Grass roots chemistry: meta-, an herbicidal nonprotein

Ce´ cile Bertin†‡, Leslie A. Weston†, Tengfang Huang‡, Georg Jander‡, Thomas Owens§, Jerrold Meinwald¶ʈ, and Frank C. Schroeder¶ʈ

Departments of †Horticulture, §Plant Biology, and ¶Chemistry and Chemical Biology and ‡Boyce Thompson Institute, Cornell University, Ithaca, NY 14853

Contributed by Jerrold Meinwald, July 31, 2007 (sent for review May 28, 2007)

Fine fescue grasses displace neighboring plants by depositing large OH quantities of an aqueous phytotoxic root exudate in the soil O OH rhizosphere. Via activity-guided fractionation, we have isolated and identified the nonprotein amino acid m-tyrosine as the major

active component. m-Tyrosine is significantly more phytotoxic NH2 NH2 than its structural isomers o- and p-tyrosine. We show that m- OH O tyrosine exposure results in growth inhibition for a wide range of juglone HO O HO O plant species and propose that the release of this nonprotein tyrosine m-tyrosine amino acid interferes with root development of competing plants. O "p-tyrosine" Acid hydrolysis of total root protein from Arabidopsis thaliana OH showed incorporation of m-tyrosine, suggesting this as a possible mechanism of phytotoxicity. m-Tyrosine inhibition of A. thaliana H3CO root growth is counteracted by exogenous addition of protein O sorgoleone amino acids, with having the most significant effect. Fig. 1. Structures of juglone, sorgoleone, the protein amino acid L-p- The discovery of m-tyrosine, as well as a further understanding of tyrosine, and its isomer, L-m-tyrosine. its mode(s) of action, could lead to the development of biorational approaches to weed control. inhibition of lettuce root growth, was selected for further allelopathy ͉ festuca ͉ rhizosphere ͉ root ecology ͉ Arabidopsis fractionation (Fig. 2) by reverse-phase column chromatography on C18-coated silica gel, followed by size exclusion chromatog- oot exudation of small molecules plays a major role in plant raphy with Sephadex LH20 beads. At all stages, biological Recosystems and is often associated with the development of activity was monitored by using the same filter paper assay. competitive advantage through allelopathy (1, 2). Juglone, a More than 80% of the resulting active fraction consisted of highly phytotoxic naphthoquinone produced by black walnut one major component, which was characterized without addi- (Juglans nigra L.), and sorgoleone, a substituted quinone from tional purification via a standard set of two-dimensional NMR sorghum (Sorghum spp.), are two classic examples of potently spectra,includingDQF-COSY,(1H,13C)-heteronuclearmultiple- active allelochemicals deposited via the plant’s living root system quantum correlation spectra, and (1H,13C)-heteronuclear mul- (Fig. 1) (3). Elucidation of the structures and mode of action of tiple-bond correlation spectra (6). The NMR-spectroscopic data previously unknown root-derived phytotoxins could lead to new suggested a 3Ј-substituted phenylalanine derivative as the struc- biorational approaches to weed control. ture of the major component. Additional analyses by high- Because of their stress tolerance and disease resistance, fescue resolution positive-ion electrospray mass spectrometry yielded a (Festuca spp.) grasses are commonly used in landscape, roadside, molecular formula of C9H11NO3. In combination with the results and pasture settings, as well as for conservation purposes (4, 5). from UV and infrared spectroscopic analysis, these data indi- The unusual ability of many fine leaf fescue species to outcom- cated that 3-hydroxyphenylalanine, commonly known as m- pete other plants is well known, and previous investigations tyrosine, is the major component of the active fraction isolated suggested that fescue root exudates have phytotoxic properties from the root exudates. This structural assignment was con- (4). Here, we report the isolation, identification, and biological firmed via an NMR-spectroscopic mixing experiment, whereby activity of m-tyrosine, a potent, structurally unusual broad- a small amount of synthetic m-tyrosine was added to the isolated spectrum phytotoxin exuded by the roots of some fine leaf fescue active fraction (7). Finally, the absolute configuration of the grasses. isolated m-tyrosine was determined to be L by NMR- spectroscopic comparison of its (S)-methoxytrifluoromethylphe- Results and Discussion nylacetic acid [(S)-MTPA] derivative with the (R)- and (S)- In an initial field evaluation of 80 fine fescue cultivars, 8 cultivars MTPA derivatives of synthetic m-tyrosine (8). with strong weed suppressive potential were identified and their allelopathic potential in laboratory settings was confirmed (4). Author contributions: C.B., L.A.W., G.J., T.O., and F.C.S. designed research; C.B., T.H., G.J., Based on both field and laboratory results, we selected ‘‘In- and F.C.S. performed research; C.B., T.H., G.J., and F.C.S. analyzed data; and L.A.W., G.J., trigue,’’ a common Chewing’s fescue cultivar (Festuca rubra L. J.M., and F.C.S. wrote the paper. ssp. commutata), for further studies. The authors declare no conflict of interest. To identify the allelopathic compound(s) contained in In- Abbreviations: MTPA, methoxytrifluoromethylphenylacetic acid; MS medium, Murashige trigue root exudates, we developed an activity-guided separation and Skoog medium. scheme based on the inhibition of lettuce (Lactuca sativa L.) See Commentary on page 16729. radicle elongation in a filter paper-based assay. By using this ʈTo whom correspondence may be addressed. E-mail: [email protected] or fs31@ assay, we compared the phytotoxicity of root surface washes cornell.edu. (hexanes, dichloromethane, methanol, and water) prepared This article contains supporting information online at www.pnas.org/cgi/content/full/ from 2-week-old Intrigue seedlings grown under soil-free con- 0707198104/DC1. ditions. The aqueous root wash, which showed the strongest © 2007 by The National Academy of Sciences of the USA

