Molecular detective work
BERT PÖPPING

GM plants look, smell and taste the same as their unmodified counterparts - so how do you tell them apart? And how do you know if they're in your dinner?

Foodstuffs derived from genetically modified (GM) microorganisms such as bacteria and yeast have hit the headlines recently. But GM organisms have been used to make commodities including yogurt, cheese and certain vaccines from as early as 1982. The first product using material from a GM plant to reach the UK market was a tomato puree, which was made available in Safeway and Sainsbury's stores in 1997.¹ This was clearly labelled 'made with genetically modified tomatoes', had a price advantage and was accepted by some consumers.
Although legislation governs the release of novel foods^5,6 and a strict approval process is in place that includes the safety assessment of every new GM plant, there is still serious concern about products derived from GM plants. Public alarm in the UK was increased dramatically by TV and press coverage, as well as by the actions of Greenpeace and Friends of the Earth raising awareness of this issue.
In 1996, the first large-scale production of GM staple crops began. GM (or 'transgenic') plants were grown on about 4M acres in Canada and the US.² The projected global market for these plants in 2000 is $2bn, increasing to $6bn by 2005. In 1997, GM soya - Monsanto's Roundup Ready - already represented about a tenth of the total soya harvest in the US. In 1999 it will account for more than half.³ Unlike tomato, soya is one of the most versatile staple crops and is used as an ingredient in over 25,000 foods.⁴ Similarly, GM maize that is resistant to the corn borer - Novartis Maize Bt 176 - is used in a wide range of foods.
On 1 September 1998 the EC directive 1139/98/EC came into force, which requires the labelling of Monsanto GM soya and Novartis GM maize.⁷ On 18 March 1999, UK food safety minister Jeff Rooker announced the introduction of new powers aimed at enforcing this directive. The measures include labelling products on supermarket shelves as well as those sold in restaurants, bakeries and at delicatessen counters. However, as products containing GM ingredients do not look, smell or taste different from non-GM products, how can they be identified? To answer this question, one has to look at what makes GM plants different from other plants.

Detection techniques
All organisms are made up of a huge number of cells, each of which contains a complete blueprint of the information needed to create every single cell in the organism. This information is stored in the DNA inside each cell. Genes are sections of this DNA that contain the blueprint for a single protein. To make the protein, the genes are transcribed (synthesised) into ribonucleic acid (RNA) and then translated into a sequence of amino acids that then form the protein. This process is known as 'protein expression'.
Transgenic plants are created by inserting a new gene into a single cell, which goes on to form a whole organism. The new DNA is translated and the new protein expressed. This gives the plant a new characteristic, such as resistance to certain insects or tolerance to herbicides.
The basis of every GM detection technology is to find the difference between an unmodified and a transformed plant. This can be done by detecting the new DNA that has been inserted, the new protein expressed or, if the protein acts as an enzyme, by using chemical analysis to detect the product of the enzymatic reaction - for example, a new fatty acid or lipid.

Although chemical detection is feasible in principle, it is difficult in practice because the food matrix tends to be a complex mixture of lipids, proteins, nucleic acids and many other molecules. Identifying a specific protein or a specific fragment of DNA in a complex food matrix is very difficult without any way of amplifying the target. However, if a plant produces large amounts of a substance as a consequence of a newly introduced enzyme, this could be determined by numerous analytical techniques such as NMR or gas chromatography­mass spectrometry. However, this is difficult to achieve as one has to know exactly what to look for - each product of a new enzymatic reaction would have to be looked for individually. This is extremely labour intensive and rarely practical.
When choosing a strategy for analysis, several criteria need to be assessed including the specificity, sensitivity, feasibility and practicality of the method; the potential for validation (it must be replicable in more than one laboratory); and the cost efficiency. For the actual detection, other factors are also important. The food matrix contains a whole range of other substances such as lipids, fatty acids and polysaccharides in addition to the DNA and proteins. Some of these substances can interfere with GM detection. For example, certain plant polysaccharides can inhibit important detection reactions. In the absence of appropriate controls, this could be interpreted as a false-negative, in other words, although GM material is present, the detection is blocked because of the inhibitory effect of the plant polysaccharides.
The degree of degradation of the test material is also important. If a protein or DNA is too degraded to be recognised, there is again potential for a false-negative result. This can also be avoided by applying appropriate controls. In addition, homogeneity is crucial as in most cases only a sub-sample of the product is taken. If the GM material is inhomogeneously represented in the sample, there is also potential for a false-negative result.

