Gene-for-gene relationship in the host–pathogen system Malus × robusta 5–Erwinia amylovora



  • Fire blight is a destructive bacterial disease caused by Erwinia amylovora affecting plants in the family Rosaceae, including apple. Host resistance to fire blight is present mainly in accessions of Malus spp. and is thought to be quantitative in this pathosystem.
  • In this study we analyzed the importance of the E. amylovora effector avrRpt2EA, a homolog of Pseudomonas syringae avrRpt2, for resistance of Malus × robusta 5 (Mr5).
  • The deletion mutant E. amylovora Ea1189ΔavrRpt2EA was able to overcome the fire blight resistance of Mr5. One single nucleotide polymorphism (SNP), resulting in an exchange of cysteine to serine in the encoded protein, was detected in avrRpt2EA of several Erwinia strains differing in virulence to Mr5. E. amylovora strains encoding serine (S-allele) were able to overcome resistance of Mr5, whereas strains encoding cysteine (C-allele) were not. Allele specificity was also observed in a coexpression assay with Arabidopsis thaliana RIN4 in Nicotiana benthamiana. A homolog of RIN4 has been detected and isolated in Mr5.
  • These results suggest a system similar to the interaction of RPS2 from A. thaliana and AvrRpt2 from P. syringae with RIN4 as guard. Our data are suggestive of a gene-for-gene relationship for the host–pathogen system Mr5 and E. amylovora.


The bacterial disease fire blight is currently one of the most economically important plant diseases in pome fruit production worldwide. The disease is caused by the Gram-negative enterobacterium Erwinia amylovora, which was the first bacterium that has been identified as the causal agent of a plant disease (Winslow et al., 1920). E. amylovora was first observed in North America in the Hudson Valley of New York (Denning, 1794) and has since been decribed in more than 40 countries (Peil et al., 2009). The bacterium E. amylovora overwinters in cankers of infected trees and can be disseminated in spring of the next season by birds, insects, mites, spiders, humans, wind, water or mechanical equipment (Schroth et al., 1974). In plants, the bacteria cells are mostly localized in the xylem and intercellular spaces from where they disseminate downwards against the direction of the water flow (Bogs et al., 1998). Bacterial aggregation in the xylem causes a disruption of the vessel walls by changing the vessel pressure (Esau, 1965). Plugging of the vascular system by bacteria and capsular extracellular polysaccharide leads to wilting symptoms and necrosis of the plant tissue above the infection site (Van Alfen & Allard-Turner, 1979). Fire blight infections can cause severe economic losses in pome fruit production. In the United States, for example, the estimated annual costs as a result of fire blight infections are c. US$100 million (Norelli et al., 2003). In Switzerland, an amount of US$9 million of losses was reported between 1997 and 2000 (Hasler et al., 2002) and in Germany, losses of c. US$1.6 million were estimated for the Lake Constance region (southern Germany) after fire blight infections in 2007 (Scheer, 2009).

Disease management is possible with streptomycin or, less efficiently, with copper sprays; however, streptomycin resistance in E. amylovora occurs in the US and in other countries, including Canada, Israel, and New Zealand (McManus et al., 2002; McGhee et al., 2011). US apple growers spend c. US$2.8 million per year on antibiotic sprays (Gianessi et al., 2002). Streptomycin-containing products for fire blight control are not permitted in many European countries. Biological control is another possibility, but control can be variable and may not be effective in years with disease-conducive weather conditions (Johnson & Stockwell, 1998).

Planting of fire blight-resistant cultivars seems to be the most promising strategy, which is environmental and producer-friendly. Most of the apple cultivars in current production globally are highly susceptible to fire blight and thus justify resistance breeding to fire blight as a primary objective in many apple breeding programs. Donor genotypes for fire blight resistance in wild apple species have been described and a number of studies were performed with the aim of investigating the genetic basis of this trait. During the last decade, several quantitative trait loci (QTLs) for resistance to fire blight in different genetic backgrounds and in response to different strains of the pathogen were identified (Khan et al., 2012). For example, the QTL on linkage group 3 (LG 3) of Malus × robusta 5 (Mr5) is of particular interest for breeding (Gardiner et al., 2012). This QTL was stable during 14 yr of virulence screening, in different cross-bred populations and after inoculation with a number of different E. amylovora strains (Peil et al., 2007, 2008). In total, the QTL on LG 3 of Mr5 accounted for between 67 and 83% of the phenotypic variance, indicating the existence of one or a few major resistance genes in this genomic region. This assumption is supported by several publications, which described the existence of E. amylovora strains varying in virulence to Mr5 (Fazio et al., 2008) and strains overcoming the resistance of Mr5, (Norelli & Aldwinckle, 1986; Peil et al., 2011). Such diagnostic pathogen strains are very useful not only for future breeding programs with objectives of resistance gene pyramiding but also in studies of plant–pathogen interactions. Little is known about genes involved in the Mr5-E. amylovora host–pathogen interaction. Gardner et al. (1980) postulated a dominant resistant gene for Mr5 controlling resistance to fire blight and Peil et al. (2007) also assumed a major resistance gene.

Erwinia amylovora is known to encode a type III secretion system (T3SS), which secretes effector proteins such as DspA/E, Eop1, HopPtoCEA, HrpN, HrpW and others (Khan et al., 2012; Malnoy et al., 2012; McNally et al., 2012). One of these effectors is AvrRpt2EA, a homolog of the AvrRpt2 protein of Pseudomonas syringae (Zhao et al., 2006). AvrRpt2 activates the RPS2 resistance gene of Arabidopsis thaliana via the cleavage of RIN4, the guard of RPS2 (Axtell & Staskawicz, 2003; Mackey et al., 2003; Day et al., 2005).

Shoots of Mr5 were inoculated with a number of different E. amylovora wildtype strains and an avrRpt2EA mutant strain ZYRKD3-1 (Zhao et al., 2006) in order to investigate the role of AvrRpt2EA in the Mr5-E. amylovora host–pathogen interactions. Interestingly, four natural isolates and the avrRpt2EA mutant ZYRKD3-1 were found to overcome the resistance of Mr5, suggesting a contribution of avrRpt2EA to virulence. Sequencing of the avrRpt2EA gene of strains differing in virulence to Mr5 revealed a single nucleotide polymorphism (SNP), which resulted in a change in the amino acid sequence correlated with virulence. Artificial shoot inoculations with complemented mutant strains gave further evidence for the involvement and importance of the SNP in the avrRpt2EA gene in virulence to Mr5 and the probable existence of a gene-for-gene-relationship in the host–pathogen system Malus × robusta 5–E. amylovora.

