A sulfur-containing volatile emitted by potato-associated bacteria confers protection against late blight through direct anti-oomycete activity

Plant diseases are a major cause for yield losses and new strategies to control them without harming the environment are urgently needed. Plant-associated bacteria contribute to their host’s health in diverse ways, among which the emission of disease-inhibiting volatile organic compounds (VOCs). We have previously reported that VOCs emitted by potato-associated bacteria caused strong in vitro growth inhibition of the late blight causing agent Phytophthora infestans. This work focuses on sulfur-containing VOCs (sVOCs) and demonstrates the high in planta protective potential of S-methyl methane thiosulfonate (MMTS), which fully prevented late blight disease in potato leaves and plantlets without phytotoxic effects, in contrast to other sVOCs. Short exposure times were sufficient to protect plants against infection. We further showed that MMTS’s protective activity was not mediated by the plant immune system but lied in its anti-oomycete activity. Using quantitative proteomics, we determined that different sVOCs caused specific proteome changes in P. infestans, indicating perturbations in sulfur metabolism, protein translation and redox balance. This work brings new perspectives for plant protection against the devastating Irish Famine pathogen, while opening new research avenues on the role of sVOCs in the interaction between plants and their microbiome.


Results and Discussion
Sulfur-containing volatile organic compounds constrain late blight in potato leaf discs. Following an initial screen for P. infestans-inhibiting VOCs that revealed the high in vitro activity of sulfur-containing volatiles (sVOCs) 19 , we explored the capacity of three sVOCs, DMDS, DMTS and MMTS (see Fig. S1 for the chemical structures of these sVOCs) to inhibit late blight in planta using leaf disc assays. Airborne exposure to 1 mg of DMTS or MMTS in the Petri dish atmosphere (80 mL) led to full protection against P. infestans, while DMDS was by far less active (Fig. 1a). Binocular observation confirmed that MMTS and DMTS totally prevented the development of P. infestans at the leaf surface (Fig. 1b). Nevertheless, we could not exclude at this stage that internal leaf tissues might be colonized by the pathogen. We therefore used a fatty acid methyl esters (FAMEs) analysis to quantify the oomycete in plant tissues. P. infestans produces specific fatty acids, such as the eicosapentaenoic acid (EPA; C20:5) 27,28 that may serve as molecular markers to quantify the oomycete biomass in plant tissues, as previously demonstrated for P. sojae or Plasmopara viticola 29,30 . FAME analysis of inoculated leaf discs revealed several fatty acids that were specifically detected in heavily infested samples (Fig. S2a). A major peak confirmed by GC-MS analysis as C20:5 ( Fig. S2b) was used to quantify the pathogen in the different treatments. Our results showed that MMTS and DMTS totally prevented the proliferation of P. infestans in potato leaf discs, while DMDS only partially prevented it (Fig. 1c).
We also examined the phenotype of the sVOC-treated leaf discs without pathogen. Apart from natural colour variation possibly due to differing anthocyanin contents, the DMDS-and especially DMTS-treated leaf discs MMtS inhibits late blight development in potato plantlets. Next, we tested the protection efficiency of sVOCs in whole plants, using in vitro potato plantlets. Applying P. infestans zoospores on one leaf of potato plantlets led to successful infection, evidenced by wilting and by a white mat of hyphae and sporangiophores on the stem and on all leaves (Fig. 4). By contrast, plantlets treated with 10 to 100 μg MMTS in the tube atmosphere   (40 mL) showed no late blight symptom, which also resulted in higher biomass than the non-treated, infected controls (Fig. 4b). Here too, MMTS had no phytotoxic effect and leaves showed a normal phenotype under binocular and microscopic examination. In contrast to MMTS, 30 μg DMDS and DMTS induced only limited disease protection ( Fig. 4 and Fig. S5). At higher doses, DMTS was highly phytotoxic, inducing arrested growth and bleaching (Fig. S6), which confirmed earlier observations on leaf discs. The structurally related DMDS did not induce visible toxicity symptoms and both sulfides induced slight -but non-significant -plant growth promotion at 10 μg (Fig. S6b). Finally all sVOCs applied at 1 mg/tube induced strong phytotoxicity, indicating that a proper adjustment of MMTS dosage is required to balance plant protection vs. plant fitness (data not shown). Under these experimental conditions, the minimal active dose of MMTS was 1.75 mg.L −1 air. However, since the glass tubes represented a high humidity environment particularly conducive to late blight, we investigated whether lower doses would be sufficient in a lesser artificial setup. Indeed, when in vitro plantlets were transferred to pots and incubated in plastic boxes, 1 mg MMTS was sufficient to fully inhibit disease symptoms and did not induce any phytotoxicity (data not shown). This corresponds to a dose of 0.24 mg.L −1 air.
