Most of the GM crops in production at the moment have modified crop protection characteristics, chiefly protection against insects and from competition (herbicide tolerance). While there are some commercial virus resistant crops, there are further possibilities for improvement, especially in relation to the plant diseases that prevail in developing countries. Progress, particularly in using pathogen-derived resistance, was reviewed by Roger Beachy (Donald Danforth Plant Science Center, St. Louis; Beachy, 1997). A more difficult challenge has been to engineer resistance to a range of fungal pathogens. Nevertheless, considerable progress has been made recently in understanding the genetics of the interactions between fungi and plants. Jonathan Jones (John Innes Centre, Norwich, UK) described advances in the isolation of resistance genes and also their molecular structure (Hammond-Kosack and Jones, 1997; Noel et al., 1999; Parniske and Jones, 1999). Arabidopsis carries around 100 resistance genes, whereas crop plants may have two to three times this number. Many have been used in conventional breeding by introgression from wild sources. This is a slow process, and the durability of a single resistance gene is short when deployed in monoculture. The molecular description of the resistance genes should enable them to be moved more rapidly, either by marker-assisted breeding or by transformation into crops. It should also enable a range of different resistance genes to be assembled in different transgenic lines of the same cultivar so as to allow mosaics of resistance genes to be used within a single field. The understanding of the interaction of the Leu repeat regions of resistance genes with the recognition proteins of the pathogen, allied with techniques of gene shuffling and in vitro evolution, opens up opportunities to produce new resistance gene variants.
Whereas agrochemicals have been successfully used against biotic stresses, they have had little effect in protecting crops against abiotic stresses. There are indications that biotechnology may be more suited to achieve this goal. One of the widest ranging stresses in world agriculture is pH, with some 40% of arable land being too acidic and another 20% too alkaline for optimal crop production. Acidic soils lead to metals such as aluminum becoming toxic, and to nutrients such as phosphate becoming deficient. Herrera-Estrella described an approach to engineering stress resistance based on the observation that certain acid-tolerant plants excrete organic acids that chelate and trap aluminum in the rhizosphere (De la Fuente-Martinez and Herrera-Estrella, 1999; Herrera-Estrella, 1999). Generating maize plants that overexpress cytosolic citrate synthase leads to excretion of organic acids by their roots. In experiments using plants in pots, these plants were found to have enhanced tolerance to toxic concentrations of aluminum and an increased capacity for phosphate uptake. Although Herrera-Estrella has sufficient material to conduct field tests, he pointed out that such GM field experiments in Mexico have been blocked under pressure from Greenpeace and other environmental groups.
Water is probably the crop resource that is in shortest supply and this condition will worsen. In addition, the quality of the water used for irrigation will decline because of a greater salt load. Because plants need to have their stomata open to take up CO2 for carbon fixation, they lose water continuously through transpiration. This water needs to be replaced by the uptake of water from the soil. Can plants be created that lose less water in times of water deficit and yet carry out photosynthesis and grow? Julian Schroeder (University of California San Diego, La Jolla, CA) discussed research elucidating the genetics of some of the steps in the opening and closing of stomata. Stomata are regulated by major signal cascades involving abscisic acid, cytosolic calcium, protein kinases and phosphatases, potassium channels, and farensyl transferase (Ichida et al., 1997; Allen et al., 1999). Several of the genes involved have been cloned, and transgenic plants have been made. The results hold out the hope that modifications in stomatal control may be made that would favor more efficient water use.
The recent development that was most widely mentioned during the conference was that of the -carotene-rich, yellow rice created by I. Potrykus, P. Beyer, and colleagues (Fig. 2). Ingo Potrykus (Swiss Federal Institute of Technology-ETH, Zurich) described the science behind this advance and some others in his laboratory (Ye et al., 2000). Rice endosperm does not contain any provitamin A (-carotene). Theoretically, four enzymes complete the pathway from geranylgeranyl pyrophosphate to provitamin A, and genes for these enzymes were isolated from Narcissus and Erwinia. These genes were combined in transgenic rice, and some stable lines with yellow endosperms were produced. Biochemical analysis confirmed that the color was due to provitamin A. This was present in sufficient amounts such that a typical Asian rice diet using these lines would provide the daily requirement.
Iron deficiency has also been approached by a multigene strategy in which genes for phytase, ferritin, and a Cys-rich metallothionin-like protein were transferred and expressed in rice endosperm. Current lines have around twice as much iron as the wild type. This modified rice now has to go through the normal biosafety and agronomic tests prior to release into Asian fields. These are significant hurdles. However, there may be more obstacles, as de Greef noted that the EU rice buyers, who buy only a small proportion of Thai rice, had notified Thailand that its rice risked rejection if any of it was found to contain GM material. Thus, the EU and the powerful multinational buyers of cereals are exerting pressure that could block the introduction of what he described as "one of the most significant biotechnological developments of the last decade."
