British Society for Plant Pathology 25th Anniversary

25 years of plant pathogen interactions

by Richard Cooper, BSPP President 2007

The author 25 years ago.

In 1981 I went on sabbatical in the USA to work on fireblight, Erwinia amylovora. This was my first hands-on experience with a bacterial pathogen and I soon realised why the wealth of information at that time on interactions had come from fungal diseases. In short one could track fungal infections by light microscopy, antimicrobial compounds from plants were generally more active against fungi, spores as inocula were easily counted and fungi seemed more amenable to bioassays. Now, thanks to molecular genetic techniques with bacteria, the situation is reversed, but not entirely. I think it would surprise many current molecular-trained and -driven young scientists how much was known twenty five years ago. That knowledge of the biochemistry, physiology and ultrastructure of interactions still provides some of the basis for teaching, but major gaps there certainly were, as I will attempt to outline. Much of the contrast can be found by comparing the state of play in two multiauthored textbooks: Biochemical Plant Pathology 1983 edited by J. A Callow (Wiley) and Plant Pathogen Interactions 2004 edited by N. J. Talbot (Blackwell). I will attempt to bring the subject more or less up to date, but it is not meant to be comprehensive. Some key references of more recent work are included in the text for those that may wish to explore more deeply.

What was known then?

Host defences centred on the hypersensitive response (HR) for which there was excellent information on ultrastructure, sequence and timing of pathogen inhibition, and associated triggering and localisation of phytoalexins. The long list of phytoalexins from most families, their chemistry and toxicity was being added to with preformed compounds or phytoanticipins. Physical changes linked with defence were largely cell wall based. Pathogenesis-related (PR) proteins had been found in viral diseases but functions were unknown. Non-host resistance was ignored by most as too obscure; we knew little enough above host resistance based on major genes, let alone that conferred by minor genes, and in the "wrong" host. Triggers of defence were well recognised as elicitors from fungi, such as glucan and chitin oligomers and LPS from bacteria. No serious claims for race-specific triggering of HR existed. Pathogenicity was well represented with extensive information on toxins. Much was available for their chemistry, role and in a few cases, mode of action. Classical genetic information on host and pathogen underlined some Cochliobolus toxins as prerequisite for disease and rare explanations of specificity. Barrier-degrading enzymes (cutinases, pectinases, hemicellulases and cellulases) were well characterised and seemed to be involved in many diverse diseases involving necrotrophs and some hemibiotrophs. The highly destructive action of pectin hydrolases and lyases on primary walls along with their ability to kill plant cells, justifiably, gave them the spotlight. This mechanism is perhaps the most common means of cell killing by necrotrophs, but remarkably, is still unknown. Later we implicated fungal xylanases rather than pectinases in pathogenicity to graminaceous monocots as an adaptation to the xylan-rich primary wall matrix of cereals. Bacterial extracellular polysaccharides (EPS) also appeared to be key in some diseases. Siderophores and phytohormones were under investigation. Fungal biotrophy was well described ultrastructurally, such as the haustorial-host membrane interface, and in terms of nutrient uptake following the isolation of haustorial complexes and analysis of carbohydrates therein. Suppressors of recognition or of host defences were anticipated, but not identified. In spite of all this wealth of information, the bottom line of genetic proof was mostly lacking. Evidence often seemed overwhelming, yet remained circumstantial.

Host physiology. Much was known about the deleterious effects of disease on respiration, photosynthesis, phytohormone balance, nitrogen metabolism and water stress. It was realised eventually however that such information was not going to reveal the early critical events in interactions that we were seeking. These changes largely reflected downstream cascade effects. Efforts to study total biochemistry were thankfully waning. One might bear this in mind with regard to the current upsurge in metabolomics.

What we did not know then?