Materials and Methods

Plant material

We used the following plant material for artificial shoot inoculations: 4- to 6-month-old single- and double-shooted branched plants of Malus × robusta 5 (Mr5), single-shooted Malus baccata jackii (accession no. MAL0419), Malus floribunda 821, Malus baccata (accession no. MAL0004), Malus fusca (accession no. MAL0045), Malus × domestica Borkh. cvs ‘Idared’, ‘Royal Gala’, ‘Prima’ and ‘Pinova’ and the advanced breeding clone 181 of the Apple Breeding Collection at the Institute for Breeding Research on Horticultural and Fruit Crops in Dresden, Saxony, Germany. Graft sticks of each genotype were cut and grafted onto Malling 9 (M9) rootstocks. We also used double-grafted plants. Mr5 was grafted onto ‘Idared’/M9 interstem/rootstock combinations as scion or as an interstem between ‘Idared’ (as scion) and M9 (as rootstock). All plants were grown in the glasshouse at temperatures between 10 and 15°C under a natural photoperiod with extension of day time in spring.

Bacterial strains and fire blight inoculation

Erwinia amylovora (Burrill) Winslow et al. wildtype strains, the avrRpt2EA mutant strain ZYRKD3-1 (Zhao et al., 2006) and the complemented mutant strains ZYRKD3-1 (pZYR2-415-S) and ZYRKD3-1 (pZYR2-415-C) used for artificial inoculation are listed in Table 1. E. amylovora strains were denoted according to our strain nomenclature system; in some cases, strains have been given multiple designations (Table 1). Each strain received from different suppliers was recorded separately; for example, strain CUCPB 265 received from France was denoted as Ea3049, whereas strain E2002A, synonym for CUCPB 265, delivered from USA was denoted as Ea395.

Table 1. Description of the Erwinia amylovora strains used in this study
Strains (alternative name)OriginReferenceAA156Virulence
  1. AA156 is the amino acid at position 156, *avrRpt2EA gene was sequenced; virulence was tested on Malus × robusta 5: +, virulent; −, not virulent.

Ea7Pyrus, Brandenburg, 1972H-J. SchaeferC*
Ea91Pyrus, Potsdam, 1985H-J. SchaeferC
Ea115Crataegus, Eisleben, 1989K. RichterC
Ea237Malus, Baden-Württemberg, 1994E. MoltmannC 
Ea250Malus, Langenweddingen, 1995D. BeymeC
Ea269Pyrus, Baden-Württemberg, 1996E. MoltmannC
Ea270Pyrus, Baden-Württemberg, 1996E. MoltmannC
Ea273Cydonia, Rieder, 1995K. RichterC
Ea401 (Ea1/79)Cotoneaster, 1979W. ZellerC* 
Ea402 (Ea7/74)Cotoneaster, 1974W. ZellerC* 
Ea627Pyrus, Dresden, 2003SMUL, DresdenC* 
Ea662Pyrus, Tundersleben (Magdeburg), 2003D. BeymeC* 
Ea763Pyrus, Baden-Württemberg, 2006E. MoltmannC* 
Ea782Crataegus, Quedlinburg, 2007K. RichterC* 
Ea797Malus, Baden-Württemberg, 2007E. MoltmannC 
Ea815Malus, Freising, 2008G. PoschenriederC* 
Ea839Pyrus, Baden-Württemberg, 2008E. MoltmannC* 
Ea898 (Ea1189) Malus Burse et al. (2004)C* 
Ea77 (CFBP 3050, CUCPB 266, E4001A, Ea396, Ea3050)Malus, OntarioW.G. BonnS+
Ea395 (CFBP 3049, CUCPB 265, E2002A, Ea3049)Malus, Ontario, obtained from USAW.G. BonnS+
Ea396 (CFBP 3050, CUCPB 266, E4001A, Ea77, Ea3050)Malus, Ontario, obtained from USAW.G. BonnS+
Ea3049 (CFBP 3049, CUCPB 265, E2002A, Ea395)Malus, Ontario, obtained from FranceW.G. BonnS* 
Ea3050 (CFBP 3050, CUCPB 266, E4001A, Ea77, Ea396)Malus, Ontario, obtained from FranceW.G. BonnS* 
United States
Ea78 (CFBP 3051, CUCPB 273, Ea3051)Malus, New York, 1980S.V. BeerC
Ea3051 (CFBP 3051, CUCPB 273, Ea78)Malus, obtained from France C* 
Ea110Malus, MichiganA.L. JonesS*+
Ea400 (Ea581)Malus, Kearneysville, 1998T. van der ZwetS+
PFB4 (INRA 2653-1)Prunus, Idaho, 1995McManus & Jones (1995)C* 
PFB15 (INRA 2655-1)Prunus, IdahoK. Mohan via J. P. PaulinC* 
Czech Republic
Ea222 (50/92)Cotoneaster, Havlickuv Brod, 1992V. KudelaC*
Ea717Crataegus, Slany, 1997J. KorbaC* 
New Zealand
Ea789Malus, 2007M. HornerC* 
Ea180Cotoneaster, 1993ACW WädenswilC
Ea842Malus, 2008ACW WädenswilC* 
Ea846Malus, 1986P. SobiczewskiC* 
Ea847 (CFBP 1430)Crataegus, 1973J.-P. PaulinC* 
ZYRKD3-1 Zhao et al. (2006)  
ZYRKD3-1 (pZYR2-415-S) This study  
ZYRKD3-1 (pZYR2-415-C) This study  

Bacteria for use in fire blight inoculation experiments were cultured on bouillon glycerin agar at 28°C for 48 h. Antibiotics were added to growth media for the mutant ZYRKD3-1 (20 μg ml−1 chloramphenicol) and the complemented mutants containing pZYR2-415-C or pZYR2-415-S (15 μg ml−1 tetracycline). The plants were inoculated by cutting tips of the upper two leaves with scissors dipped in the bacterial suspension (109 colony-forming unit (cfu) ml−1). Inoculation was performed on shoots with a minimum length of 25 cm. The inoculated plants were incubated in the glasshouse at 25–27°C (day) and 22°C (night). Necrosis of each inoculated shoot was measured 28 d post inoculation (dpi). The length of necrotic shoot tissue relative to the total shoot length averaged over all replicates was recorded as a percentage (%).

