Leaf nodule endosymbiotic Burkholderia confer targeted allelopathy to their Psychotria hosts

After a century of investigations, the function of the obligate betaproteobacterial endosymbionts accommodated in leaf nodules of tropical Rubiaceae remained enigmatic. We report that the α-d-glucose analogue (+)-streptol, systemically supplied by mature Ca. Burkholderia kirkii nodules to their Psychotria hosts, exhibits potent and selective root growth inhibiting activity. We provide compelling evidence that (+)-streptol specifically affects meristematic root cells transitioning to anisotropic elongation by disrupting cell wall organization in a mechanism of action that is distinct from canonical cellulose biosynthesis inhibitors. We observed no inhibitory or cytotoxic effects on organisms other than seed plants, further suggesting (+)-streptol as a bona fide allelochemical. We propose that the suppression of growth of plant competitors is a major driver of the formation and maintenance of the Psychotria–Burkholderia association. In addition to potential agricultural applications as a herbicidal agent, (+)-streptol might also prove useful to dissect plant cell and organ growth processes.

The leaves of discrete genera of Rubiaceae and Primulaceae host Burkholderia symbionts in their mesophyll and form stoma-derived structures, termed nodules, in a limited number of species [1][2][3] . Despite the early description of these associations 4,5 , the bacterial contribution to the plant host remains obscure. Leaf-nodulating Psychotria are restricted to ca. 80 monophyletic tropical African species and have been thoroughly investigated 1,6,7 . A pool of host-specific endosymbionts is maintained into shoot meristems; this bacterial population colonizes reproductive organs and leaves as they emerge. Mature flowers (hence seeds) inherit a small Burkholderia population that perpetuates the symbiotic cycle; a unique case of vertical transmission in higher plants 1,8 . The incongruent phylogenies of the Burkholderia symbionts and their Rubiaceae hosts advocate for occasional, presumably insectdriven, horizontal transmission; however, extant partners do not survive outside of the symbiosis 9,10 . Burkholderia cells proved unviable in axenic cultures and aposymbiotic plants die within months due to the collapse of the shoot apical meristems. The transition from a free-living lifestyle to an obligate symbiont was accompanied by a massive erosion of the Burkholderia genomes 1,11,12 . Genomic and proteomic evidence ruled out historical speculations about the role of the Psychotria endosymbionts, including nitrogen fixation or phytohormones production 11,12 . Yet, neither the molecular mechanisms triggering and maintaining the symbiosis nor the reason of the obligate nature of the symbiosis have been elucidated.
Interestingly, in the context of genetic drift, the genome of Candidatus Burkholderia kirkii has retained a 140kb low copy number plasmid (pKIR01) 11 , which is absent from free-living Burkholderiaceae. However, pKIR01 orthologous genes are also found in other Psychotria leaf nodule symbionts 10 . Their conservation underlines their importance for the symbiosis. The plasmid pKIR01 encodes enzymes with significant homology to genes involved in the biosynthesis of C 7 cyclitol-containing secondary metabolites in Actinobacteria 11 . Bioactivities of such compounds include glycosidase inhibition (acarbose), and antitumoral (pericosine A), antifungal (validamycin A), and insecticidal (validoxylamine A) properties 13,14 .
In contradiction to former reports suggesting that Burkholderia cells shift to a dormant state in mature nodules 15 , the pKIR01 2-epi-5-epi-valiolone synthase (EEVS), catalysing the committed step in cyclitols biosynthesis, was shown to be highly induced in nodules relative to the apical population 12 . Furthermore, the high abundance of pKIR01 proteins assigned to cyclitols production suggests that Candidatus B. kirkii dedicates its metabolism to the production of bioactive compounds once nodules are established. Indeed, two C 7 cyclitols found in high amounts in the nodulated leaves of Psychotria kirkii, but absent from aposymbiotic plants, were recently isolated and synthesized 10,16,17 . The N-acetyl glucosamine analogue kirkamide is toxic to arthropods, thus hinting towards a protective role against folivorous insects 16 . In addition, kirkamide and related compounds inhibit multiple N-acetyl-d-glucosaminidases 18 . The pseudodisaccharide (+)-streptol-β-glucoside (SG) and its aglycone moiety (+)-streptol were shown to inhibit lettuce germination 10,17,19 . The exclusive distribution and conservation of the cyclitol biosynthetic genes in leaf-nodule Burkholderia, supported by the presence of the compounds in host plants 10 , argues for positive selection of these traits and suggests a defensive symbiosis.
Separating chemical interference from other mechanisms, such as resources or spatial competition, has proven difficult and the role of secondary metabolites-based allelopathy in evolutionary ecology is currently under debate 20 . In this study, we investigate the contribution of C 7 cyclitols to the P. kirkii chemical arsenal against plant competitors. This is the first report of a natural antagonist that targets root growth of seed plants.

Results
Allelopathic activity of P. kirkii leaf extracts. We used the high sensitivity of lettuce seedlings towards symbiotic P. kirkii leaf extracts 10 as a benchmark for evaluating the inhibitory activity within distinct plant organs. Lettuce germination was greatly hindered by filtered water extracts of P. kirkii apical shoots, flowers, fruits, hypocotyls and roots ( Supplementary Fig. 1a). In addition, lettuce seeds did not grow in the presence of intact mature drupes but developed when mesocarps were removed from pyrenes ( Supplementary Fig. 1b). Previous work has quantified high concentration of kirkamide and SG, but low concentration of streptol in P. kirkii leaves, shoots and roots 10,17 . Thus the cyclitols appear systemically distributed throughout P. kirkii. Allelochemicals can be released into the environment as volatiles, root exudates, leachates of aboveground parts or by decomposition of plant material. We detected streptol (0.14 ± 0.001 µg ml −1 ) in the root exudates of axenically grown symbiotic P. kirkii, but not of aposymbiotic plants. Leachates obtained by bathing intact P. kirkii leaves in water contained sufficient inhibitory activity to prevent lettuce root establishment, but did not completely block germination, as embryo roots protruded from the seed coat and hypocotyls developed similarly in treated and untreated seedlings ( Supplementary Fig. 1c). In fact, lettuce samples germination displayed a binary response towards P. kirkii leaf extracts, suggesting that inhibition of root development requires a threshold concentration of the inhibitor. We further evaluated the allelopathic potential by growing black mustard seeds on soil watered with P. kirkii leaf leachates or crude extract dilutions. Surprisingly, the seedlings germinated and developed intact aerial organs when the soil was kept moist; yet, root length was drastically affected in both treatments ( Supplementary  Fig. 2a,c). Closer inspection of these roots revealed swollen tissues above the root apical meristem (Supplementary Fig. 2b). In vitro mustard germination rates were not affected by the leachate but the success of seedlings to establish in soil greatly dropped with lower water supply ( Supplementary Fig. 2d). Similar results were obtained with mustard seeds sown on soil amended with mulched P. kirkii leaves. These combined results indicated that symbiotic P. kirkii plants contain and release a potent allelochemical that inhibits root development.