In view of the emergence of fungicide resistance in P. infestans strains, it is urgent to find new solutions to control this versatile pathogen 2 . Here, we show that the sVOC MMTS diffuses through the air and inhibits infection at doses that are not toxic to plants. Importantly, short exposure to MMTS can stop late blight development and this compound even shows protective potential when applied after inoculation of the pathogen. In order to better understand the mode of action of this efficient late blight inhibitor, we next investigated whether the protection originated form direct anti-oomycete activity or from an induction of plant defences. protection against pathogens conferred by MMtS is independent from plant defences. Many beneficial plant-associated microbes, including Pseudomonas, protect plants against pathogens by triggering ISR 11 . Microbe-associated molecular patterns (MAMPs) are recognized by plants and induce plant defences. Among other bacterial determinants, volatiles (e.g. 2,3-butanediol) were shown to induce the expression of defence genes in Arabidopsis 17,32 . Giving the strong protective effect of MMTS against late blight, we assessed whether this volatile elicited the plant defence responses. First, MMTS was applied two days before the pathogen to allow putative induction of defences. This preventive treatment did not lead to a lesser infection, suggesting that MMTS did not induce plant defences in this experimental setup (Fig. S7). Next, we investigated whether sVOCs induced the accumulation of reactive oxygen species (ROS) in plant tissues. MAMP perception by plant cells induces an "oxidative burst", i.e. a rapid and transient accumulation of ROS 33 . We applied a luminol-based chemiluminescent assay to detect ROS production and used the synthetic peptide flg22 (from flagellin) as positive control of MAMP response 34 . Preliminary assays with potato leaf discs showed a large inter-replicate variation and we therefore performed this analysis on Arabidopsis. Leaf discs treated with 1 μg MMTS, DMTS and DMDS  exhibited no detectable increase in luminescent signal, such as the one observed with flg22 ( Fig. 5a). We concluded that these sVOCs did not trigger the typical MAMP-induced oxidative burst. Interestingly, the samples treated with DMTS and MMTS showed significant luminescence reduction after flg22 treatment (Fig. 5a). This effect was dose-dependent and specific for these two sVOCs, as DMDS and another sulfur volatile, bis(methylthiomethyl) sulfide (BMTMS) had no significant effect on the flg22-induced oxidative burst (Fig. S8). As sVOCs might easily oxidize 35 , we suspected that they compromised the chemical reaction of luminol oxidation used to detect ROS production. In assays where the various sVOCs were supplemented few minutes after elicitation with flg22, DMTS and MMTS (but not DMDS) quickly decreased the luminescent signal (Fig. S9). The question if sVOCs directly affected the oxidation reaction remains open, as we cannot exclude that these compounds might exhibit some toxicity to plant cells when applied directly into solution. Nevertheless, this finding is interesting as previous studies have proposed that sulfane sulfur (sulfur atoms that are bonded covalently in chains to other sulfur atoms) volatiles might carry antioxidant properties 36 . Our current data indicate that the sulfane sulfur-carrying DMTS and MMTS might interfere directly with the ROS produced by plant cells upon flagellin perception.
To further investigate the potential of these two protective sVOCs to activate plant immunity, transcript levels of defence-related genes were analysed in potato leaf discs exposed or not to sVOCs. Genes encoding Ethylene Response Factor 3 (ERF3), linoleate 9S-lipoxygenase 2-like (LOX), Pathogenesis-related protein 1b precursor (PR1-b), and Thaumatin-like protein (PR-5) were selected as defence markers as previously described in potato 37 . When applied without P. infestans, MMTS and DMTS did not induce significant changes in transcript levels compared with the controls. While gene expression increased significantly upon infection with P. infestans, it was similar in control and volatile-exposed leaf discs (Fig. 5b), showing that MMTS and DMTS did not affect the expression of defence-related genes in potato. Altogether, our data point to the conclusion that the protection conferred by MMTS against last blight is not mediated by the typical plant defence pathways, but rather by a direct anti-oomycete activity.
Global changes in the phytophthora proteome after sVoc-treatment. A quantitative proteomic approach was used to get insights into the biological pathways affected by MMTS and other sVOCs on P. infestans. In this experiment, we compared the proteome changes induced by 24 h exposure to 300 μg of each of five individual sVOCs detected in the volatile blends of potato-associated Pseudomonas and differing in their anti-oomycete activity 19 : MMTS, DMTS, DMDS, bis(methylthiomethyl) sulfide (BMTMS), and S-methyl butanethioate (SM). DMTS and MMTS led to strong inhibition of P. infestans mycelial growth, while the effect was less strong for BMTMS and only marginal for DMDS and SM (Fig. S10) 19 . Stringent quality filters were applied and only proteins identified in at least two of three biological replicates were considered. We detected 3348 P. infestans unique proteins, corresponding to 19% of the total proteome (Supplementary Table 1). Label-free quantification allowed to semi-quantitatively assess their expression and identify "regulated proteins", i.e. those detected in lower or higher amounts in the treatment vs. control samples, with two-fold change used as threshold (Supplementary Table 1). Similar proportions of proteins were found regulated by DMDS and DMTS (around 3.3%) on the one hand, and by BMTMS, SM and MMTS (4.5 to 5.4%) on the other hand. A striking observation was that 80% of the MMTS-regulated proteins were downregulated or undetectable, which likely reflects the strong anti-oomycete activity of this volatile. This massive downregulation contrasts with the effect of DMDS, which mainly induced upregulation of proteins. We observed strong specificity in the proteome changes caused by exposure to the individual sVOCs, with only few proteins commonly regulated by DMDS, DMTS and MMTS ( Fig. 6) or by the 5 sVOCs (data not shown). This specificity is consistent with their differential activities on hyphae and spores 19 . A larger proportion of shared upregulated proteins were found between DMTS and DMDS (n = 23) than between MMTS and DMDS (n = 3) or between MMTS and DMTS (n = 4) (Fig. 6). By contrast, the overlap of downregulated proteins was low between DMDS and DMTS, but higher between DMTS and MMTS, which again might relate to their stronger anti-oomycete activity.
