Presence/absence, differential expression and sequence polymorphisms between PiAVR2 and PiAVR2-like in Phytophthora infestans determine virulence on R2 plants


Author for correspondence:
Paul R. J. Birch
Tel: +44 (0)1382 562731


  • A detailed molecular understanding of how oomycete plant pathogens evade disease resistance is essential to inform the deployment of durable resistance (R) genes.
  • Map-based cloning, transient expression in planta, pathogen transformation and DNA sequence variation across diverse isolates were used to identify and characterize PiAVR2 from potato late blight pathogen Phytophthora infestans.
  • PiAVR2 is an RXLR-EER effector that is up-regulated during infection, accumulates at the site of haustoria formation, and is recognized inside host cells by potato protein R2. Expression of PiAVR2 in a virulent P. infestans isolate conveys a gain-of-avirulence phenotype, indicating that this is a dominant gene triggering R2-dependent disease resistance. PiAVR2 presence/absence polymorphisms and differential transcription explain virulence on R2 plants. Isolates infecting R2 plants express PiAVR2-like, which evades recognition by R2. PiAVR2 and PiAVR2-like differ in 13 amino acids, eight of which are in the C-terminal effector domain; one or more of these determines recognition by R2. Nevertheless, few polymorphisms were observed within each gene in pathogen isolates, suggesting limited selection pressure for change within PiAVR2 and PiAVR2-like.
  • Our results direct a search for R genes recognizing PiAVR2-like, which, deployed with R2, may exert strong selection pressure against the P. infestans population.


Oomycetes comprise a major group of eukaryotic microbial pathogens that cause devastating diseases on dicotyledonous plants (Kamoun, 2003). One notorious representative is Phytophthora infestans, the cause of late blight, the most significant global disease of potato. The genetic flexibility of Phytophthora (Brasier, 1992) and the coevolution of P. infestans populations in Central and South America with wild Solanum species have yielded a remarkable source of genetic diversity in populations of this pathogen. P. infestans thus possesses an alarming, and demonstrated, adaptibility, able to respond rapidly to selection pressures within agricultural systems, resulting in global late blight epidemics (Fry, 2008). Breeding efforts to control this disease by introgression of resistance from wild Solanum species have had limited success, probably because of the genetic diversity within pathogen populations (Hein et al., 2009a). Indeed, the genome sequence of P. infestans reveals striking potential for genetic change (Haas et al., 2009). A major scientific goal is thus to develop a detailed understanding of how disease resistance to P. infestans has been overcome to date. Such knowledge is critical to combating this and other economically important oomycete plant pathogens.

Inducible disease resistance in plants is based on detection of two distinct classes of pathogen molecules. Recognition of secreted or surface-exposed pathogen/microbe-associated molecular patterns (PAMP/MAMPs) by pattern recognition receptors (PRRs) in the host leads to broadly effective PAMP (or, more generally, Pattern)-triggered immunity (PTI). Pathogens deploy effector proteins that suppress this response (effector-triggered susceptibility; ETS). Effectors are a second class of molecules that can be detected by plants, often by nucleotide-binding leucine-rich repeat (NB-LRR) resistance (R) proteins. When detected, effectors are termed avirulence (AVR) proteins, and the consequent disease resistance is referred to as effector-triggered immunity (ETI), or the hypersensitive response (HR) (Jones & Dangl, 2006; Chisholm et al., 2006).

In recent years, a number of AVR genes have been identified from oomycete plant pathogens. These include AVR3a (Armstrong et al., 2005), AVR4 (Van Poppel et al., 2008), AVR-blb1 (Vleeshouwers et al., 2008; Champouret et al., 2009) and AVR-blb2 (Oh et al., 2009) from P. infestans; Avr1b (Shan et al., 2004), Avr3c (Dong et al., 2010), Avr3a, Avr1a (Qutob et al., 2009) and Avr4/6 (Dou et al., 2008) from the soybean pathogen P. sojae; and ATR13 (Allen et al., 2004) and ATR1 (Rehmany et al., 2005) from the Arabidopsis thaliana pathogen Hyaloperonospora arabidopsidis. All are members of the extensive, diverse RXLR class of effectors (Birch et al., 2006, 2008, 2009; Kamoun, 2006, 2007; Hein et al., 2009b; Schornack et al., 2009). These effectors are so named for the amino acid motif, RXLR (Arg-any amino acid-Leu-Arg), often closely followed by the motif EER (Glu-Glu-Arg), which is required for their entry into living plant cells (Whisson et al., 2007; Dou et al., 2008).

Each of these effectors has enhanced our understanding of how oomycete plant pathogens can evade ETI and thus overcome disease resistance. Single nucleotide polymorphisms (SNPs) within allelic forms may give rise to proteins with amino acid changes that evade recognition. This has been well documented for AVR3a from P. infestans; only two alleles reported within the pathogen population encode proteins differing in two amino acids (K80E and I103M) which dictate recognition by R3a (Armstrong et al., 2005; Bos et al., 2006). As AVR3a is an essential pathogenicity determinant (Bos et al., 2010), deployment of an R gene that targets the virulent form, AVR3aEM, in combination with R3a, which targets the avirulent form, AVR3aKI, would potentially impose strong selection pressure on the pathogen population. In addition to AVR3a, SNPs that encode alternative, virulent alleles have been reported for ATR1 and ATR13 from H. arabidopsidis (Allen et al., 2004; Rehmany et al., 2005) and Avr1b and Avr3c from P. sojae (Shan et al., 2004; Dong et al., 2009). In addition to amino acid polymorphisms, which can retain the virulence function of the effector (Bos et al., 2010), virulence on plants containing some R genes has been achieved by loss of a functional AVR gene. Frame-shift mutations, resulting in truncated versions of AVR4, have been reported in P. infestans isolates that infect potato expressing R4 (Van Poppel et al., 2008). Moreover, differential gene expression, sometimes associated with gene deletion or gene copy number variation, has been reported for Avr1b (Shan et al., 2004), Avr1a and Avr3a (Qutob et al., 2009) from P. sojae. Presumably, loss of an effector gene, or of its expression, may be compensated for by functional redundancy in the effector complement (Birch et al., 2008).

In theory, durable disease resistance may be achieved with an R protein, or a judicious combination of R proteins, that target all of the effectors (or effector forms deriving from alleles) contributing to an essential pathogenicity function, as hypothesized earlier for AVR3a from P. infestans. This is apparently the case for the resistance protein Rpi-blb2 from Solanum bulbocastanum, which recognizes multiple members of a closely related P. infestans RXLR effector family, of which there are seven paralogues within the genome sequence of clone T30-4 (Oh et al., 2009). Rpi-blb2 has so far proven durable.

