Pseudomonas Strains Naturally Associated with Potato Plants Produce Volatiles with High Potential for Inhibition of Phytophthora infestans

Abstract

Bacteria emit volatile organic compounds with a wide range of effects on bacteria, fungi, plants, and animals. The antifungal potential of bacterial volatiles has been investigated with a broad span of phytopathogenic organisms, yet the reaction of oomycetes to these volatile signals is largely unknown. For instance, the response of the late blight-causing agent and most devastating oomycete pathogen worldwide, Phytophthora infestans, to bacterial volatiles has not been assessed so far. In this work, we analyzed this response and compared it to that of selected fungal and bacterial potato pathogens, using newly isolated, potato-associated bacterial strains as volatile emitters. P. infestans was highly susceptible to bacterial volatiles, while fungal and bacterial pathogens were less sensitive. Cyanogenic Pseudomonas strains were the most active, leading to complete growth inhibition, yet noncyanogenic ones also produced antioomycete volatiles. Headspace analysis of the emitted volatiles revealed 1-undecene as a compound produced by strains inducing volatile-mediated P. infestans growth inhibition. Supplying pure 1-undecene to P. infestans significantly reduced mycelial growth, sporangium formation, germination, and zoospore release in a dose-dependent manner. This work demonstrates the high sensitivity of P. infestans to bacterial volatiles and opens new perspectives for sustainable control of this devastating pathogen.

INTRODUCTION

During the last decade, it has become evident that bacteria communicate with other organisms through the emission of volatile compounds. Highly significant volatile-mediated effects of bacteria have been reported for various target organisms, including bacteria themselves (1–5), plants (5–9), and fungi (10–12). The research carried out to understand the nature of this volatile-mediated interaction of bacteria with plants and with other bacteria has focused so far on model organisms (e.g., Arabidopsis thaliana and Escherichia coli) and has enabled identification of some of the active compounds involved in the respective interactions, such as indole, 2,3-butanediol, dimethyl disulfide, hydrogen sulfide, and ammonia. The research on model organisms has also contributed to understanding of the mechanisms underlying the observed phenotypic changes of increased (13–15) or decreased (16, 17) plant biomass and increased antibiotic resistance in bacteria (2–4, 18).

As far as fungi are concerned, most studies investigating their response to bacterial volatiles have focused on potential application and have thus largely neglected deeper investigation of the chemical nature of the active compounds and/or of the mode of action of these molecules. In addition to the inorganic volatiles hydrogen cyanide (19) and ammonia (20), few volatile organic compounds, such as sulfur compounds and long-chain ketones, have been unequivocally shown to inhibit the growth of phytopathogenic fungi when applied at biologically relevant concentrations (12). With the ultimate prospect of using the antifungal potential of bacterial volatiles for crop protection, most studies investigating fungal response to bacterial volatiles have used phytopathogenic fungi of agronomical relevance, such as Fusarium, Rhizoctonia, Verticillium, and many others (12, 21). Remarkably, oomycetes, a group of organisms phylogenetically related to algae but morphologically and physiologically close to fungi, have largely been ignored in previous studies that characterized the response of phytopathogenic organisms to bacterial volatiles, despite the heavy losses they cause in a wide range of crops (22, 23). For instance, the late blight-causing agent, Phytophthora infestans, which is responsible for the most devastating disease of potato worldwide, has so far not been tested for sensitivity to bacterial volatiles. The aims of this work were (i) to investigate the response of the oomycete P. infestans to bacterial volatiles, (ii) to compare this response to that of other selected potato pathogens belonging to different kingdoms (fungi and bacteria), and (iii) to identify active molecules responsible for the observed effects. To achieve these aims, we isolated a collection of bacterial strains from the rhizosphere and the phyllosphere of field-grown, late-blight-infested potato plants and assessed their volatile-mediated effects on the oomycete. We compared the response of P. infestans to that of other potato pathogens belonging to fungal and bacterial taxa, including Rhizoctonia solani, Helminthosporium solani, Fusarium oxysporum, and Dickeya dianthicola. Moreover, we analyzed the volatile profiles of a selected subset of strains to elucidate the chemical nature of the molecules responsible for the observed effects.

MATERIALS AND METHODS

Chemicals and culture media.

Chemicals were purchased from Sigma-Aldrich (Switzerland) unless otherwise specified. Luria-Bertani (LB) medium was prepared by dissolving 20 g liter⁻¹ of Difco LB broth, Lennox (BD), and adding 15 g liter⁻¹ agar (Agar-agar; ERNE surface AG). For 10-fold-diluted LB (1/10 LB), 2 g liter⁻¹ of LB broth was used. Actinomycete isolation agar (AMA) contained 22 g liter⁻¹ Difco actinomycete isolation agar (BD) and 5 ml liter⁻¹ glycerol (Sigma-Aldrich). Rye agar (RA) was prepared by simmering 200 g rye grains (winter rye cv. Picasso95) in 1.5 liters of tap water for ca. 1 h. The liquid was then filtered through a sieve (1.5-mm mesh) and made up to a final volume of 1 liter with tap water. Agar (20 g liter⁻¹) was added. Malt agar (MA) contained 15 g liter⁻¹ Difco malt extract agar (BD) and 12 g liter⁻¹ agar. Potato dextrose agar (PDA) was used at 39 g liter⁻¹ (Oxoid). Petri dishes were filled using a plate-pouring machine (Mediajet; Integra Biosciences) with 18 ml of medium for standard petri dishes (94 by 16 mm; Greiner Bio-One) and 5.6 ml of medium for small dishes (55 by 14.2 mm; Gosselin, Semadeni). Medium for two-compartment petri dishes (94 by 15 mm; Greiner Bio-One) was poured manually, aiming at ca. 10 ml per compartment.

