To test potential activities of P. nodorum VOCs, four fungi were used for growth antagonist assays: P. nodorum, Fusarium oxysporum f. sp. lycopersici, Eutiarosporella tritici-australis and Zymoseptoria tritici. Escherichia coli, Pseudomonas syringae, Sinorhizobium meliloti, and three other bacteria isolated from within surface sterilized wheat seeds, Bacillus cereus, Sphingobacterium multivorum, and Flavobacterium sp. were also used for the growth assays.

In one side of a segmented Petri dish (9 cm diameter), 25 μl of a P. nodorum spore solution (1 × 10⁶ spores/ml) was inoculated on Fries agar (1.5%) (Supplementary Table S1) and incubated at 22°C in 12-12 h dark and light. After two weeks, the other compartment of the segmented Petri dish was inoculated with the test organisms. The fungi were inoculated onto potato dextrose agar (PDA), the bacteria on Lysogeny broth (LB) agar. Water agar (1%) was used for the germination of wheat (Triticum aestivum cv. Grandin) and Medicago truncatula for which 8 and 10 seeds per plate were used respectively (Supplementary Table S1). Plates were sealed with parafilm after the test organisms were inoculated. The effect of the VOCs was then visually monitored daily.

To assess the phytotoxic and fungistatic effect of 2-methyl-1-propanol, 2-methyl-1-butanol, 3-methyl-1-butanol and 2-phenylethanol, each of the compounds was placed on a 1 cm² filter paper on a section of a segmented Petri dish. On the other half of the dish was placed either wheat seeds or P. nodorum was inoculated, on the appropriate medium as described above. An additional treatment was also performed containing a mix of these compounds in the proportions found in the chromatographic analysis of P. nodorum VOCs. 1 mM of each of the pure compounds (74 to 122 ppm) and 100 ppm for the mix were used in these assays, considering a free internal volume of the petri dish of 48.9 cm³ (the total volume minus 15 ml of test medium). Plates were sealed using parafilm immediately after the compound was added, however it cannot be excluded that the molecules did not undergo some degree of diffusion through the membrane.

Each experiment was repeated twice using at least three Petri dishes per treatment. The number of seeds germinated per plate as well as radicle and coleoptile lengths were recorded and the average and standard deviation were calculated. T-tests were performed comparing each treatment against the control to determine statistical significance. Qualitative data was not statistically evaluated.

To identify the individual components of the VOCs, slanted Fries agar head-space (HS) vials (20 ml) were inoculated with P. nodorum. Vials with cotton stoppers were incubated for one week at 22°C in 12-12 h dark and light cycles. Vials were sealed with silicon/Teflon septa crimp caps 24 h prior to the analysis. Three mock-inoculated vials were used as controls. To calculate the retention indices, 5 μl of an alkane mix at 20 ppm in CH₂Cl₂ was added to a HS vial (Kováts, 1958; van Den Dool and Dec Kratz, 1963). To confirm the identity of ethyl acetate, 2-methyl-1-propanol, 2-methyl-1-butanol, 3-methyl-1-butanol and 2-phenylethanol within the fungal VOCs, a mix following the proportions found in solid phase micro-extractions in line with a gas chromatography-mass spectrometry (SPME-GC-MS) analysis of the fungal cultures (10:14:26:53:11 respectively) was prepared using pure compounds and 1 μl of the mix was added to a HS vial.

To evaluate the in planta production of sesquiterpenes, the distal 5 cm of the second leaf of 2-week old wheat seedlings were excised and sprayed with a 1 × 10⁶ P. nodorum spores/ml solution containing 0.02% tween 20. The cut end of each leaf was embedded in a HS vial containing 2 ml of water agar (1%). Vials were closed with silicon/Teflon septa crimp caps and incubated for 3 days at 22°C in 12-12 h dark and light cycles. Three mock-inoculated samples were used as controls.