16964–16969 ͉ PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0707198104 Downloaded by guest on October 7, 2021 A A 1.8

1.6 SEE COMMENTARY )mc( htgnel toor ecutteL toor htgnel )mc( 1.4 160 µM 1.2 40 µM 80 µM 20 µM 1.0 control

0.8

0.6 Fresh roots crude extract 0.4 synthetic m-tyrosine

0.2 B H2O CH3OH CH2Cl2 Hexanes ** * o o 0.0 0 204080160 m-Tyrosine concentration (µµµM) Reversed-phase chromatography B 250 )lortnoc fo %( h %( fo )lortnoc H2O CH3OH:H2O CH3OHo ** * 200

Sephadex column 150 chromatography o-tyrosine t

gn p-tyrosine

e m-tyrosine

l 100

1 ... … 8 9 10 t o

o o ec or

** ** ECOLOGY HPLC + NMR/MS 50 u m-Tyrosine tteL 0 0 50 100 150 200 250 300 Fig. 2. Fractionation of fescue root exudates. (A) F. rubra rubra cv. Intrigue m-/o-/p-Tyrosine concentration (µµµM) roots. The slight yellowish discoloration near the root tips (arrows) indicates accumulation of root exudates on actively growing roots. (B) Activity-guided Fig. 3. Effect of m-tyrosine on lettuce. (A) Comparison of the effect of F. fraction of cv. Intrigue root exudates. Fractions are classified as strongly active rubra cv. Intrigue aqueous root exudate extract and authentic m-tyrosine on ( ), slightly active ( ), and not active (G) in our filter paper phytotoxicity assay ** * lettuce (L. sativa) seedling root growth (9). (Inset) Photograph of 3-day-old (7). lettuce seedlings exposed to various concentrations of aqueous root exudate extract, showing stunted growth and discoloration of root tips. (B) Effect of o-, m-, and p-tyrosine on lettuce seedling radicle elongation (9). Analysis of aqueous extracts of root exudates from multiple fine fescue cultivars and related species showed that all Arizona fescue (F. arizonica), creeping red fescue (F. rubra ssp. rubra), tyrosine accounted for the majority of observed toxicity (Fig. and Chewing’s fescue (F. rubra ssp. commutata) cultivars pro- 3A). In assays using enantiomerically pure samples of m- duced large amounts of m-tyrosine, whereas hard (Festuca tyrosine, the D and L enantiomers proved equally effective in longifolia), sheep (Festuca ovina), and Idaho (Festuca idahoensis) inhibiting lettuce root growth, with the concentrations required fescues did not produce detectable amounts of m-tyrosine (9). to achieve 50% reduction of lettuce root growth (IC50) being 17 m-Tyrosine has been detected in only one other plant species, and 21 ␮M, respectively. Whereas lettuce roots exposed to Euphorbia myrsinitis (10), and we are not aware of any reports of m-tyrosine showed stunting and brownish, discolored roots, this nonprotein amino acid in natural soils. shoot growth of 5-day-old lettuce seedlings was only marginally HPLC analyses showed that m-tyrosine constitutes 33–43% of affected (Fig. 3A Inset). This could indicate that roots are the dry weight of Intrigue aqueous root exudate extract and is uniquely sensitive, or that m-tyrosine is not transported to the also an abundant metabolite in intact plant tissue. In 1-week-old shoots. After 7 days of m-tyrosine treatment, lettuce seedlings seedlings, m-tyrosine was 10-fold more abundant in roots (6,500 showed a significant reduction in shoot growth, possibly as a pmol/mg wet weight; mean Ϯ SD of n ϭ 6) than in leaves (590 Ϯ result of insufficient root development. In contrast to the root 160 pmol/mg wet weight; mean Ϯ SD of n ϭ 5). Free m-tyrosine growth inhibition by m-tyrosine, p-tyrosine and o-tyrosine ac- was present at much lower concentrations in seeds (24 Ϯ 8 tually stimulated lettuce root growth at concentrations as low as pmol/mg; mean Ϯ SD of n ϭ 6), suggesting that the biosynthesis 50 ␮M (Fig. 3B). of this metabolite is initiated after germination. Because me- Because our previous studies suggested that fine leaf fescue tabolite production is often highly dependent on plant age and may impact growth of a wide range of plant species (12), we environmental factors (11), the absolute amounts of m-tyrosine explored the m-tyrosine susceptibility of a selection of monocot produced by the various fescue species and cultivars will vary and dicot plants by measuring root growth in a filter paper assay with plant growth stage and in response to changes in growth [Fig. 4 and supporting information (SI) Table 1]. The IC50 of conditions. m-tyrosine ranged from 10 to 260 ␮M for the tested species. To determine whether the phytotoxicity of aqueous extracts of Root elongation of fescue species that produce m-tyrosine was fine fescue root exudate can be attributed to m-tyrosine, the not affected by 20 to 160 ␮M synthetic m-tyrosine, whereas activity was compared with that of authentic DL-m-tyrosine nonproducing fescue species were strongly sensitive (Fig. 4C). solutions. By using an array of equivalent concentrations of Further experiments to characterize m-tyrosine toxicity aqueous root exudates extracts and DL-m-tyrosine, nearly iden- were conducted with the model plant Arabidopsis thaliana tical dose–response curves were obtained, indicating that m- (Arabidopsis). Whereas D- and L-m-tyrosine were equally toxic