Fungal strains and culture conditions.

A P. infestans polyspore isolate obtained in 2001 (provided by H. Krebs, Agroscope) was used for all experiments. This isolate had been maintained as a mycelial culture on RA supplemented with 5 g liter⁻¹ d-glucose and regularly transferred to potato slices for host passages. The fungi Rhizoctonia solani, Helminthosporium solani, and Botrytis cinerea were obtained from P. Frei (Agroscope). A Fusarium oxysporum strain was isolated from infected tubers in 2013. The fungi were kept in 25% glycerol at −80°C and routinely grown on either MA (R. solani and B. cinerea) or PDA (R. solani and H. solani). Petri dishes were sealed with Parafilm M (Bemis flexible packaging) and stored in the dark at ca 20°C (fungi) or 18°C (P. infestans).

Bacterial strains and culture conditions.

Most bacteria were newly isolated (see below). In addition to these newly isolated strains, we used Pseudomonas protegens CHA0 and CHA77, as well as Pseudomonas chlororaphis MA 342 (commercially available as Cerall), which were kindly provided by C. Keel and K. Lapouge from the University of Lausanne. Dickeya dianthicola strain 88-23 was kindly provided by S. Schaerer (Agroscope). Bacterial strains were routinely grown on LB and kept at −80°C in 25% glycerol for long-term storage.

Isolation of bacteria from field-grown potato plants.

In October 2012, three potato plants were collected with their root systems and adhering soil from a field that had been previously artificially inoculated with P. infestans (experimental field from Agroscope, site Zurich-Reckenholz). For each of the three collected plants, leaves and stems (referred to here as “shoots”) as well as root tissues were separated and treated as follows: shoots were ground in a disinfected ceramic mortar using ca. 5 ml of sterile water; roots and adhering soil were shaken in sterile water to collect the rhizosphere soil, while the roots themselves were discarded. The samples were homogenized by shaking and pipetted into a test tube with a cut tip. These suspensions were 10-fold serially diluted in sterile water and plated on three different isolation media: 1/10 LB, MA, and AMA (see above). Dilutions of 10⁻² to 10⁻⁷ were plated on 1/10 LB and MA, and dilutions of 0 to 10⁻⁵ were plated on the selective medium AMA. All plates were incubated at ca. 20°C for at least 6 days. In order to cover as much of the cultivable diversity as possible, single colonies with different morphologies (per plant and sample type [rhizosphere versus phyllosphere]) were picked for isolation. A total of 137 bacterial strains were isolated from the different plants (79 from rhizosphere soil and 58 from the phyllosphere). Strains were then transferred to 1/10 LB and kept on this medium. A minority of the strains could not be grown in liquid culture and were harvested from LB plates to be resuspended in 0.9% NaCl. They were then treated in a manner similar to that used for the liquid LB cultures.

Phylogenetic identification of the strains using 16S and rpoD gene sequencing.

Actively growing cells were picked with a sterile toothpick, which was vigorously stirred in a tube containing 50 μl of lysis buffer (50 mM KCl, 0.1% Tween 20, 10 mM Tris-HCl [pH 8.3]). After 99°C incubation for 10 min, 1 μl of the cell lysate was used as the template for PCR. Reaction mixtures contained 1× Go Taq Flexi buffer (Promega), 3 mM MgCl₂, 0.2 mM concentrations of deoxynucleoside triphosphates (dNTPs), a 0.4 μM concentration of each primer, and 1 U Go Taq Flexi polymerase. The universal 16S rRNA primers F (5′-AGAGTTTGATYMTGGCTCAG-3′; E. coli 16S rRNA gene positions 8 to 27; forward primer) and R (5′-CAKAAAGGAGGTGATCC-3′; E. coli 16S rRNA gene positions 1529 to 1545; reverse primer) were used (24). The PCR protocol included an initial denaturing step of 5 min at 94°C, followed by 35 cycles of 30 s at 94°C, 30 s at 58°C, and 90 s at 72°C and a final elongation step of 10 min at 72°C. The reaction was performed with a thermocycler iCycler (Bio-Rad). Correct and specific amplification was verified by gel electrophoresis, and samples showing a single 1.5-kb band were purified using a Nucleo-Spin gel and PCR cleanup kit (Macherey-Nagel), following the manufacturer's instructions. Thereafter, a sequencing PCR (total volume, 5 μl) was performed utilizing 2 μl of the purified PCR product as the template, either one of the primers used for initial PCR (5 μM), BigDye sequencing buffer (1×), and 1 μl of BigDye Terminator from the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems). The sequencing PCR started with 10 s denaturation (96°C) followed by 25 cycles of 5 s at 50°C and 4 min at 60°C. PCR products were subsequently purified with a BigDye Terminator kit according to the manufacturer's instructions (Applied Biosystems). Twenty microliters of purified DNA (supernatant) was finally pipetted into a 96-well plate and analyzed using a Prism 3300xl genetic analyzer (Applied Biosystems). The obtained sequences were analyzed with BioEdit (Tom Hall, Ibis Biosciences). The BLAST (basic local alignment search tool; http://blast.ncbi.nlm.nih.gov) interface and databank were used for identification of the strains to the genus level (see Table S1 in the supplemental material). Active strains affiliated with the genus Pseudomonas were characterized further using rpoD sequencing. To this end, a similar PCR protocol was applied as for the 16S PCR, with the following modifications: the primers used were rpoDf (5′-ACTTCCCTGGCACGGTTGACCA-3′) and rpoDr (5′-TCGACATGCGACGGTTGATGTC-3′) (25), the annealing temperature was 60°C, and the elongation time was 1 min. Purification, sequencing, and BLAST searching were performed as described above, and the results are shown in Table S2 in the supplemental material.