Psychotria kirkii extract specifically impedes root elongation. To clarify the biological activity of P. kirkii extracts, we took advantage of the well-studied model plant Arabidopsis thaliana. In vitro, Arabidopsis seedlings germinated and developed to maturity in the presence of up to 15 ppm P. kirkii extracts (Supplementary Fig. 3a). Aerial organs did not show defects or changes in stature when compared to controls (Supplementary Fig. 3b); however, the root system appeared extremely branched and stunted. Doubling the concentration of P. kirkii extracts in the medium was sufficient that Arabidopsis seedlings formed dwarves ( Supplementary  Fig. 3a). Time-lapse monitoring of Arabidopsis germination and growth on vertical plates supplemented with P. kirkii extracts confirmed that the active compounds rapidly inhibited root elongation, but did not impede seed germination or lateral root emergence ( Supplementary Fig. 4a,b). We noticed that this inhibition was diminished when 1% sucrose was added to the growth medium ( Supplementary Fig. 4d). Transferring control-grown seedlings to medium supplemented with 15 ppm P. kirkii extracts quickly stalled root growth ( Supplementary  Fig. 4c), confirming the organ as the prime target. Thus, the active compounds act as pre-emergence root elongation inhibitors. Microscopic inspection of treated roots revealed a radial swelling of the epidermal and cortical tissues above the meristematic zone (Fig. 1a). Measurements of longitudinal root cell lengths from the quiescent centre in the cortical files showed that cells treated with P. kirkii extracts transitioned earlier from the meristematic (MZ) to the elongation zone (EZ) compared to controls (Fig. 1b). The 10-15 most distal meristematic cells showed equivalent dimensions in controls and treatment, however treated cells entering the transition zone (TZ) quickly elongated while enlarging. By contrast, treated cells protruding the EZ rarely grew longer than 100 µm at maturity. To dismiss that the swelling was due to an increased number of actively dividing cells, we monitored the expression pattern of the cell cycle marker pCYCB1;1::DB-GUS 21 in seedlings transferred to P. kirkii extracts. Symptomatic roots exhibited similar numbers of GUS-positive cells and lengths of pCYCB1;1::DB-GUS expression regions as control roots; no ectopic cell divisions were detected (Supplementary Fig. 5a). In order to assess if the observed cell enlargement is caused by incomplete cytokinesis, we carefully monitor primary cell wall integrity and nuclear partitioning along the root apex zones ( Supplementary  Fig. 5b,c). Symptomatic seedlings did not present abnormal ploidy or incomplete cell plates in the affected EZ or in the MZ, suggesting that cytokinesis occurred normally despite isotropic growth. Also, we did not observe differences in the levels or the distribution of auxin in treated root tips versus control seedlings ( Supplementary  Fig. 5d). Taken together, these data indicate that P. kirkii extracts do not primarily impact cell proliferation nor root patterning but appears to disturb root growth anisotropy at the onset of cellular elongation. www.nature.com/scientificreports/ The α-d-glucose analogue (+)-streptol is the active pharmacophore. We next tested the bioactivity of pure naturally-produced as well as synthetic derivatives of cyclitols 17 (Fig. 2a). The natural products (+)-streptol, (+)-streptol-β-glucoside and (+)-streptol-α-glucoside (A-79197-2 22 ) inhibited root development with apparent IC 50 values of 4.7 ± 1.1 s.e., 9.4 ± 1.1 s.e. and 16.7 ± 1.2 s.e. µM, respectively. Importantly, (−)-streptol, (−)-streptol-β-glucoside and (−)-streptol-α-glucoside (diastereomer of A-79197-2) did not exhibit bioactivity, suggesting that the active compounds may target a specific protein. High concentration of kirkamide (100 µM) had also no effect (Fig. 2b). The dose-dependent radial expansion of root tips upon (+)-streptol treatments confirmed that 5 µM was sufficient to mimic the effects of P. kirkii extracts (Fig. 2c,d). Beside a two-fold change in activity, both (+)-streptol and SG showed comparable effects on Arabidopsis photo-and skotomorphogenic development and treatments with these compounds led to root growth dose-response curves with steep slopes (Supplementary Fig. 6a). Etiolated Arabidopsis hypocotyls did not show abnormalities in the rapidlyelongating apical hook region, suggesting that (+)-streptol targets cells entering into the root EZ. We previously hypothesized that streptol glycosylation might lower the toxicity of the compound for the host plant 17 . As checkerboard assays did not reveal synergism between (+)-streptol and SG ( Supplementary Fig. 6b), we speculated that SG could act as a prodrug. Similar amounts of (+)-streptol and SG were detected in the roots of Arabidopsis mature plants grown on 15 ppm P. kirkii extracts (0.44 ± 0.01 and 0.53 ± 0.02 µg ml −1 extracts, respectively); however, only (+)-streptol (0.31 ± 0.01 µg ml −1 ) was detected in inflorescences. To support these . Dataset represents fifteen seedlings per treatment, curves were fitted with the Gompertz model, circles indicate data distribution. Dotted lines delimit the distinct apical root zones. Middle, cortical cell length in the given root apical zone. Significant differences between treatment and water control were analysed in two-sided unpaired t-tests (p-values are indicated). Right, cortical cell length in function of the position from QC initials. Data represent mean values; curves were fitted with the exponential growth model, standard errors are indicated. Significant differences between treatment and untreated control for each position were analysed in two-sided unpaired t-tests (*p-value < 0.05). MZ meristematic zone, TZ transition zone, EZ elongation zone. www.nature.com/scientificreports/ results, we bathed mature Arabidopsis roots in a solution of 250 µM pure SG for 48 h. After careful washing, we detected substantial amounts of (+)-streptol (0.85 ± 0.06 µg ml −1 ) in these roots, but not of SG. Similarly, we found that the etiolated hypocotyls of Arabidopsis grown on medium supplemented with 250 µM SG accumulated 10 times more (+)-streptol than SG (3.98 ± 0.03 and 0.44 ± 0.01 µg ml −1 , respectively). Interestingly, when controlgrown mature Arabidopsis roots were transferred to solid medium supplemented with 100 µM SG, a significant amount of (+)-streptol was found in the gelose medium after 10 days (0.76 ± 0.03 µg ml −1 ). (+)-Streptol was not detected in the medium in the absence of roots. Collectively, these data suggest that (+)-streptol and/or SG can be taken up by the roots and that at least (+)-streptol is systemically transported in planta. It