Bacterial medium-chain 3-hydroxy fatty acid metabolites trigger immunity in Arabidopsis plants

See allHide authors and affiliations

Science  12 Apr 2019:
Vol. 364, Issue 6436, pp. 178-181
DOI: 10.1126/science.aau1279

A fatty acid triggers immune responses

Plants and animals respond to the microbial communities around them, whether in antagonistic or mutualistic ways. Some of these interactions are mediated by lipopolysaccharide—a large, complex, and irregular molecule on the surface of most Gram-negative bacteria. Studying the small mustard plant Arabidopsis, Kutschera et al. identified a 3-hydroxydecanoyl chain as the structural element sensed by the plant's lectin receptor kinase. Indeed, synthetic 3-hydroxydecanoic acid alone was sufficient to produce a response. A small microbial metabolite may thus suffice to trigger immune responses.

Science, this issue p. 178


In plants, cell-surface immune receptors sense molecular non–self-signatures. Lipid A of Gram-negative bacterial lipopolysaccharide is considered such a non–self-signature. The receptor kinase LIPOOLIGOSACCHARIDE-SPECIFIC REDUCED ELICITATION (LORE) mediates plant immune responses to Pseudomonas and Xanthomonas but not enterobacterial lipid A or lipopolysaccharide preparations. Here, we demonstrate that synthetic and bacterial lipopolysaccharide-copurified medium-chain 3-hydroxy fatty acid (mc-3-OH-FA) metabolites elicit LORE-dependent immunity. The mc-3-OH-FAs are sensed in a chain length– and hydroxylation-specific manner, with free (R)-3-hydroxydecanoic acid [(R)-3-OH-C10:0] representing the strongest immune elicitor. By contrast, bacterial compounds comprising mc-3-OH-acyl building blocks but devoid of free mc-3-OH-FAs—including lipid A or lipopolysaccharide, rhamnolipids, lipopeptides, and acyl-homoserine-lactones—do not trigger LORE-dependent responses. Hence, plants sense low-complexity bacterial metabolites to trigger immune responses.

Plant cell–surface pattern recognition receptors sense characteristic microbe- or pathogen-associated molecular patterns (MAMPs or PAMPs) and host-derived damage-associated patterns (DAMPs) and activate pattern-triggered immunity (1). Lipopolysaccharide, a glycolipid consisting of lipid A, core oligosaccharide, and O-polysaccharide, is the major outer membrane component of most Gram-negative bacteria (2). In humans and mammals, lipid A triggers inflammation and pyroptosis, potentially culminating in life-threatening sepsis (2, 3). Lipopolysaccharide also elicits defense responses in various plant species (4). In Arabidopsis thaliana (hereafter Arabidopsis) and other crucifers, lipopolysaccharide preparations from Pseudomonas spp. and Xanthomonas campestris, but not from Escherichia coli, Salmonella enterica, or Burkholderia spp., activate characteristic immune responses through the plant-specific bulb-type lectin receptor kinase LIPOOLIGOSACCHARIDE-SPECIFIC REDUCED ELICITATION/S-DOMAIN-1-29 (LORE/SD1-29) (5). Pseudomonas aeruginosa lipid A is sufficient to trigger immunity, whereas removal of the ester-linked acyl chains from lipid A abolishes its activity (5). Thus, lipid A with a distinctive acylation pattern appeared to be critical for immune sensing in Arabidopsis.

To decipher the structural features of LORE-dependent lipid A sensing, we screened lipopolysaccharide with different acylation patterns from various bacterial species and P. aeruginosa lipid A biosynthesis mutants for activation of cytosolic calcium ([Ca2+]cyt) signaling as an indicator of immune response in Arabidopsis (table S1 and figs. S1 to S4). Structurally, all tested, elicitor-active lipopolysaccharide preparations comprise a medium-length acyl chain, mostly 3-OH-decanoyl, at positions C-3 and/or C-3′ of lipid A, whereas lipopolysaccharide with 3-OH-tetradecanoyl chains at these positions was inactive. This suggests that a 3-OH-decanoyl chain is a key structural feature for LORE-dependent lipopolysaccharide sensing.

