The ability to cause plant disease is a complex trait that occurs in only a small subset of bacterial and fungal species. Understanding the developmental and physiological adaptations of pathogens that allow them to invade plants, colonize tissues, and subvert plant metabolism is therefore a considerable challenge. A task that will be equally demanding will be developing an understanding of the molecular basis of pathogen recognition by plants, which underlies the evolution of disease resistance. Functional genomic analysis is poised to revolutionize our understanding of these complex biological systems (Brown, 2003; Wren, 2000), and reviews in this Special Issue of New Phytologist highlight recent advances in the application of functional genomics to the study of plant–pathogen interactions. These illustrate very effectively how organisms that were in the past experimentally intractable can now be investigated in far greater detail than was previously possible, and also highlight similarities and differences in comparison with the situation in other plant–microbe interactions, including legumes/nitrogen-fixing bacteria (Colebatch et al., 2002; Sprent, 2002) and mycorrhizas (Franken & Requena, 2001; Marsh & Schultze, 2001; Martin, 2001; Tunlid & Talbot, 2002)
The philosophy – reversing reductionism
Functional genomics stems from the availability of genome sequence information from an organism. The generation of genome sequence data from two model plant species, Arabidopsis and rice, a number of phytopathogenic bacteria and most recently a phytopathogenic fungus, has provided the raw information necessary for these new approaches. So, what is functional genomics and why does it offer such promise? A philosophical definition might be, ‘a holistic or systems-based approach to studying information flow within a cell’ (Brown, 2003). This is in marked contrast to molecular biology which has been a profoundly reductionist discipline, investigating the individual actions of genes in an organism and allowing metabolic or signalling pathways to be characterised in an empirically determined step-by-step manner. A second, more practical definition of functional genomics would be ‘the application of high throughput methods using automated technologies to biology allowing functional analysis of the genome, proteome and metabolome of an organism’ (see Box 1; also Colebatch et al., 2002; Tunlid, 2003). Genomics therefore embraces the inherent complexity of biological systems, allowing insight on the interplay of a large number of gene products and the consequences of this communication to the physiology of a cell. The reviews presented here give a glimpse of what can be achieved by applying functional genomics to plant–microbe interactions and the type of studies that are likely to be carried out in forthcoming years.
Obtaining the information: genome analysis of pathogenic species
The first step in functional genomics obviously requires generation of a high quality DNA sequence. The review by Mitchell et al. (pp. 53–61) shows the progress made towards generating a full genome sequence of the pathogenic ascomycete Magnaporthe grisea. This fungus, which causes a devastating disease of cultivated rice, is widely studied because of the relative ease with which it can be genetically manipulated and the fascinating developmental biology that it displays (Tucker & Talbot, 2001). The authors describe how a physical map of the M. grisea genome has been generated by anchoring short sequence reads from the end of BAC clones of a large insert genomic library of the fungus. In this way they have provided the skeleton outline upon which a full genome sequence, generated using information from a whole genome shotgun approach at the Whitehead Institute, can be constructed. The second draft assembly of the M. grisea sequence is already available (http://www-genome.wi.mit.edu/annotation/fungi/magnaporthe) and is in the process of being annotated. The recent description of the first genome sequence of a filamentous fungus, Neurospora crassa, has provided a glimpse of the treats in store in such a resource (Galagan et al., 2003). The 10 000 or so identified genes have already revealed significant differences in signal transduction apparatus in N. crassa compared with the relatively closely related yeasts, Saccharomyces cerevisiae and Schizosaccharomyces pombe, and the presence of an extensive array of secondary metabolic pathways. The N. crassa genome also contains the smallest number of duplicated genes in any eukaryotic genome, a sign of the extensive array of mechanisms – including, most notably, repeat-induced point mutations – that prevent duplicated gene sequences being tolerated in the fungus, probably as a defence against invading intracellular parasites, such as viruses or other forms of foreign DNA. The M. grisea genome sequence contains some of the same features, but analysis so far already indicates important differences in gene number, secondary metabolism and extracellular enzyme production that might be clues to the traits that determine its success as a pathogen. Dean and coworkers emphasise the need for bioinformatic resources and have established a central data repository for all forms of gene functional analysis of the fungus, which will serve the international community that works on the fungus. Similar resources to allow comparative analysis of phytopathogenic fungi are also being established and will allow the whole phytopathogen research community to benefit from the M. grisea genomic information (Soanes et al., 2002).
The use of bioinformatics to identify genes of interest in a pathogenic microorganism is also described in the review in this issue by Bos et al. (pp. 63–72). They have used the available expressed sequence tag collections for the potato late blight oomycete pathogen Phytophthora infestans very effectively to identify secreted proteins that may act as virulence factors, or as a consequence of such a role have become recognised as avirulence gene products by the products of plant resistance genes. An algorithm called Pexfinder was devised which allows putative signal peptides to be identified in raw cDNA sequences. This was used to select potential secreted proteins, which were then expressed in host plants using the Potato virus X expression system. This has enabled identification of a group of necrosis-inducing peptides in P. infestans that can now be studied in a systematic manner. Development of gene silencing strategies for P. infestans will provide an effective route for gene functional analysis in an organism that was previously difficult to study using genetics (van West et al., 1999). In a novel extension of their strategy, Bos et al. have identified polymorphic sequences in this group of genes putatively encoding low molecular weight secreted proteins and used linkage disequilibrium to determine if they correspond to known avirulence gene loci, or whether they represent novel avirulence gene specificities for which resistance genes have yet to be identified.