is worth noting that the lack of toxicity on aerial organs is not due to the absence of (+)-streptol in these tissues. Given (+)-streptol structural proximity to α-d-glucose, the translocation might occur through dedicated plasmalemma transporters. In Arabidopsis, the major facilitator superfamily SUGAR TRANSPORT PROTEIN (STP) symporters mediate the uptake of various hexoses 23,24 . We found single and double mutants in the root expressed STP1, 4 and 13 insensitive to restrictive doses of (+)-streptol (Fig. 3). Exogenous supplementation of glucose restored root growth in STP-defective lines and substantially alleviated streptol symptoms (Fig. 3a). Similar results were obtained using 10 µM SG. We concluded that (+)-streptol primarily enters and translocates into the plant body via hexose transporters. Streptol defines the P. kirkii active pharmacophore. (a) Arabidopsis root growth inhibition by synthetic, conjugated and enantiomeric streptol derivatives. Root micrographs illustrate the apical phenotype at the IC 50 or highest tested dose. Bar = 100 µm. Boxplots represent root length relative to water controls. + indicates the mean values of three biological replicates (10 seedlings each); dots indicate outliers. (b) Root length of 10 days old Arabidopsis seedlings grown in the presence 100 µM kirkamide or without supplementation. Significant differences between treatment and water control were analysed in a two-sided unpaired t-test. Values represent 12 seedlings, p-value is indicated. (c) Arabidopsis root diameter in the elongation zone (EZ) after 5 days of growth in increasing concentrations of streptol. Significant differences between treatment and water control were analysed in a one-way ANOVA with Dunnett's post-hoc test (***p-value < 0.001; ns, not significant). Data represent the values of three biological replicates (10 seedlings each). (d) Root apex diameter along the longitudinal axis of 5 days old seedlings grown in increasing concentrations of streptol. Data represent the mean values of 12 seedlings per treatment; standard errors are indicated. Dotted lines delimit the distinct apical root zones. QC apical quiescent centre. www.nature.com/scientificreports/ Streptol specifically disturbs root cell wall organisation without affecting cellulose biosynthesis. Several carbohydrate-related genetic lesions in Arabidopsis cause extensive swelling of the root apex. In particular, mutations and pharmacological interference obstructing cell wall biosynthesis and remodelling trigger phenotypes resembling those observed in (+)-streptol conditions [25][26][27][28] . The primary plant cell wall is a polysaccharidic extracellular matrix synthesized and rearranged to support cell morphogenesis and growth under vacuole-generated turgor pressure 29 . The major carbohydrate polymers of primary walls are cellulose, hemicelluloses and pectins 30 . Cellulose, a linear β-(1,4)-linked glucan chain, is synthesized at the plasma membrane by multi-heteromeric complexes termed Cellulose Synthase Complexes (CSCs) 31 . Primary wall CSCs are composed of at least three distinct CELLULOSE SYNTHASE (CesA) isoforms 32,33 , CESA1 and CESA3 operating as core units, and context-specific CESA6-like proteins complementing the structure [34][35][36] . CSCs are superorganized in hexameric structures known as "rosettes" 31,37 that are guided along cortical microtubules to achieve directional cellulose deposition into the apoplast [38][39][40] . The multiple cellulose chains produced by CSCs arrange into higher order, H bond-driven microfibrils perpendicular to the growth axis, thus defining anisotropic elongation 41,42 . These microfibrils are tethered by hemicelluloses and embedded into a pectic matrix that confers optimal viscoelasticity to the cell wall. These polymers are produced in the Golgi apparatus and released to the apoplast by exocytosis.
In order to understand the mechanism of (+)-streptol-mediated growth inhibition, we assessed the composition of (+)-streptol-treated apical cell walls through an array of histological stains. We first observed higher degrees of dyeing for all histochemicals, suggesting either greater wall polymers contents or better accessibility for the dyes (Supplementary Fig. 7). Under (+)-streptol conditions, Wiesner's reaction, ruthenium red and basic fuchsin colorations indicated increased and ectopic lignin contents in the EZ, a common trait of cellulose-deficient mutants 43 . Next, outer epidermal cellulose microfibrils arrays in the EZ were observed after Direct Red 23 staining 41 under the confocal microscope (Fig. 4). Although (+)-streptol treatments did not radically affect the apparent abundance of microfibrils, the characteristic microfibrils transverse orientation of anisotropically growing cells was disordered, which may explain the isotropic expansion of symptomatic cells. The aberrant punctuate distribution of Direct red 23 signals under (+)-streptol treatment (Fig. 4a) strikingly resembles the fluorescence patterns of pectic rhamnogalacturonan-I obtained in root EZ epidermal cells after metabolic labelling with a fucose alkyne analogue 44 . Fourier transformation of scanning electron micrographs of outer epidermal walls in the EZ confirmed that the net directionality of cellulose bundles patterning is heavily altered in the presence of (+)-streptol, with an overall loosening of the matrix lattice (Fig. 4b). www.nature.com/scientificreports/ The transversal, CesA-mediated, directional deposition of cellulose microfibrils required for root cell elongation is sustained by cortical microtubules (CMT) alignments 45 . In untreated wild-type roots, CMTs display a predominant transversal orientation throughout the apex 46 . However, the random reorientation of CMTs in the EZ has been consistently reported in cellulose-deficient mutants or upon wall biogenesis chemical perturbation 45 . Further, this aberrant CMT behaviour occurs in root-swelling mutants and pharmacological CMT interference triggers swollen roots 47,48 . We thus sought to assess the status of epidermal CMTs orientation by monitoring seedlings expressing pUBQ10:EYFP-TUB6 in (+)-streptol conditions. In line with the absence of obvious defects in the MZ, (+)-streptol-treated cells in this zone displayed the canonical transverse CMT orientation (Supplementary Fig. 8a). Contrary to untreated seedlings, cells transitioning to the EZ in the presence of (+)-streptol exhibited a progressive shift of CMT orientations from transverse to random, resulting in a significantly decreased anisotropy ( Supplementary Fig. 8a). No differences in CMT behaviour in aerial organs epidermal cells were observed between treatment and control seedlings ( Supplementary Fig. 8b).