Bertin et al. PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 ͉ 16965 Downloaded by guest on October 7, 2021 A higher than that of m-tyrosine, but the overall phenotypic 120 effects were similar. p-Hydroxyphenylpyruvate, the deami- )%( htgneL tooR dezilamroN tooR htgneL )%( nated form of p-tyrosine, was nontoxic at concentrations up to 40 ␮M. 100 We have considered several mechanisms by which m-tyrosine could inhibit root growth, including direct interference with 80 amino acid metabolism, inhibition of cell wall formation, and alteration of plant hormone signaling. In addition, we considered 60 the possibility of m-tyrosine interfering with soil microbial ecology. However, growth of Escherichia coli, Bacillus cereus, 40 Bacillus subtilis, and the soil fungus Metarhizium anisopliae was Intrigue not inhibited at concentrations as high as 25 mM. Given the Oxford Ͼ 20 Blue Fescue 1,000-fold higher susceptibility of Arabidopsis (and most other Sandpiper monocots and dicots), it seemed likely that m-tyrosine toxicity 0 results from direct interference with plant metabolism. --4∞,5-4,0-3,5-3,0 Other root-deposited allelochemicals such as juglone and sorgoleone appear to interfere directly with photosynthesis and B other processes in the plant cell or cell membrane (13, 14).

)%( dezila )%( However, m-tyrosine affected neither photosynthetic efficiency 100 nor production in lettuce seedlings, suggesting a different mode of action. The structures of both juglone and 80 sorgoleone contain a quinonoid system, and their mode of action

m appears to be directly related to the chemical properties of this r

oN h oN 60 structural feature (15). Although m-tyrosine could serve as a biosynthetic precursor of a quinonoid amino acid, this is unlikely t

gneL to play a significant role for its toxicity because o-tyrosine, which 40 also is a potential precursor for the same para-quinone, is devoid

too µ large crabgrass (I50 = 38 M) of herbicidal activity (Fig. 3B and SI Fig. 6). Furthermore, 20 µ white clover (I50 = 47 M) L-DOPA, which could represent a downstream metabolite and R µ dandelion (I50 = 39 M) intermediate in the conversion of m-tyrosine to a corresponding 0 ortho-quinone, is at least 15-fold less toxic to Arabidopsis than -∞ -5.0 -4.5 -4.0 -3.5 m-tyrosine (data not shown). Our observations that low concentrations of m-tyrosine inhibit C 100 the primary root and promote lateral root elongation in Arabi- dopsis and some lettuce isolates suggested that interference with ) plant growth hormones, in particular auxin (indole-3-acetic %( de %( 80 acid), could be a mechanism of m-tyrosine toxicity. Generally,

zil high concentrations of auxin inhibit root growth, whereas very

a 60 low concentrations stimulate root development (16). Further- mr more, the chemical structures of m-tyrosine and especially its oN h oN metabolite 3-hydroxyphenylpyruvic acid suggest that they could

t 40 interfere with auxin-dependent growth regulation. However, gneL expression of an auxin-responsive DR5-GUS fusion (17) was sorghum (sudex) (I = 151 µM)

50

tooR unaffected by m-tyrosine treatment, and six Arabidopsis auxin- 20 µ tobacco (I50 = 40 M) response mutants (ilr1-1, aux1-7, tir1-1, axr1-3, axr2-1, and µ tomato (I50 = 76 M) axr3-1) did not show altered sensitivity to m-tyrosine, suggesting 0 that m-tyrosine does not interfere directly with auxin metabo- -∞ -5.0 -4.5 -4.0 -3.5 lism or activity. Log(m-tyrosine concentration (M)) Given the chemical structure of m-tyrosine, it seemed possible that this compound would interfere with plant amino acid Fig. 4. Effect of m-tyrosine on other plant species. (A) Effect of m-tyrosine metabolism. The toxicity of 3 ␮M DL-m-tyrosine for Arabidopsis on fescue. m-Tyrosine-producing F. rubra ssp. commutata (Intrigue, Sand- root growth was counteracted to some extent by the addition of piper) is unaffected, whereas m-tyrosine nonproducing F. longifolia (Oxford) ␮ and F. ovina var. glauca (blue fescue) show strong root growth inhibition. (B 14 of the 20 protein amino acids at 40 M concentrations (Fig. and C) Effect of m-tyrosine on seedling radicle elongation for several common 5B). Addition of charged amino acids caused little or no im- weed and crop species. For additional species examples, see SI Table 1. provement in root growth, which may indicate that aromatic and neutral amino acids compete with m-tyrosine for uptake or transport within the roots. In control experiments, the protein toward lettuce in the filter paper bioassay, L-m-tyrosine (IC50, amino acids by themselves did not significantly improve root 0.6 ␮M) was 10 times more toxic than the D enantiomer (IC50, growth at these concentrations (data not shown). 9.3 ␮M) in an agar-based Arabidopsis bioassay. Different Although the pattern of amino acid rescue (Fig. 5B) is similar uptake profiles of D- and L-m-tyrosine in lettuce and Arabi- to the profile of amino acids that are substrates for the AAP1 dopsis could account for this difference in the relative toxic- transporter, two Arabidopsis land race Wassilewskija aap1 mu- ities of the two enantiomers. Alternatively, conversion of tants (18) did not show elevated resistance to m-tyrosine (data D-m-tyrosine to the presumably more active L-m-tyrosine by an not shown). However, the Arabidopsis genome encodes a large amino acid racemase or by an aminotransferase, with m- number of predicted and proven amino acid transporters, and it hydroxyphenylpyruvate as an intermediate, might occur more is possible that other such proteins play a nonredundant role in readily in lettuce than in Arabidopsis. The m-hydroxy- the uptake or within-plant movement of m-tyrosine. For in- phenylpyruvate IC50 for Arabidopsis root elongation was 5-fold stance, the recently identified LHT1 transporter, which contrib-