Inoculation of double-branched plants

We used double-branched Mr5 (referred as 2xMr5) and Mr5 grafted as scion or as interstem to investigate the intercellular spread of the activated resistance signal. The inoculation was performed on nine to 17 plants of each combination. The first shoot of 2xMr5 plants as well as the resistant shoot (Mr5) of Mr5/‘Idared’/M9 and of ‘Idared’/Mr5/M9 plants were inoculated with strain Ea898 to induce resistance. The second shoot of 2xMr5 was inoculated 24 h later with ZYRKD3-1 and with Ea898 in the case of the scion/interstem/rootstock combination. All shoots were inoculated using scissors dipped in suspensions of the relevant bacterial strains. The length of necrotic shoot tissue relative to the total shoot length averaged over all replicates was recorded as a percentage (%).

avrRpt2EA-gene sequencing

The avrRpt2EA gene was amplified from single colonies of different E. amylovora strains by PCR using the primers avrRpt2-1 and avrRpt2-2 (Table 2). The PCR reaction was performed in a 50 μl standard volume consisting of 5 μl dNTPs (2 mM), 2.5 μl forward and reverse primer (10 pmol), 0.2 μl DreamTaq DNA polymerase (Fermentas, St. Leon-Rot, Germany) and 5 μl 10× buffer. DNA sequencing of the avrRpt2EA gene of different E. amylovora strains was conducted by Eurofins MWG Operon. The primers avrRpt2-1, avrRpt2-2, avrRpt2-5 and avrRpt2-6 were used for direct sequencing of the PCR amplification products. Primer sequences are shown in Table 2. The alignment of the resulting sequences was performed using MUSCLE software (Edgar, 2004) and BioEdit software version (Hall, 1999). Translation was performed using the EMBOSS Transeq software (Rice et al., 2000).

Table 2. Sequences of primers used in this study
PrimerSequence 5′ → 3′ directionTA (°C)
  1. TA, annealing temperature used for PCR; nonbinding bases are underlined; HA, hemagglutinin.

  2. a

    Sequencing primer.


SNP marker development

Polymerase chain reaction primers 680 (specific for both alleles), 682 (specific for the S-allele) and 683 (specific for the C-allele) (Table 2) were designed to distinguish strains with the SNP in the avrRpt2EA gene at position 644. The base variation was included in the critical 3′ end position of the primer to exclude product formation for the mismatch allele. PCR reactions were performed in 25 μl volume reactions using amplification conditions as described previously (Gehring & Geider, 2012).

Complementation of ZYRKD3-1

The avrRpt2EA genes from E. amylovora strains differing in the SNP at position 644 (Ea222 and Ea3049) were amplified using the primers avrRpt2-4-Bam5′ and avrRpt2-3-Eco5′ (Table 2) containing appropriate restriction sites (BamHI or EcoRI, respectively) for further cloning. The amplification products were ligated into the pCR®2.1-TOPO® vector and transformed into One Shot® Chemically Competent TOP10 E. coli cells by using the TOPO TA Cloning® Kit (Invitrogen). Transformants were screened on medium containing 20 μg ml−1 kanamycin and further verified to contain the plasmid by PCR (primers M13for and M13rev). Plasmid DNA from positive clones was isolated using the GeneJET Plasmid Miniprep Kit (Fermentas). The inserts were excised with BamHI and EcoRI and purified by gel extraction using the QIAquick Gel Extraction Kit (Qiagen). The inserts were ligated into the expression vector pRK415 (Keen et al., 1988) and transformed into Library Efficiency® DH5α Competent Cells (Invitrogen). Both vector gene constructs were sequenced to ensure the correct sequence of the cloned fragments. The transformation into different E. amylovora strains was done by electroporation at 800 Ω, 25 μF, and 2.5 kV using the Gene Pulser® II Electroporation System (BioRad, Munich, Germany) with 1 mm gap 90 μl cuvettes (VWR, Dresden, Germany).

Expression analysis

Expression of the avrRpt2EA gene was verified by isolation of bacterial total RNA using the GeneJET RNA Purification Kit (Fermentas) followed by cDNA synthesis with RevertAid Reverse Transcriptase (Fermentas) and transcript content of avrRpt2EA analyzed by PCR using the SNP primers 680, 682 and 683 (Table 2). RNA and DNA were used as controls in the PCR reaction.

RIN4 disappearance assays, expression of AvrRpt2EA in Nicotiana benthamiana

avrRpt2EA from Ea3049 and Ea222 were amplified via PCR from vector pRK415 containing the appropriate allele using primers designed to add an N-terminal SalI site (avrRpt2_SalI, Table 2) and a C-terminal hemagglutinin (HA) tag and SacI site (avrRpt2_HA-SacI, Table 2). The native GTG start codon was changed to ATG to ensure proper translation in Agrobacterium tumefaciens. The resulting PCR products were subcloned into the TA cloning vector pGEM-T-Easy (Promega) and their sequences were verified. Plasmids were digested with the appropriate enzymes and products were ligated into the pMD1 expression vector. pMD1 plasmids were transformed in A. tumefaciens C58C1 for use in transient expression assays. Infiltration and transient expression in N. benthamiana using A. tumefaciens were performed on 4- to 6-wk-old plants. A. tumefaciens strains were grown overnight at 28°C on Luria-Bertani (LB) plates containing 50 μg ml−1 rifampicin and 25 μg ml−1 kanamycin. A. tumefaciens clones were resuspended in induction buffer (10 mM MES (2-(N-morpholino)ethanesulfonic acid), pH 5.6, 10 mM MgCl2, 150 mM acetosyringone), incubated at room temperature and shaking (200 rpm) was performed in the dark for 2 h before infiltration. A. tumefaciens suspensions of single constructs were infiltrated at a final concentration of OD600 = 0.8 (OD, optical density). A. tumefaciens pMD1 T7:RIN4 was infiltrated in a 1 : 1 ratio (OD600 = 0.4 : 0.4) with each of the AvrRpt2:HA constructs for disappearance assays. Phenotypes were assessed at the time of infiltration and 3 dpi as described by Axtell et al. (2003).