It is therefore likely that (+)-streptol, directly or indirectly, affects CMT-cell wall organization. However, in contrast to established cellulose biosynthesis inhibitors 49,50 (CBIs), no experimental evidence supports a direct www.nature.com/scientificreports/ inhibition of cellulose microfibrils deposition by the cyclitol. The group I isoxaben 51 , group II 2,6-dichlorobenzonitrile 52 and auxinic quinclorac 53 CBIs did not cause (+)-streptol-like responses in seedlings (Fig. 4c). At permissive concentrations, these inhibitors produced mild to no relaxation of the EZ epidermal matrix, contrasting with (+)-streptol-triggered defects. Higher restrictive CBIs doses completely impeded root development and often caused additional abnormalities to hypocotyls, symptoms that are absent in (+)-streptol-treated cells (Fig. 4c). Checkerboard assays with increasing doses of (+)-streptol and isoxaben resulted in additive effects in the inhibition of root elongation and EZ swelling (Fig. 4d), suggesting that the two drugs act independently. Moreover, well-investigated CSC mutants such as the CBI-resistant procuste 35 (prc1, cesa6), constitutive expression of VSP1 54 (cev1, cesa3), and the anisotropy mutant radially swollen 1 55 (rsw1, cesa1) were found fully sensitive to (+)-streptol, indicating that the cyclitol targets a different component of the cell wall biosynthesis apparatus ( Supplementary Fig. 9a). We also evaluated the crystalline cellulose content in cells from the symptomatic EZ and found no compelling evidence of significant changes between control and treatment ( Supplementary  Fig. 9b). Furthermore, radial swelling associated with cellulose-defective mutants was reported to be triggered or enhanced by high temperature 26,55 ; however, the (+)-streptol phenotype was not enhanced in seedlings transferred to 28 °C ( Supplementary Fig. 10). Finally, Arabidopsis root cell suspensions cultured for 7 days in presence or absence of 10 µM (+)-streptol did not show significant differences in growth rates or cell shape ( Supplementary  Fig. 11). Hence, the cyclitol presumably does not directly target the plasma membrane CesAs but interferes with distinct primary cell wall biogenesis or remodelling processes in a cell type-or tissue-specific manner.
Streptol does not target glycoside hydrolase activity. Glycoside hydrolases (GH) are primarily involved in cell wall degradation 56 and represent obvious candidate targets of (+)-streptol inhibition. However, several lines of evidence argue against this hypothesis. First, neither (+)-streptol, SG, or their isomers showed inhibition of α-or β-glucosidase activities in vitro (not shown). Blockage of such enzymes, especially starch degrading GHs, would rapidly induce starvation and greater tissue disorganization than observed with (+)-streptol 57,58 . Second, mutations in GHs engaged in N-linked glycan processing such as the Arabidopsis alpha-glucosidase II mutant rsw3 28 , the embryo defective alpha-glucosidase I mutant knf-14 59 or the α-mannosidase I mns multiple mutants 60 , can cause strong reduction in cellulose contents and radial root swellings but exhibit morphological changes that are distinct from streptol treatment. Likewise, micromolar doses of the α-mannosidase I inhibitor kifunensine was shown to trigger acute root swelling in Arabidopsis more severe than (+)-streptol effects 60 . By analogy, high micromolar doses of the iminosugars 1-deoxynojirimycin and isofagomine, respectively generic α-and β-glucosidase inhibitors, impede root growth and stimulate radial root expansion but do not alter wall patterning at growth permissive concentrations 57,58 (Supplementary Fig. 12). Also, the glucose analogue conduritol-β-epoxide (CBE), a covalent inhibitor of animal glucocerebrosidase targeting the Arabidopsis glucosylceramidase GCD3 61 , affects root growth without drastically disturbing cell wall organization. Third, Arabidopsis single mutants in major cell wall or cytosolic invertases, which are sucrose-degrading GHs, did not respond differently to SG treatments than wild-type plants ( Supplementary Fig. 13a). Although no particular cell wall defects were associated with invertase single mutants, the isotropic root growth of a cinv1 cinv2 double mutant was attributed to an impaired cellulose microfibrils patterning distinct from the one observed under (+)-streptol conditions 62 . Mutant seedlings of the bidirectional sucrose synthase 63 (SUS1) responded to (+)-streptol and SG similar as the wild-type plant ( Supplementary Fig. 13b), in line with the lack of a cell wall phenotype in the SUS quadruple mutant 64 . In conclusion, our data suggest that the two cyclitols do not act as generic GH inhibitors in planta.

Streptol is not a generic antimetabolite and affects only higher plants.