Dual-culture experiments with selected active strains.

Of the 92 strains identified, 32 caused mycelial growth inhibition in a prescreen using P. infestans, R. solani, and B. cinerea as target organisms and were selected for further investigation. These 32 candidate strains were tested for antioomycete, antifungal, and antibacterial activity together with three well-characterized Pseudomonas strains used as controls (P. protegens CHA0, its isogenic cyanide mutant CHA77, and P. chlororaphis MA342). Five different target organisms were used: the oomycete P. infestans, the fungi R. solani, H. solani and F. oxysporum, and the bacterium D. dianthicola. Split petri dishes were used to analyze the volatile-mediated effect of the isolates on the target organisms. LB was poured on one half (isolated strains) and the respective medium (RA for P. infestans and F. oxysporum, PDA for H. solani, and R. solani or LB for D. dianthicola) on the other half (targets). To take the different growth speeds into account, the targets were inoculated at different time points: on the same day as bacteria (D. dianthicola and R. solani), 1 day before (P. infestans), 4 days before (F. oxysporum), or 14 days before (H. solani). Three drops of 10 μl of overnight bacterial culture adjusted to an optical density at 600 nm (OD₆₀₀) of 1 were spotted on one half of the split petri dish (LB for control plates), while the targets (a plug of mycelium or a drop of overnight culture for D. dianthicola) were inoculated on the other half. Plates were sealed with Parafilm, incubated at 20°C in the dark and photographed after 3 days for R. solani, 7 days for D. dianthicola and F. oxysporum, 14 days for P. infestans, and 28 days for H. solani. The plates were photographed from below with a reflex camera mounted on a stand. The obtained pictures were analyzed with the digital imaging software ImageJ (http://imagej.nih.gov/ij/). The mycelium area was determined with the freehand (or circle, for D. dianthicola) area measurement tool of ImageJ. Growth was determined by subtracting the initial mycelial surface from that obtained after the given incubation time. This growth value was then compared to that measured in control plates (targets exposed to LB only), and a percentage was calculated. For D. dianthicola, integrated density (pixels) was used as a measure of colony density. These dual culture assays were performed with 3 or 4 biological replicates (petri dishes) per strain and target.

Analysis of HCN and NH₃ emission.

The emission of volatile hydrogen cyanide (HCN) was detected by inoculating the candidate strains as one 10-μl drop (OD₆₀₀ = 1) in a compartment of a split petri dish containing LB and placing a sterile piece of filter paper (1- by 1-cm² cuts) soaked in 5 mg/ml copper(II) ethyloacetate (Strem Chemicals, Newburyport, MA, USA) and 5 mg/ml 4,4-methylenebis-N,N-dimethylaniline (Fluka, Switzerland) in chloroform, which was air dried for 2 to 3 min, in the other compartment (left empty). Pseudomonas protegens strain CHA0 was used as a positive control, and its isogenic cyanide mutant CHA77 and LB were used as negative controls. After 1 day, filter papers were checked. A blue color indicated cyanide production. Ammonia was detected using the same split petri dish assay (LB on one side and nothing on the other) and a commercially available colorimetric reaction kit (MQuant Ammonium [NH₄⁺] test; Merck, Darmstadt, Germany).

Collection of volatiles and GC-MS analysis.

The volatiles of eight selected strains were collected and analyzed by gas chromatography-mass spectrometry (GC-MS) using closed-loop-stripping analysis (CLSA) as described earlier (5). The strains as well as uninoculated medium, as a control, were pregrown at 30°C in 6 ml LB liquid medium for 24 h. LB agar plates were inoculated with 300 μl of the preculture, cultivated for 24 h at 30°C, and then analyzed by CLSA at room temperature as described in reference 26. For each strain, three biological replicates were analyzed. In this system, air is continuously pumped (MB-21E; Senior Flexonics, Bartlett, IL) through the closed system containing an activated charcoal filter (precision charcoal filter, 5 mg; Chromtech GmbH, Idstein, Germany) and the agar plate for 24 h. Trapped volatiles were extracted from the charcoal filter by rinsing the filter three times with 15 μl dichloromethane (≥99.8%; Merck, Germany). The headspace extracts were subsequently analyzed by GC-MS. Media were analyzed without inoculation as control. GC-MS analyses were performed on a HP7890A GC connected to a HP5975C mass selective detector fitted with an HP-5ms fused silica capillary column (30 m; 0.22-mm inside diameter [i.d.]; 0.25-μm film; Agilent Technologies, USA). Conditions were as follows: inlet pressure, 67 kPa; He, 23.3 ml/min; injection volume, 1 μl; transfer line, 300°C; injector, 250°C; electron energy, 70 eV. The gas chromatograph was programmed as follows: 5 min at 50°C, then increasing 5°C/min to 320°C. Linear retention indices were determined from a homologous series of n-alkanes (C₈ to C₃₂). Compounds were identified by comparison of mass spectra to database spectra (from the Wiley Registry of Mass Spectral Data [7th ed.], NIST MS Library [2008 edition], and our own database created from synthesized reference compounds) and by comparison of the retention index data to standards (our own database and NIST Chemistry WebBook [http://webbook.nist.gov/chemistry; accessed 2013]).