Western blotting

Plant tissue was collected at 1 dpi and frozen immediately in liquid nitrogen. Tissue was ground to a fine white powder and resuspended in lysis buffer (50 mM Tris, pH 8.0, 1 mM EDTA, 0.1% Triton) containing complete Mini protease inhibitor (Roche). Samples were centrifuged at 15.300 g at 4 °C and the supernatant retained. Protein was quantified using the Bradford assay and 50 μg total protein was run for each sample on an 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel. Gels were transferred to nitrocellulose membrane, blocked overnight at 4°C, and probed with anti-T7-HRP (horseradish peroxidase) conjugated antibody (Novagen, Madison, WI, USA).

Isolation of RIN4 homologs in Mr5

RIN4 homologs were amplified from genomic DNA and cDNA of Mr5 using primers RIN4_F and RIN4_R (Table 2) containing the restriction site for BamHI and EcoRI. PCR reaction was performed in a 50 μl standard volume consisting of 5 μl dNTPs (2 mM), 2.5 μl forward and reverse primer (10 pmol), 0.5 μl Phusion® High-Fidelity DNA Polymerase (Fermentas) and 10 μl 5X HF Buffer. The products were digested with BamHI and EcoRI and purified by gel extraction using the QIAquick Gel Extraction Kit (Qiagen). The fragments were ligated into expression vector pGADT7 (Clontech, Heidelberg, Germany) and transformed in Library Efficiency DH5a Competent Cells (Invitrogen). To ensure the correct sequence of the construct, sequencing was done at Eurofins MWG Operon. Translation was performed using the EMBOSS Transeq software (Rice et al., 2000). The alignment of the obtained protein sequences was performed with BioEdit software version (Hall, 1999).


Individual strains of E. amylovora differ in their virulence to Mr5

During a period of 3 yr, the virulence of 13 different E. amylovora strains was examined by artificial shoot inoculation on Mr5, ‘Prima’ and the fire blight resistant breeding clone 181. Two of these strains originated from Cotoneaster, one from Crategus, one from Cydonia, five from Malus and four from Pyrus (Table 1). Whereas all strains resulted in more or less strong infection of ‘Prima’ and breeding clone 181, only three strains Ea77 (Ea396 is a synonym of Ea77), Ea395 and Ea400, all isolated from Malus, were able to overcome the resistance of Mr5 (Fig. 1). The mean necrosis rate of the three strains over the years was 67.8% on Mr5 compared with 68.1 and 44.3% on ‘Prima’ and 181, respectively. All strains virulent to Mr5 induced stronger symptoms on the resistant clone 181 than the other isolates. The situation was different for ‘Prima’, as strains Ea180, Ea222, Ea250 and Ea270 caused higher mean necrosis rates than Ea400.

Figure 1.

Evaluation of virulence in different Erwinia amylovora strains. Malus × robusta 5 (Mr5), ‘Prima’ and the advanced breeding clone 181 were inoculated with 13 different E. amylovora strains (Ea77 and Ea396 are both synonyms of E4001a, but were received from different sources; Table 1). Percentage lesion length : length of necrotic shoot / shoot length *100%, on average 16 shoots per strain and clone were inoculated during a 3 yr experiment. Error bars, + SD.

The avrRpt2EA mutant strain ZYRKD3-1 is virulent to Mr5

Shoots of the fire blight resistant wild apple accessions M. baccata, M. fusca, Mr5 and the susceptible cv ‘Idared’ were inoculated with the E. amylovora wildtype strain Ea898 (synonym of Ea1189) and its avrRpt2EA mutant strain ZYRKD3-1. Both were virulent to ‘Idared’ with mean shoot necrosis of 69.8% for Ea898 and 90.7% for ZYRKD3-1, but avirulent to M. baccata and M. fusca. Using the wildtype strain Ea898, no symptoms were detectable on Mr5, but the avrRpt2EA mutant induced necrosis with a mean length of 52.4% (Figs 2, 3).

Figure 2.

Virulence of the Erwinia amylovora wildtype strain Ea898 and the avrRpt2EA mutant strain ZYRKD3-1. Shoots of the fire blight-susceptible apple cv ‘Idared’ and the three resistant wild apple accessions Malus × robusta 5 (Mr5), Malus fusca, Malus baccata were inoculated with both E. amylovora strains (109 colony-forming units (cfu) ml−1). Percentage lesion length: length of necrotic shoot / shoot length *100% of 16 inoculated shoots per strain and clone on average. Error bars, + SD.

Figure 3.

Shoots of Malus × robusta 5 (Mr5) infected with the wildtype strain Ea898 (left) and the avrRpt2EA mutant strain ZYRKD3-1 (right). Photographs were taken 28 d after inoculation.

Fire blight resistance of Mr5 is not systemic

One shoot of double-branched plants of Mr5 was inoculated with Ea898, which was avirulent to Mr5. No shoot necrosis was detectable on the first shoot. The second shoot was inoculated with avrRpt2EA mutant ZYRKD3-1 24 h after the initial inoculation. As shown in Fig. 4, the mutant ZYRKD3-1 was virulent to Mr5 when applied to the second branch. An average mean length of shoot necrosis of 42% was detected.

Figure 4.

Double-branched plants of Malus × robusta 5 (Mr5) (top) and scion–interstem combinations Mr5/'Idared' (middle) and ‘Idared’/Mr5 (bottom) were inoculated on the Mr5 shoot with the avirulent strain Ea898 to induce a resistance signal (Idared, red arrow; Mr5, blue arrow). Twenty-four hours later, the second shoot was inoculated with the avrRpt2EA mutant strain ZYRKD3-1 (double-branched plants of Mr5 (left)) or Ea898 (‘Idared’ of the scion–interstem combinations (middle and right)). Percentage lesion length: length of necrotic shoot / shoot length *100% of 16 inoculated shoots per strain and clone on average. Error bars, + SD.