Recent work demonstrated the inhibition of 3-dehydroquinate (DHQ) synthase by the cyanobacterial pseudosugar 7-deoxy-sedoheptulofuranose (7dSh) 65 . This antimetabolite blocks the shikimate pathway and thereby triggers growth arrest or cell death in a wide range of prototrophs. Arabidopsis development was clearly affected by micromolar doses of 7dSh, although no defects in the root organization were reported 65 . The pentose phosphate pathway intermediate sedoheptulopyranose 7-phosphate is a prominent precursor of cyclitol moieties found in natural products such as acarbose or validamycin A and, interestingly, the EEVS sugar phosphate cyclase is homologous to the DHQ synthase 66 .
In order to evaluate if (+)-streptol features a similar mechanism of action, we first cultured a panel of bacteria in presence of the compounds. Growth kinetics of Gram-negative and Gram-positive organisms were not impacted by 250 µM (+)-streptol in rich or minimal medium (Supplementary Fig. 13a). When grown on solid media, none of the tested bacterial species showed significant changes in colony sizes or extracellular polysaccharides production in the presence of either P. kirkii extracts, SG or (+)-streptol ( Supplementary Fig. 14a). Finally, the establishment of polysaccharide-rich Burkholderia and Pseudomonas biofilms was not disturbed by up to 1 mM (+)-streptol ( Supplementary Fig. 14b). In our experimental setup, the cyclitols did not demonstrate bactericidal nor bacteriostatic properties.
We next challenged the stramenopile Phytophthora infestans with (+)-streptol. The cellulose-based cell wall of P. infestans is sensitive to 2,6-dichlorobenzonitrile applications 67 ; however, hyphal growth, spore formation and germination were not affected by the presence of P. kirkii extracts or micromolar concentrations of (+)-streptol ( Supplementary Fig. 15). Similarly, we observed no significant antifungal activity by treatment with P. kirkii extracts or with SG and (+)-streptol even at millimolar concentrations ( Supplementary Fig. 16). These results argue against streptol-mediated inhibition of a central metabolic pathway in these organisms.
Hatching Artemia salina dehydrated eggs in increasing micromolar concentrations of (+)-streptol did not influence cyst decapsulation nor further nauplii growth and development (Supplementary Fig. 17a). (+)-streptol-treated specimens showed no changes in coloration or visible abnormalities. Likewise, we did not observe www.nature.com/scientificreports/ alterations of human colon carcinoma LS174T cells viability upon (+)-streptol treatment, even at concentrations two orders of magnitude higher than that required for root inhibition ( Supplementary Fig. 17b), thus discarding acute cytotoxicity towards animals. To test whether the inhibitory activity of (+)-streptol is restricted to certain plant lineages, we cultured the single-celled Chlorophyta Chlamydomonas reinhardtii and Gloeotilopsis planctonica as well as the filamentous Charophyta Klebsormidium flaccidum and Zygnema circumcarinatum in K-medium supplemented with 10 µM (+)-streptol. After one week of growth, (+)-streptol-treated cultures were indistinguishable from controls. Closer inspection of cell shape and Calcofluor White (CFW) staining of cell walls did not reveal significant alterations in the (+)-streptol-treated populations ( Supplementary Fig. 18). We also measured individual cells area from micrographs, as a proxy for cell volume, but did not observe significant increase or decrease in cell sizes for the four tested species (Supplementary Fig. 18). We next extended our survey to mosses. Physcomitrella patens, Funaria hygrometrica and Physcomitrium pyriforme grown for one month on solid medium supplemented with 10 µM (+)-streptol appeared identical to controls in colony size and coloration. No changes were observed in shape and size of organs, including rhizoids, and CFW staining did not reveal alterations in cell wall architecture ( Supplementary Fig. 19). Noteworthily, none of the lower plant species tested displayed evident chlorosis, ruling out that streptol blocks central phototrophic pathways.
To assess if streptol requires components of the complex tissue organization of vascular plants for activity, we germinated mature spores of the Polypodiopsida Ctenitis submarginalis on solid medium containing micromolar concentrations of (+)-streptol until the photosynthetic gametophyte stage. No defects in cell division, dimensions or organ patterning were observed and rhizoids grew as much as in untreated samples ( Supplementary  Fig. 20). However, treatment of seeds from four distinct gymnosperms failed at establishing seedlings when grown in the presence of P. kirkii extracts and their roots showed swellings similar to those observed in Arabidopsis. To confirm the restriction of streptol effects to spermatophytes, we expanded the germination assays to various angiosperms (Fig. 5). The seeds from all magnoliids, monocots and eudicots tested responded to streptol treatments with swollen and stunted roots. Within the Gentianales order, the seeds of some Rubiaceae and of Psychotria capensis, which does not belong to the nodulated Psychotria clade, failed at growing in the presence of the inhibitor. Thus far, P. kirkii is the only seed plant found to be insensitive to streptol (Fig. 5). We propose that www.nature.com/scientificreports/ the cyclitol specifically inhibits the biogenesis or remodelling of cell wall constituents during the elongation of root cells. However, a quick survey of the various polysaccharides present in the species included in our assays did not readily allow the identification of a cell wall polymer that would be unambiguously associated with the observed streptol-induced defects (Fig. 5).

Discussion
This report describes the first natural cyclitol that specifically affects root growth of seed plants through cell wall disorganisation. The strong in vivo activity of (+)-streptol against potential plant competitors of nodulated Psychotria is congruent with the concept of allelopathy. Although difficult to demonstrate, the production of streptol by the bacterial symbionts may provide a fitness benefit to the host plant in the competition for space and nutrients. An α-d-glucose analogue disrupting cell wall integrity in growing roots would not only hamper water and nutrient uptake, but also increase the susceptibility to pests 68 . However, the ongoing adaptive role of the symbiosis is challenged by the recent placement of two non-nodulated Psychotria species within the nodulated clade 69 and the pseudogenization of cyclitol biosynthetic genes in several symbionts 10 , indicating that cyclitol production is not the only contribution to the obligate character of the symbiotic association. In a current scenario 9 , climate change events of the Miocene resulted in arid conditions that compelled the rainforest to survive into restricted, hence highly competitive habitats. Remarkably, with the exception of P. eminiana, extant savannah-adapted African Psychotria associate with nodulating Burkholderia symbionts 69,70 . It would be pertinent to monitor the presence of cyclitols in the leaves of non-nodulating symbiotic Rubiaceae, where the bacterial symbionts invade the mesophyll, including several African species of the genera Fadogia, Vangueria, and about 20 Psychotria that do not belong to the nodulated clade 2,71 . The production of allopathic compounds may extend to other plant families and the acquisition of bacteria producing novel secondary metabolites may form a common basis for the establishment of plant-microbe symbioses.