16966 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0707198104 Bertin et al. Downloaded by guest on October 7, 2021 A However, because significant growth inhibition is observed at Ͼ100-fold lower concentrations in Arabidopsis than in B. subtilis (23) or Chinese hamster ovary cells (24), plant proteins would SEE COMMENTARY 20 80 5 10 have to be uniquely sensitive to m-tyrosine incorporation. 320 In summary, our results show that several fine fescue species 2 release large amounts of m-tyrosine into the rhizosphere, and that this nonprotein amino acid functions as a broad-spectrum 1 ?µ 0.5 M m-tyrosine phytotoxin. Although several hundred nonprotein amino acids, 0.25 including several with potential allelopathic properties, have been identified in plants (29), m-tyrosine is unique in being both 0 root-exuded and inhibitory to root growth of other plants at low (micromolar) concentrations. Given the increasing public con- cern about the use of synthetic , there is great need for B 45 new approaches to weed management. Therefore, the identifi- 40 * cation of m-tyrosine as a naturally produced phytotoxin may 35 contribute to the development of effective and more environ- mm ni htgnel toor htgnel ni mm 30 mentally friendly weed management systems. 25 * * Materials and Methods 20 Plant Material and Chemicals. Seeds of fine leaf fescue cultivar * * * 15 * Intrigue (F. rubra spp. commutata) were donated by Preferred * ****** 10 * Seed Company (Buffalo, NY). Other fescue seeds were obtained 5 from Scott’s (Marysville, OH), Turf Merchants (Tangent, OR), Lebanon Seaboard (Lebanon, PA), and Seed Superstore (Buf- 0 l l l teM teM rhT rhT re re ryT ryT 3 ylG siH ylG siH syC syC yln yln syL syL orP prT orP nlG prT nlG ns ns ps ulG ps ulG ehP e ehP e ueL al g ueL al g ortnoc ryT r ortnoc l a l a falo, NY). Lettuce (L. sativa L.) seeds were purchased from r r I I ON V V A A S S A A A A A A o o ryT-mM ryT-mM ryT-mM 4 Johnny’s Selected Seed (Winslow, ME). Seeds of A. thaliana land HN HN