Therefore, synthetic 3-OH-FAs of varying chain length (table S2) were tested for activation of Arabidopsis immune responses. Free synthetic 3-OH-C10:0 (1; see table S2 for assignment of compound numbers; Fig. 1A) elicits increases in [Ca2+]cyt comparable to those of Pseudomonas lipopolysaccharide preparations (Fig. 1B and fig. S5A). Medium-chain (mc)-3-OH-FAs (bearing 8 to 12 carbon atoms, 1 to 5), particularly 3-OH-C10:0 (1), but not long-chain (lc)-3-OH-FAs (13 to 16 carbon atoms, 6 to 8) activated LORE-dependent [Ca2+]cyt signaling, biphasic production of reactive oxygen species (ROS), transcript accumulation of defense-related genes, and phosphorylation of mitogen-activated protein kinases (Fig. 1, C and D, and fig. S5, B to F). In plants, local MAMP application induces systemic resistance to pathogen infection (6). Wild-type Arabidopsis plants pretreated with 3-OH-C10:0 (1) were more resistant to P. syringae infection, but this systemic resistance was not activated in Arabidopsis lore-1 mutants or when 3-OH-C14:0 (7) was used (Fig. 1E and fig. S5G). Solanaceous Nicotiana benthamiana plants are insensitive to Pseudomonas lipopolysaccharide preparations but gain responsiveness upon transient expression of a LORE–green fluorescent protein (GFP) fusion (5). mc-3-OH-FAs (but not lc-3-OH-FAs) also induced ROS production in N. benthamiana leaves expressing LORE-GFP but not leaves expressing a kinase-inactive control (Fig. 1F and fig. S5H). LORE ectodomain expressed and purified from insect cells directly binds 3-OH-C10:0 (1) in vitro. Under these conditions, the observed binding affinity was relatively low, however (Fig. 1G and fig. S6). Taken together, this supports that LORE is a receptor for mc-3-OH-FAs.

Fig. 1 The 3-OH-FAs trigger LORE-dependent immune responses in a chain length–specific manner.

(A) Chemical structure of 3-OH-C10:0 (1). (B and C) Maximum (Max.) increases in [Ca2+]cyt in Arabidopsis Col-0AEQ seedlings treated with (B) different concentrations of 3-OH-C10:0 (1, stock concentration determined by quantitative nuclear magnetic resonance spectroscopy) or (C) 1 μM 3-OH-FAs of various chain length [mean ± SD, n = 12 (B), n = 3 (C)]. Kinetics are shown in fig. S5, A and B. (D) Defense gene expression in Arabidopsis seedlings 4 hours after elicitation with 3-OH-FAs (1 μM) relative to Col-0 MeOH control. Individual data (symbols) and means (bars) of three experiments are shown. (E) Pst DC3000 titer 4 days after infection in Arabidopsis plants pretreated with 10 μM 3-OH-FAs or MeOH for 3 days before infection. Individual data (symbols) and means (bars) of two experiments are shown (n = 6). Different letters indicate significant differences [one-way analysis of variance (ANOVA) with Tukey’s post hoc test applied separately to Col-0AEQ (small letters) and lore-1 (capitals), P < 0.01]. Similar results occur with 2 days of pretreatment (fig. S5G). CFU, colony-forming units. (F) Maximum ROS accumulation in leaf disks from N. benthamiana transiently expressing LORE-GFP or kinase-inactive LOREm-GFP elicited with 1 μM 3-OH-FAs (mean ± SE, n = 8). LOREm, kinase-inactive LORE variant (Lys516 mutated to Ala); RLU, relative light units. (G) Quantification of binding between 3-OH-C10:0 (1) and fluorescently His-tag-labeled LORE ectodomain by microscale thermophoresis (MST). Measurements were performed at a light-emitting diode (LED) power of 90% and MST power of 80%. Data points indicate the fraction of labeled LORE ectodomain bound to 3-OH-C10:0 (1) (mean ± SE, n = 4). EC50, half-maximal effective concentration. Full figures including controls and additional data with alternatively labeled LORE ectodomain are provided in fig. S6. Experiments were repeated two (F and G), three (B), or four times (C) with similar results.