Hunting for proteins
Protein detection is one option for analysis. The most promising system is the enzyme-linked immunosorbent assay (ELISA), in which the extracted protein is bound to a plastic plate. An antibody - essentially a protein that binds specifically to the extracted protein - is then added. The antibody is coupled to an enzyme, which catalyses a colour reaction. The intensity of the colour produced is proportional to the amount of protein detected. The ELISA system is widely used and many laboratories are familiar with the technique. It can determine the amount of protein present and is a high-throughput system that can simultaneously analyse up to 96 assays on one plastic plate, including controls.
The drawback of the system is the lack of appropriate antibodies for the proteins in question. So far, only antibodies for Roundup Ready are commercially available.^8,9 Also, the ELISA system has only been validated in Europe for testing flour, not for more highly processed foods where the protein is usually denatured and would not be recognised by the antibody. Another drawback is the fact that some proteins are not expressed at the same level in all parts of the plant. In Novartis Bt 176 maize, the expression of the Bt protein in the kernels, the part that is used in foods, is about 1000 times lower than in the leaves.¹⁰ Detecting Bt protein in kernels by ELISA is
not feasible. Nevertheless, ELISA could prove useful for transgenic plants when samples are taken at early stages of processing.

Focusing on DNA
The most promising target for GM analysis is the inserted DNA. This molecule is much more stable than proteins and is present in the same concentrations throughout the plant. Although individual DNA molecules are impossible to detect, they can be amplified using the technique known as the polymerase chain reaction (PCR). The inserted gene usually has at least three elements: a 'start' signal (the promoter), the gene which encodes the new characteristic of the plant (herbicide resistance, for example) and a 'stop' signal (the terminator). All three can form targets for analysis using PCR technology (see Figure 1). The most widely used systems are standard PCR and nested PCR.
Standard PCR seeks to amplify a very specific region of the DNA, which is defined by primers. Primers are short stretches of DNA (oligonucleotides) that bind very selectively to complementary parts of the DNA to be amplified. They are extended by the enzyme Taq polymerase to produce copies of the DNA fragment. This process is repeated and the newly produced copies are copied again and again, doubling in number each time (this is why the process is called a 'chain reaction'). In 30 of these cycles, depending on the number of target DNA molecules present at start of the process, up to 10⁹ copies can be generated.
The resulting molecules are separated according to their size by electrophoresis on an agarose gel. The gel is stained with a fluorescent dye, ethidium bromide, which intercalates with the DNA and glows orange under UV light (see Figure 1). Careful choice of primers can distinguish between sequences with minute differences, avoiding amplification of unwanted, similar sequences which could give rise to false-positive results.

To improve specificity further when non-specific amplification is expected, nested PCR is used. This involves an initial amplification of a longer target fragment by PCR, followed by a second PCR where primers are used that bind to a section within the amplified product, to give a shorter fragment. This dramatically increases the specificity as both sets of primers have to be complementary to the target DNA sequence. The drawback of this method is that it is more time-consuming than standard PCR.
In fact, standard PCR is sufficient in most cases to identify GM-specific sequences. This has been shown in a successful EU-wide validation study organised by the Joint Research Center in Ispra, Italy, where primers were used that specifically bind to the promoter and terminator sequences of transgenic plants including Monsanto's Roundup Ready soya.¹¹

Making DNA count
Unlike ELISA, PCR is not a quantitative technique. To allow quantitation of PCR results, two technologies are available: competitive PCR and Taqman analysis.¹² Competitive PCR is based on the assumption that if there are two sequences present that have the same complementary DNA sequence for the primers to bind, they compete to bind with the primers. First, a known amount of a very short DNA fragment that has the same sequence for the primers to bind to is added to the sample. This shorter fragment is known as the 'internal standard'. PCR is then carried out and the products are separated by size on an agarose gel (see Figure 2). Since the intensity of their fluorescence is proportional to the amount of amplified DNA, the internal standard and the target DNA appear as two distinct bands on the gel with different brightnesses. A dilution series is run to reach the point where the brightness of both bands is equal, and since the number of internal standard molecules added to the PCR is known, the number of target DNA molecules can be calculated. That this system works has been shown in a validation study by Swiss government laboratories¹³ and it will undergo further trials in Europe later this year. This method is less expensive than Taqman analysis, but producing the dilution series is more time-consuming.

Taqman technology uses fluorescent probes that bind to the DNA to measure the amount of target DNA amplified. This way, the number of molecules at the start of the reaction can be calculated. Although the Taqman system is relatively expensive to buy, it can analyse up to 96 assays at a time, much like ELISA.