Effect of 1-undecene on P. infestans mycelial growth and sporulation.

The effect of pure 1-undecene and undecane on mycelial growth and sporangium formation of P. infestans was assessed as follows. Five-millimeter agar plugs from a growing mycelium were placed facing downward in the center of new RA plates. Silicone septa (5 mm, GR-2; Supelco) were soaked with definite quantities of the test compounds and placed in the center of the petri dish lid. Plates were sealed with Parafilm M and stored upside down in the dark at 18°C for 10 days. Mycelial growth was monitored by taking photographs and further analyzed using ImageJ. At the end of the incubation period, the plates were opened and sporangia were collected through the half cross-section of the plate by applying clear adhesive tape to the mycelium. Samples were directly mounted on glass slides, and snapshots were taken under the microscope every 3 mm from the center of the plate throughout the cross-section. Total numbers of closed and open sporangia were assessed using ImageJ. To assess the effect of pure 1-undecene on zoospore release, sporangia were collected from 1-week-old mycelial plates and pipetted onto 30-μl 0.2% agar drops containing various concentrations of 1-undecene. After 24 h of incubation at 4°C to trigger zoospore release from sporangia, closed, open, and germinating sporangia were counted. To investigate the effect of 1-undecene on direct sporangium germination (rather than zoospore release), the sporangium suspensions were vortexed for 30 s and amended with definite amounts of 1-undecene. They were then incubated for 30 min at room temperature with gentle agitation. The suspensions were then vortexed for another 30 s before being plated on 1.5% water-agar. The number of closed, open, and germinating sporangia were counted after 24 h incubation at 20°C.

Data analysis.

Data were analyzed using the GraphPad Quikcalcs tools (http://www.graphpad.com/quickcalcs/), GraphPad Prism 5 software, and Microsoft Excel software.

Nucleotide sequence accession numbers.

The sequences determined in this study have been deposited in GenBank under accession numbers KP067092 to KP067180 for 16S sequences and KP067181 to KP067198 for rpoD sequences.

RESULTS

Isolates with volatile-mediated antagonistic activity.

Isolation of morphologically distinct bacteria from potato plants on three different media yielded 137 strains, of which 92 could be phylogenetically identified to the genus or species level. A first screen of their putative antifungal activity yielded 32 isolates of interest (see Table S1 in the supplemental material). Pseudomonas strains were the most frequent among the identified isolates, and this genus also contained the highest proportion of active isolates (Table 1). In contrast, only two of 18 identified Arthrobacter and one of 12 identified Microbacterium strains showed antagonistic activity. These 32 active candidate strains were selected for further investigation to compare their volatile-mediated effects on five different potato pathogens.

TABLE 1.

Taxonomic identity and antagonistic activity of 92 isolated strains^a

Phylum	Genus	No. of strains in:
		Rhizosphere		Phyllosphere
		Total	Active	Total	Active
Actinobacteria	Agromyces	0	0	1	1
	Arthrobacter	9	2	9	0
	Curtobacterium	0	0	4	2
	Frigoribacterium	0	0	2	0
	Microbacterium	6	1	6	0
	Nocardioides	1	0	0	0
	Plantibacter	1	0	3	0
	Rathayibacter	0	0	1	0
	Rhodococcus	2	1	0	0
	Streptomyces	3	0	1	1
Bacteroidetes	Flavobacterium	1	1	0	0
Firmicutes	Bacillus	7	3	1	0
	Sporosarcina	1	1	0	0
Proteobacteria	Acidovorax	1	0	0	0
	Citrobacter	1	0	0	0
	Enterobacter	2	0	0	0
	Janthinobacterium	1	0	0	0
	Methylobacterium	0	0	1	0
	Pseudomonas	14	10	10	9
	Rahnella	0	0	1	0
	Variovorax	2	0	0	0
Total		52	19	40	13

^aFifty-two bacterial strains isolated from the rhizosphere and 40 bacterial strains isolated from the phyllosphere of field-grown potato plants were identified to the genus level by sequencing of the 16S rRNA gene. A first screen for mycelial growth inhibition of P. infestans, R. solani, and B. cinerea enabled the selection of 32 candidate antagonistic strains (see Table S1 in the supplemental material). Strains were considered active when they reduced mycelial growth of at least one target pathogen by more than 25%.