race Columbia-0 (Col-0) and mutant lines were obtained from ECOLOGY -m oN -m -

μ the Arabidopsis Biological Resource Center (www.arabidop-

3 3 sis.org). The following weed seeds were purchased from Herbi- Fig. 5. m-Tyrosine inhibition and amino acid rescue of Arabidopsis root seed (Twyford, U.K.): dandelion (Taraxacum officinale Weber in growth. (A) Photograph of 5-day-old Arabidopsis seedlings exposed to a series Wiggers), large crabgrass (Digitaria sanguinalis L.), black medic of m-tyrosine concentrations. (B) Rescue of 3 ␮M m-tyrosine toxicity by 40 ␮M (Medicago lupulina L.), cress (Lepidium sativum L.), barnyard- or individual amino acids or NH4NO3. Root length after 1 week of growth on grass [Echinochloa crus-galli (L.) Beauv.], annual bluegrass (Poa MS agar with 3 ␮M m-tyrosine and other amino acids at 40 ␮M. Shown is annua L.), birdsfoot trefoil (Lotus corniculatus L.), broadleaf mean Ϯ SD of n ϭ 16–23. *, P Ͻ 0.01; t test, relative to the 3 ␮M m-tyrosine-only plantain (Plantago major L.), mouse-ear chickweed (Cerastium treatment. vulgatum L.), common chickweed (Stellaria media L.), velvetleaf (Abutilon theophrasti Medicus), purslane (Portulaca oleracea L.), utes to the root import of several amino acids (19, 20), might also and white clover (Trifolium repens L.). be involved in m-tyrosine uptake from the medium. Seeds used for the capillary mat system or for Petri dishes bioassays were surfaced sterilized by suspension for 1 min in 5% Inhibition of the shikimate pathway 3-deoxy-D- ethanol, followed by three rinses with distilled water. Seeds were arabino-heptulosonic acid 7-phosphate synthase by m-tyrosine used in the bioassays immediately after sterilization. Unless other- in E. coli (21) suggested that similar effects could produce wise specified, all chemicals and solvents were purchased from toxicity in plants. However, we did not observe significant Sigma–Aldrich (St. Louis, MO). A sample of L-m-tyrosine was changes in the abundance of aromatic amino acids in Arabidopsis kindly donated by Albany Molecular Research (Syracuse, NY). seedlings growing on growth-inhibiting levels of m-tyrosine. The majority of phenylalanine flux in plants passes through phenyl- Preparation of Fescue Root Exudate Extracts. Fine leaf fescue alanine lyase, and some plant phenylalanine ammonia species were grown on a capillary mat system (30). Seeds were lyase can convert m-tyrosine to m-hydroxycinnamate sterilized by suspension for 1 min in 50% ethanol, followed by (22). However, m-hydroxycinnamate concentrations up to 80 ␮ three rinses with distilled water. Approximately 50 g of sterilized M did not inhibit root growth (data not shown), indicating that fine leaf fescue seeds were placed between two layers of wet enzymatic conversion by phenylalanine ammonia lyase probably cheesecloth (40 ϫ 50 cm) arranged on the capillary mat system does not contribute to m-tyrosine toxicity. and grown for 14 days. m-Tyrosine is incorporated into proteins in place of phenyl- Method A. Roots were harvested by separation from the adjacent alanine in bacteria (23) and mammalian cells (24), where it is screen with a razor blade. The fresh roots were carefully associated with increased protein turnover (25). Vigna radiata submerged for 15 min in 200 ml of hexanes, 200 ml of methanol, (mung bean) phenylalanine tRNA synthase accepts m-tyrosine 200 ml of dichloromethane, or 200 ml of water. with 25% of the efficiency of phenylalanine (26), suggesting that Method B. Uninjured roots were carefully rinsed with water by m-tyrosine might also be misincorporated into plant proteins. placing the entire mat system into a shallow water bath. After Two days after transfer of Arabidopsis seedlings to agar plates filtration, the exudate extracts obtained through methods A or with 10 ␮M m-tyrosine, the roots contained low but measurable B were evaporated to dryness in vacuo by using a rotary amounts of incorporated m-tyrosine, representing Ͻ1% of the evaporator at ambient temperature. The dried root exudate total phenylalanine in the protein fraction (0.88 Ϯ 0.006%; extracts were weighed and then stored at Ϫ20°C until further use. mean Ϯ SD of n ϭ 3). This misincorporation of m-tyrosine into plant proteins could cause structural disruptions, or might Filter Paper Bioassays with Lettuce. Hexanes, dichloromethane, interfere with what are normally p-tyrosine-specific functions methanol, and aqueous fescue root exudate extracts were used such as the formation of tyrosine cross-links in cell walls (27) or in Petri dish bioassays to assess their effect on lettuce growth. regulation of protein function by tyrosine phosphorylation (28). Whatman no. 1 filter paper (Whatman, Middlesex, U.K.) in