The KOG (euKaryotic Orthologous Groups) database was used for functional classification of all identified proteins. Proteins were grouped into 26 categories according to their putative functional classes (Supplementary Table 1). Between 30% (BMTMS) and 42% (DMDS) of all identified regulated proteins had no functional assignment, which might be linked to the poor functional annotation of the P. infestans proteome. Most of the others had putative functions associated with i) intracellular trafficking, secretion, and vesicular transport (predominant class for all sVOCs except DMDS), i) post-translational modification, protein turnover, chaperones (predominant for all sVOCs except DMTS), iii) signal transduction (predominant for DMDS, DMTS, MMTS), and iv) transcription. Voronoi treemaps illustrating global changes in protein expression patterns upon exposure to each of the five sVOCs are shown as supplementary data. In general, no dramatic increase/decrease in proteins enabled to point to specific biological processes (max induction fold: 5; max reduction fold: 0.07), although differences were statistically significant (Supplementary Table 1). These results might indicate that sVOCs have multiple targets in P. infestans, as suggested for the sulfur sulfane structure shared by DMDS, DMTS and MMTS 35 . Nevertheless, some functional classes were more specifically affected by individual sVOCs, such as "amino-acid metabolism" for DMTS and "translation, ribosomal structure and biogenesis" for BMTMS. The biological relevance of the changes detected after BMTS, DMTS and MMTS exposure is discussed below.

BMTS treatment affects the abundance of proteins involved in ribosome biogenesis. The most
obvious specific effect of individual sVOCs on the P. infestans proteome occurred upon exposure to the moderately active BMTMS 19 (Fig. S10). Manual inspection of BMTMS-regulated proteins and Gene Ontology enrichment analysis revealed a group of proteins related to ribosomes (GO:0003735 Structural constituent of ribosome; GO:00006412 Translation) ( Table 1). The ribosome of P. infestans is composed of two subunits (40S and 60S) 38 . Five 60S proteins and two 40S proteins showed decreased abundance upon exposure to BMTMS, but not to other sVOCs. Each subunit comprises proteins associated to ribosomal RNAs (rRNA). Ribosomal RNA are modified by pseudouridylation, which is thought to regulate the stability and translational function of the ribonucleoprotein complex 39 . BMTMS affected the quantity of three important members of the H/ACA ribonucleoprotein complex, involved in the pseudouridylation of rRNA. The link between these and the regulated 40S/60S proteins remains unclear, but we could assume that if the ribonucleoprotein complex is less stable due to impaired rRNA pseudouridylation 39 , other components might be incorrectly stabilized and thus degraded by the cell. Altogether, our data suggest that BMTMS inhibits protein translation. This finding is of particular interest in view of the higher translation activity observed at particular, infection-relevant stages of the oomycete cycle: germinating cysts of P. sojae and P. ramorum exhibited a strong increase in proteins involved in ribosome structure, biogenesis and translation 40 and similar observations were reported for the fish pathogen Saprognelia parasitica 41 . This increased translation activity may also represent the requirement to build the necessary machinery for host invasion, e.g. appressoria or effectors. In this respect, identifying a natural compound interfering with such processes might open promising research avenues. Although BMTMS had only modest effects on P. infestans development in vitro 19 (Fig. S10), it might be interesting to re-analyse its activity on the ability of zoospores to encyst on plant tissue. If verified, such partial protective effect might be valuable enough in combination with other compounds of differing modes of action, e.g. acting on mycelial growth or sporulation. www.nature.com/scientificreports www.nature.com/scientificreports/ Proteins of the sulfur metabolism are differentially regulated upon DMTS and MMTS treatments. Sulfur volatile compounds are highly reactive chemical species, which have been shown to affect many organisms such as bacteria 42 , plants and fungi [43][44][45] . In terms of modes of action of these sVOCs, plants were shown to take up DMDS emitted by bacteria and use it as sulfur source 43 . Whether fungi or oomycetes are also capable of integrating sVOCs into their sulfur metabolism, or to which extent this metabolism would be affected by exposure to different sVOCs, is however so far unknown. To investigate this question, we first drafted a scheme of two main sulfur metabolism pathways 46 in P. infestans, i.e. sulfate reduction and synthesis of cysteine/methionine. Since these pathways have not yet been studied in P. infestans, this scheme was based on the KEGG and Uniprot databases, and we used our proteomic data to complement the information.