(+)-Streptol causes intriguing root growth inhibition that can be mitigated by exogenous carbohydrate supply. The contrast between the structural CMT and cell wall defects observed in elongating root cells and the intact germination and aerial organs growth in (+)-streptol conditions might indicate a specific alteration of carbon partitioning into the EZ extracellular matrix. The complexity of plant cell wall biogenesis, that involves numerous enzymes, metabolic intermediates and dynamic membrane trafficking is not yet fully resolved 72 . However, the tightly controlled incorporation of monosaccharides into the primary cell wall of actively growing sink organs is central to plant life. Given its analogy to α-d-glucose, it is likely that (+)-streptol interferes with key glucose processing enzymes. A handful of mutations in such genes, metabolically upstream of CSCs or hemicellulosic and pectic synthesis machineries were reported to impact wall organisation to various degrees. The Arabidopsis cinv1 cinv2 double mutant 64 is impaired in sucrose-derived UDP-glucose generation, which is required for cellulose synthesis, resulting in anomalous cellulose patterning and reduced anisotropy 62 . The ugp1 ugp2 double mutant, which is devoid of UDP-glucose pyrophosphorylase activity, displays a drastic reduction in UDP-glucose and severe growth defects that can be rescued by exogenous sucrose supply 73 . Despite lower contents in downstream nucleotide-sugars and down-regulation of discrete CesAs transcription, ugp1 ugp2 was, however, not reported to display prominent cell wall defects. Interestingly, cytosolic phosphoglucomutase pgm2 pgm3 double mutants accumulate glucose-6-phosphate and only display mild alterations of the cell wall matrix, while the starch content increases 74 . Cytosolic UDP-glucose is therefore pivotal to the carbon channelling required for proper wall biosynthesis, as most nucleotide-sugars depend on its multi-step conversion to reach the Golgi synthetic routes to hemicelluloses and pectins. Further investigations of nucleotide sugar metabolism in Arabidopsis will be required to elucidate (+)-streptol specificity. Though the tolerance of P. kirkii towards the cyclitol remains enigmatic, we are currently investigating the genomes of P kirkii and non-nodulating sister species in search for possible (+)-streptol resistance mechanisms.
It is surprising that carbasugars displaying potent and specific activity against carbohydrate-active enzymes were so far rarely exploited in cell wall research. These compounds have several advantages: they are often nonmetabolizable, do not undergo isomerization and show transition state analogy to virtually all natural substrates. We foresee (+)-streptol as a novel in vivo pharmacological tool to dissect plant cell anisotropic expansion beyond the well-characterized mechanisms of cellulose deposition.
The most exciting perspective of the natural product (+)-streptol, however, is its herbicidal potential, which may open novel strategies for sustainable agricultural management.

Methods
Preparation of P. kirkii extracts and leachates. Crude extracts were obtained from 100 g of fresh, 70% EtOH-and 1% hypochlorite surface-cleaned symbiotic P. kirkii leaves ground in liquid N 2 and extracted with 1 l methanol:H 2 O (80:20) for 24 h in the dark. Solvent extract was filtered, mixed with 1 volume of chloroform and the methanol fraction was collected with a separatory funnel and lyophilized using a rotary evaporator under vacuum. The resulting 3.5-4 g viscous residual was dissolved in 12 ml pure H 2 O. Water extracts were obtained by reducing washed P. kirkii tissues (roots, 10 g; shoots, fruit flesh and flowers, 2 g) to powder in liquid N 2 , macerated 2 h in pure H 2 O and centrifuged. Resulting supernatants were filtered with 0.22 µm polyethersulfone filters (Millipore). P. kirkii leachates were obtained by bathing 100 g fresh, surface-cleaned leaves into 1 l sterile tap water for 48 h at room temperature. Leachates were filter-sterilized prior to in vitro use.
Quantification of streptol and streptol glucoside by UHPLC-MS. The concentration of streptol and streptol glucoside was quantified using two different analytical methods using a UHPLC system (Ultimate 3000 LC, Thermo Fisher Scientific) coupled to a mass spectrometer (TSQ Quantum Ultra, Thermo Fisher Scientific). The first method consisted on analysing the samples using a Rezex RPM-monosaccharide Pb + (300 × 7.  Table 1. Non-commercial seeds were kindly provided by Peter Enz (University of Zurich's Botanical Garden) with relevant permissions or ordered with relevant permissions from the SAG (Goettingen University, Germany) and the IMSC (University of Freiburg, Germany). All the experimental work on plant material described in this study complies with the relevant institutional, national, and international guidelines and legislation.

Plants culture conditions and growth inhibition assays. Unless otherwise stated, all plant species
were grown in a Conviron growth cabinet with 16/8 h, 20/18 °C day/night cycles, 60% relative air humidity and a photon flux of 100 μmol m −2 s −1 during light periods.