Here, we report the map-based cloning of PiAVR2 from P. infestans. This gene (PITG_22870) from P. infestans clone T30-4 was previously shown to trigger R2-dependent cell death (Lokossou et al., 2009). Transformation of a virulent isolate with PiAVR2 conferred a gain-of-avirulence phenotype when inoculated on to the R2 potato differential, indicating that it is a dominant gene specifying R2-mediated recognition and disease resistance. We show that, whereas all avirulent isolates possess PiAVR2, virulent isolates lack either PiAVR2 or its expression. By contrast, virulent isolates express a divergent form, PiAVR2-like, which is not recognized by R2 or R2-like orthologues from wild Solanum species. Remarkably, while there is little sequence diversity within either PiAVR2 or PiAVR2-like in the sampled P. infestans populations, the predicted PiAVR2 and PiAVR2-like polypeptides differ from each other by 13 amino acids in the mature protein. Given that each form, though markedly divergent, nevertheless appears to be highly conserved in diverse P. infestans isolates, and that all isolates tested possess one or other, or both forms, we propose a strategy for durable late blight disease resistance.

Materials and Methods

Plant and microbial strains and growth conditions

Potato plants and P. infestans isolates and transgenic strains were maintained, and infection assays performed, as described in Whisson et al. (2007). Nicotiana benthamiana and potato genotypes were grown as in Bos et al. (2010). Escherichia coli strain DH10B and Agrobacterium tumefaciens strain AGL1 were used for cloning. Agrobacterium strains for transient expression of potato R2 and its orthologues R2-like, BLB3 and ABPT are described in Lokossou et al. (2009). All A. tumefaciens cultures were grown at 27°C at 200 rpm for 2–3 d in LB (Luria Bertani broth), spun at 4000 rpm and the pellet resuspended in sterile 10 mM 2-(N-morpholine)-ethanesulphonic acid (MES) and 10 mM MgCl2 buffer with 200 μM acetosyringone, to OD600 = 0.5 for each construct.

Map-based cloning of PiAVR2

Amplified fragment length polymorphism (AFLP) markers E + GA/M + ATs513 and E + TG/M + CTs338, which span the Avr2 locus of P. infestans (Van der Lee et al., 2001; Fig. 1a), were used to screen the pooled bacterial artificial chromosome (BAC) library of P. infestans T30-4 as described previously (Whisson et al., 2001). BAC clones identified with E + GA/M + ATs513 (23O22 and 39L24) and with E + TG/M + CTs338 (8C12 and 5D19) were end-sequenced using SP6 and T7 primers as described (Whisson et al., 2001), and these sequences were positioned on the genome sequence of T30-4 (Haas et al., 2009), in supercontig 1.16 using BlastN. The region delimited by the outer BAC sequences was screened for annotated RXLR-EER coding sequences, revealing PITG_22870, PITG_08943 and PITG_08949.

Figure 1.

Identification of PiAVR2 candidates by map-based cloning. (a) Genetic interval (left) showing amplified fragment length polymorphism markers flanking the AVR2 locus. Bacterial artificial chromosome (BAC) clones containing these markers were end-sequenced and the DNA sequences positioned in supercontig 1.16 of the Phytophthora infestans T30-4 genome sequence, spanning a region of 742.6 kb. Within a 36.6 kb region of this (right), three candidate RXLR-dEER effector genes were identified: PITG_08949, PITG_08943 and PITG_22870. This diagram is not drawn to scale. (b) Alignment of the predicted protein sequences of PITG_22870 and PITG_08949 revealed considerable similarity over the first 79 amino acids, and subsequent divergence, as a result of a likely DNA recombination event (Supporting Information, Fig. S2). The pale box represents the signal peptide and the dark boxes indicate the RLLR and EER motifs.

Transient expression of P. infestans genes in plants to assess R gene responses

PITG_08949 and PITG_22870 (PiAVR2) sequences were initially amplified from genomic DNA of P. infestans isolate 88069 using Asc1-For and BamH1-Rev primers (Supporting Information, Table S1) and cloned into vector pGRAB using these restriction sites and transformed into A. tumefaciens strain AGL1 cells by electroporation as described in Bos et al. (2010). Positive transformants were subsequently grown in LB supplemented with rifampicin and kanamycin for transient expression in potato and N. benthamiana. For gateway cloning® (Invitrogen), PITG_22870 and PITG_08949 were amplified from sequences encoding from the cleavage site of the signal peptide (SP) to the stop codon from genomic DNA of P. infestans isolate T30-4 with gene-specific primers (Table S1). The N-terminus-encoding region of PiAVR2N31 was cloned from the SP cleavage site to the last amino acid in the EER motif. The C-terminus-encoding region of PiAVR2 was cloned from the first amino acid after the EER motif to the stop codon (see Table S1 for PCR primers). AttB recombination sites were added by a second PCR using the AttB1/AttB2 to all sequences and recombined into pDNR221 using BP clonase® (Invitrogen). Piavr2 DNA encoding from the SP cleavage site to the stop codon (357 bp), and the C-terminus-encoding region from the first amino acid after the EER motif until the stop codon, or without the stop (222 bp), was synthesized and delivered in pUC57 (Genscript, Piscataway, USA).

LR clonase® (Invitrogen) was used to recombine DNA sequences into pMDC32 plant expression vector (Curtis & Grossniklaus, 2003), transformed into E. coli by electroporation, sequenced, and plasmids with confirmed inserts were transformed into A. tumefaciens strain AGL1, pSoup, pVirG cells by electroporation. Positive transformants were subsequently grown for 2 d at 27°C in LB supplemented with rifampicin, chloroamphenicol, tetracycline and kanamycin for transient expression in potato and N. benthamiana. For plant inoculations, all A. tumefaciens cultures were resuspended to OD600 = 0.5. Cultures carrying RXLRs were mixed 1 : 1 with one of the cultures carrying R2 or an R2 orthologue so the final OD600 of each is 0.25. Cultures not mixed with either an RXLR or an R gene construct were diluted with an equal volume of buffer to a final OD600 = 0.25. Cultures were infiltrated with a 1 ml syringe without a needle through the abaxial leaf surface superficially wounded with a needle. Three to four leaves on at least four plants were used for each biological replicate. HRs were recorded and photographed between 2 and 5 d postinfiltration depending on the expression vector and plant species. An individual inoculation was counted as positive if > 50% of the inoculated area developed a clear HR. Data graphs present the mean percentage of total inoculations per plant developing a clear HR with error bars representing ± standard error (SE) of combined data from at least three biological replicates. Co-bombardment assays with GUS were performed and assayed as described in Armstrong et al. (2005) using pGRAB::PiAVR2.