The major harvested products of plants are polysaccharides, particularly starch. Lothar Willmitzer (Max Planck Institute of Molecular Plant Physiology, Golm bei Potsdam, Germany) discussed approaches he and his colleagues have taken to modify carbohydrate metabolism. They have made transgenic plants that produce modified starches that might be useful in industry. Producing such starches in plants removes the need for certain chemical modifications that have environmental side effects (Lloyd et al., 1999). However, to achieve this has not been easy and the detailed story shows many examples of transgenic plants that failed to behave in the manner predicted from the biochemical hypotheses on which the choice of transgenes were based (e.g. Veramendi et al., 1999). The research has also led to identification of proteins and genes in starch granule breakdown that were not known nor predicted by biochemistry.
Biopharming is a term used to describe the use of transgenic plants to produce pharmaceuticals. Mitch Hein (EPIcyte Pharmacetical, San Diego, CA) described the production of antibodies in plants. The first success led to the synthesis in tobacco leaves of antibodies effective against dental caries. Subsequently, it has been possible to synthesize high-affinity monoclonal secretory antibodies that can prevent microbial infection in humans. The technology can thus be considered for other immunotherapeutic uses, especially in mucosal tissue (Ma et al., 1999). Plants could be particularly valuable as commercial production systems for antibodies that are needed in large amounts. Current pharmaceutical production is expensive and limited; the total Western hemisphere fermentation production capacity is 500 kg. If antibodies are to be used for prophylaxis, production would need to be in excess of 5,000 kg/year. Plant production systems could provide this, but the current limitation is devising suitable purification procedures. Hein identified a number of potential targets, such as contraception and sexually transmitted herpes.
Although there have been a range of commercially successful transgenic crops, the technologies used for transformation have been relatively crude. There is a need to improve the efficiency of transformation, to limit the presence of unnecessary genes in the products, to direct insertion to specific sites, and to give more control over when, where, and how much expression of the transgene occurs. Not all of these topics were addressed, but Nam-Hai Chua (Rockefeller University, New York) described improvements in transformation based upon the utilization of genes promoting endogenous hormone production under the control of chemical signals. One system uses the ipt (isopentenyltransferase) gene from the Ti plasmid of Agrobacterium tumefaciens to increase cytokinin levels, leading to the generation of shoots from transformed plant cells (Kunkel et al., 1999). However, these shoots retain the shooty phenotype and result in sterile plants. Using a construct in which the ipt gene is only active in the presence of a chemical inducer, the transformation and early cultivation of the cultures is carried out in the presence of the inducer. When shoots have formed, the inducer is removed and whole fertile plants can be recovered with high efficiency. The use of chemical induction cassettes in conjunction with the CRE-lox recombination system (Ow, 1996) has allowed Chua and colleagues to trigger the removal of transferred DNA from transgenic plants. This has considerable potential for crop biosafety through the ability to remove transgenes, which are no longer needed, before the product reaches the fields and the market.
Pal Maliga (Waksman Institute, Rutgers University, Piscataway, NJ) pointed out that using the plastid genome as a site for transgenes has several advantages over the nuclear genome. These include transfer of the genes only via the female line (for many but not all plant species), thus preventing the movement of the transgene to wild species via pollen; very high levels of expression (so far up to 20%-25% of total cellular protein); and targeted homologous recombination into the plastid genome. The technology is challenging and only works routinely with tobacco, although Maliga is optimistic that it will soon work well in rice (Maliga and Nixon, 1998; Khan and Maliga, 1999).
Markers and Crop Improvement by Non-Transgenic Methods
Recombinant DNA technology has given rise to a range of methods that allow the genome to be tagged with DNA markers (Karp et al., 1997; Davis et al., 1999; Vuylsteke et al., 1999). Some of these methods and the new developments that may be expected, were reviewed by Mark Zabeau (University of Gent, Gent, Belgium). The crucial drive is to provide systems that can be automated using robots, chips, sophisticated analyses (e.g. MALDI-TOF), and computers so that large populations of plants can be handled in a cheap and routine manner. Based on the discussions at the conference and elsewhere, it would seem that considerable success is on the way. Such technology has a number of uses, the most important being the identification of agronomic trait loci and their movement into adapted cultivars using marker assisted breeding. This method of crop improvement is not (currently) subject to criticism and even received the vocal approval of Haerlin. Once loci have been identified, map-based cloning can be used to isolate the genes involved. Marc van Montagu (University of Gent) pointed out that this could be of value to mine for genes of importance (e.g. those responsible for the synthesis of pharmacologically active compounds) from exotic species.
The practical value of marker technology was exemplified by the talk of Susan McCouch (Cornell University, Ithaca, NY), in which she described an experiment done with rice involving scientists in many centers in different countries, which was designed to discover quantitative trait loci important in crop performance and to recombine favorable alleles at those loci. Oryza rufipogon, a wild relative of rice, was crossed with three elite cultivars adapted to different growing environments in China, Korea, and Columbia, and 300 backcross lines were derived. These lines were tested in a wide range of environments and evaluated for 12 key agronomic traits. Several lines showed superior performance to the parents (Xiao et al., 1998). Subsequent marker analyses showed that, surprisingly, not all of the favorable alleles were in the adapted cultivar. O. rufipogon contributed alleles that were consistently associated with improvements in yield, quality, maturity, and plant height. This work, together with other examples, suggest that reservoirs of genetic diversity that reside in the wild relatives of crop species may contain numerous alleles that can provide the key to future increases in the productivity of a number of crops (Tanksley and McCouch, 1997). The great advantage of this approach is that it overrides preconceived notions of what lines or which traits might be valuable in a cross, instead allowing the genes to "speak for themselves." This is a humbling notion for scientists, and McCouch found considerable difficulty in getting breeders interested in participating in the analysis of a cross involving such an apparently useless parent as O. rufipogon.
