Characterization of bacterial isolates from rotting potato tuber tissue showing antagonism to Dickeya sp. biovar 3 in vitro and in planta

Authors

  • R. Czajkowski,

    1. Plant Research International, PO Box 16, 6700 AA Wageningen
    2. Netherlands Institute of Ecology, Netherlands Royal Academy of Art and Science (NIOO-KNAW), Boterhoeksestraat 48, 6666 GA Heteren
    Search for more papers by this author
  • W. J. de Boer,

    1. Plant Research International, PO Box 16, 6700 AA Wageningen
    Search for more papers by this author
  • J. A. van Veen,

    1. Netherlands Institute of Ecology, Netherlands Royal Academy of Art and Science (NIOO-KNAW), Boterhoeksestraat 48, 6666 GA Heteren
    2. Institute of Biology, University of Leiden, Sylviusweg 72, 2333 BE Leiden, The Netherlands
    Search for more papers by this author
  • J. M. van der Wolf

    Corresponding author
    1. Plant Research International, PO Box 16, 6700 AA Wageningen
    Search for more papers by this author

E-mail: Jan.vanderWolf@wur.nl

Abstract

Possibilities for biocontrol of biovar 3 Dickeya sp. in potato were investigated, using bacteria from rotting potato tissue isolated by dilution plating on nonselective agar media. In a plate assay, 649 isolates were screened for antibiosis against Dickeya sp. IPO2222 and for the production of siderophores. Forty-one isolates (6·4%) produced antibiotics and 112 isolates (17·3%) produced siderophores. A selection of 41 antibiotic-producing isolates and 41 siderophore-producing isolates were tested in a potato slice assay for control of the Dickeya sp. Isolates able to reduce rotting of potato tuber tissue by at least 50% of the control were selected. Isolates were characterized by 16S rDNA analysis as Bacillus, Pseudomonas, Rhodococcus, Serratia, Obesumbacterium and Lysinibacillus genera. Twenty-three isolates belonging to different species and genera, 13 producing antibiotics and 10 producing siderophores, were further characterized by testing acyl-homoserine lactone (AHL) production, quorum quenching, motility, biosurfactant production, growth at low (4·0) and high (10·0) pH, growth at 10°C under aerobic and anaerobic conditions and auxin production. In replicated greenhouse experiments, four selected antagonists based on the in vitro tests were tested in planta using wounded or intact minitubers of cv. Kondor subsequently inoculated by vacuum infiltration with an antagonist and a GFP (green fluorescent protein)-tagged biovar 3 Dickeya sp. strain. A potato endophyte A30, characterized as S. plymuthica, protected potato plants by reducing blackleg development by 100% and colonization of stems by Dickeya sp. by 97%. The potential use of S. plymuthica A30 for the biocontrol of Dickeya sp. is discussed.

Introduction

Soft rot and blackleg diseases of potato caused by pectinolytic bacteria belonging to Pectobacterium and Dickeya spp. are a continuous threat to (seed) potato production worldwide. Potato plants and tubers are affected by the presence of pectinolytic bacterial pathogens in virtually all phases of tuber production including storage. In western and northern Europe, Dickeya spp. in particular cause increasing economic losses in seed potato production (http://www.eppo.org/QUARANTINE/bacteria/Erwinia_chrysanthemi/ERWICH_ds.pdf). Dickeya spp. cause pre-emergence seed piece decay, blackleg and aerial stem rot, and soft rot of progeny tubers (Pérombelon, 2002). Increased losses in seed potato production in Europe resulting from Dickeya spp. infection are related to the recent occurrence and spread of a new Dickeya sp. genetic clade belonging to biovar 3 which probably constitutes a new Dickeya species (Tsror et al., 2008).

Potato (seed) tubers are the primary source of soft rot and blackleg inoculum (Pérombelon, 1974). Production of pathogen-free seed tuber lots or eradication of the existing bacteria in seed lots is consequently of major interest. Selection for blackleg and soft rot resistant potato cultivars has not resulted in seed lots completely free from Pectobacterium and Dickeya spp. bacteria. Soft rot and blackleg control is therefore presently based on an integrated strategy which includes the use of pathogen-free initial propagation material and practices which avoid the wounding of tubers and plants, the occurrence of oxygen depletion by free water on tuber surfaces and smearing and dissemination of the pathogens within or between seed lots by mechanical equipment (Pérombelon, 2002; Czajkowski et al., 2011).

Simultaneously, the use of physical (e.g. hot water treatment) and chemical disinfection procedures for controlling blackleg and soft rot have resulted in a reduction, but not in an elimination, of the bacterial inoculum (Pérombelon & Salmond, 1995). During chemical and/or physical sterilization, bacteria present on or in the periderm may be destroyed, but not those present inside the tubers. It has already been reported that the highest densities of pectinolytic Pectobacterium and Dickeya spp. are located in the vascular system at the stolon end of tubers (Czajkowski et al., 2009). Inoculum present inside tuber tissue will not be affected by superficial sterilization procedures.

Bacterial antagonists can be considered as an alternative to chemical and physical control strategies to reduce Dickeya spp. populations in plant vascular tissue and in tubers during storage. However, successful commercially available biocontrol agents have only been developed for a limited number of bacterial phytopathogens. Agrobacterium radiobacter K84 was developed to control A. tumefaciens, the causative agent of crown gall (Vicedo et al., 1993), and a product based on Erwinia herbicola (Pantoea agglomerans) successfully controlled E. amylovora, responsible for fire blight in apple and pear (Stockwell et al., 1998).

Several publications describe promising results for biocontrol of bacterial pathogens in potato. For Clavibacter michiganensis subsp. sepedonicus, the causal agent of bacterial ring rot, reasonable levels of control have been obtained with Arthrobacter histidinolovorans and Serratia sp. in field experiments. Both strains reduced tuber infections by 35–70% and infection in plants by 60–80% (Gamard & De Boer, 1995). Wilt symptoms caused by Ralstonia solanacearum could be reduced by up to 83% and 73% using Bacillus subtilis and Pseudomonas fluorescens strains, respectively (Aliye et al., 2008). Streptomyces scabies, causing common scab, was antagonized by a Bacillus sp. strain in pot assay experiments and reduced infections in potato tubers by up to 35% (Han et al., 2005). Fluorescent Pseudomonas spp., applied to tubers, reduced populations of blackleg and soft rot bacteria on potato roots and inside progeny tubers. They were also effective in controlling post-harvest soft rot on potato (Kloepper, 1983). Finally Bacillus subtilis, producing antibiotics active against a broad spectrum of bacteria, showed a high level of control towards Pectobacterium spp. in vitro and on potato tubers (Sharga & Lyon, 1998). However, none of the bacterial biocontrol agents has been used in commercial applications.

Only limited attempts have been made for the biocontrol of Dickeya spp. Various Gram-positive and Gram-negative bacteria isolated from different hosts and environments, producing antibiotics against Dickeya spp. or siderophores competing with Dickeya spp. for iron, have been characterized (Kastelein et al., 1999; Jafra et al., 2006). Bacteria isolated from the potato tuber surface (Kastelein et al., 1999) or the potato rhizosphere (Jafra et al., 2006) proved useful in the protection of tuber tissue from P. atrosepticum and Dickeya spp. However, to date, these bacteria have not been tested for the control of Dickeya spp. in potato plants.