Gene expression analyses

Standard RT-PCR to examine gene expression in a number of isolates at 48 h postinoculation (hpi) on potato was performed using primers diagnostic for expression of PiAVR2 (AVR2F4 and AVR2R4) and PiAVR2-like (avr2diagF1 and avr2diagR1) (Fig. S1) at an annealing temperature of 61°C for 30 cycles. Quantitative gene expression analyses of RXLR genes were performed as described in Whisson et al. (2007), with Power SYBR® Green (Applied Biosystems, Warrington, UK), using a Chromo4® Real-Time Detector (Bio-Rad). ActA was used as an endogenous control gene as described previously (Bos et al., 2010). General PiAVR2 and PiAVR2-like expression was quantified using qRT-PCRfwd and qRT-PCRrev primers (Fig. S1). PITG_08949 expression was quantified using For 5′-AGGAATCTGAGACCGAGGAA-3′ and Rev 5′-GGGGGTTAATGGGATTGAG-3′. Data are presented as fold change relative to normalized expression in sporangia growth stage of each individual P. infestans isolate, calculated by the ΔΔCt method with error bars representing ± SE.

Presence/absence and sequence diversity in P. infestans isolates

PiAVR2 and PiAVR2-like were PCR-amplified from P. infestans isolates detailed in Table 1. Primers AVR2F1 and AVR2R1 (Fig. S1) were designed against PITG_22870 and amplified the predicted size product (541 bp) from PiAVR2. However, they repeatedly failed to generate PCR products from nine isolates (Fig. 2c; Table 1). The reverse primer AVR2R1 also proved unsuitable for initiating DNA sequencing of the PCR products as a result of a 1 bp indel adjacent to the primer (Fig. S1). Additional primer pairs AVR2F2/AVR2R2 and AVR2F4/AVR2R4 were designed (Fig. S1) to amplify this region from the nine isolates that failed previously and to provide improved sequencing primers. Both pairs amplified the expected product sizes (480 and 340 bp, respectively) from isolates that yielded products with the primer pair AVR2F1/R1 but neither amplified products from the nine isolates that failed previously. The primer pair AVR2F2/R2 proved optimal for PCR and sequencing of PiAVR2 and was used in all subsequent work. The use of the primer pair AVR2F2/R2 at a lower annealing temperature of 58°C generated weak PCR products from three isolates (01/29, MP618 and 06_3928A; Table 1). The sequences of these products differed from PITG_22870 and PITG_08943 at 25 bases, revealing PiAVR2-like (Fig. S1). Primer pairs specific for the PiAVR2-like sequence (avr2F6/avr2R6 and avr2F7/avr2R7 (Fig. S1) amplified PCR products of the predicted size (392 and 472 bp, respectively) from the nine isolates that had failed to amplify PiAVR2 but also from 12 isolates from which PiAVR2 was also amplified. Primers NitRedF (5′-GGACCGCTGGGCCACTTCAC-3′)/NitRedR (5′-CGCTGGCTTGCAGGCGTACT-3′) were used as control reactions to amplify a 435 bp region from the nitrate reductase gene (GenBank accession U14405; PITG_13012).

Table 1.   Details of isolates used in this study and details of PCR product amplification with primer sets (shown in Supporting Information, Fig. S1) specific to each of PiAVR2 or PiAVR2-like
GenotypeaIsolate nameOriginR2b virulenceAmplification with primers belowAmplification with primers belowSNPe within PiAVR2-like
AVR2 F2 and R2cAVR2 F1 and R1AVR2 F4 and R4SNPd within PiAVR2avr2 F7 and R7avr2 F6 and R6
  1. SNP, single nucleotide polymorphism; n/a, not applicable.

  2. aPhytophthora infestans genotypes are based on defined simple-sequence repeat marker profiles that will be reported in detail elsewhere.

  3. b‘+’ indicates ability to infect R2 plants and ‘−’ indicates the isolate triggers HR on R2 plants.

  4. c‘+/−’ indicates PCR amplification only when the annealing temperature was decreased.

  5. dSNP results in amino acid polymorphism N31K in PiAVR2. K/N is heterozygous.

  6. eTwo SNPs result in amino acid polymorphisms M10T and I92V in PiAVR2-like. MI/TV is heterozygous

  7. Protein alignments from the isolates above, indicating the N31K, M10T and I92V polymorphisms, are shown in Fig. S4.

8_A1SC_95 17_3_2UK+++K++MI
Misc88069The Netherlands+++K++MI
Figure 2.

PITG_22870 is PiAVR2. (a) qRT-PCR showed that, relative to the control gene ActA, the expression of both PITG_22870 (orange) and PITG_08949 (purple) was up-regulated in Phytophthora infestans isolate 88069 in zoospores (Z), germinating cysts (C), and 12 and 72 h postinoculation of susceptible cv Bintje (B12 and B72), relative to expression in sporangia (S), which was given a value of 1. Error bars indicate ± SD for three biological replicates. (b) Whereas expression of PITG_22870 (orange), PITG_08943 (purple) and, as controls, AVR3aEM, AVR3aKI and empty pGRAB vector (all white) yielded no response in cv Bintje (left), AVR3aKI alone caused a hypersensitive response (HR) in transgenic Nicotiana benthamiana expressing R3a (right), and only PITG_22870 triggered an HR in the R2 potato differential 1512 c(16) (middle). (c) A P. infestans transformant expressing PiAVR2 (PITG_22870)::Tdtomato fusion. The left panel shows the accumulation of the fusion protein in haustoria (H), which can be seen in relation to mycelium (M) in the middle panel. The right panel shows a higher magnification of haustorial accumulation of PiAVR2::Tdtomato fluorescence (H). White bars, 50 μm (left panel); 10 μm (right).

All PCR reactions were carried out with the same reagents in a 20 μl reaction volume using a Primus 96plus Thermalcycler (MWG-Biotech, Ebersberg, Germany). Each reaction contained 1 × GoTaq® Flexi buffer, 20 μg BSA, 1.5 mM MgCl2, 100 μM dNTPs, 0.8 unit of Taq polymerase (GoTaq® DNA polymerase; Promega), 0.2 μM of primers and c. 20 ng of template DNA. Amplification conditions consisted of one cycle of 94°C for 1 min, 30 cycles of 95°C for 30 s, 60°C for primer pairs AVR2F1/R2, F2/R2 and F4/R4, 58°C for primer pairs avr2F6/R6 and F7/R7, and 55°C for the primer pair NitRedF/R for 30 s, 72°C for 30 s and a final cycle of 72°C for 5 min.