Host-pathogen specificity was a black box (other than for a handful of interactions involving host-selective toxins, for which host-pathogen genetics does not fit the usual pattern for most diseases). How are pathogens recognised (or not) and by what? Speculation included claims for differential synthesis or release of elicitors and existence of cultivar-specific elicitors. According to transmission electron microscopy bacteria bound to cell walls and there followed claims for specific binding involving host lectins and pathogen oligosaccharides. The binding later appeared to be artefactual as infiltrated water from bacterial suspensions is withdrawn from the apoplast. Resistance genes were being used to study the expression of defences, but this told us little of the early critical events. Pathogen avirulence genes clearly existed, but why and what were they? How did they interact with products of R genes? There were many models proposed but no molecules. With hindsight, we had no facile model systems to study the bigger picture. We knew little or nothing of the genes behind pathogen virulence and host defences. Also signalling molecules and pathways in hosts and pathogens were another barren area.

So what's new and why?

Transposons, cloning, site directed mutagenesis. Around this time, molecular genetics was established in model bacterial systems and being adapted for plant pathogens. Transposons with selectable markers were being used to induce mutations in Rhizobium and Pseudomonas syringae, and DNA cloning systems were becoming routine. Labs such as Staskawicz's started the search for avirulence and virulence genes. Perhaps the first example of a bacterial virulence factor came from Comai and Kosuge in 1982 cloning tryptophan 2-monoxygenase and introducing this into a plasmid-cured IAA minus strains of P. syringae pv savastanoi to restore virulence (but is was not essential for infection, just degree of symptoms). Many groups in the 1980s such as labs of Keen and Collmer investigated the multiple pectinases of soft rot erwinias with their relatively simple E. coli-type genetics. This was only possible with site-directed mutagenesis to remove stepwise the 4-6 genes to evaluate the individual or combined roles of the iso-enzymes. Here hangs a tale: pathogens turned out to be far more versatile and complex than might have been predicted (or were we just desperate for answers by then?). Many gene knockouts of putative pathogenicity factors had no effect on phenotype. For example most wall degrading enzymes are represented by several genes such that deletion of one seems to be compensated by others (e.g. Bindschedler et al. 2003, Fung. Genet. Biol. 38, 43). Walton's group have deleted many genes such as xylanases and glucanases from Cochliobolus carbonum with no alteration to pathogenicity. Detoxification of antimicrobial compounds can occur by different routes. Multiple genes, previously silent genes and functional redundancy has severely limited the number of clear answers given by this approach. There are clear links established however for some factors as diverse as avenacinase of Gaeumannomyces graminis (to detoxify the oat phytoanticipin avenacin) and HC toxin synthase of C. carbonum.

Forget the past; random mutants reveal hrp genes. A more progressive approach was to create random mutants and screen for non-pathogenicity or reduced virulence. Genes responsible could then be located, cloned and complemented back to restore normal function. A classic example and truly major advance came from a search by random mutagenesis for bacterial genes linked with HR. Many HR-defective mutants were readily obtained; unexpectedly and at that time confusingly, most mutants also lost pathogenicity. How could avirulence and pathogenicity be linked? These so called HRP genes (HR and Pathogenicity) mostly coded for type III secretion apparatus, known from animal pathogens like Salmonella and Yersinia to be responsible for secretion of virulence proteins. We now know that a long, filamentous HRP pilus transcends the plant cell wall and delivers HRP effectors directly to host cytoplasm or nucleus; the proteins are required for HR in incompatible interactions and pathogenicity in compatible responses and some are the once elusive avirulence gene products. Mutagenesis of fungi remains more problematic. REMI (restriction enzyme mediated integration) provided hope but has been surpassed by Agrobacterium-mediated transformation, used now routinely on a wide range of pathogens for insertional and targeted mutagenesis, as its T-DNA integrates on chromosomes usually as a single copy.