Scion/interstem combinations of Mr5/‘Idared’ and ‘Idared’/Mr5 were also tested for systemic-acquired resistance in response to inoculation with Ea898 wildtype. Ea898 was applied to Mr5 and, 24 h later, ‘Idared’ was inoculated using the same strain. No symptoms were detected on Mr5, whereas ‘Idared’ showed shoot necrosis of 72% (Mr5/Idared) and 98% (Idared/Mr5) (Fig. 4).

Sequencing of the avrRpt2EA gene of various strains of E. amylovora

The avrRpt2EA gene of 22 E. amylovora strains of different origins was amplified from genomic DNA by PCR using the primers avrRpt2-1 and avrRpt2-2 and sequenced. The nucleic acid sequence of the avrRpt2EA gene (669 bp) was identical for 19 strains. Three strains (Ea110, Ea3049 and Ea3050) contained a single nucleotide deviation in which guanine at position 644 of the DNA sequence was substituted by cytosine (G644C). This mutation resulted in a change of the amino acid sequence from cysteine at position 156 (C-allele) to serine (C156S). A partial alignment of the AvrRpt2EA amino acid sequences of the 22 E. amylovora strains is shown in Fig. 5. Strains Ea3049 (Ea395), Ea3050 (Ea396) and Ea110, all of which contained the C156S substitution (S-allele), were virulent to Mr5 (Figs 1, 6).

Figure 5.

Partial alignment of derived amino acid sequences of the effector protein AvrRpt2EA from different Erwinia amylovora strains. The highlighted strains contain a single substitution of cysteine by serine at position 156 (C156S).

Figure 6.

Virulence of the Erwinia amylovora wildtype strains Ea898, Ea110 and Ea3050, the avrRpt2EA mutant strain ZYRKD3-1 as well as the complemented mutant strains ZYRKD3-1 (pZYR2-415-S) and ZYRKD3-1 (pZYR2-415-C) at 109 cfu ml−1. Percentage lesion length: length of necrotic shoot / shoot length *100% of 16 inoculated shoots per strain and clone on average. Error bars, + SD. *, ‘Pinova’, M. baccata jackii and ‘Royal Gala’ were not included into evaluations using strain ZYRKD3-415-C.

SNP marker-based detection of additional strains containing the single base mutation

Twenty-two E. amylovora strains used for sequencing of the avrRpt2EA gene and 53 additional strains were tested to distinguish strains with and without the SNP in the avrRpt2EA gene by colony PCR using the primer combination 680–682 for the guanine variant, and 680–683 for the cytosine variant. Primers 682 and 683 contain the SNP base as the discriminating base at the 3′ end. Under stringent PCR conditions, a PCR product was only obtained for the respective SNP variant. Identical results were obtained by PCR and sequencing of the 22 strains (Fig. 5 and Supporting Information, Fig. S1). This indicates that the appropriate primers are a useful tool for fast screening of strains differing in the SNP at position 644. To summarize, 70 E. amylovora isolates contained guanine at nucleotide position 644, whereas five strains contained cytosine (Tables 1, Fig. S1). It should be noted that most E. amylovora strains also recently isolated in the Ontario region of Canada carried the C-allele such as Ea CaV8, CaV15, Tp3 or Tp9 (Table S1).

Development of complemented mutant strains

The two alleles of the avrRpt2EA gene were used for complementation of the mutant strain ZYRKD3-1. The gene was amplified from the E. amylovora strains Ea222 (containing the C-allele) and Ea3049 (containing the S-allele) and cloned into the expression vector pRK415 as described. The resulting plasmids were designated pZYR2-415-C (containing the C-allele) and pZYR2-415-S (containing the S-allele). The plasmids were transferred to the avrRpt2EA mutant ZYRKD3-1 to generate the complemented mutant strains ZYRKD3-1 (pZYR2-415-S), referred to as ZYRKD-415-S, and ZYRKD3-1 (pZYR2-415-C), referred to as ZYRKD-415-C. Next, to confirm the generation of the desired complemented strains, the expression of the avrRpt2EA gene was tested by reverse transcriptase PCR (RT-PCR) on cDNA. E. amylovora wildtype strains (Ea898, Ea110 and Ea3050), the avrRpt2EA mutant ZYRKD3-1, and the complemented strains ZYRKD-415-S and ZYRKD-415-C were compared. RNA was included as negative control. As shown in Fig. S2, all constructs were validated as being complemented with the desired alleles.

The effect of the single base mutation onto virulence

An inoculation experiment was performed using the E. amylovora wildtype strains Ea898, carrying the C-allele, as well as Ea110 and Ea3050, both carrying the S-allele, in order to study the virulence effect of the SNP in the avrRpt2EA gene on apple shoots. Furthermore, the avrRpt2EA mutant ZYRKD3-1 and the complemented strains ZYRKD-415-S and ZYRKD-415-C were included. These strains were used for inoculation of several fire blight-resistant wild apple genotypes M. baccata jackii, M. floribunda 821, Mr5, M. baccata and M. fusca. Three fire blight-susceptible Malus × domestica cvs ‘Idared’, ‘Pinova’ and ‘Royal Gala’ were used as control genotypes. ‘Pinova’, ‘Royal Gala’ and M. baccata jackii were not tested with the complemented mutant ZYRKD-415-C, as the number of successfully grafted plants of these genotypes was too low. The results obtained in this experiment are shown in Fig. 6. As shown, all three apple cultivars were susceptible to the applied strains, while, by contrast, none to very little necrosis was detected on M. baccata, M. floribunda 821 and M. fusca. On both M. baccata jackii and Mr5, considerable necrosis was detected after inoculation with ZYRKD3-1, ZYRKD-415-S and the two S-allele strains Ea110 and Ea3050. Neither the wildtype strain Ea898 nor the complemented mutant strain ZYRKD-415-C could overcome the resistance (Fig. 6) of Mr5, indicating a possible role for avrRpt2EA in resistance activation.