The purpose of this study was to select and characterize bacterial biocontrol agents against the recently described genetic clade of Dickeya sp. biovar 3 in potato. Bacteria were isolated from rotten potato tuber tissue to gain isolates able to compete with Dickeya spp. in this harsh environment. Selection of antagonists involved antibiotic production, the ability to compete for iron with the pathogen by the production of siderophores, the production of (antibacterial) surfactants, the ability to interfere with quorum sensing and the ability to grow under different environmental conditions. The selection also included greenhouse experiments with treated tubers.

Materials and methods

Bacterial isolates and media used for cultivation

Bacterial isolates used in this study are listed in Table 1. Tryptic soya agar (TSA) (Oxoid), King’s B agar (Fluka), nutrient agar (NA) (Difco) and R2A (Difco) agar media supplemented with 200 μg mL−1 of cycloheximide (Sigma) to prevent fungal growth were used to isolate bacteria from rotting tuber tissue. NA or TSA and/or nutrient broth (NB) (Difco) were used for bacteria maintenance. Bacteria were grown on or in media for 24–48 h at 28°C. Growth of bacteria in liquid media was done by shaking at 200 rpm during incubation. Dickeya sp. biovar 3 IPO2222 (Tsror et al., 2008) was grown on TSA or in NB and Dickeya sp. IPO2254 pPROBE-AT-gfp (Czajkowski et al., 2010) was grown on TSA or in NB supplemented with 150 μg mL−1 of ampicillin.

Table 1.   Characterization of 23 bacterial isolates from rotting potato tubers able to reduce tuber decay caused by Dickeya sp. on the basis of motility, AHL production and degradation, production of auxin, siderophores and biosurfactants, antibiosis, pectinolysis and growth in a medium at pH 4·0 and 10·0 and at 10°C, under aerobic and anaerobic conditions
IsolateaOriginMotilitycAHL productiondAHL degradationeAuxinfPectinolysisgSiderophorehAntibiosisiBiosurfactantjGrowth in NBkGrowth at 10°Cl
Potato cultivarRotting typebCVO26/JB534HHLOHHLOHLOOHL−Trp+TrpAt pH 4·0At pH 10·0AnaerobicAerobic
  1. aBacterial isolates from rotting potato tissue; 12 isolates were used in a first greenhouse experiment (*) and four were used in a second (**). Only 23 isolates belonging to risk category 1(+) and protecting potato from maceration caused by Dickeya sp. IPO2222 in a potato slice assay were characterized in detail.

  2. bPotato tubers were rotted in loamy sand, a rich soil collected in the region of Wageningen, NL (S1) (Garbeva et al., 2003), or in potting compost (S2) or wrapped in plastic film. Tubers were kept for 5–10 days under low-oxygen conditions and 90% relative humidity at 28°C in closed humid boxes.

  3. cMotility was tested in a motility agar assay. −: colony diameter <5 mm, +: colony diameter 5–15 mm, ++: 16–30 mm, +++: 31–50 mm, ++++: >50 mm.

  4. dAHL production was tested in a bioassay with two indicator strains C. violaceum (CVO26) and E. coli (JB534). −: isolate does not produce AHLs, +: isolate produces AHLs.

  5. eAHL degradation was determined using an E. coli reporter strain responding to the presence of AHLs by production of a green fluorescent protein. Degradation was determined for HHL (hexanoyl-l-homoserine lactone), OHHL (3-oxo-hexanoyl-l-homoserine lactone), OHL (octanoyl-l-homoserine lactone) and OOHL (3-oxo-octanoyl-l-homoserine lactone). +: degradation of the tested signal molecule, −: lack of degradation, ND: activity not determined.

  6. fAuxin concentrations were spectrophotometrically determined in media supplemented with l-tryptophan (+Trp) (auxin precursor) or without (−Trp). −: no production of auxins, ±: concentration 0·1–0·4 μg mL−1, +: 0·5–1·0 μg mL−1, ++: 1·1–2·0 μg mL−1, +++: 2·1–5·0 μg mL−1, ++++: 5·1–15·0 μg mL−1, +++++: >15·1 μg mL−1.

  7. gPectinolysis was determined on a polygalacturonic minimal medium on which pectinolysis results in a white halo. −: no halo, +: halo diameter 0·1–10 mm, ++: 11–15 mm, +++: 16–25 mm.

  8. hSiderophore production was tested in a CAS agar plate assay, in which siderophore production resulted in a pink halo. −: no halo, +: halo.

  9. iAntibiosis was tested in vitro in an overlay assay with Dickeya sp. IPO2222 as indicator strain. −: no inhibition, +: inhibition.

  10. jBiosurfactant production was tested using an oil spreading method in which surfactants enlarge the diameter of the oil droplet. −: no enlargement, +: diameter of the droplet 1–6 mm, ++: 7–15 mm, +++: 16–30 mm, ++++ >30 mm.

  11. kGrowth in NB broth at pH 4·0 and 10·0 by comparing the optical density (OD600) at T = 0 and T = 48 h. −: difference in OD600 < 0·050, +: difference in OD600 > 2.

  12. lGrowth at 10°C under aerobic and anaerobic conditions tested in liquid PDB broth, comparing the optical density (OD600) at T = 0 and T = 120 h. −: difference in OD600 < 0·050, +: difference in OD600 between 2 and 5 times, ++: difference in OD600 between 5 and 10 times, +++: difference in OD600 ≥ 10 times.

A3* (Lysinibacillus sphaericus)KondorS1++−/+NDNDNDND++++++++++++++++
A6* (Bacillus simplex)Kondor (minitubers)S1−/−+++++++++++
A10* (Rhodococcus erythropolis)KondorS2++++−/+NDNDNDND±+++++++±+++
A12* (B. subtilis)AgriaFilm−/−++++±+++++++
A13* (Pseudomonas brassicacearum)ArcadeS1++++−/−++++±+++++
A17 (B. subtilis)ArcadeFilm−/−++++++++++
A19** (B. simplex)KonsulS1−/−+++++++++++
A20 (B. simplex)ArcadeFilm−/−+++++++
A21 (B. subtilis)ArcadeFilm++−/−±++++++
A23 (Serratia plymuthica)ArcadeFilm++++−/+NDNDNDND±++++++++++++
A30** (S. plymuthica)ArcadeFilm++++−/+NDNDNDND±+++++++++++++++
A34 (S. plymuthica)ArcadeFilm++++−/+NDNDNDND±++++++++++++++
A36 (B. subtilis)AgriaFilm−/−+++++++++++++
S3* (P. putida)AgriaFilm++++−/+NDNDNDND+++++++
S9* (Obesumbacterium proteus)ArcadeS1++++−/+NDNDNDND+++++++++++
S10 (P. putida)KonsulS1++++−/+NDNDNDND+++++++++++
S20** (P. putida)KonsulS2++++−/−++++++++++++
S21 (P. putida)KonsulS1++++−/+NDNDNDND++++++++++
S23* (P. fulva)AgriaFilm+++++/+NDNDNDND++++++++++++++++
S26 (P. putida)ArcadeFilm++++−/−++++++++++++++++++
S31** (P. fulva)AgriaS2++++−/+NDNDNDND++++++++++
S37 (P. putida)ArcadeFilm++++−/−+++++++++
S38 (P. putida)ArcadeFilm++++−/−++++++++++++