mRNA and proteins in planta. Other open-minded routes to unravel pathogenicity factors include: seeking in planta pathogen transcripts, such as revealed mpg1 coding for the hydrophobin of Magnaporthe grisea on rice leaf surface; or isolation of pathogen proteins in planta which led to identification of avirulence proteins from Cladosporium fulvum and Fusarium oxysporum in tomato leaf apoplast and xylem fluids respectively; development specific transcripts or proteins showed wall-degrading enzymes in infection structures and hexose and amino acid transporters of haustoria of Uromyces viciae faba, confirmation of their role alluded to above as feeding structures. Flax rust provided the original gene for gene concept of plant-pathogen interaction. Now the molecules involved are being revealed. AvrL567 rust genes were identified by map-based cloning and the small, secreted proteins are recognised by the L5, L6 anor L7 flax resistance proteins (Dodds et al. 2004, Plant Cell 16, 755). The avr genes are expressed in haustoria and another 20 haustorial expressed proteins were found from an isolated haustorium cDNA library; at least d/two are avr genes (avrM and avrP4) and their transient expression triggered HR in plants containing the corresponding R genes M or P4. Notably, even relatively intractable fungi such as obligate biotrophs and Oomycetes are open to these approaches. The transcriptome of surface structures of Blumeria graminis on barley revealed some cDNAs homologous to fungal pathogenicity and virulence genes; other unidentified homologues in this cluster might be proved "guilty by association". This biotroph appears able to carry out most primary metabolic processes and it appears unlikely that auxotrophy will explain its obligate nature; reliance on host cues is more likely (Both et al. 2005. Mol. Plant-Microbe Interact. 18, 125). For Phytophthora spp. there are intensive inputs as sequencing projects from interaction transcriptomes. For example, novel necrosis-inducing proteins crn1 and crn2 have been revealed from P. infestans. Also serine protease inhibitors were found, of which EP11 interacts with tomato P69 subtilisin-type proteases suggesting defence-counter defence crosstalk. EP11 also protects another protease inhibitor from plant proteases in intercellular fluids. The knowledge of pathogen (and related saprotroph) genomes is often required for identification of what are sometimes gene fragments (expression sequence tags or ESTs), which are being generated in large numbers from infection or starvation libraries. Soames & Talbot (2006. Mol. Plant Pathol. 7, 61) describe a comparative genomic analysis of fungi from ESTs including differences between free-living yeasts, more complex saprotrophic filamentous fungi and pathogenic fungi. One revelation was that B. graminis with its biotrophic lifestyle requires many gene products not found in even hemibiotrophs and possesses many unisequences of unknown function (68%) of which some are expressed during infection.