The SPME-GC-MS analyses were performed in an Agilent 7890A gas chromatograph coupled to a single quadrupole Agilent 5975 mass spectrometer using a Gerstel MPS 2XL autosampler. The column for the analyses was a Varian CP9013-1Factor 4 5 m s 350°C: 40 m x 250 (μm x 0.25 (μm. Elution was performed with He flow at 1.5 ml/min and temperature programed from 40°C (hold 3 min) to 180°C at 4°C/min and then to 220°C (hold 5 min) at 10°C/min. The mass spectrometer was operated in the electron ionization (EI) mode with ionization energy of 70 eV and scanning the mass range of m/z 40-600. Temperatures were set to: GC inlet, 240°C; GC transfer line, 240°C; MS source, 200°C; and quadrupole 250°C. Volatiles were adsorbed onto a SPME fiber coated with divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) (1 cm, 23 Ga, 50/30 μm film thickness, Supelco) for 120 min at 30°C after a 5 min equilibration. The fiber was desorbed in the injector at 240°C (splitless mode 2 min). The fibers were conditioned by keeping them in the GC injector at 240°C for 10 min.

All GC-MS experiments were done in triplicate. Data was acquired using MSD ChemStation E.02.01.1177 (© Agilent Technologies, Inc.). Analysis of the data was performed using ChemStation and MS Search NIST Mass Spectral Search Program [Version 2.0g] for the NIST/EPA/NIH Mass Spectral Library [NIST Standard Reference Database 1A Version NIST 11] build May 19 2011 (© National Institute of Standards and Technology).

Parastagonospora nodorum sesquiterpene synthases (Sts), SNOG_03562 (Sts1), and SNOG_04807 (Sts2), were individually disrupted in P. nodorum wild type (SN15) by split marker homologous recombination of a phleomycin resistance cassette. 1.5 Kb 5’ and 3’ flanking regions for each gene were amplified from P. nodorum genomic DNA (primers in Supplementary Table S2). The phleomycin resistance gene was amplified as two overlapping amplicons, Phl and Leo, from the pAN8-1 plasmid (primers in Supplementary Table S2) (Mattern et al., 1988). 5’ flanks were PCR fused to Leo while 3’ flanks were PCR fused to Phl (primers in Supplementary Table S2). P. nodorum was then transformed by a PEG-protoplast method as previously reported (Solomon et al., 2004).

To assess the copy number of the phleomycin cassette in the transformants, qRT-PCR primers were designed for the phleomycin resistance gene; elongation factor 1α, actin and SnToxA primers were used to normalize the data (primers in Supplementary Table S2). As a phleomycin single copy reference, a tox3 knock out strain was used (Tan et al., 2014).

To isolate and characterize the product of Sts1 and Sts2, the coding sequences were cloned into the linearized plasmid backbone (XW55) from YEplac-ADH2p (primers in Supplementary Table S2) (Lee and DaSilva, 2005, Chooi et al., 2015a). In vivo yeast recombination cloning using each gene and XW55 was performed with the Frozen-EZ Yeast Transformation II^TM Kit (Zymo Research, Irvine, CA) and competent Saccharomyces cerevisiae BJ5464-NpgA according to manufacturer’s protocol. Positive transformants were selected by PCR from colonies grown on synthetic dropout agar lacking uracil.

Medium scale P. nodorum fermentations for the isolation of α-cyperone were performed using 8 l of liquid Fries medium inoculated with 4 × 10⁶ spores. Cultures were incubated in the dark during 10 days at 22 °C and shaking at 120 rpm.

For the isolation of acora-4,9-diene, and eudesma-4,1-diene and β-elemene, transformed S. cerevisiae harboring Sts1 or Sts2 were inoculated in in 6 ml synthetic dropout agar lacking uracil (Supplementary Table S1) for 72 h at 28°C 200 rpm. Each of these seed cultures was used to inoculate 5 l YPD broth (Supplementary Table S1) and incubated 90 h at 22°C at 200 rpm.