Isolate 06_3928A genome sequence and alignment to T30-4 genome

For the genomic DNA extraction, Phytophthora infestans strain 06_3928A was cultured in rye sucrose agar (RSA) plates at 18°C for 12 d. Plugs with mycelium of P. infestans strain 06_3928A were transferred to modified plich medium (Kamoun et al., 1993), grown for another 2 wk at 18°C and then harvested for genomic DNA isolation using Omniprep kit (G-Biosciences, Maryland Heights, MO, USA; catalogue number 786-136) with minor modifications. For sequencing, the flow cells were prepared according to the manufacturer’s instructions using the Illumina pair read cluster generation kit, PE-203-4001. Sequencing reactions were performed mostly on 2G GAs (Illumina Inc., Chesterford Research Park, Essex, UK). The reference genome sequence of the P. infestans strain T30-4, annotation and gene/exon locations was downloaded from (GenBank project accession number AATU01000000). The generated raw reads with abnormal lengths and reads containing Ns were removed from the datasets. Filtered reads were use to align to the reference genome strain T30-4. Alignments were obtained with BWA software v0.5.7 (Li & Durbin, 2010) using as parameters a seed length (l) of 38 and a maximum of mismatches (M) of 3.

Western analyses of protein stability

PiAVR2, PiAVR2-like and PITG_08949 sequences were cloned into pDNR221 using almost identical primers as described earlier except with one nucleotide of TAA stop codon changed to make a sense codon AAA. LR clonase® (Invitrogen) was used to recombine correct sequences into pB7FWG2.0 (Karimi et al., 2002) C-terminal GFP-tagged plant expression vector, transformed into E. coli by electroporation, sequenced, and plasmids with confirmed inserts were transformed into A. tumefaciens strain AGL1, pSoup, pVirG cells by electroporation. Positive transformants were grown for 2 d at 27°C in LB supplemented with rifampicin, chloroamphenicol, and spectinomycin. For transient expression in N. benthamiana, the conditions described earlier were used. A final OD600 = 0.25 of pB7FWG2.0 containing cultures was achieved by mixing with an OD600 = 0.2 pJL3-p19 containing culture in a 2 : 1 ratio. Whole leaves were infiltrated with culture to allow 1 cm2 leaf discs to be cut out at 3 dpi for the protein extraction protocol. Leaf discs of c. 100 mg were ground in liquid N2, 200 μl extraction buffer was added (20 mM HEPES, 13% sucrose, 1 mM EDTA, 1 mM DTT, proteinase inhibitor cocktail tablet, 0.1% Triton – the DTT (dithiothreitol), Triton and proteinase inhibitor cocktail tablet were added fresh each time) and left to thaw on ice. Twenty microlitres of the whole lysate was mixed with 20 μl sodium dodecyl sulphate loading buffer. Samples were boiled for 5 min at 95°C, loaded on to a 12% Bis-Tris NuPAGE® Novex® Mini gel (Invitrogen) and run at 200 V, 120 mA and 25 W for 1 h, then membrane-blotted for 1 h at 130 V. Four per cent% milk powder was used to block the membrane. The primary antibody was monoclonal mouse GFP antibody (Sigma-Aldrich) at 1 : 10 000 dilution. The membrane was washed with PBS-T 0.1% before addition of secondary goat antimouse immunoglobulin (Ig) horseradish peroxidise antibody (Sigma-Aldrich) at 1 : 2000 dilution. ECL-Plus Western Blotting Detection Reagents (Amersham) were used for detection, according to the manufacturer’s instructions.

P. infestans transformations

For expression of PiAVR2K31 in P. infestans isolate 06_3928A, the gene was cloned from genomic DNA of isolate 88069 using primers shown in Fig. S1. The ClaI and SacI restriction sites generated were used to clone PiAVR2K31 into P. infestans expression pTOR, and transformation of isolate 06_3928A was carried out as previously described (Bos et al., 2010). To generate a P. infestans strain expressing PiAVR2::Tdtomato, PiAVR2 was amplified from isolate 88069 using an alternative reverse primer (Fig. S1) altering the stop codon to a sense codon, and cloned into pTOR::Tdtomato (based on pTOR::mRFP described in Whisson et al., 2007; but with the fluorescent protein gene Tdtomato in place of mRFP) using ClaI and SacI restriction sites. This was transformed into isolate 88069 as described in Bos et al. (2010). Isolate 06_3928A and this genotype expressing PiAVR2K31 and isolate 88069 were inoculated on to potato cv Craigs Royal and R2 differential 1512 c(16) as described previously (Bos et al., 2010) and lesions were observed 1 wk later. Inoculation of the 88069 strain expressing PiAVR2K31::Tdtomato on to N. benthamiana, and confocal microscopy were conducted as described in Whisson et al. (2007), but to image Tdtomato, it was excited with the 561 nm laser line, and the emissions were collected between 570 and 600 nm.


Map-based cloning of PiAVR2 gene candidates

Previously, F1 progeny of a cross between P. infestans parental isolates 80029 (race 2.4.7; A1 mating type) and 88133 (race1.; A2 mating type) were found to segregate for six dominant Avr genes, allowing these to be positioned within a genetic linkage map (Van der Lee et al., 1997, 2001). To facilitate positional cloning of AVR genes, a BAC library was constructed from T30-4, an F1 individual from this cross that contains all six segregating avirulence genes (AVR1, AVR2, AVR3, AVR4, AVR10 and AVR11) (Whisson et al., 2001). The BAC library was screened with AFLP markers E + GA/M + ATs513 and E + TG/M + CTs338 (Fig. 1a), defining the AVR2 map location. Four positive BAC clones that contained either one or other of the two markers were assembled into two contigs that did not completely span the AVR2-containing region (Fig. 1a). BAC end-sequencing allowed the contigs to be anchored to supercontig 1.16 of the P. infestans isolate T30-4 genome sequence (; Haas et al., 2009). Sequences at the outer ends of the BAC contigs indicated that the cloned fragments spanned 742.6 kb, within which was a 36.6 kb region containing three predicted RXLR-dEER effector-encoding genes (PITG_08943, PITG_08949 and PITG_22870) (Fig. 1a). Two of these (PITG_08943 and PITG_22870) are identical, apparently duplicated copies 2 kb apart. We confirmed that this is a genuine duplication in the T30-4 genome using PCR primers AVR2F10 and AVR2R10 (Fig. S1), which amplified between the duplicated gene copies to generate a PCR product of the expected 2.2 kb size (results not shown). As they are identical in T30-4, we refer only to PITG_22870 in the following. The third gene (PITG_ 08949) encodes a predicted protein that is similar across the first 79 amino acids, but which diverges considerably at the C-terminus (Fig. 1b), probably as a result of a DNA recombination event (Fig. S2).