Pathogen genomes. The first plant pathogenic bacterial genome sequence was Xylella fastidiosa from citrus in 2000. Now there are around 80 completed or ongoing projects, including many fungi, as sequencing centres can now handle their larger genomes of the range 30-50Mb for ascomycetes and <250 Mb for Oomycetes. The list includes: Agrobacterium, Clavibacter subspp., Erwinia spp., Ralstonia, Pseudomonas syringae pvs., Fusarium spp., Magnaporthe, Phytophthora spp., Puccinia, Stagonospora, Streptomyces. Other than functional genomics, which interaction libraries are revealing, comparative genomics provides much fundamental information. In bacteria genomic islands differing in GC content and codon usage pattern from the rest of the genome reveal horizontal gene transfer. The regions are often pathogenicity islands containing type III secretion and associated effector protein genes, or toxins, and show the potential for rapid evolution to new pathogenicity. For example Xylella is a highly specialised, xylem-invading citrus pathogen introduced by an insect vector. Coincidentally and unusually, it lacks type III secretion and the ability to deal with lipid metabolism, yet its carbohydrate utilisation is evident. In contrast Ralstonia has much greater flexibility coincident with its mode of pathogenicity and need for survival outside the host. In particular, many of its genes control attachment, polysaccharide production and type III system and effectors (Salanoubat et al. 2002, Nature 415, 497). Who would have predicted that the archetypal necrotrophic, pectinase producing, cell macerating Pectobacterium (syn. Erwinia) atroseptica uses type III effectors, fixes nitrogen and needs the toxin coronatine (Toth & Birch 2005. Curr. Opin. Plant Biol. 8, 424)? This type of information really makes one think again and discard dogmas. Likewise, why does M. grisea carry nine putative cutinases when the fungus is well known for generating enormous turgor pressure to penetrate its hosts mechanically? Perhaps one should not be surprised when fungi such as Fusarium are predicted to produce 350-450 secreted degradative enzymes, from pectinases to nucleases. Nitric oxide is now known as an animal and plant signalling molecule but for what purpose does M. grisea contain four NO synthases? Mimicry of host molecules to subvert key host defences is evident from coronatine production by Pectobacterium and by P. syringae, as coronatine structurally and functionally mimics methyl jasmonate, a plant defence signalling molecule. Also Boucher's group showed that Ralstonia solanacearum produces GALA effectors. These seem to mimic plant F-box proteins and target plant cellular components for proteolysis through the ubiquitin ligase-mediated pathway. Interspecific gene transfer has recently been shown in fungi. The genome sequence of Stagonospora nodorum has a predicted gene with 99.7% similarity to ToxA of Pyrenophora. Mixed infection of wheat leaves is common and a population of P. tritici-repentis with significantly enhanced virulence appears to have arisen around 1941 and spread "tanspot" worldwide (Friesen et al. 2006 Nature Genetics 38, 953). Evidence for longer term evolution to pathogenicity comes from genome comparisons of Phytophthora sojae and P. ramorum with stramenophile photosynthetic algal ancestors. The expansion and diversification of protein families is linked with infection of plants, notably hydrolases such as cutinases, ABC transporters, toxins, proteinase inhibitors, and especially a superfamily of putative avirulence genes; data mining of the genomes and from functional genomics has identified >200 genes likely to code for secreted effectors. Secreted proteins have evolved significantly more rapidly than the overall proteome (Tyler et al. 2006. Science 313, 1261). Many bacterial virulence traits are controlled by quorum sensing in which bacterial cell populations act via diffusible signals molecules, such as homoserine lactones (OHHL) in E. carotovora and P. syringae. Other than the well established control of wall degrading enzymes and EPS, another gene activated or repressed by OHHL is nip. The Nip protein induces necrosis and is homologous to various necrosis-inducing toxins from Phytophthora, some true fungi and a streptomycete (Pemberton et al. 2005. Mol. Plant-Microb. Interact. 18, 343). Unculturable phytoplasmas have remained poorly understood despite their importance, until the recent sequencing of three of their genomes. AY-WB (Asters Yellows strain Witches' Broom) encodes >58 secreted proteins of which several are predicted to target cell nuclei as confirmed by fluorescent protein fusions; also gene transcripts were detected in infected plants (Bai et al. 2006. J. Bacteriol. 188, 3682).

Model systems, essential, but mind the gaps. Arabidopsis and its range of pathogens have enabled major advances in understanding interactions. All will be aware of the reasons, such as rapid life cycle, sequenced genome, tagged mutations and small size. We should though bear in mind its limitations. It does not harbour Rhizobium or mycorrhizas. Some pathogens appear to me forced rather than representing true diseases; these include most necrotrophic fungi and vascular fungi. For example Fusarium oxysporum invasion of leaves hardly mimics xylem invasion. Medicago truncatula as a legume model covers some of these deficiencies and is being groomed accordingly.