Bertin et al. PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 ͉ 16967 Downloaded by guest on October 7, 2021 10-cm-diameter Petri dishes was treated with 1.0 ml of a solution fraction showed strong phytotoxic activity, whereas the other consisting of 0.125, 0.25, 0.5, and 1.0 mg of root exudate extracts three fractions were inactive. The active fraction was subjected per milliliter of the solvent. To avoid toxic effects of the solvents, to a series of two-dimensional NMR-spectroscopic experiments, filter paper treated with hexanes, methanol, and methylene including phase-sensitive DQF-COSY, heteronuclear multiple- chloride extract solutions were placed in a fume hood for1hto quantum correlation spectroscopy, and heteronuclear multiple- allow complete solvent evaporation. Subsequently, 1.0 ml of bond correlation (SI Table 2). water was added to each filter paper disk. Filter paper from After NMR-spectroscopic analysis, the sample was subjected which 1.0 ml of solvent had been evaporated before addition of to mass spectrometry, using a Micromass (Manchester, U.K.) 1.0 ml water, as well as filter paper with 1.0 ml of water alone, Quattro I tandem mass spectrometer operated in positive-ion was used for control experiments. In additional experiments, the electrospray mode with direct infusion of the sample dissolved in filter paper was treated with 1 ml of aqueous solutions of a 50:50 (vol/vol) solution of methanol and water containing 1% L-p-tyrosine, L-o-tyrosine, D-m-tyrosine, L-m-tyrosine, and DL- ϩ ␮ formic acid. Molecular mass calculated for C9H12NO3 m-tyrosine at 10, 20, 40, 80, 160, and 320 M concentrations. Ten ϩ ϩ lettuce seeds were placed on the moist paper filter in each Petri (M H ) was m/z 182.07 and found was m/z 182.07. dish. After 5 days in a controlled environment (22°C, 45 ␮mol mϪ2 sϪ1 photosynthetic photon flux density), radicle and shoot Determination of the Absolute Configuration of Fescue-Produced length of experimental and control plants were measured. m-Tyrosine. Reference samples of the (S)- and (R)-2-methoxy- 2-trifluoromethyl-2-phenylacetic acid [(S)- and (R)-MTPA] Ј Filter Paper Bioassays with Weed and Crop Species. The weed and derivatives of L-3 -hydroxyphenylalanine were prepared as crop species listed in the plant material section in SI Table 1,as follows. To a well stirred solution of 0.5 mg of L-3Ј- well as the fescue species F. rubra ssp. commutata (Chewing’s hydroxyphenylalanine in 0.5 ml of water at 0°C were added 0.5 fescue cv. Intrigue and Sandpiper), F. rubra ssp. rubra (creeping ml of aqueous NaHCO3 solution, 1 ml of acetone, and 4 ␮lof red fescue cv. Jasper) seeds, F. longifolia (hard fescue cv. either (R)- or (S)-2-methoxy-2-trifluoromethyl-2-phenylacetic Oxford), and F. ovina (sheep fescue), were tested for their acid chloride [(R)- and (S)-MTPA-Cl]. The resulting mixture sensitivity to m-tyrosine by using the filter paper assay described was stirred for 1 h at 20°C. Subsequently, the acetone was above. Seeds were sterilized with 50% ethanol as described evaporated in vacuo by using a rotary evaporator, and the above and were subsequently placed on filter paper treated with aqueous residue was acidified by addition of 1 M aqueous 0, 10, 20, 40, 80, or 160 ␮M DL-m-tyrosine (SI Table 1). HCl and extracted with 1 ml of ether. The organic extract was filtered over a pad of anhydrous Na2SO4 and evaporated Agar Plate Bioassays with Arabidopsis. To assess effects of m- to dryness in vacuo. The residue was dissolved in 0.6 ml tyrosine and other compounds on Arabidopsis root growth, of acetone-d6, and the resulting solution was analyzed by formulated solutions of each were added to half-strength Mu- 1H-NMR spectroscopy. The diastereomeric (S)- and (R)- rashige and Skoog medium (0.5ϫ MS) (31), 1% Phytagar MTPA derivatives of L-3Ј-hydroxyphenylalanine showed sig- (Invitrogen, Carlsbad, CA), and 1% sucrose in Petri dishes. nificant differences in their 1H-NMR spectra. Characteristic Arabidopsis seeds were sterilized by shaking in 30% bleach, 0.3% Triton X-100 for 10 min, followed by three rinses with sterile signals include protons 3-H␣ and 3-H␤ {(R)-MTPA derivative [from (S)-MTPA-Cl] ␦ 3.09 ppm and 3.24 ppm; (S)-MTPA distilled water. Petri dishes with seeds on agar medium were ␦ cold-stratified for 24 h at 4°C, and were subsequently placed derivative [from (R)-MTPA-Cl] 3.02 ppm and 3.19 ppm}. vertically in Conviron (Winnipeg, MB, Canada) growth cham- Subsequently, a portion of fescue-produced m-tyrosine (iso- bers at 23°C, 180 ␮mol mϪ2 sϪ1 photosynthetic photon flux lated from the aqueous fescue root exudate extracts) was density, and a 16:8 h light/dark cycle. After 5 days of growth, the reacted with (R)-MTPA-Cl in the same manner as described root lengths of 10 seedlings per plate were measured. Experi- above. NMR-spectroscopic analysis of the resulting (S)-MTPA ments were repeated three times, and each replicate consisted of derivative produced an 1H-NMR spectrum showing signals for three agar plates. By using this assay, the effects of DL-m-tyrosine protons 3-H␣ and 3-H␤ at 3.02 and 3.19 ppm, indicating that the on Arabidopsis root growth were assessed at concentrations root exudate fraction contained the L-isomer of 3Ј- ranging from 0 to 320 ␮M (Fig. 3B). Rescue of 3 ␮M m-tyrosine hydroxyphenylalanine (m-tyrosine). toxicity was assessed by adding the 20 protein amino acids individually at 40 ␮M concentration to the assay. Phytotoxicity Determination of m-Tyrosine Concentrations. The m-tyrosine con- of L-, D-, and DL-m-tyrosine was compared at 0.25, 0.5, 1, 2, 5, centration in aqueous Intrigue root exudate extracts prepared and 10 ␮M concentrations. Toxicity of L-p-tyrosine, DL-o- according to methods A and B (see above) and in active fractions tyrosine, and m-hydroxycinnamate and L-dopa (L-DOPA) was collected during purification was determined by HPLC analysis, ␮ assessed at concentrations ranging from 1.25 to 40 M. using an Agilent 1100 HPLC system (Agilent, Palo Alto, CA) equipped with a diode-array detector and a Supelco (Bellefonte, Activity-Guided Fractionation of Aqueous Root Exudate Extracts and PA) RP-18 Discovery column (length, 250 mm; diameter, 10 mm), Chemical Analysis. Crude aqueous root exudate extracts (method which was eluted with methanol–water mixtures (starting with a A) were subjected to reverse-phase column chromatography on mixture of 3% methanol, 67% water, and 30% of 0.05% aqueous C -coated silica gel, using a methanol–water solvent gradient 18 trifluoroacetic acid for an initial period of 3 min, followed by a for elution, increasing the methanol content from 0 to 100%. Three fractions were collected for the filter paper bioassay linear gradient reaching 50% methanol, 20% water, and 30% of described above (SI Fig. 7). The aqueous fraction, which showed 0.05% aqueous trifluoroacetic acid at 30 min, at a constant flow of by far the strongest toxicity, was subjected to size-exclusion 3.4 ml/min). In preparation for HPLC analysis, aqueous root column chromatography on Sephadex LH20, using a 1:1 mixture exudate extracts were evaporated to dryness, and the residue was of methanol and water as solvent. Ten fractions were collected, reconstituted at a concentration of 13 mg/ml water. Monitoring which were evaporated separately and then submitted to 1H absorption at 280 nm, peaks at 12.7 min were integrated and NMR spectroscopic analysis using a 600 MHz Varian (Palo Alto, compared against a calibration curve obtained from serial dilutions CA) INOVA spectrometer. Fractions with similar 1H NMR of commercial m-tyrosine in water. Aqueous root exudate extracts spectroscopic profiles were combined, which resulted in four prepared according to methods A or B contained similar amounts fractions that were tested in the filter paper bioassay. One single of m-tyrosine.