Expression of AvrRpt2EA in N. benthamiana

The two alleles of the avrRpt2EA gene, the P. syringae effector AvrRpt2, and the catalytically inactive P. syringae AvrRpt2 mutant, AvrRpt2 C122A and T7:RIN4 as negative controls were transiently expressed using A. tumefaciens-mediated transformation in N. benthamiana in order to test whether the avrRpt2EA alleles are able to cause cell death. As shown in Fig. 7, both alleles of E. amylovora AvrRpt2EA, the S-allele from strain Ea3049 (AvrRpt2EA 3049:HA) and the C-allele from Ea222 (AvrRpt2EA 222:HA) induced cell death in N. benthamiana 3 d after infiltration. The same reaction was induced by the P. syringae effector AvrRpt2 (PsAvrRpt2:HA), whereas infiltration with the negative controls PsAvrRpt2:HA C122A and T7:RIN4 did not elicit cell death (Fig. 7). These data suggest recognition of the avirulence activity of AvrRpt2EA, possibly mediated in a manner similar to that previously observed in A. thaliana (Axtell & Staskawicz, 2003).

Figure 7.

Infiltration of Nicotiana benthamiana with Agrobacterium tumefaciens suspensions of the double constructs, containing the alleles of AvrRpt2EA (S-allele, AvrRpt2EA 3049:HA (hemagglutinin); and C-allele, AvrRpt2EAa 222:HA), the Pseudomonas syringae AvrRpt2 (PsAvrRpt2:HA) or the catalytically inactive P. syringae AvrRpt2 mutant, PsAvrRpt2:HA C122A and RIN4. As control, RIN4 was separately expressed. Images were collected at the time of infiltration and at 3 d post inoculation (dpi).

AvrRpt2EA coexpression with Arabidopsis RIN4 results in partial elimination in a catalytically dependent manner

Previous data have demonstrated that P. syringae AvrRpt2 cleaves RIN4, resulting in the elimination of RIN4; this mechanism is hypothesized to release the negative regulation RIN4 imposes on the resistance (R) protein RPS2 (Day et al., 2005). To determine if coexpression of AtRIN4 and AvrRpt2EA results in the elimination of RIN4 in a catalytically dependent manner, we expressed AtRIN4 and AvrRpt2EA in N. benthamiana and monitored the cleavage (i.e. disappearance) of AtRIN4. As shown in Fig. 8, wildtype P. syringae AvrRpt2:HA (PsAvrRpt2:HA) coexpressed with AtRIN4 resulted in the complete elimination of RIN4, while the catalytically inactive PsAvrRpt2:HA C122A did not. Interestingly, we observed the partial elimination of T7-tagged AtRIN4 by AvrRpt2EA:HA 3049, suggesting some degree of conservation in both substrate (i.e. RIN4) recognition and activity of AvrRpt2EA with P. syringae AvrRpt2. On the other hand, the coexpression of AvrRpt2EA:HA 222 with AtRIN4 seemed to have no effect on the abundance of RIN4 (Fig. 8).

Figure 8.

Disappearance assay via western blot with T7Rin4 antibody. Both alleles of AvrRpt2EA (S-allele, AvrRpt2EAa 3049:HA; and C-allele, AvrRpt2EAa 222:HA), the Pseudomonas syringae AvrRpt2 (PsAvrRpt2:HA) and the catalytically inactive P. syringae AvrRpt2 mutant, PsAvrRpt2:HA C122A, were transiently coexpressed via Agrobacterium tumefaciens with RIN4 in Nicotiana benthamiana. RIN4 was also separately expressed as control. Ponceau staining of the blot as shown in the lower panel confirms equal protein loading.

Isolation of RIN4 sequences from Mr5

The Arabidopsis RIN4 gene is part of a large gene family of nitrate-induced (NOI) domain-containing proteins, many of which are capable of being cleaved by the P. syringae effector AvrRpt2 (Afzal et al., 2011). To determine if AtRIN4 homologs exist in apple, thereby supporting our hypothesis that AvrRpt2EA recognition and activity are transduced in a similar fashion to that in Arabidopsis, we sought to identify an MrRIN4 homolog. As shown in Fig. 9, we were successful in identifying two nucleic acid sequences of putative RIN4 homologs in the genome of Mr5 and assigned them to the homolog chromosomes 5 and 10 of the Golden Delicious genome (Velasco et al., 2010). The deduced protein sequences differ in 17 amino acids and share an overall amino acid identity of 92%. Additionally, we found that the RIN4 sequence on chromosome 10 has a deletion at position 98–99, resulting in a deletion of two amino acids. MrRIN4-1 and MrRIN4-2 are 241 and 239 amino acids in total, respectively. An alignment of the A. thaliana RIN4 and MrRIN4-1 and MrRIN4-2 is shown in Fig. 9. Sequence scanning of the two sequences identified two widely conserved areas, one in the N-terminal and the other in the C-terminal region. Both contain the cleavage sites of AvrRpt2 from P. syringae pv. tomato (Chisholm et al., 2005).

Figure 9.

Protein alignment of RIN4 from Arabidopsis thaliana and Mr5 from chromosome 5 (Mr5_RIN4_Chr5) and 10 (Mr5_RIN4_Chr10). The arrows indicate the cleavage sites of AvRpt2 from Pseudomonas syringae pv. tomato.


The crabapple species Malus × robusta (Carrière) Rehder (synonyms Malus microcarpa var. robusta Carrière, Pyrus baccata var. cerasifera Regel) is referred to as a hybrid between the two primary wild apple species M. prunifolia and M. baccata (Jefferson, 1970; Ignatov & Bodishevskaya, 2011). One of the most famous accessions of this species is selection M. × robusta No. 5 (Mr5), which was grown from seed obtained in 1927 from Russia through arrangements with the Arnold Arboretum (Jefferson, 1970). Mr5 has been described as tolerant to common viruses of eastern Canada, resistant to collar rot, woolly apple aphid (Watkins & Spangelo, 1970), powdery mildew (Wan & Fazio, 2011) and fire blight (Watkins, 1971; Van Der Zwet & Keil, 1974). Differential interaction between Mr5 and various E. amylovora strains were reported by Norelli & Aldwinckle (1986), who inoculated 25 apple cultivars, including Mr5, with E. amylovora strains Ea273 (Ea78 and Ea3051 in the author database) and Ea266 (Ea77, Ea396 and Ea3050 in the author database). Mr5 was resistant to Ea273 but susceptible to Ea266. The Canadian isolate Ea3049 could also overcome the resistance of Mr5 and resulted in 97% average shoot necrosis (Peil et al., 2011). In the present study, two additional E. amylovora isolates virulent to Mr5 have been identified, Ea400 and Ea110 (Figs 1, 6). The results obtained by Norelli & Aldwinckle (1986), Peil et al. (2011) and the present study demonstrate that the resistance of Mr5 to fire blight is highly strain-specific. Furthermore, it was shown that all E. amylovora strains able to overcome the resistance of Mr5 were highly virulent to susceptible and moderately resistant apple genotypes as well (Fig. 1).