Next, we tested whether a 3-hydroxyl and/or a carboxyl group are molecular determinants for activating LORE-mediated immunity (Fig. 2). Compared with mc-3-OH-FAs (1 to 5), mc-2-OH-FAs (11 to 13) and nonhydroxylated mc-FAs (15 to 17) induced only limited immune responses in Arabidopsis and LORE-GFP-expressing N. benthamiana (Fig. 2, A and B, and fig. S7, A to E). Nonhydroxylated or Δ2-unsaturated (19,20) mc-FAs induced weak LORE-independent responses (Fig. 2, A and B, and fig. S7, A to F). The 4-OH-C10:0 (21) and 5-OH-C10:0 (22) did not elicit LORE-dependent ROS production (Fig. 2C and fig. S7G). The 3-oxodecanoic acid (3-oxo-C10:0, 23), or its keto-enol tautomer 3-OH-decenoic acid, induced a moderate LORE-dependent [Ca2+]cyt response (Fig. 2D and fig. S7H). Modification of the 3-hydroxyl group of 3-OH-C10:0 (1) with a methyl ether (24) or acetyl ester (25) or substitution with chlorine (27) abolished elicitor activity (Fig. 2, D and E, and fig. S7, H and J). The synthetic 3-OH-FAs used here are racemates, but Pseudomonas lipopolysaccharide comprises typically (R)-3-OH-decanoyl as primary ester-linked acyl chains (7) (table S1 and figs. S1 and S2). (R)-3-OH-C10:0 (28) induced stronger immune signaling than (S)-3-OH-C10:0 did (29) (Fig. 2F and fig. S8). Collectively, a free 3-hydroxyl group, preferably in the (R)-configuration, is critical for LORE-mediated immune sensing. 3-Decanol (31) or 1,3-decandiol (32) was inactive (Fig. 2G and fig. S9, A and B), indicating that the carboxyl function is also required for activating LORE-mediated immunity. Ester- or amide-linked moieties at the carboxyl group of 3-OH-C10:0 (26, 33 to 43) gradually impaired elicitor activity with increasing hydrophobicity and size, with bulky moieties, such as tert-butyl (35) or glucosamine (43), rendering the 3-OH-C10:0 derivatives inactive (Fig. 2, H to J, and fig. S9, B to E). Thus, mc-3-OH-FAs are the minimal motif for LORE-mediated immune activation, with free (R)-3-OH-C10:0 (28) showing the strongest activity in Arabidopsis.

Fig. 2 The 3-OH-FAs in their free form show the strongest elicitor activity in Arabidopsis.

(A) Maximum increases in [Ca2+]cyt in Arabidopsis seedlings treated with FAs (5 μM) or MeOH (mean ± SD, n = 6). (B) Defense gene expression in Arabidopsis seedlings 4 hours after elicitation with FAs (1 μM) relative to Col-0 MeOH control. Individual data (symbols) and means (bars) of three experiments are shown. (C) Maximum ROS accumulation in Arabidopsis leaf disks treated with the indicated compounds (5 μM; mean ± SE, n = 8). (D, E, and G to J) Maximum increases in [Ca2+]cyt in Arabidopsis seedlings treated with the indicated compounds (5 μM) or MeOH [mean ± SD, n = 6 (D and G to J), n = 3 (E)]. (F) Maximum increases in [Ca2+]cyt in Arabidopsis seedlings treated with different concentrations of (R)-3-OH-C10:0 (28) or (S)-3-OH-C10:0 (29) (mean ± SD, n = 6). (A, C to J) Kinetics are shown in figs. S7 to S9. Experiments were repeated two (A, C, and G to J), three (D and E), or four times (F) with similar results.

The latter indicates that free mc-3-OH-FAs are sensed by Arabidopsis rather than lipid A–bound mc-3-OH-acyl chains. Indeed, unbound 3-OH-C10:0 (1) was detected and quantified by means of stable isotope dilution analysis combined with ultrahigh-performance liquid chromatography–mass spectrometry (SIDA-UHPLC-MS/MS) in all lipopolysaccharide and lipid A preparations that activated LORE-dependent immunity (table S3 and fig. S10). As this includes samples from different sources, including commercial suppliers, or isolated by different multistep procedures (table S3), mc-3-OH-FAs appear to copurify with lipopolysaccharide and lipid A. Detergent-based repurification depleted unbound mc-3-OH-FAs from lipopolysaccharide samples (table S3 and figs. S10 and S11). mc-3-OH-FA–depleted lipopolysaccharide samples did not activate immune signaling, whereas typical responses were induced by subsequent application of synthetic 3-OH-C10:0 (1) or the peptide MAMP flg22 (8) (Fig. 3, A and B, and figs. S12 and S13). mc-3-OH-FA–depleted lipopolysaccharide samples also did not induce systemic resistance to P. syringae infection in Arabidopsis (Fig. 3C). This hints at a separate source of free, elicitor-active mc-3-OH-FAs in bacteria rather than rapid release of mc-3-OH-FAs from lipopolysaccharide in the plant apoplast.