Rotting of potato tubers and isolation of potato endophytes

Dickeya spp.-free minitubers of cv. Kondor, and seed tubers of cvs Arcade, Agria, Konsul and Kondor were used to isolate bacteria from rotting potato tissue. To isolate bacterial endophytes indigenously present in potato tubers, tubers were sterilized by immersion in 70% ethanol twice for 10 min each time followed by washing in running tap water. Surface-sterilized tubers were firstly mechanically wounded and subsequently enclosed in plastic film and kept for 5–10 days at 28°C. Isolation of bacteria able to colonize rotting tuber tissue from soil was done using tubers buried either in loamy sand, a rich soil collected in the region of Wageningen, NL (S1) (Garbeva et al., 2003), or in potting compost (S2) (basis potgrond nr. 4, Hortmea Group) and kept for 5–10 days under low-oxygen conditions obtained by immersion of pots in water to 80% soil saturation capacity and 90% relative humidity at 28°C in closed humid boxes. Rotting tubers in soil were washed with tap water before processing. Approximately 20 g of the rotting potato tissue, collected if possible from the inside of each tuber, was resuspended in 50 mL of a quarter strength (1/4) Ringer’s buffer (Merck) supplemented with 0·02% diethyl-dithiocarbamic acid (DIECA) (Acros Organics) as an antioxidant. Rotting tuber suspensions were incubated at room temperature for 1 h with shaking (200 rpm). Serial dilutions of the rotting tissue extracts were made in a Ringer’s buffer and 100 μL of 10−4, 10−5 and 10−6 dilutions were plated on King’s B, R2A, TSA and NA plates supplemented with 200 μg mL−1 cycloheximide. Plates were incubated at 28°C for 24–48 h. Single bacterial colonies of different morphologies were collected from plates inoculated with material from different potato tubers and different treatments for further analysis. Bacteria were grown to pure cultures on TSA or NA and stored at −80°C in NB/glycerol (60/40%) stocks.

Pre-screening of candidate antagonists against Dickeya sp. IPO2222

Pure cultures of collected isolates were tested for the ability to produce antibiotics against Dickeya sp. IPO2222 or siderophores. In each case, two independent replications were made for each tested isolate and the experiment was independently repeated once. Results from both experiments were averaged.

Antibiotic production assay

Production of antibiotics was tested in an overlay assay with Dickeya sp. IPO2222 as indicator strain. Two hundred microlitres of an overnight culture of Dickeya sp. (approximately 109 CFU mL−1) in NB was mixed with 20 mL of soft top agar (NB supplemented with 0·7% agar) at 45–50°C and poured onto square (15 × 15 cm) TSA plates. Once the agar had solidified, 2·5 μL of an overnight culture of the antagonist grown in NB (approximately 109 CFU mL−1) was spotted on the surface of the agar plate. Plates were incubated for 24–48 h at 28°C. Isolates inhibiting growth of Dickeya sp. IPO2222 and causing a clear ‘halo’ around their colonies were collected for further analysis.

Siderophore production assay

The availability of iron ions is crucial during plant infection. In an iron-poor environment the virulence of Dickeya spp. is seriously reduced (Expert, 1999). The presence of bacteria able to produce strong siderophores and sequester iron from the environment will in theory reduce disease development caused by Dickeya sp.

A CAS plate assay was used to evaluate siderophore production by test isolates (Schwyn & Neilands, 1997). Two and a half microlitres of overnight isolate culture in NB (approximately 109 CFU mL−1) was spotted on the surface of the CAS agar plate. Plates were incubated for 24–48 h at 28°C. Isolates producing an orange/pink halo around colonies, indicating siderophore production, were collected for further analysis.

16S rDNA sequence analysis

To identify the bacterial isolates, a 16S rDNA fragment between 968 and 1401 bp (numbering based on the Escherichia coli genome) was sequenced with primers F968 (5′-AACGCGAAGAACCTTAC-3′) and R1401 (5′-CGGTGTGTACAAGGCCCGGGAACG-3′) in both directions. A colony PCR procedure was used to amplify DNA using the same primers as used for sequencing. Individual colonies were collected from TSA plates using a sterile toothpick and suspended in 50 μL of 5 mm NaOH (Sigma). Suspensions were boiled at 95°C for 5 min and kept at 4°C. One microlitre of the cell lysate was used per PCR reaction. PCR products were purified with the PCR purification kit (QIAGEN) according to the manufacturer’s protocol. Sequencing reactions were performed using the Big Dye Terminator Cycle Sequencing kit (Applied Biosystems). DNA sequences were compared with available sequences deposited in GenBank (http://www.ncbi.nlm.nih.gov/Genbank) using nucleotide-nucleotide Basic Local Alignment Search Tool (blast) for nucleotides (blastn) alignments accessed from the website (http://www.ncbi.nlm.nih.gov/BLAST/).

Determination of species in risk groups

The DSMZ (German Collection of Microorganisms and Cell Cultures; http://www.dsmz.de/) database was used to classify bacterial isolates into risk categories on the basis of their ability to cause disease in humans (http://www.absa.org/riskgroups/index.html).

Screening for antagonism in a potato slice assay

A potato slice assay was used to evaluate the ability of selected isolates to reduce potato tissue maceration by Dickeya sp. IPO2222. Bacteria were centrifuged (5 min, 6000 g) and washed twice with 1/4 Ringer’s buffer. Densities of test isolates were adjusted to approximately 108 CFU mL−1 and Dickeya sp. to approximately 106 CFU mL−1 with sterile water. Potato tubers of cv. Agria (Agrico) were rinsed with tap water, surface sterilized with 70% ethanol for 10 min, rinsed with tap water again and dried with tissue paper. Tubers were cut into 0·7 cm transverse slices using a sterile knife. Three wells (5 × 5 × 5 mm) per slice were made using a sterile cork borer and were filled with a 50 μL suspension containing 108 CFU mL−1 of test isolate and 106 CFU mL−1 of Dickeya sp. For each test isolate, three potato slices derived from three different tubers were used. The negative control was 50 μL of sterile water instead of bacterial suspension. The effect of test isolates on potato tissue was measured by comparing the ratio of the average diameter of rotting potato tissue around wells inoculated with Dickeya sp. and an antagonist, with the average diameter of rotting potato tissue around wells of the positive control (Dickeya sp. only). Two replications were made for each tested isolate and the experiment was independently repeated once. Results from both experiments were averaged.

Characterization of selected isolates for features potentially involved in antagonism

Isolates showing at least 50% decrease of rotting ability of Dickeya sp. in the potato slice assay were screened for motility, AHL (acyl homoserine lactone) production and degradation, auxin production with and without supplementation of l-tryptophan, biosurfactant production, growth at pH 4·0 and 10·0 and at 10°C in anaerobic and aerobic conditions and for pectinolytic enzyme production. For each assay, two independent replications were made for each tested isolate and the experiment was independently repeated once. The results from both experiments were averaged.