Fungal cultures were lyophilized and low to medium polarity compounds were extracted with dichloromethane. Yeast cultures were centrifuged and the cells subjected to acetone extraction. Both, dichloromethane and acetone were evaporated in a rotary evaporator. The sesquiterpenes from the extracts were isolated by acetonitrile/hexane partition. Sesquiterpenes were purified by SiO₂ hexane flash chromatography followed by C18 water/acetonitrile flash chromatography. Purity of terpenes was assessed by GC-MS. Isopentane was used to recover the terpenes from the water-acetonitrile mixture.

GC-MS² was performed to identify β-elemene and α-cyperone by comparison with standards in a Finnigan TraceGC ultra (Thermo Scientific) coupled to an iontrap Finnigan Polaris Q (Thermo Scientific) mass spectrometer. β-elemene was injected onto a BPX70 30 m × 0.25 mm id (SGE Analytical Science) while α-cyperone in a Varian CP9013-1Factor4 5 ms column, which were eluted with He (inlet pressure 15 psi; injection port 200°C; interface 250°C; source 200°C). For β-elemene the column was temperature programed from 60°C (hold 1 min) to 100°C at 25oC/min, then to 150 at 10 oC/min, and finally to 240°C at 10°C/min (hold 3.5 min); for α-cyperone the program started at 60°C (hold 1 min) to 200°C at 30°C/min, then to 220 at 3°C/min, and finally to 325°C at 30°C/min (hold 1 min). The mass spectrometer was operated in the electronic ionization (EI) mode with ionization energy of 70 eV, scanning the mass range of m/z 50–450. For MS/MS experiments, the precursor ions were selected with a peak width of 1.0 amu over 12ms. The ions were excited at 1 V for 15 ms with q = 0.3 and the products scanned over a mass range of m/z 100–250. Data analysis was performed employing Xcalibur^TM 1.4 (©Thermo Scientific) software.

The eudesma-4,11-diene (1) and the acora-4,9-diene (4) were dissolved in CDCl₃ and analyzed by NMR. ¹H NMR, ¹³C NMR, HSQC, and HMBC were performed in an Avance^TMIII HD 300 MHz NanoBay NMR device (Bruker).

To test the requirement of Sts genes for infection, the mutants of P. nodorum lacking the Sts1 and Sts2 genes were inoculated onto the second leaf of two-week-old wheat seedlings (cv. Axe) which was attached to a styrofoam platform using double sided sticky tape and spayed with a spore solution (1 × 10⁶ spores/ml containing 0.02% tween 20). 0.02% tween 20 was used as control. Inoculated seedlings were incubated for 48 h at 22°C in a dark moisture chamber. After the initial 48 h incubation, the inoculated plants were grown at 85% humidity, 20°C during the day and 12°C at night with 16-8 h light/dark cycles. The leaves were collected at five days post inoculation to evaluate the disease. These assays were performed twice with a minimum of four leaves per treatment used.

To assess if volatile emissions of P. nodorum harbor bioactive VOCs, split plate assays were performed to evaluate a series of biological activities including phytotoxicity, fungitoxicity and bactericidal (Figure 1). Segmented Petri dishes were used to prevent molecules diffusing through the media and ensure that any observable activities could be solely attributed to the volatile compounds. To assay for phytotoxicity, seeds from the host of P. nodorum, wheat cv. Grandin, and a dicotyledonous plant, Medicago truncatula, were used. P. nodorum VOCs had a strong effect on the radicle elongation and hypocotyl or coleoptile growth which were significantly reduced when the seeds were germinated in the presence of P. nodorum but appeared unaffected in the control plates (no fungal inoculation). The effect on bacteria was mixed as there was no observable impact on the growth of a variety of different strains including Escherichia coli, Pseudomonas syringae, Bacillus cereus, or Flavobacterium sp. in the presence of the fungal VOCs (Supplementary Figure S1). In contrast, there was a strong reduction in the growth of the nitrogen-fixing bacterium Sinorhizobium meliloti and also Sphingobacterium multivorum when P. nodorum was cultured in the same Petri dish (Figure 1). There was no apparent impact on the growth of any of the fungi tested when grown with P. nodorum with the exception of the apparent self-inhibition of P. nodorum growth (by its own VOCs) (Figure 1 and Supplementary Figure S1).