In 2006, Zhao et al. identified an analog to the effector protein AvrRpt2 from P. syringae pv. tomato in the genome of E. amylovora and annotated it as AvrRpt2EA. Inoculation of immature pear fruits with an avrRpt2EA deletion mutant (ZYRKD3-1) resulted in reduction of disease symptoms, indicating that AvrRpt2EA is a potential virulence factor in the host–pathogen system pear–E. amylovora. To study the role of AvrRpt2EA in the Malus × robustaE. amylovora host–pathogen relationship, the fire blight resistant wild apple genotypes M. baccata, M. fusca, Mr5 and the susceptible apple cv ‘Idared’ were inoculated with the wildtype strain Ea898 and ZYRKD3-1. Results obtained on ‘Idared’ could not support the hypothesis of Zhao et al. (2006), that avrRpt2EA seems to act as a virulence gene in susceptible cultivars, because the deletion mutant caused a similar number of symptoms as the wildtype strain. This indifferent reaction was confirmed by inoculation of the susceptible apple cvs ‘Royal Gala’ and ‘Pinova’, indicating that avrRpt2EA is not indispensable for a strong infection of apple cultivars. The results obtained on Mr5 were completely different. Whereas the wildtype strain Ea898 was not able to infect Mr5, the resistance of Mr5 was broken by the mutant strain which caused a percentage shoot necrosis of > 50% (Figs 2, 3). These results suggest that AvrRpt2EA acts as an avirulence factor in the host–pathogen relationship Mr5 and E. amylovora. Interestingly, the other two resistant wild apple accessions, M. baccata and M. fusca, showed resistance to both the wildtype and the mutant strain, suggesting a different mode of resistance for M. baccata and M. fusca.

The fact that the deletion of the avrRpt2EA gene of E. amylovora results in successful infection of Mr5 is a strong indication of a gene-for-gene relationship in the host–pathogen system Mr5–E. amylovora. In general, plants recognize pathogen effectors, also called avirulence proteins, by resistance proteins and activate a defense cascade. This mode of interaction was first described by Flor (1971) and later supplemented by the guard model (Van Der Biezen & Jones, 1998; Dangl & Jones, 2001), explaining indirect interactions where the avirulence gene is targeted/recognized by a guard. Effectors such as the AvrRpt2EA protein are translocated into plant cells by T3SSs (Collmer et al., 2002). Three T3SSs which are encoded by so-called pathogenicity islands (PAIs) are known in E. amylovora (He et al., 2004; Oh & Beer, 2005). The hypersensitive response and pathogenicity (Hrp) T3SS PAI1 is essential for virulence of fire blight (Bellemann & Geider, 1992; reviewed in Oh & Beer, 2005), whereas the PAI2 and PAI3 T3SSs are dispensable for virulence (Zhao et al., 2009). Besides avrRpt2EA, four more potential effector genes have been identified in the genome of E. amylovora: eop1 (orfB or eopB), eop3 (hopX1Ea), dspA/E and hopPtoCEa (McNally et al., 2012). DspA/E is known as essential pathogenicity factor, as mutants were not able to induce disease symptoms or to grow on host plants (Barny et al., 1990; Gaudriault et al., 1997; Bogdanove et al., 1998). Deletion mutants of eop1, eop3 and hopPtoCEa were also tested on immature pear fruits but did not differ in virulence from the wildtype strain (Zhao et al., 2005; Asselin et al., 2011).

The apparent central function of AvrRpt2EA in the defense mechanism of Mr5 gave cause for further investigations. For that reason, the avrRpt2EA gene of 22 E. amylovora strains (virulent and avirulent to Mr5) was sequenced and the amino acid sequences were deduced. Only one SNP was detected among the 22 sequences that resulted in an amino acid exchange at position 156 from cysteine to serine (C156S). SNP analysis showed that only five out of 75 strains encoded serine at position 156 (Tables 2, Fig. S1). Only strains containing serine on position 156 were able to overcome the resistance of Mr5, whereas strains with cysteine at position 156 were not. Therefore, both were considered as different alleles (C-allele and S-allele). The ability of cysteine to form disulfide bridges could result in a different tertiary structure for the two alleles, thereby modifying the recognition process in Mr5. Results obtained for the P. syringae AvrRpt2 protein in the A. thalianaP. syringae host–pathogen system support the hypothesis that position 156 of the AvrRpt2EA protein seems to be important for avirulence activity (Lim & Kunkel, 2004).

ZYRKD3-1 was complemented with the C156 and the S156 avrRpt2EA allele driven by its own promoter to verify whether the C156S substitution affects the avirulence activity. The complementation with the C-allele should result in the recovery of resistance in Mr5 if there is a gene-for-gene relationship. On the other hand, ZYRKD3-1 complemented with the S-allele should overcome the resistance of Mr5. RT-PCR of the complemented strains showed that both alleles were expressed (Fig. S2). Virulence analysis was done with the wildtype strain Ea898 (carrying the C-allele), ZYRKD3-1, both complemented versions of ZYRKD3-1 and strains Ea110 and Ea3050 (both carrying the S-allele) on ‘Pinova’, M. baccata jackii, M. floribunda 821, M. baccata, M. fusca, Mr5, ‘Idared’ and ‘Royal Gala’. The mutant strain complemented with the C-allele was not virulent to Mr5, whereas the mutant strain complemented with the S-allele resulted in an average shoot necrosis of 70%, thus breaking down resistance. These results confirm a gene-for-gene relationship in the pathogen system Mr5 and E. amylovora.

The wild apple accessions of M. fusca, M. floribunda and M. baccata are highly resistant to all tested strains, indicating another resistance mechanism. M. baccata jackii has shown a similar pattern to that of Mr5, the mutant strain, as all strains carrying the S-allele were able to overcome the resistance of M. baccata jackii, indicating a similar mechanism. Since M. × robusta is a hybrid between M. prunifolia and M. baccata, this is not surprising.