Susceptibility to bacterial volatile and diffusible compounds varied greatly between the different targets, F. oxysporum being generally the least affected and P. infestans the most susceptible target organism (Table 2). Strikingly, P. infestans was the only target whose growth was completely arrested when it was exposed to the volatiles of the five most active strains, all belonging to the genus Pseudomonas. Transferring mycelial plugs exposed to the volatiles of the Pseudomonas strain R47 to fresh plates did not allow the oomycete to resume growth, suggesting a lethal effect (data not shown). In addition to these producers of lethal volatiles, other, overall less active strains, such as the Pseudomonas strains S35 and R75, the Arthrobacter strain S27, and the Flavobacterium strain R96, emitted volatiles that caused highly significant mycelial growth reduction in P. infestans without affecting the other targets. In contrast, susceptibility to the volatiles of the Bacillus strain R29 was shared with H. solani. The growth of the latter fungus was specifically inhibited by the volatiles of a few members of the Actinobacteria that did not inhibit any other target, e.g., Streptomyces strain S01, Curtobacterium strain S46, and Microbacterium strain R42. The fast-growing species R. solani was in general little affected by the volatiles of the isolated strains: even the most strongly inhibiting strain, S04, still allowed the fungus to grow to almost 60% of its normal growth (Table 2). F. oxysporum proved extremely resistant to bacterial volatiles, with hardly any significant growth inhibition upon exposure to any of the 32 strains. The phytopathogenic bacterium D. dianthicola was highly inhibited by the eight most active Pseudomonas strains (Table 2).

TABLE 2.

Volatile-mediated activity of 32 strains isolated from rhizosphere (R) or phyllosphere (S) of field-grown potato plants^a

Strain	Phylogenetic affiliation	Growth (% of control)					HCN emission	NH3 concn (mg liter−1)
		P. infestans	H. solani	R. solani	F. oxysporum	D. dianthicola
R32	P. vranovensis	0	53	76	88	41	+	55
R84	P. marginalis	0	40	87	97	37	+	ND
R01	P. moraviensis	0	33	92	100	39	+	ND
R47	P. chlororaphis	0	59	96	99	38	+	35
R82	P. marginalis	0	71	73	102	53	+	ND
R76	P. fluorescens	13	50	69	98	70	ND	55
S50	P. moraviensis	3	39	109	98	52	+	ND
S49	P. fluorescens	3	55	104	101	56	+	ND
R29	Bacillus sp.	32	39	91	101	99	ND	ND
S04	P. frederiksbergensis	58	77	59	101	77	ND	205
S24	P. frederiksbergensis	21	70	86	99	100	ND	105
R74	P. frederiksbergensis	35	79	77	99	105	ND	355
S35	P. marginalis	30	72	90	97	113	ND	ND
R02	P. veronii	61	93	82	96	91	ND	ND
S06	P. frederiksbergensis	49	81	77	98	121	ND	ND
R75	P. frederiksbergensis	15	73	91	96	151	ND	35
S27	Arthrobacter sp.	50	86	95	97	106	ND	35
R96	Flavobacterium sp.	57	90	98	101	93	ND	255
R31	Sporosarcina sp.	82	56	97	101	106	ND	ND
R60	Arthrobacter sp.	75	59	98	103	113	ND	ND
R95	P. lini	87	60	92	99	121	ND	55
R42	Microbacterium sp.	77	39	102	101	145	ND	ND
S25	Curtobacterium sp.	96	100	82	101	91	ND	ND
S19	P. frederiksbergensis	62	63	83	98	164	ND	ND
R61	Arthrobacter sp.	66	72	115	100	123	ND	15
S22	P. syringae	86	42	110	101	136	ND	ND
R54	Bacillus sp.	64	85	108	98	123	ND	ND
S34	P. jessenii	76	87	98	97	121	ND	ND
S01	Streptomyces sp.	98	65	95	101	133	ND	205
R73	Bacillus sp.	79	80	109	101	128	ND	35
S46	Curtobacterium sp.	108	62	103	103	128	ND	ND
R85	Rhodococcus sp	75	89	96	99	181	ND	ND
Cerall	P. chlororaphis	2	69	92	99	51	+	nt
Cha0	P. protegens	2	50	76	95	42	+	ND
Cha77	P. protegens	68	80	77	93	200	ND	ND

^aStrains were tested against the oomycete P. infestans, the fungi H. solani, R. solani, and F. oxysporum, and the phytopathogenic bacterium D. dianthicola. Growth values are averages for 3 ot 4 replicates, expressed as a percentage of the control value. Bold values are significantly different from the control according to Student's t test (see Table S3 in the supplemental material for standard errors and significance levels). ND, not detected; nt, not tested. Strains are ordered according to their overall activity (most active first). Reference Pseudomonas strains (P. chlororaphis MA 342 [Cerall] and P. protegens CHA0 and its isogenic HCN-deficient mutant CHA77) are included for comparison.

Emission of hydrogen cyanide and of other bioactive volatiles.

Among the five target organisms tested, the oomycete and the bacterium responded more strongly to the volatile blends of the eight most active Pseudomonas strains than the three fungi (Table 2), suggesting greater sensitivity to the emitted volatiles. The capacity to emit the respiratory toxin hydrogen cyanide (HCN) was assessed in all strains and detected in seven of the eight most active Pseudomonas strains that triggered the highest inhibitory effects on both P. infestans and D. dianthicola. Comparing the effects of the volatiles from P. protegens CHA0 and from its isogenic cyanide-deficient mutant (CHA77) confirmed that inhibition of D. dianthicola was mostly, if not entirely, due to cyanide emission, given the growth promotion obtained upon exposure to the volatiles of CHA77 (Table 2).