PITG_22870 triggers R2-dependent HR

Transcripts of RXLR effector genes accumulate predominantly during the early, biotrophic stage of P. infestans infection (Whisson et al., 2007; Haas et al., 2009). Real-time qRT-PCR was used to investigate transcript accumulation of PITG_22870 and PITG_08949 during potato infection by isolate 88069, which is avirulent on R2 plants and possesses gene sequences that are identical to those in T30-4. Transcripts of both genes accumulated significantly in preinfection stages and during the biotrophic phase (1–3 d postinoculation), showing the characteristic expression (Whisson et al., 2007) of RXLR effector genes (Fig. 2a).

To test whether PITG_22870 and PITG_08949 are potentially PiAVR2, each was expressed, minus SP-encoding sequences, in potato cv Bintje, which lacks known late blight R genes; in the potato R2 differential 1512 c(16); and, as a further control, in transgenic N. benthamiana expressing R3a. PITG_08949 failed to elicit an HR in any of these plants, whereas AVR3aKI elicited an HR, as expected (Armstrong et al., 2005), only in N. benthamiana expressing R3a. PITG_22870 elicited an HR exclusively in the R2 potato differential (Fig. 2b). Both PITG_22870 and PITG_08949, fused at the C-terminus to GFP, are stable when expressed in N. benthamiana (see Fig. 4).

To verify R2-dependent recognition of PITG_22870, the full-length construct (minus SP) was expressed, via co-bombardment with GUS as a marker of cell vitality, in cv Bintje and the R2 differential clone 1512 c(16). Whereas no difference in GUS activity was seen following co-bombardment with PITG_22870 in Bintje, a considerable reduction in GUS activity, consistent with triggering the HR, was seen following co-bombardment with PITG_22870 exclusively on the R2 differential (Fig. S3). Independently, both PITG_22870 and PITG_08949 were coexpressed in N. benthamiana with the R2 gene, or with the orthologous genes Blb3, ABPT and R2-like, each of which provides a similar spectrum of resistance to P. infestans isolates. HR was observed with each R2 orthologue only when coexpressed with PITG_22870 (Lokossou et al., 2009). Taking these results together, PITG_22870 was tentatively renamed PiAVR2.

Previously, effector AVR3a was shown to accumulate at the site of haustorium formation during the biotrophic phase of late blight infection (Whisson et al., 2007). Similar localization patterns have been observed for AVR4 and AVR-blb1 (Van Poppel, 2009). To investigate whether this was also the case for PiAVR2, isolate 88069 was transformed to express PiAVR2, translationally fused at its C-terminus to the fluorescent protein Tdtomato. Confocal microscopy revealed that PiAVR2 exclusively accumulated at haustoria during infection (Fig. 2c), a property consistent with a potential virulence role during biotrophy.

Presence/absence and sequence polymorphisms in PiAVR2

To investigate polymorphisms that may explain P. infestans virulence on R2 potato plants, three PCR primer pairs, one located in 3′ and 5′-flanking regions (primers AVR2F1 and R1), one spanning flanking and coding regions (AVR2F2 and R2; Fig. S1) and one set located within the coding region (AVR2F4 and R4; Fig. S1), were used to amplify PiAVR2 from a diverse collection of 29 P. infestans isolates, 12 of which are virulent on R2 plants and 17 of which are avirulent. Strikingly, no PCR amplification was observed with any of the three primer sets from nine of the 12 virulent isolates, suggesting that PiAVR2, and its flanking regions, are either absent or highly divergent in these genomes. In the case of primer set AVR2F2/R2, lowering the annealing temperature resulted in weak PCR amplification products of the expected size from three of the nine virulent isolates that showed no amplification with AVR2F1/R1 and AVR2F4/R4 primers (Table 1). By contrast, all three primer sets amplified PCR products of the expected size from the remaining three virulent isolates and from all 17 avirulent isolates (Table 1).

The strong PCR amplification products generated with the AVR2F2/R2 PCR primer set from 20 of the tested isolates were sequenced. Only a single amino acid polymorphism, N31K, was observed in the predicted mature protein sequences derived from these PiAVR2 genes. Seven isolates were heterozygous for PiAVR2N31 and PiAVR2K31, and 13 isolates were homozygous for PiAVR2K31 (Table 1; Fig. S4). Both PiAVR2N31 and PiAVR2K31 sequences were represented within the three virulent and 17 avirulent isolates, indicating that this polymorphism is unlikely to specify differential recognition by R2. The PiAVR2 sequence was PCR-amplified using the AVR2F2/R2 primer set and sequenced from a further 30 European and North American isolates of undetermined virulence on R2 plants. Again, only the single N31K polymorphism was observed; 20 of these isolates were heterozygous for PiAVR2N31 and PiAVR2K31, and 10 isolates were homozygous for PiAVR2K31 (Table S2). Intriguingly, out of the 50 tested isolates containing this gene, no PiAVR2N31 homozygotes were observed.

The C-terminal effector domain of PiAVR2 is recognized by R2

To test whether the N31K polymorphism in PiAVR2 proteins determined recognition by R2, each form, minus SP-encoding sequences, was coexpressed with R2 in N. benthamiana. R2 was also coexpressed with the N-terminal half (from the cleavage site of the SP to the end of the EER motif) or C-terminal half (from the first amino acid after the EER motif to the stop codon) coding regions of PiAVR2N31 (the sequence within the assembled genome of avirulent clone T30-4; Haas et al., 2009) (Fig. 3a). We found that R2-dependent HR occurred, to a similar extent, with both PiAVR2N31 and PiAVR2K31 full-length (minus SP-encoding) sequences (Fig. 3b,c). Given the presence of the PiAVR2N31 and/or PiAVR2K31 forms in three isolates that can infect R2 plants, additional factors are needed to explain the virulence of these isolates. In agreement with this, no recognition of the N-terminus-encoding half of PiAVR2N31 was observed, whereas a stronger and faster HR was observed with the C-terminus-encoding half alone (Fig. 3b,c). Western analyses with each PiAVR2-derived construct, fused at the C-terminus to GFP, revealed that each fusion protein was detectable following agroinfiltration and expression in N. benthamiana, albeit PiAVR2K31 was apparently less stable than PiAVR2N31 (Fig. 3d; see also Fig. 4g). These results indicate that, similar to AVR3aKI (Bos et al., 2006), recognition of PiAVR2 is not dependent on the N-terminal half of the protein.

Figure 3.