Arabidopsis assays. Arabidopsis seeds were surface-sterilized and kept for imbibition in the dark at 4 °C for at least 24 h. Long term growth experiments were carried out with seeds vertically grown on half strength Murashige and Skoog (½MS, pH 5.7) medium 1.5% agar with or without 1% sucrose supplement. For timelapse series, square plates were mounted on a vertical Epson Perfection 1240U flatbed scanner adapted with a 100 μmol m −2 s −1 white LED light source (16/8 h day/night cycles) and growth was recorded every 4 h using the VueScan 9.4.32 software (Hamrick Software, USA; https:// www. hamri ck. com). Short-term experiments were carried out in 96-well microtiter plates in sterile tap water. The pure inhibitors were supplemented at given concentrations. Adult Arabidopsis plants were grown in the same conditions in Magenta boxes onto 0.8% agar ½MS and supplemented with P. kirkii extracts at given concentrations. Arabidopsis root cells suspension cultures were obtained from 1 g finely-chopped, mature, axenically grown roots in ½MS 3% sucrose supplemented with 0.5 mg l −1 benzylaminopurine, 1 mg l −1 napthalene acetic acid, 1 mg l −1 indole acetic acid, 1 mg l −1 2,4-dichlorophenoxyacetic acid and maintained by subculturing at 4-week intervals. Fresh subcultures were grown for 2 weeks prior to 10 µM streptol treatment in the exponential growth phase. Growth and fitness of cells were frequently evaluated under a Leica DM600B dissecting microscope (Leica, Germany). Quantitative analysis of Arabidopsis root gravitropism was performed as a proxy for evaluating auxin distribution disturbances. Surfacesterilized seeds were grown vertically on ½MS 1% sucrose 1% agar for 5 days and were next vertically aligned www.nature.com/scientificreports/ onto fresh plates containing increasing doses of P. kirkii extracts and allowed further growth in the light for 2 h. Gravistimulation was achieved by tilting the plates to 90° and letting the seedlings grow for 24 h in the dark. Flatbed scanner acquisitions prior to and after stimulation were used to determine root growth and tip reorientation in the gravity field.
Allelopathy assays. Surface-sterilised lettuce seeds (Lactuca sativa var. acephala) and black mustard seeds (Brassica nigra) were treated after imbibition on 55 mm Petri dishes lined with sterile filter paper (Whatman, diameter 50 mm) soaked with 1 ml symbiotic P. kirkii crude extracts or filtered water extracts from roots, shoots, flower and seed flesh or P. kirkii leaf leachate and incubated in the dark at room temperature for 6 days. Using the same setting, lettuce seeds were grown in the vicinity of intact, surface-sterilized mature P. kirkii drupes or pyrenes. Control treatments were performed with sterile tap water. Seeds with radicles as long as seed diameter were considered germinated. For soil assays, black mustard seeds were sown after imbibition on individual peat-soil plastic pots (15 seeds each) and cultured in a growth cabinet as described for 10-15 days. Treatments consisted of either watering the pots with P. kirkii leachates, crude extract dilutions (0.1% v/v) or germinating the seeds on soil amended with mulched P. kirkii leaves (100 g leaves per litre of soil); tap water was used for controls. Pots were watered to their water holding capacity once per week. Successfully established seedlings were counted 7 days after sowing.
Other plant species. Algae were axenically cultured and maintained on K medium as indicated by the Culture Collection of Algae at the university of Göttingen, Germany (http:// sagdb. uni-goett ingen. de/ cultu re_ media/). Inhibitors treatments were performed in 2 ml liquid K medium in 6-well plates supplemented with 100 μl of P. kirkii crude extract or 10 µM streptol and incubated at 22  Histology and microscopy. Unless specified otherwise, samples were observed under a Leica DM5500Q confocal microscope fitted with a TCS SPE confocal unit (Leica, Germany), an ACSAPO 40× oil-immersion objective (NA = 1.15, Leica) or an ACSAPO 10× dry objective (NA = 0.3, Leica), and laser lines set at 405, 488, and 532 nm. The images were acquired using the LAS software (Leica, Germany; https:// www. leica-micro syste ms. com).
In order to measure root cells dimensions, live Arabidopsis seedlings were stained for 2 min in 1 mg ml −1 propidium iodide, rinsed twice with tap water and optical sections of primary root apices were acquired with the confocal microscope. Measurements were performed with FiJi. Root, hypocotyl and inflorescence transversal semi-thin sections were obtained from 3% agarose-embedded samples. Deeper root tissue investigations were carried out using the modified pseudo-Schiff propidium iodide (mPS-PI) method. In short, whole seedlings were fixed in 50% methanol, 10% acetic acid at 4 °C for at least 12 h. Samples were transferred to 80 °C 80% ethanol for 5 min, transferred back to fixative for 1 h; rinsed with ddH 2 O and incubated in 1% periodic acid at room temperature for 40 min. Samples were rinsed with ddH2O and incubated in 100 mM Na 2 S 2 O 5 , 0.15 N HCl with 1 mg ml −1 propidium iodide for 2 h. Samples were transferred onto microscope slides and covered overnight with a chloral hydrate solution (4 g chloral hydrate, 1 ml glycerol, and 2 ml H 2 O). Excess chloral hydrate was removed and samples mounted in Hoyer's solution (30 g gum arabic, 200 g chloral hydrate, 20 g glycerol, and 50 ml H 2 O). Images were acquired with under the confocal microscope.
Other cell wall-related histochemical staining procedures were essentially performed as described 76 . For epidermal cellulose microfibrils visualisation, whole seedlings were stained with 1 mg ml −1 Direct Red 23 in sterile tap water for 30 min, washed thrice and mounted in water. Image acquisition was performed under the confocal microscope. www.nature.com/scientificreports/ To qualitatively assess cytokinesis, nuclear staining and cell wall counterstaining of 5-day old Arabidopsis seedlings grown in presence or absence of 5 μM streptol was achieved with either 2.5 μM SYTO 9 (Invitrogen, USA) and 5 μM propidium iodide or with 10 μg ml −1 DAPI (0.1% Triton-X100, Thermo Fisher Scientific, USA) and 1 mg ml −1 Direct Red 23 for 20 min in sterile tap water. The presence of polyploid cells or cell plate stubs was monitored under the confocal microscope (20 plants per treatment).
In order to visualize and quantify cortical microtubules orientations, pUBQ10::EYFP-TUB6 signals from 5-day old seedlings grown in presence or absence of 5 μM streptol were acquired in 0.5-1 μm step z-stacks under a the confocal microscope throughout outer epidermal surfaces. Maximum z-projections were analysed with FibriTool 77 .
DR5::GFP signals in Arabidopsis were imaged under the confocal microscope after transferring 5 days old seedlings onto the given inhibitor concentrations for 24 h. Histochemical GUS staining was performed on whole pCYCB1;1::DB-GUS seedlings essentially as described 78 , fixed in 96% ethanol and mounted in chloral hydrate.