Fig. 3 Lipopolysaccharide, rhamnolipid, lipopeptide, and homoserine-lactone comprising 3-OH-acyl building blocks but devoid of free 3-OH-C10:0 (1) do not activate LORE-mediated immune signaling.

(A and B) [Ca2+]cyt kinetics in Arabidopsis seedlings treated with (A) P. cichorii S400/S200 lipopolysaccharide (50 μg/ml; repurified by gel permeation chromatography using a desoxycholate-containing buffer (DOC-GPC), pool 3; table S3 and figs. S10 and S11, A to C) or (B) heat-detergent-repurified P. aeruginosa H4 lipopolysaccharide (100 μg/ml; table S3 and figs. S10 and S11D). After 30 min, synthetic 3-OH-C10:0 (1) (left) or flg22 (right) was added and measurements continued for 30 min (mean ± SD, n = 3 each). Full figures including controls and additional repurified lipopolysaccharide pools are provided in figs. S12 and S13, A and B. (C) Pst DC3000 titer 4 days after infection in Arabidopsis plants pretreated with P. aeruginosa H4 (25 μg/ml; LPS H4), heat-detergent-repurified P. aeruginosa H4 lipopolysaccharide (25 μg/ml; RP-LPS H4; table S3 and fig. S11D), or water for 3 days before infection. Individual data (symbols) and means (bars) of two experiments are shown (n = 6). Different letters indicate significant differences (one-way ANOVA with Tukey’s post hoc test, P < 0.01). (D and E) Maximum increases in [Ca2+]cyt in Arabidopsis seedlings treated with the indicated compounds (5 μM; table S3) or solvent control (mean ± SD, n = 6). Kinetics are shown in fig. S14. EtOH, ethanol; HSL, homoserine-lactone; RL, rhamnolipid. Experiments were repeated two (D and E) or three times (A and B) with similar results.

Several bacterial compounds, such as polyhydroxyalkanoates (9), rhamnolipids (10), lipopeptides (11), and N-acyl-homoserine-lactones (12), comprise 3-OH-acyl moieties, in Pseudomonas commonly 3-OH-decanoyl moieties (11, 12). The 3-OH-decanoyl-comprising rhamnolipids (13) or synthetic 3-OH-C10:0- (45) and 3-oxo-C10:0-homoserine-lactones (46) did not activate LORE-mediated immune signaling in Arabidopsis (Fig. 3, D and E, and fig. S14). Massetolide A, a 3-OH-decanoyl-containing lipopeptide (14), triggered LORE-independent ROS production and a weak LORE-dependent increase in [Ca2+]cyt (Fig. 3D and fig. S14, A and B), most likely caused by the 3-OH-C10:0 (1) contamination detected in this sample (table S3 and fig. S10). Taken together, this suggests that the receptor kinase LORE senses mc-3-OH-FAs in their free form but not 3-OH-decanoyl-comprising bacterial compounds.

Pseudomonas releases free (R)-3-OH-C10:0 (28) during synthesis of penta-acylated lipid A through PagL-catalyzed (R)-3-OH-C10:0 removal from hexa-acylated lipid A in the outer membrane (15, 16). However, lipopolysaccharide preparations from Pseudomonas ΔpagL mutants still contain free 3-OH-C10:0 (1) and activate LORE-mediated immunity (tables S1 and S3 and figs. S3 and S4, F and G), indicating additional sources of free mc-3-OH-FAs in bacteria. Acyl carrier protein (ACP)– and coenzyme A (CoA)–bound mc-3-OH-FAs are intermediates of fatty acid metabolism. Free mc-3-OH-FAs may be released from ACP and CoA precursors through thioesterase activities (17) or nonenzymatically upon cell lysis in bacteria as well as in plants. In crucifers, mc-3-OH-FAs released upon tissue damage could function as DAMPs. Outer membrane vesicles, constantly released by bacteria (18), may deliver membrane-located or intracellular mc-3-OH-FAs. Firmicutes (19), honey bees (20), and leaf-cutter ants (21) also produce free mc-3-OH-FAs. Thus, mc-3-OH-FAs could be delivered through multiple metabolic sources. We propose that plants evolved to sense a ubiquitous low-complexity metabolite instead of different, complex microbial end products.