Homologues of genes encoding most enzymes involved in sulfate reduction in fungi 47 and plants 46 (ATP sulfurylase, APS kinase and PAPS reductase) were identified in the P. infestans genome (Fig. 7), although some differences might exist in oomycetes 27 . The capacity of P. infestans to reduce plant sulfate is crucial for infection, because the low methionine and cysteine levels in the apoplast are likely insufficient to sustain oomycete growth. Interestingly, two central sulfur reduction enzymes were less abundant upon MMTS treatment (Fig. 7): the PAPS reductase (D0N1L8), which reduces sulfate into sulfite; and the 3′(2′),5′-bisphosphatenucleotidase (D0N678), which -in plants -detoxifies 5′-phosphoadenosine 3′-phosphate (PAP) produced during sulfation of compounds by the PAPS reductase 48 . Moreover, DMTS reduced the amount of the sulfite reductase β subunit, involved in the final reduction of sulfite into sulfide (Fig. 7). The fold reduction of 0.56 was not statistically significant (Supplementary Table 1) but the decrease is worth mentioning. These first observations indicated that sulfur reduction, the first step in sulfur acquisition, might be affected in P. infestans mycelium exposed to MMTS and DMTS.
After reduction, sulfide is normally integrated into amino acids to form methionine and cysteine (Fig. 8). We identified several candidates controlling the synthesis of cysteine/homocysteine/methionine, which were differentially regulated by DMTS and MMTS (Supplementary Table 1; recapitulated in Fig. 8 and Table 2). First, DMTS led to higher abundance of enzymes involved in the synthesis of both cysteine (D0NAM5, 2.5x) and methionine (D0NMR7, 3.9x). Both amino acids are not only protein constituents, but also precursors of important sulfur-compounds (cysteine), such as glutathione, coenzyme A or Fe-S clusters, or methyl group donors (methionine) such as S-adenosyl methionine (SAM) used for DNA or histone methylation. DMTS led to lower levels of a cysteine desulfurase (D0NRJ8), which synthetizes cofactors involved in respiration such as Fe-S clusters 49 or the molybdenum cofactor essential for the sulfite oxidase activity 50 . Desulfurases are also part of the sulfur relay system governing the thiolation of tRNA 51 . Thus, the lower content in cysteine desulfurase caused by DMTS might inhibit several biological processes essential for cell survival. In addition, upon treatment with DMTS or DMDS, we could detect a protein (D0NSN9) with homology to a L-cysteine desulfhydrase from Arabidopsis lyrata (XP_020879589.1; BlastP: 27% identity, 45% positive, with 81% coverage), which was absent in control samples. Cysteine desulfhydrase regulates the homeostasis of cysteine, which can be toxic at high doses in plants 52,53 and algae 54 . This reaction generates H 2 S, which is also a by-product of both DMTS-upregulated enzymes mentioned above (cysteine desulfhydrase, cystathionine γ lyase) 55 . Therefore, it is tempting to speculate that DMTS exposure leads to accumulation of H 2 S in P. infestans. While the effect of H 2 S on oomycetes has not yet been reported, high doses of this compound are known to be toxic to plants and might also interfere with the oomycete biology. Further supporting a role of H 2 S in the sVOC-induced P. infestans inhibition, a protein (D0P2H7) with partial homology to mercaptopyruvate sulfurtransferase (MST) was detected in all sVOC-exposed samples but not in controls ( Table 2). MST is involved in persulfidation of proteins, glutathione and cysteine in animal cells 56 , and regulates the thiolation of tRNA together with cysteine sulfhydrase. During these reactions, MST also produces www.nature.com/scientificreports www.nature.com/scientificreports/ H 2 S and other polysulfides. Moreover, another sulfide extracted from garlic, diallyl trisulfide (DATS), was also reported to induce the activity of MST in animal cells 57 . DATS is anti-carcinogenic and this recent study indicated that it might act as sulfur donor for the persulfidation of the Bcl2 protein by MST, which correlated with the inhibition of cell proliferation regulated by Bcl2 57 . Similarly, bacterial sVOCs might provide the sulfur for the oomycete MST, inducing its stability and therefore its higher abundance in sVOC-treated mycelium. The consequences of increased levels of MST on cell viability are unclear, since MST was more abundant in all sVOC-treated P. infestans independently of their inhibitory activity. While higher MST abundance is therefore unlikely responsible for the toxicity of DMTS and MMTS, it could be a general target of sVOCs and its effect might be additive to other defects caused by the individual sVOCs, e.g. H 2 S production upon DMTS treatment. A last interesting observation from the DMTS-induced changes in P. infestans proteome was the increased methionine R-sulfoxide reductase B (MsrB; D0MUU0), an enzyme involved in reverting the ROS-induced oxidation of thiomethyl groups on protein surface methionine 58 , thereby contributing to protein maintenance and cell survival. This observation suggests that DMTS treatment induced an oxidative stress in P. infestans, which is corroborated by the regulation of several other oxidative stress markers (Table 3), as discussed below.