In contrast, P. infestans was still significantly inhibited by the volatiles of the cyanide-deficient mutant, suggesting the emission of other volatiles with antioomycete potential. Emission of NH₃ was verified in the strains, and no correlation between the strains' antioomycete activity and their NH₃ production could be found (Table 2), suggesting that other compounds were responsible for the antioomycete activity of noncyanogenic strains. In order to elucidate the chemical nature of these compounds, the volatiles of eight Pseudomonas strains were collected and analyzed by GC-MS (Fig. 1; Table 3). These strains were chosen according to their origin of isolation (four from the rhizosphere and four from the phyllosphere) and to their global volatile-mediated activity, including highly active strains (R01, R47, and R76), moderately active ones (R02, S04, and S24), and practically inactive ones (S22 and S34) (Table 2). The volatile-mediated effects of these same eight Pseudomonas strains on P. infestans are depicted in Fig. 2. Of these strains, two produced HCN (R01 and R47), while the others did not. When the chromatograms of the eight selected strains were compared, a striking difference appeared between rhizosphere and phyllosphere strains, namely, the emission of massive amounts of 1-undecene in all four rhizosphere isolates (Fig. 1). These isolates were members of different species, as revealed by rpoD sequencing (see Table S3 in the supplemental material): Pseudomonas moraviensis (R1), Pseudomonas veronii (R2), P. chlororaphis (R47), and P. fluorescens (R76). In the phyllosphere isolates, smaller amounts of this compound were retrieved from the headspace of both strains identified as Pseudomonas frederiksbergensis (S04 and S24), while it was undetected in Pseudomonas syringae (S22) and in Pseudomonas jessenii (S34) (Table 3; Fig. 1). Few volatiles were commonly found in all investigated strains, such as the sulfur compounds dimethyl disulfide (most abundant in the two P. frederiksbergensis strains) and dimethyl trisulfide. 2-Acetylfurane was ubiquitous, while a group of nitrogen-containing compounds were found only in the headspace of P. veronii (R02), a rhizosphere strain of moderately antagonistic activity (Fig. 1 and 2; Table 3). The volatile profile of P. syringae S22 clearly differed from the seven others analyzed. This strain emitted less alkenes than the others, but a collection of ketones was detected in its headspace. Volatiles emitted mostly by strains showing antioomycete activity and less by those showing less antioomycete activity revealed dimethyl disulfide, 1-decene, 1-undecene, 1-dodecene, and an undecadiene with an unknown location of double bonds as putative mediators of antioomycete activity (Fig. 1; Table 3).

FIG 1.

Overlays of total ion chromatograms of four rhizosphere (R) and four phyllosphere (S) Pseudomonas strains. Headspace sampling and GC-MS analysis were performed in triplicate, and one representative example is shown. The list of detected volatiles can be found in Table 3. The chemical structures of selected volatiles are shown. a, dimethyl disulfide; b, 1-decene; c, 1-undecene; d, 2-nonanone; e, 1-dodecene.

TABLE 3.

Volatile compounds identified in the headspace of R01, R02, R47, R76, S04, S22, S24, and S34

RT (min)a	Compoundb	m/z	I (exp.)c	I (lit.)d	Relative amt ine:
					R01	R02	R47	R76	S04	S22	S24	S34
2.83	Dimethyl disulfide	94, 79, 45	774	777	x	x	x	xx	xxx	x	xxx	x
4.81	4-Hydroxy-4-methyl-2-pentanone	43, 59, 101, 39	844	842 (DB-5)		o				o
5.22	2-Furanmethanol	39, 41, 98, 81	857	857	o		xx	o	o		o
6.96	2,5-Dimethylpyrazine (A)	108, 42, 39	913	914	xx	o	x	o	x	o	x	x
7.10	2-Acetylfuran	95, 110, 38	916	916	o	o	x	o	o	o	o	o
8.18	Valine methyl ester	72, 55, 88	946				o
8.76	Benzaldehyde	106, 77, 51	962	961		o				x		xxx
8.79	4-Methyl-4-butanolide	42, 56, 39, 100	963		o				o
9.13	Dimethyl trisulfide	126, 79, 45, 64	971	972	o	xx	o	x	xx	x	x	xx
9.83	1-Decene	41, 55, 39	991	993	x			x
9.84	S-Methyl butanethioate	43, 118, 75, 61	992							x
10.22	Trimethylpyrazine (A)	42, 122, 39, 81	1,002	1,001	o	o	x	o	o	o	o	o
10.82	2-Acetylthiazole	43, 99, 58, 127	1,020	1,021	o	o			o		o
11.40	Benzyl alcohol	79, 77, 108, 107	1,037	1,037						o
12.40	S-Methyl methanethiosulfonate	45, 47, 63, 81	1,067	1,068			o	o		o		o
12.43	Acetophenone	77, 105, 51, 120	1,067	1,065	o	o	o		o	o		o
13.11	Tetramethylpyrazine (A)	54, 136, 42, 39	1,087	1,086			o		o	x	o	o
13.14	Undecadiene	41, 54, 67, 39	1,088		x	x		x			o
13.28	1-Undecene	41, 55, 43, 56	1,092	1,092	xxx	xxx	xxx	xxx	x		o
13.33	2-Nonanone	43, 58, 41, 57	1,094	1,093						x		x
13.38	Methyl benzoate	105, 77, 51, 136	1,095	1,094						x
13.52	Methyl (E)-3-(methylthio)-2-propenoate	101, 45, 73, 58, 132	1,099							x
14.31	Methyl (methylthio)methyl disulfide	61, 45, 140	1,124	1,123		o		o	o		o
15.03	Unknown S-containing compound	57, 41, 120, 92	1,148							x
16.06	Naphthalene	128, 102, 50	1,181	1,182						o
16.40	1-Dodecene	41, 55, 43, 56	1,192	1,192	x	o		x
16.47	2-Decanone	43, 58, 71	1,194	1,194								o
17.01	Dimethyl tetrasulfide	79, 45, 158	1,213	1,215					x
18.17	Unknown S-containing compound	43, 71, 41, 134	1,254							x
18.86	Tridecadiene	41, 67, 55, 81	1,278					o
19.34	2-Undecanone	43, 58, 41	1,294	1,294								o
19.87	1,2-Epoxyundecane	41, 55, 71, 43	1,314	1,307		o
23.50	Geranylacetone (A)	43, 41, 69, 67	1,454	1,455						o
23.96	Unknown	43, 67, 41, 54	1,472							o	o
24.38	Unknown N-containing compound	41, 97, 55, 43	1,490			o	o
25.36	Dihydroactinidiolide	111, 109, 43	1,531	1,537						o
26.83	Unknown N-containing compound	41, 55, 43, 97	1,592			o
27.51	Diphenylamine (A)	169, 51, 65	1,624	1,622 (DB-5)	o	o		o		o	o	o
28.66	Unknown N-containing compound	41, 55, 122, 136	1,675			x
28.70	Pentadec-8-en-2-one	43, 41, 55, 71	1,677							x
29.13	Unknown N-containing compound	41, 43, 97, 55	1,696			x
29.17	2-Pentadecanone	43, 58, 41	1,698	1,698						x
31.27	Unknown N-containing compound	41, 43, 55, 97	1,796			x
32.23	Hexahydrofarnesylacetone	43, 58, 41, 55	1,845	1,845						x
32. 95	Heptadecen-2-one	43, 55, 41, 71	1,880							x
32.98	Unknown N-containing compound	41, 55, 69, 122	1,881			xx
33.36	2-Heptadecanone	43, 58, 41, 71	1,901	1,901						xx
33.40	Unknown N-containing compound	41, 43, 55, 97	1,902			xx
36.85	Unknown N-containing compound	55, 41, 69, 43	2,085			o