Dissecting the process of plant infection
Plant diseases caused by fungi normally start with a propagule landing on the leaf or root surface of a compatible host plant. Many fungal pathogens produce specialised infection structures to breach the outer cuticles of their plant hosts, while other locate natural openings such as stomata. The development of infection structures such as appressoria has proceeded rapidly in model species such as M. grisea, and related fungi that produce melanin pigmented appressoria such as Colletotrichum spp. (Tucker & Talbot, 2001). In these species, appressorium formation is regulated by a cyclic AMP response pathway and requires a mitogen-activated protein kinase (MAPK) encoded by the PMK1 gene in M. grisea (Xu & Hamer, 1996). This signalling cascade leads to cellular differentiation and production of appressoria that develop turgor and subsequently breach the cuticle using, predominantly, physical force (Thines et al., 2000). Less is known, however, regarding the molecular genetics of appressorium development in obligate pathogenic fungi such as the rusts and powdery mildews because of their relative intractability to investigation, although some genetic components are clearly shared (Bindslev et al., 2001). In soil borne fungi such as Fusarium oxysporum, which has a broad host range and is responsible for many economically important crop diseases, a wide variety of both forward and reverse genetic approaches have been carried out, as reviewed by Recorbet et al. (pp. 73–91). They highlight how comparative analysis of F. oxysporum with fungi such as M. grisea and the corn pathogen Cochliobolus carbonum has been made possible by carrying out a logical reverse genetic series, but in contrast how insertional mutagenesis is offering fresh insights into the attributes required for this soilborne fungus to be a successful pathogen.
Following infection of plants, biotrophic fungi such as the rusts and powdery mildews elaborate specialised intracellular structures, called haustoria – the review in this issue by Voegele & Mendgen (pp. 93–100) showcases how a multidisciplinary approach utilising high resolution microscopy, differential gene expression studies and biochemistry can provide very significant advances in our understanding of these specialised structures. Long thought to be simple feeding structures, Voegele & Mendgen show that haustoria fulfil other biosynthetic functions and may act to suppress plant defences and to alter the flow of metabolites within plants, dramatically altering the sink–source relationships in the plant in favour of the invading pathogen.
Host defence mechanisms
The host of course, is far from passive during plant infection and launches an orchestrated set of cellular and biochemical responses to pathogen invasion. Plants have a set of constitutive chemical barriers to infection, including preformed antimicrobial compounds such as saponins and cyanogenic glucosides. The biosynthesis and biological function of these compounds is described in the review by Osbourn et al. (pp. 101–108). They emphasise the potential for defence compounds to act in a variety of different ways to protect plants from pathogen attack. The use of genomic resources and gene functional analysis is also well documented in the ongoing investigation into the pathway for saponin biosynthesis in oats and its potential for providing engineered protection from root pathogens in other cereal crops. A recent study also shows that saponin detoxification, a counter-attacking process carried out by some pathogenic fungi during plant infection, may not simply inactivate preformed antimicrobial compounds, but additionally the degradation products generated by saponin detoxification may themselves act to suppress inducible plant defence responses (Bouarab et al., 2002). This highlights the fact that seemingly straightforward enzymatic functions with predictable substrates can lead to very unpredictable cellular consequences in the context of plant–microbe interactions.
Cultivar-specific resistance, resulting from the recognition of a pathogen-associated molecular pattern (an avirulence gene product), by the product of a plant resistance gene, can often result in a form of localised cell death called the hypersensitive reaction, which is very widely studied and shows parallels (and important differences) with programmed cell death, which occurs in metazoan development. The complexity of the hypersensitive response makes it a good target for genomic analysis using transcriptional profiling and proteomic analysis. An example of how gene expression profiling can lead to identification of novel regulatory gene networks can be seen by the study of systemic acquired resistance carried out using microarrays in Arabidopsis (Maleck et al., 2000). This study shows how genomics can rapidly define hitherto unrecognised associations between genes in a manner which fully embraces the complexity of the cellular response, while also clarifying the transcriptional response into a subset of regulatory networks that can individually be subjected to further experimentation. A review in this issue by Dale Walters pp. 109–115 also considers the potential involvement of polyamine catabolism in the generation of hydrogen peroxide prior to hypersensitive cell death and the biochemical mechanism by which this might take place.
The way ahead
The potential for investigating plant–microbe interactions using genomic approaches is very clear and it seems likely that genome sequence will become available for a number of phytopathogenic and mutualistic fungal species in the next few years. Harnessing this information effectively and learning lessons in how systematically to analyse genes in a high throughput manner (Oliver, 2002) will be the key challenges for the future.
Reviews highlighted here emerged from the 10th New Phytologist Symposium, ‘Functional Genomics of Plant–Microbe Interactions’, held in Nancy, France in October 2002.