The demonstrated gene-for-gene relationship and the high degree of similarity between the two effector proteins, AvrRpt2 from P. syringae pv. tomato and AvrRpt2EA from E. amylovora, support the hypothesis of a resistance mechanism in Mr5 similar to the one in A. thaliana. AvrRpt2 is activated in the plant host cytosol by ROC1, a cyclophilin, via prolyl isomerization (Coaker et al., 2006) and is able to cleave RIN4 (RPM1 interacting protein 4). RIN4 is physically associated with RPS2. The cleavage of RIN4 results in the activation of RPS2 (Resistance to P. syringae protein 2) and thereby in the activation of the pathogen defense of the plant (Mackey et al., 2002; Axtell & Staskawicz, 2003; Kim et al., 2005).

We were also able to show in a coexpression assay of both AvrRpt2EA alleles with A. thaliana RIN4 in N. benthamiana that the S-allele was capable of partial elimination of RIN4, whereas the C-allele had no effect on the abundance of the RPS2 target. Two alleles of the avrRpt2EA gene were transiently expressed via A. tumefaciens in N. benthamiana in order to test whether they are able to cause cell death. Similar to P. syringae AvrRpt2, both alleles of E. amylovora AvrRpt2EA elicited a cell death-like response in N. benthamiana. These data are in agreement with the hypothesis that, like that of previous studies (Lim & Kunkel, 2004; Chisholm et al., 2005), RIN4 is not the only target of AvrRpt2EA. Indeed, similar results were obtained by Lim & Kunkel (2004) and Axtell et al. (2003), who demonstrated that mutated avrRpt2 strains were also unable to eliminate RIN4 but still virulent to the host.

In addition to AvrRpt2, RIN4 is known to interact with the effectors AvrRpm1 and AvrB (Mackey et al., 2003), which in turn activate resistance through a second coiled coil-nucleotide-binding site-leucine rich repeat (CC-NBS-LRR) resistance gene, RPM1 (Grant et al., 1995). Thus, RIN4 appears to be a multi-functional target for the regulation and activation of gene-for-gene resistance to a variety of phytopathogenic effector proteins. A putative homolog of AtRIN4 was identified in the transcriptome of unchallenged Malus × robusta 5 (Fahrentrapp, 2012). Two nearly identical RIN4-like genes on the homologous apple chromosomes 5 and 10 have been detected in further investigations (Fig. 9). Furthermore, a putative fire blight resistance gene, FB_Mr5 from Mr5, located at the QTL on LG3 (Peil et al., 2007) was recently published, showing a certain similarity to RPS2 of A. thaliana (Fahrentrapp et al., 2012). This CNL (CC-NBS-LRR) gene was recently identified using a map-based cloning approach (Fahrentrapp et al., 2011). Parravicini et al. (2011) could identify two genes at the fire blight resistance locus of ‘Evereste’, showing high homology to the Pto/Prf complex, indicating a similar resistance mechanism.

This study gives a first indication of how fire blight resistance could function in the pathosystem Malus × robusta 5 and E. amylovora. We demonstrated that AvrRpt2EA, an analog to the effector protein AvrRpt2 from P. syringae pv. tomato, plays an important role in the resistance mechanism of Mr5 and that it is part of a gene-for-gene relationship. The existence of a homologous gene of RIN4 and an Rps2-like resistance gene in the genome of Mr5 assumes a similar resistance mechanism to the one in A. thaliana. This hypothesis is supported by the fact that AvrRpt2EA is able to cleave RIN4 in Arabidopsis. As AvrRpt2EA is not essential for virulence on Malus types other than Mr5, it is of interest that the respective gene is not found in other closely related Erwinia species. While avrRpt2EA is highly conserved among the E. amylovora analyzed here, and even those present in various Rubus isolates (see Table S1), no homolog is present even in closely related Erwinia species. While synteny of the respective genomic region is well conserved in E. amylovora, no similarity to the avrRpt2EA sequence is found in the pear blight pathogen E. pyrifoliae, and the nonpathogenic E. billingiae and E. tasmaniensis. It can be concluded that avrRpt2EA is a recently acquired gene in an evolutionary context.

Further investigations will reveal whether a cleavage of Mr5 Rin4 by the E. amylovora effector AvrRpt2EA is dependent on the respective allele. The functionality of the candidate resistance gene FB_Mr5 has to be proved in a complementation assay. The hypothesis of a similar system as described for A. thalianaP. syringae could be confirmed by yeast two-hybrid experiments to examine the possible interactions of Fb_Mr5, MrRIN4 and AvrRpt2EA. Another interesting task would be to investigate the resistance mechanism of the two highly resistant wild species M. fusca and M. baccata, which are obviously different from the one of Mr5.


The resistance of the wild apple species Malus × robusta 5 (Mr5) to E. amylovora is highly strain-specific and can be broken only by a few strains originating from North America or the strain isolated from loquat in Israel. Furthermore, an avrRpt2EA deletion mutant, ZYRKD3-1 (Zhao et al., 2006), was able to overcome the resistance of Mr5, which gave cause for further investigations. Subsequently, 22 avrRpt2EA genes from different E. amylovora strains were sequenced and compared. Only one SNP has been identified which resulted in an exchange of cysteine to serine at position 156 in the amino acid sequence. The C-allele was responsible for the diverging resistance responses, because only strains carrying the S-allele were able to overcome the resistance of Mr5. Inoculation of Mr5 with the mutant ZYRKD3-1 complemented with the C- or S-allele resulted in receiving or breaking resistance and gave evidence of a gene-for-gene relationship. The defense mechanism appears to be similar to the one in A. thaliana triggered by the homolog effector AvrRpt2. In a disappearance assay, the S-allele of AvrRptEaA was able to cleave nearly all of the RIN4 of A. thaliana. Besides RIN4, an RPS2-like gene could also be identified in the genome of Mr5. In contrast to Arabidopsis (Cameron et al., 1994), the resistance found in Mr5 is not systemic.


We thank John L. Norelli (USDA-ARS-AFRS, Kearneysville, WV, USA) for sharing E. amylovora strains and Marina Gernold for part of the PCR assays. This work was financially supported by the Deutsche Forschungsgemeinschaft (DFG) project numbers AOBJ574457 and AOBJ577770 and by Michigan AgBioResearch.