Motility assay

Motility is an important feature in active colonization of plant surfaces (e.g. roots, leaves), migration in soil and in establishment of high bacterial populations on plants (Andersen et al., 2003). Motile antagonistic bacteria will have an advantage in competition with motile Dickeya spp. in the plant and soil environment.

The motility of selected isolates was tested using a motility agar assay (NB supplemented with 0·3% agar) (S. Jafra, University of Gdansk, Gdansk, Poland, personal communication). Briefly, 2·5 μL of an overnight culture of the test isolate (109 CFU mL−1) grown in NB and diluted 10 times in NB (approximately 108 CFU mL−1) was spotted on the surface of the motility agar plates and incubated for 24 h at 28°C. The colony diameter after incubation time was measured. Bacterial motility was assessed using the following indexation based on the diameter of the colony: −: ≤5 mm, +: 6–15 mm, ++: 16–30 mm, +++: 31–50 mm, ++++: ≥51 mm.

Production and degradation of AHLs

The AHL based quorum sensing mechanism is involved in regulation of different physiological processes in bacteria including production of pathogenicity factors. The ability to interfere with this mechanism by degradation of signal molecules (so-called quorum quenching) proved useful in the biocontrol of plant pathogenic bacteria (Uroz et al., 2003). First, the AHL production in the selected isolates was tested, as the bioassays employed in this study for testing AHL degradation cannot be used if an isolate produces its own AHLs. All the isolates able to produce their own signal molecules were excluded from the AHL degradation assay.

Production of AHLs was tested using two AHL reporter strains, Chromobacterium violaceum CVO26 and Escherichia coli JB534 (Andersen et al., 2001). These strains produce a purple pigment (CVO26) or GFP (JB534) in the presence of exogenous AHLs. Overnight cultures of the reporter strains grown on agar plates were streaked on the surface of TSA plates in two lines at a distance of 3 cm using a sterile 1 μL inoculation loop. Thereafter, 2·5 μL of an overnight culture of the test isolate grown in NB was spotted at 0·5 cm distance from reporter strain lines. Prepared plates were incubated at 28°C for 24–48 h and visually inspected for the presence of purple pigment (CVO26) and for GFP fluorescence (JB534). The following indexation was used: +/+: isolates positive for both reporter strains, +/−: isolates positive for one reporter strain, and −/−: isolates negative for both reporter strains.

To test the ability of bacterial isolates to degrade AHLs, a fast screening method was used with AHL reporter strain Escherichia coli JB534. AHL degradation was studied for four synthetic signal molecules: hexonoyl-l-homoserine lactone (HHL), 3-oxo-hexonoyl-l-homoserine lactone (OHHL), octanoyl-l-homoserine lactone (OHL) and 3-oxo-octanoyl-l-homoserine lactone (OOHL) (Sigma). Fifty microlitres of 10 μg mL−1 of the individual AHL and 20 μL of the test isolate (109 CFU mL−1 in NB) was added to 100 μL of NB buffered to pH 6·4 with 1 m MOPS (3-(N-morpholino)-propane-sulphonic acid) (Sigma) in the well of a 96-well microtitre plate. Plates were incubated for 4 h at 28°C. Bacteria were killed under UV light for 30 min and wells were filled with 100 μL of a suspension of E. coli JB534 (108 CFU mL−1 in NB). Plates were then incubated at 37°C for 12 h and screened for GFP fluorescence using the Fluoroscan Ascent FL (Thermo Fisher Scientific) with the filter sets 495 nm (excitation) and 530 nm (emission).

Auxin production

Antagonists were tested for their ability to produce auxins. Microbial plant hormones may contribute to plant development and fitness (Leveau, 2010). To screen for l-tryptophan (l-Trp)-independent or l-Trp-dependent auxin production, test isolates were grown overnight in NB or in NB supplemented with 0·5%l-tryptophan (Sigma) in shaken cultures, respectively. Cultures were centrifuged (8000 g, 10 min) and 3 mL of the clear supernatant was collected per sample. Auxin production was measured as IAA (indole-3-acetic acid) equivalents: 3 mL of bacterial supernatant were mixed with 2 mL of freshly prepared Salkowski reagent (2 mL of 0·5 m FeCl3 + 98 mL 35% HClO4). A standard curve of IAA was made by preparing solutions of 130 μg mL−1 IAA (Sigma) in NB. As a negative control, E. coli DH5α (Invitrogen), which does not produce auxins, was used. For colour development, solutions were left at room temperature for 30 min. The intensity of the colour was measured spectrophotometrically at 535 nm wavelength. The following indexation was used: −: isolates not producing auxins, ±: auxin concentration 0·1–0·4 μg mL−1, +: 0·5–1·0 μg mL−1, ++: 1·1–2·0 μg mL−1, +++: 2·1–5·0 μg mL−1, ++++: 5·1–15·0 μg mL−1, and +++++: >15·0 μg mL−1 of auxins.

Biosurfactant production

Biosurfactants decreasing the surface tension act as dispersants or detergents. In the biocontrol, the main mode of surfactant action is a destabilization of the cell membranes and lysis of the cells. Biosurfactants can also induce plant systemic resistance towards plant pathogens (Tran et al., 2007).

Biosurfactant production was investigated by using a modified oil spreading method (Youssef et al., 2004) (S. Jafra, University of Gdansk, Gdansk, Poland, personal communication). Test isolates were grown overnight in NB at 28°C. Cultures were centrifuged for 20 min at 8000 g to remove the bacteria. A 40 μL drop of mineral oil (BioRad) was spotted on the surface of 30 mL demineralized water and 5 μL volume of the supernatant or sterilized water (control) was placed on top of the oil drop. The diameter of the circle created by oil is proportional to the biosurfactant concentration. The diameter of the circle was measured for each isolate and compared with the control. The following indexation was used: −: no biosurfactant production, +: diameter of circle 1–6 mm, ++: 7–15 mm, +++: 16–30 mm, ++++: >30 mm.

Growth under different conditions

For field application of a biocontrol agent it is important that the isolate can survive under different environmental conditions. Potatoes are cultivated in a wide variety of soils, including sand, peat and clay, and soils that differ in pH, organic matter and salt content. Tests were conducted to see if the antagonists could persist under different pH and temperature conditions.

Growth at pH 4·0 and 10·0  To test the growth of isolates at low and high pH, 100 μL of a suspension of 108 CFU mL−1 of the test isolate in NB was added to 20 mL of NB buffered to pH 4·0 or 10·0 with 1 m MOPS (Sigma) or to 20 mL of NB, pH 7·0 (control) (S. Atkinson, Institute of Infection, Immunity and Inflammation, Nottingham, UK, personal communication). Shaken cultures (200 rpm) were incubated for 48 h at 28°C. The turbidity of the bacterial cultures was inspected by eye and measured spectrophotometrically at 600 nm wavelength at the start (T = 0) and end (T = 48 h) of the experiment. The following indexation was used: −: no growth (increase of OD600 < 0·05), +: growth (increase of OD600 ≥ 2).