To dissect the chemical basis of the bioactivities described above, the identities of the volatile molecules were determined using a combination of solid phase micro-extractions (SPME) from the headspace (HS) of ten days old fungal cultures in slanted Fries agar vials and subsequent analysis by gas chromatography-mass spectrometry (GC-MS) and spectral comparison against pure standard and the NIST library. Within the P. nodorum VOCs mixture, several alcohols and esters were identified as being the most prominent signals (percentage of area of the whole chromatogram) (Table 1): 3-methyl-1-butanol (representing 5.36% of the VOCs mixture), 2-methyl-1-butanol (2.6%), 2-methyl-1-propanol (1.43%) and 2-phenylethanol (1.13%). Many other volatile molecules were also identified in peaks with smaller areas. The polyketide mellein (0.9%) was also detected along with some sesquiterpenes of which two were putatively identified as β-elemene and eudesma-4,11-diene.

The activity of the four most prominent volatile molecules identified in the chromatograms from the head space of P. nodorum (3-methyl-1-butanol, 2-methyl-1-butanol, 2-methyl-1-propanol, 2-phenylethanol) were assayed to assess their impact on the growth of P. nodorum and wheat seedling development. These compounds were tested independently at an atmospheric concentration of 1 mM. Additionally, a mixture of these compounds following the in vitro proportions, was prepared and tested at 100 ppm. Neither the independent pure VOCs nor the mixture had any effect on P. nodorum growth suggesting that these molecules are not responsible for the inhibitory effect described above (data not shown). In contrast, a 27% decrease in germination was observed when wheat seeds were exposed to 3-methyl-1-butanol, although no other treatment had a significant effect on germination (Figure 2). However, both radicle and coleoptile elongation were repressed in all treatments; 3-methyl-1-butanol showed the greatest inhibition (100% and 83% respectively) while 2-methyl-1-propanol showed the least inhibition (36% and 31% respectively).

In addition to the bioactive short chain alcohols, we were also interested in the presence of the sesquiterpenes found in the axenic culture VOCs. To determine if these molecules played a potential role in disease development, the production of sesquiterpenes was assayed for during infection and compared to those produced in axenic culture. VOCs were extracted from vials containing either infected leaves or the fungus grown axenically and analyzed by HS-SPME-GC-MS. Interestingly the same sesquiterpenes produced in vitro by P. nodorum were also found in planta as well as others not previously observed (Figure 3). Eudesma-4,11-diene (sesquiterpene 1), β-elemene (sesquiterpene 2) and α-cyperone (sesquiterpene 3) were putatively identified by comparing the acquired data against the NIST database. Another interesting compound was sesquiterpene 4, the most abundant sesquiterpene detected from P. nodorum, in Fries cultures and in wheat leaves. However, despite a fragmentation pattern and molecular weight typical of a sesquiterpene, its match against entries present in the NIST library was not high enough to confidently assign a possible identity.

The P. nodorum genome encodes three sesquiterpene synthases (Chooi et al., 2014), Sts1, Sts2, and Sts3. Previous studies have demonstrated that Sts1 and 2 are expressed during infection, but not Sts3. Furthermore, analysis of the Sts3 gene sequence revealed that it appears truncated implying that it isn’t functional and thus it was not considered for further study (Ipcho et al., 2012). To directly link the molecules identified above to the genes, Sts1 and Sts2 were disrupted individually in the P. nodorum genome through homologous recombination. Disruption cassettes were constructed to independently replace Sts1 and Sts2 with a phleomycin resistance marker. P. nodorum was transformed and positive colonies were selected from phleomycin-containing plates. Correct disruption of the genes was verified by PCR and strains containing a single copy of the disruption cassette were selected by qPCR.