In contrast to DMTS, MMTS reduced the abundance of proteins involved in the cysteine/methionine metabolism (Fig. 8). Although this might be due to the high toxicity of MMTS leading to general protein downregulation, it is worth noting that two enzymes involved in the production of homocysteine, the aspartate kinase (D0NKQ0) and the homoserine O-acetyltransferase (D0P4C9) were not detectable in the MMTS-treated samples (Fig. 8, Table 2). Likewise, the methionine sulfoxide reductase (D0MUU0), which was more abundant in the DMTS-treated samples, could not be detected either, nor could another isoform of cysteine desulfurase (D0NLG7). Thus MMTS led to decreased abundance of important sulfur metabolism enzymes in P. infestans.
Altogether, our data suggest that exposure to DMTS and MMTS (but not to the three other, less active sVOCs) creates an imbalance in the sulfur metabolism of P. infestans, which might partially explain the strong anti-oomycete activity of these two sVOCs. Interestingly, two fungicides (pyrimethanil and cyprodinil) were proposed to target the methionine synthesis pathway in fungi 59 (although this was recently revisited for the effect of pyrimethanil on Botrytis 60 ). Moreover, methionine synthesis was recently shown to be essential for virulence in the rice blast fungus Magnaporthe grisea 61 . In P. infestans, the methionine synthase transcripts accumulated in the appressorium during infection and methionine concentration was shown to vary during cyst germination and appressorium formation 62 . Similarly, a proteomic approach revealed that in vitro germinated cysts and appressoria of P. infestans displayed a higher content in two isoforms of the methionine synthase, when compared to the mycelium 63 , suggesting that methionine is important for oomycete virulence. There is less information on the role of cysteine in the biology of P. infestans. However, many cysteine-rich proteins are encoded in the P. infestans www.nature.com/scientificreports www.nature.com/scientificreports/ proteome. These proteins, which are likely secreted, have been proposed to be virulence factors or toxins [64][65][66] . For example, the cysteine-rich protein SCR96 was shown to determine pathogen virulence and oxidative stress tolerance in P. cactorum 67 and other types of cysteine-rich proteins are likely to play important roles in P. infestans infection 68 . Thus an impairment of the cysteine metabolism might not only impact the oomycete growth but also affect its ability to infect plant tissues, which is of particular interest for crop protection.
Additional changes in the proteome caused by MMtS. Beside sulfur metabolism, MMTS affected many other important biological processes, as judged by the large number of regulated proteins in all functional categories (Supplementary Table 1). The 5 times higher abundance of D0NCV5, a putative Pleiotropic Drug Resistance protein (PDR1-15) from the ABC superfamily, supports the high toxicity of MMTS for the oomycete. PDR proteins function as efflux pumps to dispose of xenobiotics 69 . So far little is known about the way oomycetes detoxify xenobiotics: their genomes encode less cytochromes P450 than ascomycetes, but more ABC transporters 27,69 . Therefore, the upregulation of D0NCV5 might represent an attempt to detoxify MMTS and this transporter might constitute a target to consider when investigating the role of ABC transporters in the sensitivity or resistance of P. infestans to fungicides or antimicrobials 70 . One additional defence mechanism seems to be activated in MMTS-treated P. infestans, as indicated by the enrichment of a catalase and a peroxiredoxin-2, both involved in redox homeostasis (Table 3). Beyond MMTS, these redox changes seemed to occur upon exposure to all sVOCs, which led to an increased abundance of different antioxidants compared with the control (Table 3; Supplementary Table 1).