^aRT, retention time.

^bCompounds were identified based on comparison of mass spectrum to a database spectrum, comparison of the retention index to a published retention index on the same or similar GC fused silica capillary column or comparison to a synthetic or commercially available reference compound. A, artifact (most likely a medium constituent).

^cexp., experimental.

^dlit., from the literature. Values were taken from the NIST Chemistry WebBook (http://webbook.nist.gov/chemistry; accessed 2014) or our own database.

^eAmounts are reported as being 0 to 1% (o), 1 to 10 % (x), 10 to 30 % (xx), and 30 to 100% (xxx) of the largest peak area in the total ion chromatogram.

FIG 2.

Volatile-mediated effects of four rhizosphere (R) and four phyllosphere (S) Pseudomonas strains on mycelial growth of P. infestans. Values are expressed as percentages of the value for the nonexposed control (100%). Significant differences from the control (according to Student's t test; n = 3 or 4) are indicated (*, P < 0.05; **, P < 0.01; ***, P < 0.001). Representative pictures are shown; a picture of the nonexposed control (Ctrl) is shown in the middle between R strains (left) and S strains (right).

1-Undecene inhibits the mycelial growth of P. infestans and changes its sporulation behavior.

Since 1-undecene was produced in very high quantities in all four rhizosphere strains and was also detected in the two phyllosphere strains with moderate but significant antioomycete activity (Fig. 2), we investigated whether this volatile might be involved in the observed mycelium growth inhibition of P. infestans. When different quantities of 1-undecene and, for a comparison, of undecane (which was not detected in the strains' headspace) were added to the P. infestans medium, a dose-dependent mycelium inhibition was observed for 1-undecene which was much stronger than that obtained with its reduced form, undecane (Fig. 3a). Moreover, the total amount of sporangia was significantly reduced in a concentration-dependent manner in 1-undecene treatments compared to the control treatment, as was the proportion of open sporangia, which had released the zoospores (Fig. 3b and c). Moreover, native (untreated) sporangia spotted on water agar drops containing increasing quantities of 1-undecene revealed that this molecule directly impacted zoospore release as well as direct germination of the sporangia (Fig. 3d). When applied in large amounts (from 0.75 mg on), the proportion of open sporangia was drastically reduced and the direct germination almost completely inhibited.

FIG 3.

Growth and sporulation of P. infestans in the presence of volatile 1-undecene. (a) Mycelium growth in the presence of different quantities of 1-undecene (squares) or undecane (diamonds) applied as volatiles. Data points represent the average surface mycelium growth of four to six replicates, with standard deviation bars included. (b) Absolute number of closed and open sporangia at 3 mm from the mycelium origin under volatile 1-undecene exposure. Results are representative of biological duplicates and are expressed as means ± standard errors of the means (SEM). (c) Representative examples of closed sporangia (c) containing zoospores (z), open sporangia (o), and directly germinating sporangia (g) with the germ tube (gt). Bar = 10 μm. (d) Relative numbers of closed, open, and germinating sporangia (right, zoospore release; left, direct germination). Results shown are representative of biological triplicates and are expressed as means ± SEM. Asterisks indicate statistical significance according to one-way Analysis of variance (ANOVA) followed by Dunnett's post hoc test compared to control treatments (n = 9; P > 0.001). n.s., P > 0.05; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