Resistance protein R2 recognizes the C-terminus of PiAVR2. (a) Schematic of portions of PiAVR2 that were cloned and coexpressed with R2. (b) Coexpression of R2 with full length of PiAVR2N31 or PiAVR2K31 (both minus SP), or the N-terminal or C-terminal coding portions of PiAVR2N31, as indicated, in Nicotiana benthamiana. (c) Percentage of inoculation sites developing hypersensitive response (HR) following coexpression of R2 with full length (minus SP) of PiAVR2N31 or PiAVR2K31, or the N-terminal or C-terminal coding portions of PiAVR2N31, as indicated, in N. benthamiana. These results are the averages of three independent experiments each involving 24 inoculation sites per construct combination. Error bars indicate ± SE. (d) Western hybridization of green fluorescent protein (GFP) antibody to protein extracted 3 d postinoculation of plants expressing free GFP, full length (minus SP) of PiAVR2N31 or PiAVR2K31::GFP, the N-terminal or C-terminal coding portions of PiAVR2N31, or PITG_08949, all translationally fused at the C-terminus to GFP, in N. benthamiana. Ponceau stain (PS) and size markers (kDa) are indicated.

Figure 4.

PiAVR2-like evades detection by R2. (a) Plot of sequencing depth of coverage of Illumina reads from isolate 06_3928A aligned to the region of supercontig 1.16 from isolate T30-4 containing the two PiAVR2 paralogues (in red): PITG_22870 and PITG_08943. Arrows indicate regions where sequence reads from 06_3928A are aligned to PiAVR2 genes highlighted within grey vertical bars. The horizontal dashed line indicates the average coverage of the 06_3928A genome. Note the c. 14.8 kb subregion (from 1894.9 to 1909.7 kb) that shows reduced coverage in reads from isolate 06_3928A, indicating high sequence divergence in this isolate (dark square, repeat; red square, RXLR effector; grey square, gene). (b) Protein alignment of PiAVR2 and PiAVR2-like, revealing 13 amino acid polymorphisms between the two mature proteins. The RLLR and EER motifs (dark boxes) is indicated. (c) Presence/absence polymorphisms between PiAVR2 (PCR amplified with AVR2F2/R2 primers, annealing temperature 60°C) and PiAVR2-like (PCR amplified with AVR2F7/R7 primers, annealing temperature 58°C) across 12 virulent and 17 avirulent Phytophthora infestans isolates (arranged in the order shown, from top to bottom, in Table 1). PCR amplification of the control gene nitrate reductase is indicated. (d) Whereas the full-length (minus SP-coding) and C-terminus coding regions of PiAVR2 trigger HR when coexpressed with R2 in Nicotiana benthamiana, the equivalent regions of PiAVR2-like do not. Expression of PiAVR2 and R2 alone are indicated as controls. (e) Average percentage HR for 24 inoculation sites replicated in three experiments following expression of PiAVR2 or R2 alone, or coexpression of each construct combination in N. benthamiana (as in d) as indicated. Error bars indicate ± SE. (f) Average percentage HR across 24 inoculation sites for full-length (minus SP-coding) and C-terminus coding regions of PiAVR2, or equivalent regions of PiAVR2-like following expression in the R2 differential 1512 c(16). (g) Western hybridization of green fluorescent protein (GFP) antibody to protein extracted 3 d postinoculation of plants expressing free GFP, full length (minus SP) of PiAVR2N31::GFP or PiAVR2K31::GFP, the C-terminal coding portion of PiAVR2 (PiAVR2 C-t) fused to GFP, full-length (minus SP) PiAVR2-like::GFP, the C-terminal portion of PiAVR2-like (Piavr2 C-t) fused to GFP, or PITG_08949::GFP, in N. benthamiana. Ponceau stain (PS) and size markers (kDa) are indicated.

An alternative form, PiAVR2-like, evades recognition by R2

No PCR product was obtained using primer sets AVR2F1/R1 or AVR2F4/R4 from nine of the P. infestans isolates that are virulent on R2 potato, suggesting that PiAVR2, if present, is divergent at the DNA sequence level. Nevertheless, three of these isolates yielded a weak PCR product with the AVR2F2/R2 primer set when the annealing temperature was lowered (Table 1). One of these isolates, 06_3928A, is a representative of the 13_A2 genotype that is currently prevalent in western Europe (Cooke et al., 2009; Fry et al., 2009). The genome of this isolate has recently been sequenced using Illumina technology and will be reported elsewhere. Alignment of 06_3928A reads against the genomic region of T30-4, including the two PiAVR2 paralogues, PITG_22870 and PITG_08943, revealed that a subregion of 14.8 kb is highly divergent in this strain (Fig. 4a). Within the 14.8 kb subregion, very few reads could be aligned to the PiAVR2 coding sequence, suggesting the gene is divergent in the 06_3928A strain. The weak AVR2F2/R2 PCR products from the three P. infestans isolates were thus sequenced. They revealed a gene sequence that was conserved between the three isolates and which showed striking similarity to PiAVR2; 25 SNPs were observed between the coding regions (Fig. S1), resulting in 13 amino acid polymorphisms between the mature proteins (Fig. 4b). The combined assembly of the aligned and unaligned sequence reads from 06_3928A corresponding to the PiAVR2 gene confirmed the presence of the variant form (termed PiAVR2-like) within the 06_3928A genome. PiAVR2-like was absent from both assembled and unassembled reads of the T30-4 genome.

PCR primers (avr2F6/R6 and avr2F7/R7; Fig. S1) designed specifically to amplify this sequence variant yielded amplification products from all 12 isolates that are virulent on R2, and from nine of the 17 avirulent isolates (Fig. 4c; Table 1). PCR products were sequenced from 13 isolates that represent distinct P. infestans genotypes, as defined by simple sequence repeat markers (Lees et al., 2006). Only two polymorphisms in the predicted proteins were observed, M10T and I92V (Table 1; Fig. S4). The first is within the predicted signal peptide, which is cleaved from the mature protein during secretion, and is thus unlikely to affect recognition by R2. In the case of the I92V polymorphism, the Val residue is shared with the recognized PiAVR2 protein. If PiAVR2-like evades recognition by R2, the presence of the allele encoding a protein with the Val92 residue in virulent isolates indicates that this polymorphism is unlikely to specify recognition by R2 (Fig. S4; Table 1). Although a number of isolates contain both PiAVR2 and PiAVR2-like, all avirulent isolates contain the recognized PiAVR2 sequence, and all virulent isolates contain PiAVR2-like. This prompted us to test whether PiAVR2-like is recognized by R2.