While sulfides (DMDS, DMTS), might participate in sulfhydration of free thiols of cysteine residues, the sulfonate MMTS is certainly more potent to modify thiol groups. It is actually used in chemistry to study thiol modifications on proteins 71 , since it alkylates reversibly the free thiol groups present on cysteine residues. This might explain why MMTS led to more changes in the proteome pattern of P. infestans compared to the other tested sVOCs. The high reactivity of MMTS is due to the presence of oxygen atoms near the sulfur center, as described for the related sulfur compound allicin 72 . Allicin (diallyl thiosulfinate) is another thiol modifier, which has been extensively studied in biological research because of its health-promoting properties 73 . This volatile is very abundant in Allioidae, including garlic, and is a precursor for sulfinates, such as diallyl trisulfide (DATS), whose role in protein persulfidation was mentioned above. Allicin has long been known as potent inhibitor of many microbes 72,74 , including Phytophthora 75,76 . It is also widely studied for its anticancer properties in the medical field and has a long list of cellular targets 73 . As for allicin, the antimicrobial activity of MMTS might lie in its high reactivity to the protein thiols and our proteome data also suggest multiple cellular targets in P. infestans. More advanced approaches, e.g. analysing specifically the S-thioallylation of proteins as recently performed for allicin 77 , should help to better understand the effects of MMTS and DMTS on P. infestans physiology and cellular biology, which underlie the strong anti-oomycete activity of these two sVOCs. In plants, sulfide is incorporated by the cysteine synthase into O-acetylserine to produce cysteine (C = cysteine pathway). In fungi, the main pathway involves the synthesis of cystathionine from O-acetylhomoserine as substrate for the homoserine O-acetyltransferase. From cystathionine, the activity of the cystathionine γ lyase in the reverse trans-sulfuration pathway (R) results in cysteine production. The methionine synthesis pathway (M) involves the γ cystathionine synthase and the cystathionine β lyase, which synthetize homocysteine, the precursor for methionine. Methionine can be synthetized from homocysteine by the methionine synthase using folate as methyl donor or by the homocysteine S-methyltransferase using homocysteine and S-adenosyl methionine (SAM). All reactions depicted in this schema are controlled by enzymes annotated in the KEGG for P. infestans. The thiosulfate pathway does not seem to exist in Phytophthora. Also highlighted is the significance of cysteine as substrate for cysteine desulfurases to produce important cofactors (molybdenum cofactor, Fe-S clusters) and to thiolate tRNA (MST = mercaptopyruvate sulfurtransferase). Glutathione, coenzyme A, taurocyamine are also important sulfur compound produced from cysteine in Phytophthora. Methionine, besides being an important amino acid for proteins, is metabolized to S-adenosyl methionine (SAM), which participates in the methylation of nucleic acids or histones, or the synthesis of polyamines. Modifications in protein amounts observed upon sVOC treatments are shown in red (DMTS) and green (MMTS). (2019) 9:18778 | https://doi.org/10.1038/s41598-019-55218-3 www.nature.com/scientificreports www.nature.com/scientificreports/

conclusions and perspectives
Prior to this work, we hypothesized that bacteria naturally associated with plants might be a source for novel antimicrobial compounds 14 . We therefore characterized the volatiles emitted by beneficial potato-associated Pseudomonas strains and observed that several sVOCs inhibited the growth of the late blight causing agent P. infestans 19 . Here, a more comprehensive investigation of the activities of these sVOCs was conducted on both the plant and the pathogen, which showed that: • The trisulfide DMTS and the thiosulfonate MMTS were both capable to prevent late blight on leaf discs and potato plantlets, but only MMTS did so without inducing phytotoxicity. In agronomy, only DMDS has been   www.nature.com/scientificreports www.nature.com/scientificreports/ used so far for field application to control nematodes, various soil-borne plant disease and weeds 24,78 . However, our study showed that MMTS was a much better protectant of leaf discs and plantlets against late blight than DMDS.
• MMTS was inactive as pre-treatment and did not induce plant defences. Our data indicated that MMTS acted through direct anti-oomycete activity. • Different sVOCs induced specific changes in the proteome of P. infestans. BMTMS affected the translational machinery and DMTS perturbed many important steps of sulfur metabolism. MMTS also perturbed sulfur metabolism along with many other cellular processes, including redox balance, suggesting multi-target modes of action for this potent anti-oomycete volatile compound.
We propose that MMTS, a thiosulfonate volatile compound produced by both plants and bacteria, plays an important role in plant defence against pathogens. Although MMTS was not phytotoxic in our experimental setup, future studies shall investigate the putative toxicity of this sVOC towards non-target organisms to evaluate its suitability for crop protection. Beyond this translational research aspect, our study raises many fundamental questions related to (i) the molecular targets of sVOCs in inhibited organisms such as P. infestans, (ii) the molecular determinants underlying sVOC synthesis and regulation in plant-associated bacteria, and (iii) the primary function of sVOCs for bacterial physiology and plant-bacterial interaction. Rather than being recognized as MAMPs, MMTS and DMTS dampened the plant response to flagellin. Therefore, MMTS and other sVOCs might first act as effectors to allow the initiation of the beneficial association, before contributing to the host defence against pathogens to ensure the sustainability of this association.

Material and Method
plant material and growth conditions. Potato tubers of the cultivar Bintje were potted and grown for four to six weeks (photoperiod 18 h, relative humidity 60 to 70%, 25/20 °C during the light/dark period). Leaf discs were sampled using a cork borer of 15 mm diameter from four to five individual plants using the fourth to the fifth or the fifth to the seventh leaves depending on the age of the plant. Leaf discs were incubated overnight on water agar plates (0.8% agar, LP0011, Oxoid) before infection with P. infestans and/ or treatment with sVOCs. Disease-free in vitro potato plantlets of the Victoria cultivar were provided by JP De Joffrey (Agroscope, Changins, Nyon). They were subcultured at 25 °C (18 h light; 23 °C at night) by cutting the shoot into three pieces below the axils and transferring into fresh medium. Plantlets of ten to fourteen days were used for the VOC and infection assays. The Arabidopsis plants (Columbia Col-0) used in the ROS assays were grown as one plant per pot at 21 °C with an eight-hour photoperiod for five weeks. phytophthora infestans strain and culture. Phytophthora infestans Rec01 collected at the Agroscope station of Reckenholz 5 was used for all experiments. The isolate was maintained as mycelial culture on V8 medium supplemented with 15% agar (Agar-agar, Kobe I, Roth) and 0.1% calcium carbonate. Its virulence was preserved by regular passage on Bintje potato tubers. Petri dishes were incubated upside down in a plastic bag (no sealing) in the dark at 18 °C.