Avirulence genes and other suppressors of defences. Type III effectors are double-edged swords with the capacity to induce and suppress host defences. Their predicted function is to inhibit PAMP (see below)-induced defences. Most avr genes are likely to be fundamental to fitness or virulence or they would be shed to avoid recognition by R genes. P. syringae delivers 20-50 effector proteins into plant cells and R. solanacearum has >40 predicted effectors. Some but not all enhance virulence. Seeking their functions is a priority area and ongoing research is revealing their targets. Many comprise novel sequences that do not allow prediction of their function. Many host targets for effector proteins from mammalian pathogenic bacteria are described and involve subversion of host defences, but much less is known for plant pathogens. Programmed cell death, cell wall-based defences, hormone signalling, expression of defence genes and other basal defences are some putative targets of plant pathogen effectors (see Abramovitch & Martin 2004. Curr. Opin. Plant Biol. 7, 356). Much less is known about fungal and Oomycete avr genes although the search is on. Even Avr genes from powdery (B. graminis) and downy (Hyaloperonospora parasitica) mildews are being uncovered. Bioinformatics, through synteny allows prediction of such effectors. Oomycete avr genes share two motifs including RXLR near the N terminus. Both turn up as a superfamily in genomes of P. sojae and P. ramorum "avh" (avr homologues) genes. RXLR is conserved in the malarial parasite Plasmodium to transport proteins to the cytoplasm of human erythrocytes. RXLR has over 60 representatives and they are likely to be cytoplasmic effectors; these include AVR3a from P. infestans that triggers HR in R3a plants and suppresses HR induced by INF1 elicitin. Delivery is likely to be via infection vesicles and haustoria. Apoplastic effectors are often small and cysteine-rich, such as serine and cysteine protease inhibitors (Birch et al. 2006. Trends Microbiol.14, 8). Suppression of recognition is another likely role; for example avr4 from C. fulvum encodes a chitin-binding protein that may mask fungal chitin from recognition and from host chitinases. Evolution being what it is, this protein is recognised by a host R gene Cf-4, but the pathogen has countered this by mutating a disulphide bridge, which maintain function but avoids detection. Other fungi such as Colletotrichum and Uromyces also seem to have shielded their chitin, possibly by developmentally regulated chitin deacetylases. Colletotrichum also surrounds its infection vesicle with a matrix proline-rich protein resembling that found in plant cell walls, possibly avoiding detection by this means; the gene is switched off at the onset of necrotrophy. Current work from my lab is showing how EPSs from a wide range of bacterial pathogens block signalling and defence gene expression. Previously the abundant high molecular weight polymers were assumed to be merely protective from dehydration and UV. In fact their polyanionic nature confers another key binding property of the signalling cation calcium. Viral pathogens limit host defences by suppressing RNA silencing used by plants to target and degrade viral RNA (Baulcombe 2002, Trends Microbiol 10, 306.). Virus induced gene silencing (VIGS) is now used to study function of defence genes; the viral vector is engineered to contain part of the plant gene in the antisense orientation. The produced complementary RNA will bind to the corresponding RNA in the plant cell forming d/s RNA. This causes production of small interfering RNA molecules that target their specific sequence for destruction. VIGS characterisation of genes associated with powdery mildew resistance of barley is described by Hein et al. (2005. Plant Physiol. 138, 2155).

Resistance genes are eventually identified The five main groups comprise: [1] Pto from tomato-a serine threonine kinase which interacts directly in the plasma membrane with AvrPto from P. syringae pv. tomato. [2] Proteins with leucine rich repeats (LRR), a nucleotide binding site (NBS) and a putative leucine zipper (CC). LRR motifs are implicated in protein-protein interactions. From rice blast resistance protein Pi-ta interacts with AvrPi-ta from Magnaporthe. [3] Proteins which lack CC domain but possess Toll (receptor protein homologue from Drosophila) and interleukin-1 receptors. [4] Cf proteins targetting Cladosporium fulvum in tomato possess a transmembrane domain and an extracellular LRR region. This implies receptor activity outside the cell coincident with the apoplastic location of the pathogen. The first three classes all appear to be cytoplasmically localised, relevant to intracellular effectors of bacteria. [5] Resistance to the rice pathogen Xanthomonas oryzae is given by Xa21 which contains extracellular LRR and intracellular serine-threonine kinase domain. R genes are ubiquitous in plants; Arabidopsis has ca. 150 putative R genes, based on NBB-LRR homology. Comparison of mutant and naturally occurring alleles of R genes with loss of function or different specificities is revealing regions controlling specificity. With the two exceptions above there is a lack of evidence for direct interactions between R proteins and corresponding AVR proteins. That would be too simple. Space prevents detailed discussion here of complex issues involving complexes but see Martin et al. (2003) Annu. Rev. Plant Biol. 54, 23. The guard hypothesis proposes that effector proteins target host proteins, which regulate host defences, but plants possess guard proteins that prevent or recognise such interactions and trigger HR. In other words, R proteins may undertake surveillance of key physiological processes targeted by pathogens. For example in Arabidopsis the protein RIN4 is a negative regulator of R genes RPM1 and RPS2 and seems to play the role of a broad spectrum, molecular switch regulating at least two, probably three R protein-mediated defence pathways. These are activated when P. syringae AvrRpt2 targets and cleaves RIN4 (Day et al. 2006. Plant Cell 18, 2782).