DISCUSSION

Many fungal disease-causing agents have been evaluated for their reaction to bacterial volatiles, which are often highly target specific (summarized in reference 12). However, oomycetes have largely been ignored in those studies, although they include pathogens causing major losses in global crop production (22, 23). One example of such a pathogen is the late blight-causing agent P. infestans, which was responsible for the Great Famine of Ireland in the middle of the 19th century and is the most important potato disease worldwide. In an attempt to analyze the yet-unknown response of this pathogen to bacterial volatiles, we first isolated bacterial strains from an environment where they would have been likely to encounter the pathogen, namely, from the phyllosphere and rhizosphere of field-grown, late blight-infected potato plants. Since previous reports indicated emission of antifungal volatiles from Actinobacteria (27–29), we used, in addition to broad-spectrum media, an Actinobacteria-specific medium. We thus isolated many Actinobacteria strains (Table 1), but only a few of them (mostly Arthrobacter strains) induced high volatile-mediated inhibition of P. infestans or of the other targets, while Pseudomonas strains proved the most efficient producers of growth-inhibiting volatiles under our cultivation conditions (Table 2).

When the overall susceptibilities of our different targets to bacterial volatiles were compared, P. infestans was clearly the organism responding most strongly, followed by the silver scurf-causing fungus H. solani and the black leg-inducing bacterium D. dianthicola. This confirms earlier findings indicating high sensitivity of Pythium species or of other Phytophthora species to bacterial volatiles (21, 30). A substantial difference between fungi and oomycetes is the structure of their cell walls (chitin for fungi and cellulose for oomycetes). Although this has not yet been assessed, it is tempting to speculate that the cell wall composition and the structure of oomycetes would make them more permeable to gaseous substances than that of fungi and that this might at least partly account for the higher sensitivity of oomycetes to bacterial volatiles.

Most of our highly active isolates were cyanogenic Pseudomonas strains. The inhibitory potential of this respiratory toxin is not new, and neither is its involvement in the biocontrol efficiency of cyanogenic Pseudomonas strains (19, 31). However, different organisms might show different tolerance levels to hydrogen cyanide, as observed in this study: P. infestans and D. dianthicola reacted most strongly to volatiles from cyanogenic strains, while hydrogen cyanide showed lesser toxicity to the fungi tested and especially to R. solani and F. oxysporum, for which no difference in inhibition was observed between the model strains P. protegens CHA0 and its isogenic cyanide mutant CHA77 (Table 2).

Interestingly, our results show that noncyanogenic Pseudomonas also induced significant volatile-mediated growth reduction in P. infestans, suggesting the presence of other bioactive volatiles. The volatile profiles of the four rhizosphere Pseudomonas strains analyzed were dominated by the alkene 1-undecene, while this compound was undetected or present only in small quantities in the phyllosphere strains. In a meta-analysis of 31 studies on volatiles emitted by six common pathogenic bacteria, 1-undecene was one of the molecules that differentiated Pseudomonas aeruginosa from the other, non-Pseudomonas pathogenic bacteria (32). Within the genus Pseudomonas, it seems that the relative contribution of undecene to the volatile bouquet of the strains is highly variable, ranging from the unchallenged dominance observed in our rhizosphere strains or in other antagonistic P. fluorescens strains (33) to the absence of detection for some strains, such as P. syringae S22 and P. jessenii S34. Other studies have reported 1-undecene from the headspace of Pseudomonas strains (7, 34–36) but also from other species, including members of the genera Burkholderia, Bacillus, and Serratia (7, 37).

1-Undecene seemed a good candidate to explain the growth-inhibiting effects observed upon exposure of P. infestans to noncyanogenic strains, since it was emitted in large amounts and was mostly detected in the strains showing antioomycete activity. In addition to mycelial growth inhibition, the formation of sporangia and their zoospore release were also significantly impaired by exposure to 1-undecene. This is of particular interest considering the epidemiology of P. infestans: the pathogen undergoes multiple asexual reproductive cycles during a potato cropping season and the motile zoospores released from the sporangia are a key element in the infection of new leaf tissues and tubers. Preventing their release might thus have more significant consequences on the disease spread and on the harvest quantity and quality than solely reducing the oomycete's vegetative growth. Further studies will reveal to what extent 1-undecene is produced when the strains' growth conditions are closer to the field situation and what the actual potential in terms of potato protection is.

Very few studies have analyzed the volatile emission by strains grown directly on or in their host (33) or at least on substrates mimicking the strains' natural environment (7, 21). Overall, the results of these studies were encouraging, since they demonstrated that (i) the active volatile (dimethyl disulfide) detected in the headspace of the antagonist strain grown in in vitro cultures could be detected in vivo as well, i.e., when the strain was growing inside the plants (33), and (ii) that bacteria were able to maintain significant levels of activity even when supplied with very low levels of nutrients (7, 21).

Our isolation of bacteria from the rhizosphere and phyllosphere of potato yielded a collection of strains displaying a high potential for volatile-mediated inhibition of P. infestans. The present study shows that the oomycete is highly susceptible to the volatiles of cyanogenic and noncyanogenic strains and that the massively emitted alkene 1-undecene partly accounts for volatile-mediated growth inhibition of the late blight-causing agent. Further studies could on the one hand elucidate the mechanisms by which bioactive volatiles such as 1-undecene impede the oomycete's growth and on the other hand evaluate the potential of volatile-emitting antagonists for potato protection in experimental setups that are closer to the field situation.