Regions encoding the full length (minus SP) and C-terminal halves (from the first amino acid after the EER motif to the stop codon) of PiAVR2 and PiAVR2-like were independently coexpressed with R2 in N. benthamiana. Whereas both PiAVR2 constructs triggered R2-dependent HR, full-length PiAVR2-like did not, and the C-terminal portion very rarely triggered an HR (Fig. 4d,e). Moreover, expression of each construct in the potato R2 differential, 1512 c(16), again revealed an HR with PiAVR2, but an HR was infrequently observed with only the C-terminal half of PiAVR2-like (Fig. 4f). As before for PiAVR2, western analysis of all forms translationally fused at the C-terminus to GFP indicated that PiAVR2-like was stable on expression in N. benthamiana (Fig. 4g).

R2 orthologues from wild Solanum species show a similar spectrum of recognition

Previously, three R2 orthologues from wild Solanum species, BLB3, ABPT and R2-like, were all shown to possess the same spectrum of resistance and to respond with a strong HR when coexpressed with PiAVR2 (Lokossou et al., 2009). We coexpressed the full-length (minus SP) forms of PiAVR2N31, PiAVR2K31 or PiAVR2-like with each of these R2 orthologues in N. benthamiana. As expected, whereas both PiAVR2N31 and PiAVR2K31 triggered a clear HR with each, there was seldom a response to PiAVR2-like (Fig. 5). Thus, none of these R genes is expected to extend resistance to additional P. infestans isolates, such as 06_3928A, a representative of the prevalent genotype 13_A2.

Figure 5.

Resistance genes ABPT, BLB3 and R2-like show similar specificity to R2. (a) Panels showing coexpression of full-length (minus SP-coding) regions of PiAVR2K31 or PiAVR2-like with ABPT (left), BLB3 (middle) or R2-like (right) in Nicotiana benthamiana. Expression of the relevant R gene alone is shown in each panel. (b) Average percentage hypersensitive response (HR) for 24 inoculation sites replicated in three experiments following coexpression of PiAVR2N31 or PiAVR2K31 (combined in the graph as PiAVR2, as both forms were recognized to a similar degree; light grey bars), or PiAVR2-like (mid-grey bars) with ABPT (left), BLB3 (middle) or R2-like (right) in N. benthamiana. (R gene only, black bars) Error bars indicate ± SE.

Virulence on R2 plants is associated with presence or expression of only PiAVR2-like

Whereas PiAVR2 is present in all avirulent isolates, PiAVR2-like is present in all virulent isolates. Nevertheless, nine of the avirulent isolates, and three of the tested virulent isolates, possess both forms (Table 1). A possible explanation for the phenotypic differences in this class of isolate is differential expression of PiAVR2 and PiAVR2-like. To investigate this, PCR primers specific to the coding regions of PiAVR2 (F4/R4; Fig. S1) and PiAVR2-like (Fig. S1) were used to amplify the corresponding fragments from cDNA prepared 2 d after leaf inoculation. The following isolates were tested: T30-4 (which possesses only PiAVR2); 06_3928A (which possesses only PiAVR2-like); 06_4168B and 06_4168C (which are virulent but possess both forms); and 88069 and 06_4256B (which are avirulent but possess both forms) (Table 1). As expected, expression of only PiAVR2 was detected in T30-4, and of only PiAVR2-like in 06_3928A. In addition, expression of both PiAVR2 and PiAVR2-like was detected in the avirulent isolates 88069 and 06_4256B. However, in the virulent isolates 06_4168B and 06_4168C, expression of only PiAVR2-like was detected (Fig. 6a). Thus, both presence/absence and expression polymorphisms of PiAVR2 and PiAVR2-like explain virulence on R2 potato. qRT-PCR primers which amplify both forms (Fig. S1) were used to show that the corresponding PiAVR2 and PiAVR2-like genes show the expected biotrophic pattern of transcript accumulation early in infection in T30-4 and 06_3928A, respectively (Fig. 6b).

Figure 6.

PiAVR2 is a dominant gene specifying R2-dependent disease resistance. (a) RT-PCR showing expression of PiAVR2 (upper panel), PiAVR2-like (middle panel) and control gene ActA (lower panel) in isolates T30-4, 06_3928A, 06_4168B, 06_4168C, 88069 and 06_4256B, 24 h postinoculation of potato cv Bintje. Isolates in red are avirulent and those in black are virulent on R2 plants. (b) qRT-PCR showing up-regulation, relative to the control gene ActA, of PiAVR2 in T30-4 (upper graph) and PiAVR2-like in 06_3928A (lower graph) during the first 5 d of potato cv Bintje infection, relative to that within sporangia (given a value of 1). (c) Infection of leaves from cv Craigs Royal (upper panel; CR), which lacks R2, and the R2 potato differential 1512 c(16) (lower panel; R2), 6 d after inoculation with isolates 88069 (avirulent on R2), 06_3928A (virulent on R2) and 06_3928A transformed to express PiAVR2K31.

PiAVR2 is a dominant gene triggering R2-dependent resistance

To investigate whether R2-dependent recognition of PiAVR2 confers disease resistance, isolate 06_3928A, which is virulent on R2 potato plants and contains only the PiAVR2-like variant, was transformed to express the PiAVR2K31 allele cloned from the avirulent isolate 88069. Isolate 88069 and both transformed and untransformed 06_3928A strains infected cv Craigs Royal, which lacks R2. By contrast, the 06_3928A strain transformed to express PiAVR2 elicited a clear HR on the R2 differential 1512 c(16), whereas untransformed 06_3928A was able to infect this plant (Fig. 6c). PiAVR2 is thus a dominant gene that triggers R2-dependent HR and disease resistance.


A map-based cloning strategy was used to identify an RXLR-EER effector gene, PiAVR2, from the P. infestans T30-4 genome sequence (Fig. 1). Coexpression of PiAVR2 with the potato resistance gene R2 results in a clear HR (Figs 3, 4). Moreover, transformation of a virulent isolate of the pathogen to express PiAVR2 resulted in R2-mediated HR (Fig. 6), indicating that this is a dominant gene responsible for triggering R2-dependent disease resistance. Expression in plants of either full-length PiAVR2 protein minus the signal peptide, or of its C-terminal effector domain (from the amino acid following the EER motif to the stop codon) resulted in R2-dependent HR (Figs 3, 4), indicating that, as shown for other RXLR effectors (Birch et al., 2008; Schornack et al., 2009), recognition occurs within the host cell. Again consistent with other RXLR effector genes, such as AVR3a from P. infestans (Armstrong et al., 2005; Whisson et al., 2007), PiAVR2 transcripts accumulate during biotrophy and encode a protein that accumulates at haustoria (Fig. 2), which form an intimate association with the host cell during this stage of infection.