Pathogen perception and defence-related genes. Following R-AVR interactions leading to HR, or perception of elicitors (now more often termed PAMPs or Pathogen Associated Molecular Patterns) in hosts or in non-hosts, a high proportion of host genes are up- or down-regulated. PAMPs continue to be uncovered, notably bacterial flagellin and elongation factor (EF-Tu); both are major conserved proteins that are recognised and cannot be readily altered (Kunze et al. 2004 Plant Cell 16, 3496). However, some pathogens (X. campestris, R. solanacearum, E. carotovora, A. tumefaciens) have evolved independently inactive flagellin and evade that detection system. A surprising PAMP addition is fungal xylanase. Defence elicitors can also derive from the host; oligogalacturonides are released on cell damage or by pathogen pectinases and are potent inducers of defence. Host inhibitory wall proteins PGIPs might limit polygalacturonase activity such that eliciting oligosaccharides generated from wall cleavage are not further degraded to inactive forms. Host surveillance molecules fit into two categories: Toll like receptors (TLRs) are transmembrane proteins with extracellular LRRs and NB-LRR proteins; Arabidopsis has >400 TLRs and 100 NB-LRRs.  However, only the receptors for flagellin and xylanase have been characterised. Many host genes are clearly linked to defence and include enzymes involved in biosynthesis of phenolics and phytoalexins, enzymes that modify and strengthen cell walls, PR proteins with enzyme activity against microbial cell walls or peptides that are directly antimicrobial. For example, PR1 in tobacco is induced up to 1000-fold and reaches 2% of leaf protein. Techniques allow these to be tracked with time and are revealing other numerous and diverse genes associated with defence. Expression profiling becomes even more sophisticated when combined with Arabidopsis genetics and genomics methods. Mutations that affect salicylic acid and jasmonic acid (JA) signalling in defence linked pathways) and synthesis of the phytoalexin camalexin have been used. Even non-hosts and non-model species can be analysed for potential defences; using cDNA-AFLP we described many new defence-related genes new to cassava (Kemp et al. 2005. Mol. Plant Pathol. 6, 113).