Growth at 10°C under anaerobic and aerobic conditions  To test the growth of isolates at 10°C under anaerobic and aerobic conditions, 100 μL of 108 CFU mL−1 of bacteria in NB was added to 10 mL of potato dextrose broth (PDB) (Difco) buffered to pH 7·0 with 1 m MOPS. For anaerobic incubation, 5 mL of liquid paraffin (Sigma) was added on the top of the inoculated medium. Tubes were kept at 10°C for 3–5 days and inspected by eye daily for turbidity. On day 5, turbidity of the cultures was measured spectrophotometrically at 600 nm wavelength. The experiment was repeated once and the results were averaged. The growth of the tested isolates was evaluated by comparing the optical density (OD600) at T = 0 and T = 120 h. The indexation used was: −: increase of OD600 < 0·05, +: increase of OD600 between 2 and 5 times, ++: increase of OD600 between 5 and 10 times, +++: increase of OD600 > 10 times.

Production of pectinolytic enzymes

Antagonists should preferably not produce pectinolytic enzymes to avoid the risk of enhanced tuber decay and pre-emergence rot. Pectinolytic activity of the test isolates was determined on polygalacturonic minimal medium (PMM) (per L: 3 g KH2PO4, 7 g K2HPO4, 2 g (NH4)2SO4, 4 g polygalacturonic acid (PGA), 15 g agar; pH 7·0. Dickeya sp. IPO2222 was used as a positive control. Two and a half microlitres of the test isolate (109 CFU mL−1 in NB) was spotted on the surface of PMM. Plates were incubated for 24–48 h at 28°C. After this time, plates were washed with 10% copper acetate solution in water and left for 15 min at room temperature for halo development. The diameter of the white halo around the bacterial colony is proportional to the concentration of pectinolytic enzymes. The following indexation was used, based on the diameter of the halo: −: no halo, +: 0·1–10 mm, ++: 11–15 mm, +++: 16–25 mm.

Inoculation of potato tubers and growth of potato plants

In the first greenhouse experiment, in May and June 2010, 12 selected antagonists were tested for their ability to protect potato plants from infection by Dickeya sp. IPO2254 and disease development. Suspensions of test isolates and Dickeya sp. IPO2254 were prepared in sterile demineralized water to achieve densities of 1011 CFU mL−1 (antagonist) and 106 CFU mL−1 (Dickeya sp. IPO2254). Certified Dickeya spp.-free minitubers of cv. Kondor [Dutch Plant Inspection Service for agricultural seeds and seed potatoes (NAK)] were used. Half of the tubers were wounded by removing 0·5 cm of the stolon end with a sterile knife and the others left intact. Both types of minitubers were immersed in the antagonist suspension and vacuum infiltrated for 10 min at −800 mBar in a desiccator followed by 10 min incubation in the antagonist suspension at atmospheric pressure to allow the bacteria to penetrate the tuber lenticels and wounds. For controls, intact and wounded minitubers were vacuum infiltrated with sterile demineralized water. Tubers were dried in a flow cabinet overnight and the next day were vacuum infiltrated with Dickeya sp. IPO2254 suspension or, for the control, with sterile demineralized water and dried overnight in a flow cabinet. Tubers were planted in potting compost in 5 L plastic pots and kept at a 16/8 h photoperiod, 80% relative humidity (RH) and 28°C for 3 weeks in a greenhouse. To eliminate the bias effect of the environmental conditions, a random plot design of the pots was applied (eight blocks containing five pots from each individual isolate). In each treatment, 20 intact and 20 wounded Kondor minitubers were used per test isolate. In the subsequent second greenhouse experiment in June and July 2010, the four most promising candidates from the first greenhouse experiment were tested, using the same design and under the same conditions as in the first experiment.

Symptom development

Plants were visually inspected weekly for symptom development and were evaluated for wilting, chlorosis of leaves, black rot at the stem base, aerial stem rot, haulm desiccation and whole plant death.

Sampling of potato plants derived from vacuum infiltrated tubers

Potato plants were sampled 25 days after tuber planting. Approximately 1·5 cm long stem segments taken 5 cm above ground level were collected per plant and pooled. Stems were then surface sterilized for 1 min in 70% ethanol, washed once with tap water, disinfected in 1% sodium hypochlorite (commercial bleach) for 3 min, washed three times with tap water and dried with tissue paper. Samples were weighed and 1/4 Ringer’s buffer containing 0·02% of diethyl-dithiocarbamic acid was added to twice the weight of the sample.

Pour plating of plant extracts

Samples were crushed in Universal Bioreba Bags (BIOREBA AG) using a hammer. One hundred microlitres of undiluted and 10-fold dilutions (0, 10−1 and 10−2) of plant extract were mixed with 300 μL of pre-warmed to 45–50°C PT medium (per L: 5 g polygalacturonic acid (Sigma), 0·5 g tryptone, 1 g NaNO3, 4 g K2HPO4, 0·2 g MgSO4.7H2O, 8 g agar, 0·1 mL Tween 20 (Sigma), 17 mL 1 m NaOH, pH 7·0) containing 200 μg mL−1 of cycloheximide and 150 μg mL−1 of ampicillin in the well of a 24-well plate (Greiner Bio-One B.V.). After the medium had solidified, plates were incubated for 24–48 h at 28°C to allow the bacteria to grow. Wells were inspected for the presence of GFP-tagged Dickeya sp. IPO2254 under 495 nm blue light using an epifluorescence stereo microscope (Leica Wild M32 FL4) equipped with a mercury high pressure photo-optic lamp (Leica Hg 50W/AC) and GFP plus filter (Czajkowski et al., 2010). GFP positive bacterial colonies were counted.

Statistical analysis

Greenhouse experiments were done according to a completely randomized block design. In the experiment with 12 selected isolates, 40 plants (20 with wounded and 20 with intact tubers) were used per treatment divided over eight complete blocks containing five plants each. In the experiment with a subselection of four isolates, 40 plants (20 with wounded and 20 with intact tubers) were used per treatment divided over five complete blocks containing six plants and one complete block containing five plants. Observations were as follows: (i) incidence of tuber sprouting or not, (ii) level of emerged plants with blackleg symptoms and (iii) measurements of plant height and weight. Sprouting and the incidence of blackleg symptoms were analysed using a standard generalized linear mixed model (GLMM) with the binomial distribution and a logit link to map the discrete outcome onto the real line (Cramer, 2004). A normal distribution was assumed for plant height and weight. Effects were considered to be significant at  0·05 and pair-wise differences were obtained using the t-test. Data were analysed using the statistical software package GenStat (Payne et al., 2009).

Results

Collection of candidate bacterial antagonists

In total, 649 isolates were collected from tubers belonging to different potato cultivars and from various agar media (NB, R2A, King’s B and TSA): 289 isolates from tubers rotting in soil type I (S1), 143 isolates from tubers rotting in soil type II (S2) and 217 isolates from tubers rotting in plastic film. Bacteria with different colony morphologies were selected. In a first screening, all isolates (649) were tested for antibiosis against Dickeya sp. IPO2222 in an overlay plate assay and for production of siderophores. Forty-one isolates (6·3%) produced antibiotic factors against Dickeya sp. IPO2222 whereas 112 isolates (17·3%) produced siderophores, of which 41 isolates (6·3%) produced a large orange/pink halo on the CAS agar plates (strong siderophore producers) (data not shown). None of the isolates produced both antibiotics and siderophores. Forty-one strong siderophore producers together with the 41 isolates producing antibiotics were selected for further studies.