Electron microscopy. Whole seedlings were chemically fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, rinsed with pure water, excess liquid removed, plunged-frozen in liquid propane mounted on a 3.05 mm diameter 400 mesh copper grid and stored in liquid N 2 . Samples then were mounted on a grid holder and freezedried at − 95 °C for 1 h at 10 -6 mbar vacuum. Samples were quickly transferred to a Zeiss Auriga 40 CrossBeam focused ion beam-scanning electron microscope (Zeiss, Germany) to check preparation quality. Samples were then further dried at − 95 °C for 1 h, then coated with 2.5 nm of Pt/C moving the gun between 0° and 45° and rotating the stage at 40 rpm plus 2.5 nm Pt/C double axis rotary shadowing. Samples then were again transferred to the SEM using a Leica VCT100 cryo transfer system (Leica, Germany) and imaged at − 112 °C, at 5 kV; Inlens, 1.4 and 5 mm WD. Preferred orientation of root cell wall structures from SEM acquisitions were inferred by using the FiJi Directionality analysis v2.3.0 plugin applying the Fourier components analysis method and a 90 bins partition.

Microbial culture conditions and growth inhibition assays. Bacteria. Escherichia coli ATCC25922,
Burkholderia cenocepacia H111, Bacillus subtilis, Staphylococcus epidermidis RP62a and Pseudomonas aeruginosa PAO1 were cultured in lysogeny broth (LB) at 30 or 37 °C until reaching the exponential growth phase and cell density adjusted to OD 600nm = 0.1 in sterile saline for downward experiments. Growth curves in the presence of inhibitors were obtained in a Sirius HT 96-well microtiter plate reader (BioTek Instruments, Switzerland) at 37 °C with constant shaking in a final volume of 100 µl. Mucoid exopolysaccharide production was assessed by spotting bacterial suspensions onto yeast extract medium (0.5 g l −1 yeast extract, 4 g l −1 mannitol, 15 g l −1 agar) gradient plates with the given inhibitor concentration as the highest dose. Biofilm biomass was evaluated in the presence of inhibitors by growing 100 μl of cell cultures to the late log phase in ABC medium in sealed 96well polystyrene microtitre plates for 72 h at 28 °C; Abs 550nm was measured. Growth medium was removed and 100 μl of a 1% (w/v) aqueous solution of crystal violet added at room temperature for 30 min; excess of dye was removed and wells thoroughly washed and dried. Crystal violet was solubilized in 120 μl DMSO and Abs 570nm measured. The Biofilm Index (BI) was normalized with respect to growth as BI = Abs 570nm /Abs 550nm .
Fungi and oomycetes. Agar plugs of actively growing Fusarium solanii, Rhizoctonia solanii and Botrytis cinerea were transferred to the borders of fresh malt extract 2% agar plates. Sterile paper disks were impregnated with the inhibitors at given concentrations and placed in the middle of the plates and incubated in the dark at 22 °C for 48 h. Mycelial growth was monitored. Approximatively 10 5 spores per ml harvested from mature Fusarium solanii and Botrytis cinerea as well as hyphal cuttings from Rhizoctonia solanii were placed in 6-well plates with sterile tap water supplemented with the given inhibitors and incubated for 24 h in the dark at 22 °C with gentle shaking (110 rpm). Germination of spores and regeneration of cut hyphae was evaluated under a Leica DM600B epifluorescence microscope after 100 μg ml −1 Calcofluor White staining. Phytophthora infestans Rec01 was maintained on V8 medium in our laboratory. Agar plugs of actively growing P. infestans were transferred to the middle of fresh rye 1.5% agar plates and 8 mm wells produced with a cork borer were filled with P. kirkii crude extracts or sterile water as solvent controls. Mycelial growth was monitored. Mature sporangia were harvested with sterile H 2 O from 10 days mycelia, filtered through sterile cheese cloth and inhibitors were supplemented at the given concentrations. Sporangia germination was stimulated by vigorous vortexing and let to germinate for in the dark at 22 °C for 16 h. Spore germination was assessed under a Leica M165FC stereo microscope and germ tube length was measured from Leica DM600B dissecting microscope brightfield micrographs using ImageJ software.
Cytotoxicity of (+)-streptol on animal cell line. Brine shrimps. Artemia salina were hatched from dehydrated eggs (Artemio Pur, JBL GmbH Germany) under constant shaking and 60 µE white light illumination at room temperature in artificial sea water (NaCl 37 g/l, pH 8.5 NaOH) supplemented with the given concentrations of inhibitors. After 72 h, the different stages of development of individuals were assessed under a Leica M165FC stereo microscope. No differences in nauplii swimming motility were observed between treatments.
Human cells. Human colon Duke's type B adenocarcinoma LS 174 T cell line was purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (ACC 759) and the cells were grown in Dulbecco's Modified Eagle Medium (DMEM) with 10% heat-inactivated Fetal Bovine Serum (FBS), 0.1% penicillin-streptomycin in an incubator (37 °C, 5% CO 2 , 95% humidity). LS 174 T cells (4 × 10 3 ) were seeded in 96-well cell culture plates in a final volume of 100 µl and allowed to attach for 24 h in the incubator. To evaluate the cytotoxic potential of (+)-streptol, LS 174 T cells were treated with increasing concentrations of the compound in DMEM medium www.nature.com/scientificreports/ with 10% heat-inactivated FBS, 0.1% penicillin-streptomycin for 5 days in the same growth conditions. Cytotoxicity was determined by replacing the medium with 100 µl of DMEM containing 10% colorimetric Cell Counting Kit-8 (Dojindo Molecular Technologies, Inc.), incubating the cells for another 90 min, and determining dehydrogenase activities by measuring the absorbance at 450 nm following the manufacturer's instructions. Absorbance at 650 nm was also measured and the value was subtracted to samples' values as the background. Cells in medium were used as negative control and wells without cells were set as blank.
All photographs displayed in this work were acquired from a Nikon D90 digital camera equipped with a NIKKOR AF-S Micro 60 mm f/2.8G ED macroobjective.

Data availability
The data supporting the findings reported in this work are available within the manuscript and the associated Supplementary Information files and from the corresponding authors upon reasonable request. www.nature.com/scientificreports/