Non-host resistance. Non-host resistance accounts for the majority of disease resistance in natural situations. We knew from the 1980s that rust fungi could fail on the "wrong" host at any stage from germination to haustorium establishment-so called "switching points" of Heath. Clearly, constitutive and inducible responses are involved. Activation of the latter is probably brought about by PAMPs, somewhat analogous to activation of innate immunity in animals. PAMPS, also avirulent pathogens, mycorrhizal fungi and other molecules like SA, establish systemic protection against subsequent infection with virulent pathogens. This so called "priming" is reviewed by Mauch Mani et al. (2006. Mol. Plant Microbe Interact. 19, 1062). Non-host resistance is now being dissected by various means. Mackey's group have shown resistance of Arabidopsis to the non-pathogen P. syringae pv. phaseolicola is based on at least three pathways involving "basal defences", but not HR. When all are inhibited, growth reaches levels similar to that of compatible P. syringae pv. tomato. Forward genetic screens for Arabidopsis mutants with impaired penetration by barley powdery mildew B. graminis have revealed some novel loci. One (pen1) encodes syntaxin, which belongs to the superfamily of SNARE, proteins that mediate membrane fusion during vesicle trafficking. Pen1 mutants have delayed deposition of cell wall appositions on attempted penetration. The actin cytoskeleton and cell wall-plasma membrane interconnection seem to be important preformed but responsive elements as revealed in Arabidopsis defence to wheat powdery mildew and to rusts . Analysis of mutants impaired in the hormones JA, SA and ethylene show that they not only play key roles in cultivar-specific resistance but in non-host resistance. Inducible defence responses in non-hosts include synthesis and accumulation of phytoalexins. Searching for antimicrobial compounds is no longer a mainline activity, but by chance we discovered the first (and still only) inorganic phytoalexin. Elemental sulphur (S0) is produced as a component of induced defence against xylem-invading vascular pathogens in diverse families including tomato, tobacco, cotton, Phaseolus bean and the species in which it was first discovered, Theobroma cacao. S0 is of course highly fungitoxic and was localised by SEMEDX to xylem cells. In Arabidopsis leaves S0 is constitutive(Cooper & Williams, 2004. J. Exp.Bot. 55, 1947). According to Nurnberger & Lipka (2005 Mol. Plant Pathol 6, 335), non race-specific and race-specific defences should be considered as distinct but evolutionarily interrelated and together constitute plant innate immunity. Analogies with animal innate immunity include the FLS2 receptor of flagellin in Arabidopsis that is related to animal TLR receptors; also nitric oxide and MAPK cascades are key elements in both defence systems.

Applications of resistance? Many defence-related genes have been tried as transgenes to enhance resistance. However, engineering resistance to diseases has proved much more recalcitrant than to insects and viruses. Genes such as those coding for hydrolases are best used in combination but there are few commercial examples to date. Most hopes rest with R genes, but they cannot be readily moved between taxa. They may however be accessible to domain and specificity alterations. Placing R and cognate avr genes under an infection-inducible promoter in the same plant is an exciting strategy that could provide resistance to all pathogens that trigger that promoter. It has been shown to succeed in tomato expressing both Cf9 and avr9. Conserved domains in R genes allows searches in crop species to clone gene fragments known as resistance gene analogues or RGAs, and using them as RFLP markers or in genetic mapping. Tight genetic linkage between RGAs and R genes has been found in monocots and dicots. The approach may be especially useful for species relatively intractable for resistance screening and breeding, such as oil palm and cassava. Even where R genes are well known they can be especially vulnerable to pathogen adaptability, never more so than with potato late blight. Map based cloning with LT-PCR has provided a major R gene RB from Solanum bulbocastanum, a diploid, wild relative, highly resistant to all known P. infestans races; RB gives broad spectrum resistance to P. infestans and should therefore be durable. Certain potato varieties are favoured by the processing industry, but backcrossing S. bulbocastanum derived germplasm may not be efficient and acceptable thanks to the tetraploid and heterogeneous potato genome. Favourites such as blight susceptible Russet Burbank may be rendered resistant by engineering the RB gene (Song et al. 2003. PNAS, 100, 9128). Lack of durability of R genes has been their Achilles' heel, but some have proved long lasting. Durability can in some cases be predicted by the fitness penalty that loss of the corresponding avr gene imposes. All of these possibilities await acceptability of transgenic technology in UK and some other crop production systems.

In Conclusion

The field of plant-pathogen interactions is at a truly exciting stage. This is not only because of the technologies that can now be used, but because there is real awareness that one has to operate on both sides of the fence. This is epitomised by the questions raised when considering pathogen Avr products (what are they?) targeting plant defence pathways (what is targeted?) and suppressing PAMPS perception (which ones are produced in planta?), but plants with cognate R genes guard their targets (how?) and consequently can perceive PAMPS (what are the receptors? what then results?). In the past too many researchers were focussed on the host or the pathogen. However, mechanisms of pathogenicity and defence are inextricably intertwined as they have undoubtedly co-evolved. This is why I have chosen the theme of "Attack and Defence" for my Presidential Meeting to be held in Bath in September 2007. At the current rate of progress there will be many more conceptual breakthroughs to discuss then.