16968 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0707198104 Bertin et al. Downloaded by guest on October 7, 2021 Comparison of the Phytotoxic Activities of m-Tyrosine and Root For analysis of F. rubra seed m-tyrosine content, seeds were Exudate Extracts. Aqueous solution of commercial m-tyrosine (10, ground to a fine powder in liquid , an aqueous extract 20, 40, 80, and 160 ␮M concentrations) or 1 ml of an aqueous was prepared, and amino acids were measured as described SEE COMMENTARY solution of dried Intrigue root exudates extract adjusted to above. contain m-tyrosine at identical concentrations, as determined by HPLC, were spotted onto filter paper. Root growth of L. sativa Measurement of Protein-Incorporated m-Tyrosine. Arabidopsis land and D. sanguinalis (10 seedlings per Petri dish) were measured race Col-0 seeds were sterilized and sown on 0.5ϫ MS agar plates after 5 days (L. sativa) or 7 days (D. sanguinalis) and compared that were placed vertically in the growth chamber. After 8 days, with the lengths of root and shoot of control plants grown on seedlings were transferred to new plates with control agar or agar plates treated only with 1 ml water. Each assay was performed containing 10 ␮M m-tyrosine. Two days after transfer, plant in triplicate and repeated at three different times. Root growth roots were harvested into 1.5-ml tubes (Ϸ60 mg for each sample) inhibition was visually observed and measured. Analysis of and immediately frozen with liquid nitrogen. One 3-mm steel variance using the general linear model procedure (release 9.1; ball was placed into each tube, and plant tissue was crushed by SAS Institute, Cary, NC) was carried out on the data (lettuce and shaking on a Harbil 5G-HD paint shaker (Fluid Management, crabgrass root and shoot length), and the means were separated Wheeling, IL). Five hundred microliters of extraction buffer (1ϫ ϭ by least significant difference at the P 0.05 level. Dose– PBS, pH 7.4, with 2 mM phenylmethanesulfonylfluoride) were response curves were fitted to the following four-parameter then added to the crushed sample and mixed by using the same (Ϫ ϩ ln(x)Ϫln(a ء ϭ logistic function as follows: f (b (100 b))/(1 e ), shaker. Samples were centrifuged for 10 min at 13,000 ϫ g at 4°C. where b is the smallest value of the dependent variable (i.e., root Supernatant was transferred to Millipore YM-10 spin columns, or shoot length, respectively), x is concentration of inhibitor, and centrifuged at 13,000 ϫ g for 30 min at room temperature, I50 is the concentration for 50% inhibition of the test species. washed twice with 500 ␮l of extraction buffer, and centrifuged at 14,000 rpm. Samples were adjusted to a final volume of 100 Measurement of Amino Acid Concentrations in Arabidopsis and F. ␮l using extraction buffer. Free amino acid profiles of these rubra. Eight days after germination on 0.5ϫ MS agar plates, samples were analyzed by using the AccQ tag HPLC detection Arabidopsis seedlings were transferred to 0.5ϫ MS agar plates system (Waters) to confirm that they did not contain free with or without 10 ␮M m-tyrosine. Roots were harvested 2 days m-tyrosine. Eighty microliters of each sample were adjusted to

