent path-ways. Plant Cell 28:102–129.

32. Gruner K, Griebel T, Návarová H, Attaran E, Zeier J (2013) Reprogramming of plantsduring systemic acquired resistance. Front Plant Sci 4:252.

33. Sledz W, et al. (2015) Antibacterial activity of caffeine against plant pathogenicbacteria. Acta Biochim Pol 62:605–612.

34. Anand A, et al. (2008) Salicylic acid and systemic acquired resistance play a role inattenuating crown gall disease caused by Agrobacterium tumefaciens. Plant Physiol146:703–715.

35. Bandara MBK, Zhu H, Sankaridurg PR, Willcox MDP (2006) Salicylic acid reduces theproduction of several potential virulence factors of Pseudomonas aeruginosa asso-ciated with microbial keratitis. Invest Ophthalmol Vis Sci 47:4453–4460.

36. Huijbers MME, Montersino S, Westphal AH, Tischler D, van Berkel WJH (2014) Flavindependent monooxygenases. Arch Biochem Biophys 544:2–17.

37. Schlaich NL (2007) Flavin-containing monooxygenases in plants: Looking beyonddetox. Trends Plant Sci 12:412–418.

38. Walsh CT, O’Brien RV, Khosla C (2013) Nonproteinogenic amino acid building blocksfor nonribosomal peptide and hybrid polyketide scaffolds. Angew Chem Int Ed Engl52:7098–7124.

39. Johnstone TC, Nolan EM (2015) Beyond iron: non-classical biological functions ofbacterial siderophores. Dalton Trans 44:6320–6339.

40. Dempsey DA, Vlot AC, Wildermuth MC, Klessig DF (2011) Salicylic Acid biosynthesisand metabolism. Arabidopsis Book 9:e0156.

41. Maldonado AM, Doerner P, Dixon RA, Lamb CJ, Cameron RK (2002) A putative lipidtransfer protein involved in systemic resistance signalling in Arabidopsis. Nature 419:399–403.

42. Nagasawa HT, Kohlhoff JG, Fraser PS, Mikhail AA (1972) Synthesis of 1-hydroxy-L-proline and related cyclic N-hydroxyamino acids. Metabolic disposition of 14 C-labeled 1-hydroxy-L-proline in rodents. J Med Chem 15:483–486.

43. Mousavi SAR, Chauvin A, Pascaud F, Kellenberger S, Farmer EE (2013) GLUTAMATERECEPTOR-LIKE genes mediate leaf-to-leaf wound signalling. Nature 500:422–426.

44. Farmer E, Mousavi S, Lenglet A (2013) Leaf numbering for experiments on long dis-tance signalling in Arabidopsis. Protoc Exch, 10.1038/protex.2013.071.

45. Peyret H, Lomonossoff GP (2013) The pEAQ vector series: The easy and quick way toproduce recombinant proteins in plants. Plant Mol Biol 83:51–58.

46. Cheong MS, et al. (2014) AvrBsT acetylates Arabidopsis ACIP1, a protein that asso-ciates with microtubules and is required for immunity. PLoS Pathog 10:e1003952.

Chen et al. PNAS | vol. 115 | no. 21 | E4929

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021