Suppression of soft rot development on potato slices

Eighty-two selected isolates (41 antibiotic and 41 strong siderophore producers) were tested in a potato slice assay for their ability to reduce tuber decay caused by Dickeya sp. IPO2222. Thirty-two antibiotic producers (from 41 isolated: 78%) and 41 strong siderophore producers in CAS agar assay (from 112 isolated: 37%) were able to reduce the tuber rot to at least 50% of the control, i.e. potato slices inoculated only with Dickeya sp. IPO2222 (data not shown).

Identification of potential antagonists

Identification of the 82 isolates was achieved by partially sequencing their 16S rDNA. Alignment with sequences deposited in GenBank resulted in a classification of isolates into the following genera: Alcaliges, Lysinibacillus, Aeromonas, Bacillus, Proteus, Haemophilus, Rhodococcus, Pseudomonas, Serratia, Acinetobacter, Enterobacter, Obesumbacterium, Stenotrophomonas, Pantoea and Raoultella (data not shown). Alignments showed similarities of the 16S rDNA sequences of test isolates and reference species in the GenBank database between 95% and 100% (data not shown). Fifty-nine out of 82 tested isolates (72%) were classified as risk category 2 species (human, animal or plant pathogens) and were therefore excluded from further studies.

Characterization of features involved in antagonism

Twenty-three isolates classified as risk category 1 or 1+ bacteria belonging to Lysinibacillus sphaericus, Bacillus simplex, Rhodococcus erythropolis, B. subtilis, Pseudomonas brassicacearum, Serratia plymuthica, P. putida, Obesumbacterium proteus and P. fulva species, that did not pose a risk for human and animal health and were not plant pathogens, were characterized in detail (Table 1). Isolates were investigated for the ability to produce quorum sensing signal molecules (AHLs), biosurfactants, pectinolytic enzymes and auxins, for growth at low and high pH, for growth under anaerobic and aerobic conditions at 10°C and for motility and AHL degradation. No situation was observed in which a particular isolate reacted differently in repeated experiments. Average values of each test were calculated and distributed into categories (Table 1).

AHL production

Eleven out of 23 isolates (A3, A10, A23, A30, A34, S3, S9, S10, S21, S23 and S31) were able to produce AHLs in one or both bioassays. AHL producers belonged to the genera Lysinibacillus (A3), Rhodococcus (A10), Pseudomonas (S3, S10, S21, S23 and S31), Serratia (A23, A30 and A34) and Obesumbacterium species (S9). As expected, no AHL production was observed in the Gram-positive Bacillus spp. isolates (A17, A19, A20 and A36).

AHL degradation

Twelve isolates that were not producing AHLs were tested for their ability to degrade four synthetic signal molecules (HHL, OHL, OHHL or OOHL). Nine (A6, A12, A13, A19, A36, S20, S26, S37 and S38) degraded all four synthetic AHLs. These isolates were Pseudomonas and Bacillus species.

Biosurfactant production

Sixteen isolates produced biosurfactant in the in vitro assay (all isolates except A6, A13, A19, S3, S21, S37 and S38). The concentration of the surfactant produced varied largely per isolate as indicated by the oil drop spread assay. The largest average diameter of the oil circle was produced by P. fulva, L. sphaericus, B. subtilis and P. putida (isolates A3, A17, S10, S23, S26 and S31) and the smallest by Rhodococcus erythropolis A10. Most isolates of P. putida and P. brassicacearum were negative for biosurfactant production.

Motility

The majority of test isolates were highly motile (17 isolates out of 23), especially those identified as Pseudomonas and Serratia species. Nonmotile isolates (A6, A12, A17, A19, A20 and A36) were all characterized as Bacillus species.

Production of pectinolytic enzymes

Five isolates (A3, A10, A12, A17 and A36) out of 23 produced recordable amounts of pectinolytic enzymes under in vitro conditions, characterized as L. sphaericus, R. erythropolis, B. subtilis, B. subtilis and B. subtilis, respectively. Isolates A3 and A10 showed the highest activity of pectinolytic enzymes in the in vitro plate assay.

Auxin production

All 23 isolates produced auxins in the absence of l-tryptophan; the supplementation of l-tryptophan increased auxin production in all isolates. The highest increase was observed for isolates A23, A30, A34, S23 and S26 characterized as S. plymuthica, P. putida or P. fulva.

Growth at low and high pH

None of the isolates except O. proteus were able to grow in NB at pH 4·0. Ten out of 23 isolates showed growth at pH 10·0, namely P. putida and P. fulva species (S21, S23, S26, S31, S37 and S38), S. plymuthica (A30 and A34), B. simplex (A19) and O. proteus (S9).

Growth under aerobic and anaerobic conditions

Under anaerobic conditions, seven isolates (A3, A6, A10, A23, A30, A34 and S23) and under aerobic conditions 18 isolates (all except A12, A17, A20, A21 and S37) grew at 10°C in PDB.

Greenhouse studies with vacuum infiltrated minitubers

In the first greenhouse experiment 12 bacterial isolates (A3, A6, A10, A12, A13, A19, A30, S3, S9, S20, S23 and S31) belonging to nine different species (B. simplex, B. subtilis, R. erythropolis, P. brassicacearum, S. plymuthica, P. putida, P. fulva, O. proteus and L. sphaericus) were tested for protection of potato plants against biovar 3, Dickeya sp. IPO2254. They were selected on the basis of results in in vitro assays and features that may play a role in antagonism, in order to maximize variation of strains and modes of antagonistic actions. Antagonists were tested for their ability to reduce pre-emergence rot of tubers, blackleg symptoms and colonization of stems by Dickeya sp. IPO2254. Wounded or intact minitubers of cv. Kondor were inoculated subsequently with an antagonist and GFP-tagged biovar 3 Dickeya sp. strain (Dickeya sp. IPO2254) by vacuum infiltration. Plants were screened for pre-emergence rot 15 days after planting, and weekly for symptom development. In a second greenhouse experiment, four antagonists (A19, A30, S20 and S31) which showed the best protection in the first greenhouse trial were used.

Non-emergence of potato plants

Non-emergence of subsequently-inoculated potato tubers and controls was evaluated 15 days after planting. The percentage of tubers that generated shoots was calculated (Fig. 1a). No significant differences between treatments with intact and wounded tubers were found (data not shown), therefore treatments were analysed together. Interestingly, inoculation with Dickeya sp. IPO2254 resulted in a relatively low percentage of non-emergence, averaging at 7%. Equal percentages of non-emergence were found for treatments with P. fulva S31 and B. simplex A19. Co-inoculations of Dickeya sp. with other antagonists resulted in a higher percentage of non-emergence, from on average 19% for S. plymuthica A30 to a maximum of 95% for R. erythropolis A10.

Figure 1.