after transfer. To prepare Arabidopsis root samples for amino ECOLOGY a final volume of 400 ␮l with 1% phenol and a final HCl acid analysis, Ϸ15 mg of roots was frozen in liquid nitrogen and ground to a fine powder. An extraction buffer containing 20 mM concentration of 6 M. Samples were then transferred to Kontes HCl and norleucine as a standard was added to each sample at valved NMR tubes (Kontes, Vineland, NJ), and the tubes were the concentration of 10 ␮l/mg of fresh tissue. The homogenized flushed with argon gas. Sealed tubes were then incubated by samples were centrifuged for 10 min at 4°C and filtered by using using a 110°C oil bath for 24 h. After hydrolysis, samples were a Millipore (Billerica, MA) Filtrate Plate. The samples were dried by evaporation, redissolved in 20 mM HCl, and analyzed analyzed by using the AccQ tag HPLC detection system (Waters, by using the AccQ tag HPLC detection system as described Milford, MA). The amino acid separations were carried out by above, except that a modified gradient and a column tempera- using a 3.9 ϫ 150 mm AccQ Tag reversed-phase column (Waters; ture of 30°C were used to improve separation of the m-tyrosine Millipore) at 37°C, with a flow rate of 1.0 ml/min and detection derivative from the methionine derivative. at 254 nm. The two eluent systems used were as follows: eluent A, AccQ Tag eluent A (concentrate A: 190 g of sodium acetate, We thank Dr. Donna Gibson for assistance with bacterial and fungal assays, 17.2 g of triethylamine, 10 mg of EDTA in 1 liter of Milli-Q Drs. Kevin Vaughn and Dominick Paolillo for helpful discussions, Rose water, pH 5.02; working eluent A: 100 ml of concentrate A Harmon and Mia Akoagi for assistance with seedling bioassays, Ludmila Rehak for measurement of F. rubra m-tyrosine content, Dr. Andreas Weber diluted to 1,000 ml in Milli-Q water); eluent B, 60% acetonitrile (Michigan State University, East Lansing, MI) for providing Arabidopsis in Milli-Q water (vol/vol) (for gradients used, see SI Table 3). aap1 mutant seeds, and Dr. Silvina Garcia (Albany Molecular Research, For measurement of m-tyrosine in F. rubra, seeds were North Syracuse, NY) for providing a sample of L-m-tyrosine. This work was germinated in paper seed envelopes and harvested after 1 week. supported in part by National Institutes of Health Grant GM53830, the New Leaves and roots were separated with a razor blade, and amino York State Turfgrass Association, National Science Foundation Grant acid analysis was performed as described above for Arabidopsis. DBI-0453331, and the Triad Foundation.

1. Bais HP, Vepachedu R, Gilroy S, Callaway RM, Vivanco JM (2003) Science 16. Taiz L, Zieger E (2002) in Plant Physiology (Sinauer, Sunderland, MA). 301:1377–1380. 17. Ulmasov T, Murfett J, Hageb G, Guilfoyle TJ (1997) Plant Cell 9:1963–1971. 2. Weston LA, Duke SO (2003) CRC Crit Rev Plant Sci 22:367–389. 18. Lee YH, Foster J, Chen J, Voll LM, Weber AP, Tegeder M (2007) Plant J 3. Bertin C, Yang X, Weston LA (2003) Plant Soil 256:67–83. 50:305–319. 4. Bertin C, Paul RN, Duke SO, Weston LA (2003) J Chem Ecol 29:1919–1937. 19. Svennerstam H, Ganeteg U, Bellini C, Na¨sholm T (2007) Plant Physiol 5. Ruemmele BA, Brilman LA, Huff DR (1985) Crop Sci 35:313–316. 143:1853–1860. 6. Taggi AE, Meinwald J, Schroeder FC (2004) J Am Chem Soc 126:10364–10369. 20. Hirner A, Ladwig F, Stransky H, Okumoto S, Keinath M, Harms A, Frommer 7. Schroeder FC, Weibel DB, Meinwald J (2004) Org Lett 6:3019–3022. WB, Koch W (2006) Plant Cell 18:1931–1946. 8. Evans DA, Britton TC, Ellman JA, Dorow RL (1990) J Am Chem Soc 21. Smith LC, Ravel JM, Lax SR, Shive W (1964) Arch Biochem Biophys 105:424– 112:4011–4030. 430. 9. Bertin C (2005) PhD thesis (Cornell University, Ithaca, NY). 22. Schuster B, Retey J (1995) Proc Natl Acad Sci USA 92:8433–8437. 10. Mothes K, Schutte HR, Mueller P, Ardenne MV, Tuemmler R (1964) Z 23. Aronson JN, Wermus GR (1965) J Bacteriol 99:38–46. Naturforsch 19:1161–1162. 24. Gurer-Orhan H (2006) Biochem J 395:277–284. 11. Reigosa MJ, Pedrol N, Sanchez-Moreiras AM, Gonzales L (2002) in Allelopa- 25. Rodgers KJ, Wang H, Fu S, Dean RT (2002) Free Radic Biol Med 32:766–775. thy, from Molecules to Ecosystems, eds Reigosa MJ, Pedrol N (Sci Publ, Enfield, 26. Smith IK, Fowden L (1968) Phytochem 7:1064–1065. NH), pp 231–256. 27. Held MA, Tan L, Kamyab A, Hare M, Shpak E, Kieliszewski MJ (2004) J Biol 12. Hively WD, Cox WJ (2001) Agron J 93:308–313. Chem 279:55474–55482. 13. Netzly DH, Butler LG (1986) Crop Sci 26:775–778. 28. Luan S (2002) Proc Natl Acad Sci USA 99:11567–11569. 14. Rietveld WJ (1983) J Chem Ecol 9:295–307. 29. Bell EA (2003) J Agric Food Chem 51:2854–2865. 15. Cardellina JH (1988) in Biologically Active Natural Products: Potential Use in 30. Czarnota MA (2001) Dissertation (Cornell University, Ithaca, NY). Agriculture, ed Cutler HG (Am Chem Soc, Washington, DC), pp 305–315. 31. Murashige T, Skoog FA (1962) Physiol Plant 15:473–497.

Bertin et al. PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 ͉ 16969 Downloaded by guest on October 7, 2021