Volatile treatment on p. infestans. The effect of sVOCs on mycelial growth of P. infestans was assessed using 5 mm agar plugs from the edge of actively growing mycelial colonies, which were placed downward-faced in one compartment of bi-plates (Sarstedt n°82.1195) filled with fresh V8 medium. Defined amounts of the sVOCs were applied pure or diluted in DMSO on a dry droplet of 100 μl of water agar (1%). Plates were sealed with Parafilm M and incubated upside-down in the dark at 18 °C. Mycelial growth was monitored seven days after inoculation by taking photographs and total mycelial area was further assessed using ImageJ.

Plant infection assays with p. infestans.
Zoospores were released from sporangia developed on two to three-week P. infestans plates using a treatment with ice-cold sterile water and further incubation for two hours in the fridge. After 20 min at room temperature spores were pipetted from the water surface and counted in a Jessen cell chamber. Infection assays on leaf discs were performed using a 100'000 zoospores.mL −1 by applying a ten-microliter droplet in the center of each leaf disc (abaxial surface). Petri dishes were incubated in a polystyrene box containing wet paper for six days at 18 °C in the dark. In this assay, volatiles were applied after dilution into DMSO as two-microliter droplets loaded on PTFE/silicone septa (8 mm; n° 507784 from Sigma) placed in the center of the Petri dish. For the infection of in vitro plantlets in sterile tubes (50 mL filled with 10 mL of medium), a plug of 1% water agar was loaded inside the lid where the sVOC or solvent was applied for further treatment as two-microliter droplet. One leaf of the plantlets was inoculated with ten microliters of P. infestans zoospores (100'000 spores.mL −1 ) and incubated at 18 °C under light. Pictures were taken six days post-infection to evaluate the spreading of late blight on the whole plantlets (Fig. S11). The weight of each plantlet shoot was measured after cutting the root system. fatty acid methyl esters analysis (fAMe). FAMEs were prepared from control and P. infestans -inoculated potato leaf discs treated or not with sulfur volatiles, by acid-catalysed transesterification. Samples (3 leaf discs each) were incubated in 1 ml of 5% H 2 SO 4 in MeOH (v/v), 50 μL of 0.05% butylated hydroxytoluene www.nature.com/scientificreports www.nature.com/scientificreports/ (w/v) in MeOH and 10 μg of glyceryl triheptadecanoate (Sigma Aldrich, Buchs, Switzerland) used as an internal standard. The reaction was carried out at 85 °C for 45 min in 7-ml glass tubes. Tubes were then cooled down at room temperature, briefly centrifuged, and 1.5 mL of 0.9% NaCl (w/v) and 2 mL of n-hexane were added. Samples were thoroughly shaken for 5 min and centrifuged at 240 g for 5 min. The upper organic phase containing the FAMEs was transferred into a new glass tube and the extraction was repeated two additional times with 2 mL n-hexane each time. The pooled organic phases were evaporated with nitrogen and resuspended into 200 μL of heptane. FAME samples (2 μL) were separated and quantified by GC-FID in split mode (50:1) equipped with a 30 m x 250 μm x 0.25 μm DB-23 capillary column (Agilent technologies) as previously described 79 . The chemical identification of the P. infestans-specific fatty acid 20:5 n-3 ((5Z,8Z,11Z,14Z,17Z)-5,8,11,14,17-eicosapentaenoic acid; EPA) was originally determined by co-migration with a 20:5 n-3 authentic standard (Supelco 37 component FAME mix; Sigma) by GC-FID analysis. It was further confirmed by GC-MS-EI analysis in splitless mode with the same capillary column and oven program as for GC-FID analysis. The injection port and detector temperatures of the GC-MS were set at 250 and 230 °C, respectively. Mass spectra were obtained by electron ionization set at 70 eV with a data acquisition rate of 50 Hz. The EPA mass spectrum was compared to 20:5 n-3 authentic standard (Supelco 37 component FAME mix) and with the EPA methyl ester reference mass spectrum (http:// www.lipidhome.co.uk/ms/methesters/me-5plus/index.htm).
oxidative burst assays. The effect of sVOCs on the defence response of Arabidopsis was evaluated by measuring the production of reactive oxygen species ("oxidative burst"). Briefly, leaf discs were floated on water overnight and ROS released by the leaf tissue were measured using a luminol-based chemiluminescent assay 34 . ROS were elicited with 1 μM flg22 peptide (QRLSTGSRINSAKDDAAGLQIA, obtained from EZBiolabs) in all experiments. Mock treatments without flg22 were performed with the control solution (1% w/v BSA, 100 mM NaCl) used to solubilize the peptide. sVOCs were applied at the indicated doses (0 to 10 μg) and DMSO was used as solvent control. Luminescence emitted by the oxidized L-012 luminol (Wako Chemicals USA) was measured over a time period of 30-35 min using a luminometer (Glomax, Promega, Switzerland).