 Effect of co-inoculation of seed tubers with Dickeya sp. IPO2254 and different antagonists on the average percentage of sprouted tubers (a), average fresh weight of shoots (b) and average height of the tallest shoot (c) carried out 15, 25 and 25 days after tuber co-inoculation, respectively. Control tubers were inoculated with water or only with Dickeya sp. IPO2254. Values followed by identical characters are not significantly different (= 0·05). (a) No significant difference between treatments containing intact and wounded tubers was observed therefore treatments were analysed together. (b, c) Intact tubers and cut tubers from which the stolon end part was removed (wounded) were used. Data from two independent experiments were analysed together.

Effect of antagonistic isolates on the fresh weight of potato shoots

Twenty-five days after planting, the average fresh weight of potato shoots per treatment was determined (Fig. 1b). For intact tubers, fresh plant weight was significantly decreased by all treatments except for tubers subsequently-inoculated with both P. fulva S31 and Dickeya sp. IPO2254, for which the weight of the shoots was similar to that of the water-inoculated control plants. For plants raised from wounded tubers, only a treatment with S. plymuthica A30 resulted in an improvement on the adverse effect of Dickeya sp. on shoot weight.

Effect of antagonistic isolates on the average length of the highest shoot

Twenty-five days after planting, the average length of the highest potato shoot per plant per treatment was determined for the sprouted tubers (Fig. 1c). For both intact and wounded tubers, all antagonists except R. erythropolis A10 at least partially increased shoot length relative to Dickeya sp. Treatments with the antagonists B. subtilis A12, P. brassicacearum A13 and S. plymuthica A30 resulted in an increase in shoot length of plants derived from both wounded and intact tubers relative to the control.

Effect of antagonistic isolates on blackleg incidence

Twenty-five days after planting, the average percentage of plants showing blackleg symptoms or internal discoloration and hollowing of stem tissue was determined (Fig. 2a). A significant difference between the results of intact and wounded tubers was not found in any of the treatments (data not shown), therefore results of both treatments were analysed together. Results of the treatment with joint inoculation with R. erythropolis A10 and Dickeya spp. were removed from the analysis, because of the high percentage of pre-emergence rot. All other antagonists reduced the percentage of plants with symptoms significantly. Treatment with B. simplex A19, S. plymuthica A30 and O. proteus S9 completely prevented symptom development. Co-inoculation of P. fulva S23 with Dickeya sp. IPO2254 resulted in 47% reduction of plants with symptoms, but for plants with the other bacterial co-inoculations, blackleg incidence was reduced by at least 52–96% in all other treatments.

Figure 2.

 Effect of co-inoculation of seed tubers with Dickeya sp. IPO2254 and antagonists on the average percentage of plants showing symptoms (blackleg or browning and rotting of internal stem tissue) 25 days after incubation. In experiment 1, 12 antagonists were tested (a) and in experiment 2 a subselection of four antagonists were tested (b). Control tubers were inoculated with water or only with Dickeya sp. IPO2254. No significant differences between treatments with intact and wounded tubers were observed in both experiments therefore treatments were pooled before analysis. Values followed by identical characters are not significantly different (= 0·05).

In a second greenhouse experiment, four antagonists were selected on the basis of results from the first experiment (i.e. average percentage of sprouted tubers, height and weight of shoots, reduction of blackleg symptoms and Dickeya sp. IPO2254 densities in stems), namely S. plymuthica A30, B. simplex A19, P. putida S20 and P. fulva S31. A significant difference in the results between intact and wounded tubers was not found in any of the treatments (data not shown), therefore data for both treatments were analysed jointly. Approximately 47% of control plants inoculated with Dickeya sp. IPO2254 expressed blackleg symptoms. All four antagonists reduced the blackleg incidence significantly. Co-inoculation of potato tubers with S. plymuthica A30, P. fulva S31, P. putida S20 or B. simplex A19, resulted in 0%, 7%, 12% and 17% diseased plants, respectively.

Effect of antagonistic isolates on Dickeya sp. populations in stems

Twenty-five days after planting, the percentage of plants harbouring GFP-tagged Dickeya sp. inside stems was determined by pour plating of stem extracts and screening for typical green fluorescent colonies (Fig. 3a,b) (Czajkowski et al., 2010). A significant difference in the results between intact and wounded tubers was not found in any of the treatments (data not shown), therefore results of both treatments were analysed together. All antagonists reduced infection incidence significantly relative to the Dickeya sp. control at 70%, from 45% for P. fulva S23 to 3% for S. plymuthica A30. Dickeya sp. IPO2254 was not present in any of the water-inoculated plants.

Figure 3.

 Effect of co-inoculation of seed tubers with Dickeya sp. IPO2254 and antagonists on the average percentage of plants harbouring Dickeya sp. IPO2254, 25 days after tuber co-inoculation. In experiment 1, 12 antagonists were tested (a) and in experiment 2 a subselection of four antagonists were tested (b). Control tubers were inoculated with water or only with Dickeya sp. IPO2254.Values followed by identical characters are not significantly different (= 0·05).

In a second greenhouse experiment, as above, results of both wounded and intact tuber treatments were analysed jointly. Twenty-five days after inoculation, approximately 47% of control plants inoculated with Dickeya sp. IPO2254 harboured the pathogen. In contrast, the four antagonists reduced infection incidence significantly, with 0%, 2%, 6% and 5% infected plants grown from potato tubers subsequently-inoculated with S. plymuthica A30, P. fulva S31, P. putida S20 or B. simplex A19, respectively.

Densities of Dickeya sp. IPO2254 in both experiments varied largely per treatment and plant screened. No statistically significant differences in results between intact and wounded tubers were found (data not shown). On average in both experiments, in plants inoculated with IPO2254, 25 dpi 103–104 CFU g−1 of stem tissue were present. Co-inoculation of tubers with A19, S20 and S31 antagonists resulted in a reduction of Dickeya sp. populations on average to 10, 100 and 100 CFU g−1 stem tissue, respectively (data not shown).

Discussion

This study was conducted to assess the potential of antagonistic bacteria isolated from rotting potato tubers to control a distinct genetic clade of Dickeya sp. biovar 3, which is currently one of the dominant potato pathogens in Europe causing blackleg.

Although bacterial antagonists against Dickeya spp. have been isolated previously, this study is the first dealing with antagonists isolated from rotting potato tuber tissue. This approach was taken to acquire bacteria able to grow in an environment in which Dickeya spp. are highly active. It is generally accepted that success of a biocontrol agent in controlling a pathogen depends on the occupation of the same niches, utilization of the same carbon and nitrogen sources or adaptation and multiplication in the same environmental conditions (Völksch & May, 2001). Therefore, potential antagonists were expected to be present in rotting potato tissue where they have to face high levels of antimicrobial metabolites and oxygen depletion rather than in healthy tuber tissue. It is known that rotting tubers contain high concentrations of different plant metabolites including ethanol, acetone, 2-butanone, acetalaldehyde, methyl acetate, ethyl acetate, propanol and butanol that show antimicrobial activity towards a variety of Gram-positive and Gram-negative bacteria (Maga, 1994). The high microbial activity in rotting tubers is also expected to result in high concentrations of antimicrobial metabolites produced by microorganisms, including antibiotics and siderophores.