The Big Problem/Opportunity
Tremendous opportunities for crop improvement are likely to arise as a result of the complete sequencing of plant genomes. This is well under way and due to be completed soon for Arabidopsis (Lin et al., 1999; Mayer et al., 1999); a rice genome sequence should not be far behind. The information that we have on plant synteny (Fig. 3), particularly for the grasses (Gale and Devos, 1998), should ensure that the results are useful for a much wider range of plants than those sequenced. The problem is in deriving, in a timely way, knowledge of those genes that are important for crop function or for the synthesis of high-value molecules from the great mass of sequence information that is being generated.
Geoffrey Duyk (Exelixis Pharmaceuticals, San Francisco) outlined some general approaches to the problem both from a broad life science perspective and from the perspective of his own company. The innovative drive of the technology is important in that it has already reduced the cost of sequencing 10-fold since the beginning of the human genome project, and Duyk expects another 10-fold reduction in the future. Automated, high-capacity technologies and informatics have been recruited to the biological research laboratory to deal with the sequence information generated and the opportunities that it presents. His view is that this is leading to a shift in the paradigm of research as the gathering and presentation of data has become an end in itself, resulting in a dissociation of data acquisition from classic hypothesis-based research. As this process gains speed, many more genomes, particularly of important pests and pathogens, will be targeted. Exelexis is particularly focused on insects and nematodes in the hope of aiding the discovery of novel insecticides and nematocides.
John Ryals (Paradigm Genetics, Research Triangle Park, NC) discussed his company's approach to identifying new target sites to aid in the discovery of crop protection chemicals based on high-throughput DNA sequencing, high-throughput reverse genetics, and knowledge-based computer systems. Reverse genetics is based on generating transgenics overexpressing cDNAs in the sense and antisense orientation, and then analyzing their phenotype. The generation of mutants by this or any other approach (such as tagging) is likely to be one of the limiting steps. Phenotypic analysis will be automated as far as possible. The approach is to generate and digitize images of the plants, to conduct biochemical profiling via HPLC/mass spectroscopy for around 5,000 molecules of under 5,000 D, and to interrogate the database iteratively. This is likely to generate huge amounts of data that will need to be stored and accessed. A primary focus is thus on data management. It is estimated that 10 terabytes will be generated in the analysis of Arabidopsis. Such a program needs a large amount of resources focused on the ultimate objectives. Although the major multinationals were not represented at the conference, it is likely that similar resource-intensive approaches are being used by them.
An alternative company approach was described by Guy della-Cioppa (Genomics Biosource Technologies, Vacaville, CA). His company has developed a plant expression system based on RNA viruses such as the tobacco mosaic virus (Kumagai et al., 1995; Della-Cioppa and Grill, 1996; Baulcombe, 1999; Koo et al., 1999). These virus vectors are sprayed on a growing plant and rapidly multiply, giving very high expression of their inserts, which can be in either the sense or the antisense orientation. This approach allows the identification of genes whose expression leads to a marked change in phenotype in the leaves (for example, infection with a phytoene synthase insert rapidly leads to yellow plants); it may be more limited in cases in which the gene controls early development or does not function in leaves.
The major public sector initiatives were represented by Michel Caboche (Institut National de la Recherche Agronomique, Versailles, France). Unfortunately, for various reasons there were no speakers from the U.S. or other European countries. France has set up the Genoplante project. This is a 5-year scientific program, involving private and public laboratories and funding, which will share resources and data. So far, around 80 projects have been funded. The goal of the project is to find a way to co-operate with other genome projects. Its main goals are to develop expertise, infrastructure, and competitiveness in plant genome analysis. The program has two components, a generic part that focuses on model genomes (Arabidopsis and rice), and a more commercial part that will analyze the genomes of major crops (wheat, maize, oilseed rape, and sunflower) and their syntenies with the model species. The program aims to identify genes and alleles useful for molecular breeding by positional cloning and candidate gene approaches. Caboche sees the commercial approach generating industrial property rights and new biotechnology companies, but the program does not include the generation of GM crops. Results from the generic part will be published and placed in a Genoplante database. Within the Arabidopsis program, the French projects (Camilleri et al., 1998; Desprez et al., 1998) have, or aim to produce, 5,000 non-redundant expressed sequence tags, a yeast artificial chromosome library and physical map, and 50,000 T-DNA mutant lines produced by vacuum infiltration. Caboche outlined how some of these resources are being utilized within his own laboratory in the analysis of lipid metabolism in seeds.