PCR amplification and sequencing of PiAVR2 from a diverse set of P. infestans isolates revealed that this gene was present in all avirulent isolates and was highly conserved (Fig. 4). Only a single amino acid polymorphism, N31K, was observed between the signal peptide and the RLLR-EER motifs in the predicted proteins. Of 50 isolates containing this gene, 27 were heterozygous for the PiAVR2N31 and PiAVR2K31 alleles, and 23 were homozygous for the PiAVR2K31 allele (Table 1; Table S2). The strong bias towards the PiAVR2K31 allele is reminiscent of the bias towards AVR3aEM, rather than the AVR3aKI allele, across a similar number of diverse P. infestans isolates (Armstrong et al., 2005). However, whereas AVR3aEM evades recognition by the potato resistance gene R3a, both the PiAVR2N31 and PiAVR2K31 alleles are recognized equally by R2. Indeed, the strong recognition of the C-terminal effector domain, which is conserved between the proteins encoded by these alleles, indicates that any genotypic bias in pathogen populations is not associated with evasion of R2-mediated recognition.

Gene duplication and recombination to create a new RXLR-EER effector

PiAVR2 was apparent as two identical, adjacent copies (PITG_08943 and PITG_22870) within the genome of P. infestans isolate T30-4, the consequence of a recent duplication event. Approx. 30 kb away is a related gene, PITG_08949, which is up-regulated before and during biotrophy, a hallmark of RXLR effector genes. The predicted protein sequences of PiAVR2 and PITG_08949 are highly similar across the first 79 amino acids, strongly suggestive of a common ancestry. After this they diverge significantly, providing distinct C-termini of 37 and 20 amino acids, respectively. These differences are not the result of frame-shift mutations, but rather are likely the result of a DNA recombination event (Fig. S2).

The N-terminal signal peptide and RXLR-EER domains of RXLR effectors can be regarded as a functional unit required for secretion and delivery to their site of action within the host cell. It is therefore reasonable to expect greater sequence diversity in response to selection pressure on the C-terminal ‘effector’ domains which, in addition to their proposed roles in manipulating host defences, must evade detection by R proteins. Indeed, positive selection has been detected mainly in the C-terminal portions of RXLR genes (Win et al., 2007). Regarding the N-terminal and C-terminal regions of these effectors as functionally distinct modules, a likely mechanism by which the latter could evolve distinct virulence specificities is through recombination, and this appears to be the case to generate either PiAVR2 or PITG_08949, the latter of which evades detection by R2 (Fig. 2). Further work is needed to determine the potential host targets of PiAVR2 and PITG_08949 and to investigate whether they play distinct, or related, roles in promoting P. infestans disease development.

Virulence on R2 plants

Copy number variation has been a feature of avirulence loci in P. infestans and P. sojae and has contributed to phenotypic variation (Jiang et al., 2006; Qutob et al., 2009; Dong et al., 2009). This is well documented particularly for PsAvr1a and PsAvr3a, with some avirulent P. sojae isolates containing multiple copies and some virulent isolates revealing complete deletion of recognized forms (Qutob et al., 2009). Both PCR and genome sequencing suggest that PiAVR2 is deleted from the genomes of many virulent isolates. Alignment of sequence reads from the virulent isolate 06_3928A against the genome sequence of the avirulent clone T30-4 revealed that a 14.8 kb subregion, including PiAVR2 paralogues, is highly divergent (Fig. 4a). In addition, for both PsAvr1a and PsAvr3a, transcriptional differences have contributed to virulence on plants containing the associated R genes (Qutob et al., 2009). It is apparent that not only sequence but also transcriptional variation has occurred in PiAVR2, as no expression was detected in two virulent isolates that contain this gene (Fig. 6a).

In addition to PiAVR2, and by using the genomic and PCR sequencing data, we discovered a divergent form, PiAVR2-like, which is present in the genomes of all tested virulent isolates. The predicted protein of this sequence differs from PiAVR2 in 13 amino acids, eight of which reside in the C-terminal effector domain. As this domain alone, from PiAVR2, is detected by R2, one or more of these polymorphisms must specify the evasion of R2-mediated HR by PiAVR2-like.

We found that nine of the 12 virulent isolates tested lacked PiAVR2 but possessed Piavr2. By contrast, eight of the 17 avirulent isolates tested possessed PiAVR2 but lacked PiAVR2-like. The remaining three virulent and nine avirulent isolates contain both forms. Testing two of these three virulent isolates revealed that PiAVR2-like was expressed whereas PiAVR2 was not. Therefore, not only does a combination of presence/absence polymorphisms and transcriptional silencing explain virulence on R2 plants, but the data may also suggest functional complementarity between PiAVR2 and PiAVR2-like, as absence or silencing of one form coincides with presence or expression of the other. It will thus be interesting to determine whether these sequences share a common function in pathogenicity. We noted that, of the two avirulent isolates tested that possess PiAVR2 and PiAVR2-like, both forms were expressed during infection, albeit to different levels (Fig. 6a), indicating that expression of one does not preclude expression of the other. However, critically, every isolate investigated in this study expresses one form or the other, or both, and it will thus be important to determine whether, like AVR3a (Bos et al., 2010), these sequences provide an essential role during infection.

Proposed strategy for durable disease resistance

Although both PiAVR2 and PiAVR2-like differ from each other by 25 nucleotides (13 amino acids), within each sequence we observed little variation. A single nonsynonymous change, N31K, was observed in the PiAVR2 protein sequences from 50 isolates, and only two polymorphisms, M10T and I92V, were observed in the PiAVR2 protein from 13 isolates, one of which is in the signal peptide and therefore cleaved from the mature protein (Table 1; Fig. S4). Whereas R2, and related sequences from wild Solanum species, BLB3, ABPT and R2-like, all recognize PiAVR2, none of these resistances extends specificity to PiAVR2-like. Given the apparent conservation of each effector sequence, and the ubiquitous presence of either or both forms, we propose that a search for R genes that detect PiAVR2-like is a high priority. Deployed in combination, R2 and an R gene that recognizes PiAVR2-like may provide effective and durable late blight resistances for the European potato industry, given that the pathogen genotypes studied here are representative of the wider European population. This study highlights the importance of understanding the molecular basis not only of recognition, which leads to disease resistance, but also of how the pathogen, at the molecular level, is equipped to evade such recognition. Armed with such knowledge, we propose that the targeted search for specific resistances, and their judicious deployment, can be carefully directed to control this versatile and economically devastating pathogen.


We thank the Biotechnology and Biological Sciences Research Council and Rural and the Scottish Government Environmental Research and Analysis Directorate for funding this research. J.M. was funded by an EU Alβan programme PhD studentship awarded to SCRI.