Humans, animals, and fungi sense metabolites, including various hydroxy fatty acids, through G protein–coupled receptors (2224). The pro-inflammatory receptor GPR84 expressed in leukocytes senses mc-FAs and their 2- or 3-hydroxylated forms (22, 25). GPR84 may sense mc-FAs and mc-2/3-OH-FAs of pathogenic or commensal bacteria in addition to endogenous metabolites. Therefore, our finding that lipopolysaccharide, rhamnolipid, and lipopeptide preparations from certain bacterial species may contain free mc-3-OH-FAs (table S3) is relevant to immune studies in humans and mammals, as complete removal of highly elicitor-active compounds from such biological samples is difficult (26, 27). In conclusion, both plants and animals may monitor their microbial environments by sensing low-complexity microbial metabolites through cell-surface receptors (22, 23, 28).

Supplementary Materials

Materials and Methods

Figs. S1 to S14

Tables S1 to S3

References (2989)

References and Notes

Acknowledgments: We thank W. Bitter, B. A. Dmitriev, K. Kawahara, U. Mamat, K. Rudolph, S. N. Senchenkova, C. Galanos, M. Ongena, J. M. Raaijmakers, J. Hubert, J. H. Renault, A. Collmer, M. Knight, and the University of Oxford for providing material. We thank B. Buske, B. Kunz, U. Schombel, M. Wenig, D. Günzkofer, and K. Mühlenkamp for technical assistance and I. Hang (Vienna Biocenter Protein Technologies Facility) for help with protein purification. Funding: This work was supported by the German Research Foundation (SFB924/TP-B10 to S.R. and TP-B06 to A.C.V. and Emmy Noether program RA2541/1 to S.R.); grant Rhamnoprot, Région Grand-Est and FEDER (R.S.); Vienna Science and Technology Fund (WWTF) project LS17-047 and grants from the Austrian Academy of Science through the Gregor Mendel Institute (Y.B.). E.S.-L is supported by a Hertha Firnberg Programme postdoctoral fellowship (T-947) from the FWF Austrian Science Fund. Author contributions: S.R. initiated and managed the project and designed plant experiments. C.D., C.S., and T.H. designed chemical syntheses and analytics. C.D. and C.S. performed chemical syntheses and analytics. A.K., N.G., T.G., C.E.C., R.K.E., and S.R. prepared lipopolysaccharide. N.G., D.S., C.E.C., and R.K.E. performed structural analysis of lipopolysaccharide and lipid A samples. N.G., A.K., and S.R. designed lipopolysaccharide repurification approaches. R.S. and S.D. initiated rhamnolipid and lipopeptide experiments and prepared rhamnolipid samples. A.C.V., M.S., A.K., and S.R. designed infection experiments. A.C.V., A.K., and M.S. performed infection experiments. A.K. performed and analyzed calcium, ROS, and peroxidase assays. L.R. performed and analyzed MAPK activation and gene expression experiments. E.S.-L., Y.B., and S.R. designed binding assays. E.S.-L. performed protein expression and binding assays. A.K., C.D., C.S., N.G., L.R., A.C.V., E.S.-L., Y.B., and S.R. analyzed the data. S.R. drafted the manuscript. A.K., C.D., N.G., C.S., L.R., T.G., E.S.-L., and A.C.V. contributed methods sections. A.K., C.D., N.G., S.D., Y.B., R.H., T.H., and S.R. contributed ideas, interpreted results, and critically revised the manuscript. All authors discussed the results and approved the manuscript. Competing interests: The Technical University of Munich has filed a patent application with inventors A.K., C.D., L.R., M.S, R.H., T.H., and S.R. The authors declare no conflicts of interest in relation to this work. Data and materials availability: All data are available in the manuscript or the supplementary material. A. thaliana Col-0 expressing pCaMV35S-apoaequorin in the cytosol (Col-0AEQ) can be requested from Marc Knight (Department of Biosciences, Durham University) under a material transfer agreement with Durham University.

Stay Connected to Science

Navigate This Article