Perhaps because of the few sources from which bacteria were isolated, only a limited number of bacterial species antagonistic to Dickeya spp. were found in rotting tissue. The initial selection based on cultivation and antibiotic and siderophore production resulted in bacteria belonging to only 18 genera. It was previously mentioned that potato tubers host a limited number of culturable bacterial species such as fluorescent and nonfluorescent Pseudomonas spp., Bacillus spp., Serratia spp. and Xanthomonas spp. (Sturz, 1995). To the authors’ knowledge, only very limited information is available on bacterial species present in rotting potato tissue, and no intensive study has been conducted until now.

The majority of the 82 selected bacterial isolates were classified as (opportunistic) human pathogens (i.e. Enterobacter cloace, Bacillus cereus, Alcaligens feacalis, Proteus vulgaris, Aeromonas salmonicida and Enterobacter cancerogenus). Tuber decay can result in the occurrence of high densities (103–105 CFU g−1) of these potential human pathogenic bacteria. They can spread over tubers during harvesting, grading and packaging and may pose a health risk for people and animals exposed to them. Protection in rotting tuber tissue may result in enhanced survival periods of these pathogens. Growth of (opportunistic) human pathogens in rotting plant material has been reported for other crops. For example, it was found that vegetables (e.g. beet, broccoli, cabbage, cucumber, carrot, pepper, radish, squash and tomato) infected with pectinolytic bacteria often harboured higher densities of Salmonella spp. than healthy plant tissues (Wells & Butterfield, 1997). Carlin et al. (1995) reported an increase in the density of Listeria monocytogenes in decaying endive leaf tissue in comparison with healthy tissues.

The bacteria isolated from rotting potato tissue possessed different mechanisms that could interfere with Dickeya spp. in planta such as competition for iron ions by production of siderophores, antibiosis which may be partially based on the production of biosurfactants, degradation of quorum sensing signal molecules and competition for nutrients (e.g. pectinolysis). These features have been reported previously to play a role in biocontrol (Uroz et al., 2003).

Selected isolates possessed various other features that potentially enhance their colonization, adaptation and survival in decaying potato tissue such as spore formation, motility and the ability to grow under anaerobic conditions. Oxygen depletion is one of the factors inducing rotting of potato tubers by pectinolytic bacteria, by attenuation of the plant defence mechanism (Pérombelon & Lowe, 1975). Antagonists that are able to grow under low oxygen conditions might have an advantage when competing with Dickeya spp. during tuber decay. Spore forming strains, such as Bacillus spp., are able to persist for long periods under harsh conditions as they are tolerant to heat and desiccation (Weller, 1988). Use of these strains may result in a longer protection of tubers compared to non-spore forming strains. Motility enhances colonization of (plant) surfaces, which may therefore be an advantage in competing with the motile Dickeya spp. both in plants and in soil.

The presence of high numbers of antagonistic bacterial species isolated from rotting potato tissue may explain partially the variation in symptom expression in plants homogeneously inoculated by vacuum infiltration with Dickeya sp. (J. M. van der Wolf, Plant Research International, unpublished results) and contribute to other factors affecting blackleg development (Pérombelon, 2002). The presence of high numbers of antagonistic bacteria may also explain false negative results in isolations from diseased tubers or plants or detection procedures based on enrichment for Dickeya spp. (Degefu et al., 2009).

The strategy of selecting antagonists, based on in vitro assays and a tuber slice test, appears to be successful in gaining antagonists effective in the biocontrol of biovar 3 Dickeya sp. in planta. From a subselection of 12 isolates, 10 were able to reduce the blackleg incidence by more than 50%. This is in contradiction with former studies in which in vitro assays had only a limited value in predicting the antagonist potential for in planta tests. However, it must be noted that co-inoculation of tubers with 11 out of 12 isolates and Dickeya sp. resulted in an increased incidence of non-emergence. In particular, co-inoculation with A3, A10 and A12, characterized as L. sphaericus, R. erythropolis and B. subtilis, respectively, resulted in a high level of pre-emergence rot, which may be related to their ability to produce pectinolytic enzymes.

One isolate, characterized as S. plymuthica (A30), provided a consistent high level of blackleg disease control in repeated greenhouse experiments, even under conditions very favourable for disease development. Serratia plymuthica A30 possesses different features that may play a role in antagonism of Dickeya sp. IPO2254; the isolate produces antibiotics and surfactants, it is motile and produces auxins. Protection was found after vacuum infiltration of intact tubers and tubers from which part of the stolon end had been mechanically removed, under warm (28°C) and humid (80% relative humidity) conditions favourable for blackleg disease development. Strains of S. plymuthica have been frequently used to control fungal pathogens of plants, but not to the authors’ knowledge to control phytopathogenic bacteria (De Vleesschauwer & Hofte, 2007). Potato plants, subsequently inoculated with S. plymuthica A30 and the pathogen, were protected against systemic colonization by the pathogen and consequently also against disease development. The tubers were vacuum infiltrated with a high inoculum dose (1010–1011 CFU mL−1) of the antagonist to ensure relatively high densities in the tuber periderm. Results indicate the ability of S. plymuthica A30 as an antagonist to control blackleg caused by biovar 3 Dickeya sp. on potato.

Serratia plymuthica A30 is potentially a good candidate for developing a commercial potato crop protection product. It is classified into risk group 1 according to DSMZ, meaning that the species is not expected to pose a risk for humans and environment, and to date no human or animal-related pathogenicity factors for S. plymuthica have been described. The isolate A30 does not produce prodigiosin, a red pigment (R. Czajkowski, Plant Research International, unpublished data), an antifungal and anti-eukaryotic compound that can be produced by S. marcescens (risk group 2); S. plymuthica A30 is also susceptible to a number of antibiotics routinely used in medicine to treat bacterial infections in humans (R. Czajkowski, Plant Research International, unpublished data); finally, a commercial product named Rhizostar (E-nema GmbH) based on S. plymuthica HRO-C48 and active against Rhizoctonia solani and Verticillum dahliae is presently available on the market in Europe (European patent 98124694.5), indicating the possibility of using strains belonging to S. plymuthica species as biocontrol agents in agriculture. Preliminary in vitro and potato slice experiments with isolate A30 and the commercially available S. plymuthica HRO-C48, have shown that S. plymuthica A30 is a more efficient inhibitor of biovar 3 Dickeya sp. IPO2222 and protects potato tuber tissue better against maceration by the pathogen (R. Czajkowski, Plant Research International, unpublished data).

To fully explore the usefulness of S. plymuthica A30 in practice, additional studies are required on effectiveness and consistency of control in the field, including population dynamics in the potato ecosystem, application timing, production and formulation and eco-toxicological risks. Work with deletion (knockout) mutants is now being conducted to understand the molecular basis of the antagonistic activity of S. plymuthica A30 against Dickeya sp. IPO2222.

Acknowledgements

This project was financed by STW Foundation (Technologiestichting STW, The Netherlands) via grant no. 10306 ‘Curing seed potatoes from blackleg causing bacteria’. We thank P. S. van der Zouwen (PRI, The Netherlands) for technical assistance, S. Jafra (University of Gdansk, Poland) for helpful discussion and M. C. M. Pérombelon (ex SCRI, UK) and Mrs L. J. Hyman (ex SCRI, UK) for their comments on the manuscript and their editorial work.

Ancillary