The next challenge will be to translate this technology from a model system into a practical vaccine (Box).

Selecting a vaccine species: While tobacco is a good model system for evaluating the production of recombinant proteins, it produces toxic compounds which make it unsuitable for vaccine delivery. Clinical trials have shown the induction of immune responses with antigen expressed in potato and lettuce.13,14 Lettuce is a fast-growing species suitable for direct consumption and experimental studies. Another practical alternative may be rice, which is commonly used in baby food because of its low allergenicity. Recent studies have shown that mammalian proteins can be expressed to high levels in transgenic rice.3 Furthermore, rice is easy to store and transport, and protein expressed in rice grains is stable at room temperature.3 Rice flour can also be mixed with baby food, clean water or breast milk for delivery to infants. However, rice grows slowly and requires specialised glasshouse conditions, making it a restrictive species for preliminary studies. Future development will likely see the transformation of crop species such as rice for the delivery of vaccine antigens.

Oral antigen immunogenicity: Oral vaccination requires a higher antigen dose than either intranasal or parenteral vaccination.15 A question frequently asked is whether realistic quantities of edible plant material will be able to supply sufficient antigen to generate protective immunity. Three successful human clinical trials have shown that adequate doses of antigen can be achieved with plant-based vaccines.6,13,14 Preliminary analysis of MV-H transgenic lettuce plants produced in our laboratory suggests that 35–50 g of lettuce should be sufficient to deliver doses of MV-H protein comparable to those used in clinical trials (unpublished data).

However, as with a number of soluble protein antigens expressed in plants, the induction of consistent MV-specific immune responses following vaccination with plant extract currently requires the use of a mucosal adjuvant such as cholera toxin,2,12 which enhances immune responses at mucosal surfaces and reduces the oral dose required to induce an immune response. Inducing an immune response in the absence of adjuvants may be possible if antigen doses can be increased. Delivering the vaccine in intact plant material, rather than in plant extracts, may enhance antigen immunogenicity, as bioencapsulation of the antigen within the tough plant cell wall and membrane compartments can provide increased protection from intestinal degradation.16 High-level protein expression in seeds such as rice may also concentrate the antigen and further reduce dosing requirements. More research will be necessary to determine if enhanced antigenicity can replace the use of mucosal adjuvants.

Integration of edible vaccines into combined vaccine strategies: Immunisation strategies that combine different routes of administration or vaccine types frequently result in enhanced protective immune responses.17 Edible vaccines have considerable potential for use in such "prime-boost" strategies, particularly where multiple antigens or doses are required to induce immunity. For example, we found that a single-dose MV-H DNA inoculation followed by multiple MV-H boosters, delivered orally as a plant-derived vaccine, could induce significantly greater quantities of MV-neutralising antibodies than vaccination with a DNA- or plant-derived vaccine alone. Neutralisation titres up to 20 times greater than those considered protective in humans were achieved.18

Will an edible MV vaccine induce oral tolerance? Conventional subunit vaccines have not been associated with oral tolerance. However, repeated exposure to an oral antigen has the potential to produce immunological tolerance.19 The induction of oral tolerance is both time-dependent and dose-dependent. The antigen dose necessary to induce protection is generally smaller than that required to produce tolerance.20 In addition, repeated or continuous exposure is usually necessary to induce tolerance.19 Expression of vaccines in commonly consumed foods does not mean that they will become a component of regular diets. As medicines, edible vaccines should be administered appropriately. In this setting, we believe it is unlikely that an edible vaccine would lead to oral tolerance. However, this remains an area of vaccine development that needs to be closely monitored.

Other health issues: Recent evidence has also shown that vaccination with MV-H-subunit vaccines is unlikely to prime for "atypical" measles, as was seen after use of the formalin-inactivated, alum-precipitated measles vaccine in the 1960s.21 While current measles vaccination programs may indirectly facilitate the emergence of vaccine-modified measles as a result of waning immunity,22 a cheap, edible vaccine available to both infants and young adults as part of a revaccination program has the potential to contribute to the overall eradication of measles.

The genetically modified organism (GMO) debate: Edible vaccines are genetically modified plants (organisms), and, as such, they are linked with the public debate surrounding genetically modified food. The variety of genes that are inserted into plants means that no two GMOs are the same. The MV-H gene we used is derived from the attenuated Edmonston vaccine strain. The MV-H protein has no known inherent toxicity and MV-H-subunit vaccines have been safely trialled in mice and primates.23,24 Plant-derived MV-H protein has similar immunogenicity to MV-H protein from mammalian cell culture, and no MV-H-specific toxicity has been observed in mice or baboons fed MV-H plant extract.12 Therefore, we believe this antigen, produced by plants, is safe for human consumption and potentially safer than the measles live attenuated vaccine.

Although the risk of recombination with wild-type MV is low, if this should occur the Edmonston strain H gene would only serve to attenuate the wild-type strain. In addition, plant-derived MV-H is anchored in the endoplasmic reticulum, preventing transfer to its native position on the virus surface.25

The potential environmental impact of MV-H transgenic plants is currently unclear, although the MV-H gene is not expected to confer a selective advantage on transgenic plants. Concerns about transfer of genes to non-target organisms remain to be addressed. However, it is likely that advances in plant biotechnology over the next decade will confront many of these issues. One such advance is the development of chloroplast transformation.26 In most plant species, the chloroplast genome is maternally inherited. This means that the transgene and any protein expressed following chloroplast transformation are not present in pollen, thereby reducing the risk of transmission of the transgene to neighbouring crops or weed species by cross-pollination.26 Expression of transgenes from the chloroplast genome may also result in the accumulation of significantly greater quantities of protein.26