Q-pcR analysis of gene expression. The effect of sVOCs on defence gene expression was analysed by qPCR in 4-week old potato plants. Leaf discs from three plants were sampled and incubated on water agar plates at room temperature for 6 h. Leaves were then infected (or not) with zoospores of P. infestans, and co-treated (or not) with 1 mg of volatiles as described above. After 6 h of treatment, leaf discs were collected, snap-frozen in liquid nitrogen and stored at −80 °C. RNA were extracted from leaf discs using the phenol-chloroform extraction method using Trizol solution (38% (v/v) saturated phenol (pH 8), 0.8 M guanidine thiocyanate, 0.4 M ammonium thiocyanate, 0.1 M Na-acetate pH 5, 5% (v/v) glycerol). RNA extracts were treated with the DNase I from Sigma Aldrich and reverse transcription was performed using the SensiFAST cDNA Synthesis Kit from Bioline. Quantitative PCR reactions were performed using the SensiFAST SYBR Hi-ROX Kit from Bioline. Each reaction was carried out with 5 µL of cDNA (5 ng.µL −1 ), 7.5 µL of SYBR Hi-ROX mix, 0.5 µL of each primer and 1.5 µL of sterile water. For amplification, an initial denaturation step at 95 °C for 15 min was done, followed by 45 amplification cycles (95 °C for 15 s, 60 °C for 15 s, 72 °C for 30 s). Each reaction was run in duplicate and the experiment was repeated twice. All results were analysed using the double delta Cq method with uninfected samples treated with water as references. Primers for the defence genes ERF3, LOX, PR-1b and PR-5 from potato 37 were newly designed and are listed in Supplementary Table 2. Genes coding for the peptidyl-prolyl isomerase (CyP) and Elongation Factor α (EF1-α) were used as reference for normalization of data. A two-way ANOVA followed by Tukey-HSD post-hoc test was performed for statistical analysis with p ≤ 0.05. Proteomic analysis of sVOC-treated p. infestans cultures. P. infestans was grown on 10 mL Rye agar medium 19 at 18 °C for ten days in one compartment of a 80 mL bi-plate. Thereafter, 300 μg sVOC (or DMSO as control) were pipetted on agar plugs in the other compartment of the Petri dish. This concentration was selected as that at which even the moderately active BMTMS reduced P. infestans mycelial growth (Fig. S10). Plates were sealed with parafilm, packed in a plastic bag and incubated upside down for 24 hours at 18 °C in the dark, after which P. infestans mycelium was collected by scratching the agar surface with a glass coverslip and frozen. Samples were frozen, ground in liquid nitrogen and stored at −80 °C before protein extraction. The protein extracts of the six different treatments from the three independent biological experiments (n = 18) were prepared concomitantly. To this end fifty milligrams of frozen powder were resuspended into cold SDS-extraction buffer (50 mM Tris HCl pH 7.5, 150 mM NaCl, 1% SDS) supplemented with antiproteases (Complete, Sigma).
Concentration of protein extracts was determined using Roti Nanoquant (Carl Roth). 30 µg of protein extract per sample were separated on 1D-SDS PAGE and stained with Blue silver colloidal Coomassie 80 . Lanes were cut into 15 equidistant pieces and subjected to tryptic in-gel digestion as described earlier 81 . Resulting peptide mixes were desalted using C18 Zip Tips (Millipore). LC-MS/MS analyses were done using an EASY-nLC, coupled to an Orbitrap Velos mass spectrometer (Thermo Scientific). Peptides were separated on in-house self-packed nano-LC columns (100 µm × 20 cm) containing reverse-phase C 18 material (3.6 µm, Aeris, phenomenex) and eluted by a non-linear binary gradient of 77 minutes from 5% to 99% solvent B (0.1% acetic acid (v/v), 99.9% acetonitrile (v/v)) in solvent A at a constant flow rate of 300 nl min −1 . Samples were measured in LTQ/Orbitrap parallel mode, survey scans in the Orbitrap were recorded with a resolution of 60,000 in a m/z range of 300-1,700 and the 20 most intense peaks were subjected to CID fragmentation in the LTQ. Dynamic exclusion (30 sec) of precursor ions was enabled, single-charged ions as well as ions with unknown charge-state were excluded from fragmentation. Internal lock-mass calibration (lock mass 445.120025) was enabled.