Single copy transformants of P. nodorum strains lacking Sts1 (sts1) and Sts2 (sts2) were selected for further analysis by HS-SPME-GC-MS (Figure 4). The signals of sesquiterpenes 1, 2, and 3 along with three other unidentified sesquiterpenes present in wild-type P. nodorum were absent in the sts2 mutants indicating this gene codes for the core biosynthetic enzyme of the three putative sesquiterpenes plus some other sesquiterpene structures (same mass and similar fragmentation pattern). The mutants lacking sts1 were missing three unidentified compounds putatively identified as sesquiterpenes, including sesquiterpene 4, suggesting that Sts1 is responsible for its synthesis.

To confirm the identity of these sequiterpenes, an isolation from medium scale fermentation of P. nodorum in Fries media was undertaken. Sesquiterpene 3 was isolated by silica flash chromatography followed by C18 flash chromatography. However, the isolation of sesquiterpenes 1, 2, and 4 was not achieved. Consequently, the Sts1 and Sts2 genes were heterologously expressed in yeast to confirm 1 and 2 as eudesma-4,11-diene and β–elemene respectively. Acetone extracts from small scale cultures of the yeast strains harboring the sesquiterpene synthases were analyzed by GC-MS and the production of 1, 2, and 4 by the heterologous expression of Sts2 and Sts1 was confirmed (Figure 5). Interestingly, sesquiterpene 3 (putatively α-cyperone) was not detected, suggesting it is modified post synthesis.

Sesquiterpenes 1 and 2, and 4 were isolated by silica followed by C18 flash chromatography of acetone extracts from medium scale YPDA fermentations of yeast carrying Sts2 and Sts1 respectively. Subsequent GC-MS analysis of the sesquiterpene fractions confirmed that 4 is the major product of Sts1 but also that 14 other unidentified terpenes are synthesized by the same enzyme (Supplementary Figure S2). Similarly, the predominant sesquiterpene produced by Sts2 is 1 along with 10 other molecules including compound 2 (Supplementary Figure S3).

Commercial standards of β-elemene and α-cyperone were purchased and analyzed by GC-MS² to confirm the identities of sesquiterpenes 2 and 3. Identical retention times in addition to a comparison of the MS² fragmentation profiles that demonstrated a complete overlap of the major ions present in the standards compared to the extracted samples confirmed the identities of 2 and 3 (Figure 6 and Supplementary Figures S4, S5).

The isolated sesquiterpene 1, the putative eudesma-4,1-diene, and unknown sesquiterpene 4 were subjected to ¹H NMR, ¹³C NMR, HMBC and HSQC experiments (Supplementary Figures S6–S13). Supplementary Tables S3, S4 present the assignments and chemical shifts (δ) for all carbons and the corresponding hydrogens as well as the correlations obtained from the HMBC experiment for sesquiterpene 1 and sesquiterpene 4 respectively. The identity of sesquiterpene 1 was confirmed to be eudesma-4,11-diene, and sesquiterpene 4 is proposed to be acora-4,9-diene (Figure 6) based on comparison to previously reported NMR data (Iwabuchi et al., 1989; Chen and Lin, 1993).

A segmented plate growth assay was used to determine if the molecules derived from either Sts1 or Sts2 are involved in the biological activities described above. As previously observed, volatiles from wild type P. nodorum reduced wheat seed germination and inhibited the growth of S. multivorum as well as self-growth. The growth of both the sts1 or sts2 mutants caused the same phenotypic effects as the wild type inferring that the products of either gene are not responsible for these inhibitory activities (data not shown).

The role of the Sts1 and Sts2 terpenes in pathogenicity was evaluated by inoculating the second leaf of two-week-old wheat seedlings with wild type P. nodorum and each of the mutants. Symptom development for the sts1 and sts2 mutants was unaffected compared to the wild type suggesting that the genes do not play a role in disease as assayed in this attached leaf system (Supplementary Figure S14).