A testing job
The reasons for developing detection methods are two-fold: it allows the consumer to make an informed choice, and it lets local authorities enforce legislation. If the EU takes a decision on a threshold level of GM material for labelling, the scientific methods available to enforce this decision have to be taken into account. When testing finished products it is not possible to extrapolate the amount of transgenic material used at the start of the food processing, as the extent of DNA degradation will be different in most cases.
Although the number of starting molecules in a PCR reaction can be calculated using the fluorescent probes, this does not account for the degree of degradation the sample (or food) has undergone - if a food is treated differently, perhaps heated longer, the number of molecules in the final product can differ, although the same percentage of GM starting material was used. To avoid this, the amount of GM material is linked with an 'internal reference standard'. For example, the Roundup Ready GM sequence in soya is linked with the soya lectin gene (the internal reference standard), which is present in all soya, genetically modified or not. As the lectin and the Roundup Ready gene will be degraded to the same extent, the relative percentage will remain the same in the food, irrespective of the treatment. So the percentage of GM starting material can be calculated. The corresponding genes for maize are the Bt gene (specific to genetically modified maize) and the invertase gene, which is present in all maizes.
If the soya reference sequence is present at the start of processing in 1,000,000 copies and the percentage of transgenic soya beans is 30%, 30,000 copies of the transgenic sequence will be present at start. During processing, half of the DNA will be destroyed. This means that only 50,000 copies of the soya reference sequence will be present and 15,000 copies of the transgenic sequence. The proportion of transgenic material stays at 30% relative to the ingredient measured, irrespective of the DNA degradation occurring. Both competitive PCR¹³ and Taqman systems have been developed that use internal standards to determine the amount of GM material in an ingredient. The EU threshold level is likely to be based on ingredients, rather than finished products.
All DNA detection systems are based on amplifying specific DNA sequences. For GM testing, these sequences are specific to either the promoter, the terminator or the actual gene inserted into the plant. Problems can occur when the target sequence is the promoter or terminator sequence only. For example, the commonly used promoter 35S is derived from a naturally occurring virus, the cauliflower mosaic virus, which infects brassica plants. But if a plant has been infected, the virus and its DNA would be inside the plant cells. When the DNA is extracted from the plants, the virus DNA would also be extracted. Detecting for only the 35S sequence would give a positive result even though the plant may not be genetically modified. To avoid this, only 'truly genetically engineered' sequences - sequences that do not occur in Nature - are the targets for most detection systems.
In general, the actual PCR and subsequent analysis do not normally pose a problem. The difficulty is the extraction of DNA from complex food matrices without 'co-purifying' any substances that would interfere with detection. Several techniques have been developed that improve the method described in the official Swiss screening method (OSSM) by using DNA-binding resins. However, there remain a number of food matrices from which it is difficult to extract DNA. Among these are enzymes, sugars, hydrolysed starch products and refined oils. Here, the DNA is either present in minute quantities or too degraded to be amplified. These items are considered to be candidates for a so-called negative list, meaning that since detection of transgenic sequences is not feasible, they would not need to be tested or labelled.

Existing technologies and new developments
In general, chemical analysis for GM foodstuffs does not appear to be practical in most instances. The two favoured options are either protein analysis using ELISA technology or DNA analysis. ELISA analysis could be used at the early stages of processing if antibodies of sufficient quality become available for all new traits. ELISA technology has the potential to be developed into a 'dip-stick assay', similar to the home pregnancy test. This would allow farmers and inspectors to test for the presence of transgenic proteins on the farm or in the mills. But here, quantitation will be difficult because of the different levels of proteins in different parts of the plant, as shown by the example of the Bt maize varieties.
DNA analysis, especially the quantitative methods, seems to be the method of choice. DNA assays have the advantage that tests for new traits can be developed within 7­10 days from the date a new sequence becomes known. New ELISA assays, on the other hand, take months to develop as the target protein must be available in its purest form, and it takes about a month to produce antibodies against this protein. To produce sufficient amounts of antibodies for commercial use then takes another few months.
Both ELISA and DNA methods will have to cope with an ever-increasing number of different GM organisms. Since even multiplex DNA assays, where up to three DNA sequences can be screened for in a single PCR assay tube, will not be sufficient in the near future, new approaches will have to be developed. One of these is the development of the 'gene-chip' or microarray assay. This allows screening for several hundreds, if not thousands, of different sequences at the same time. The chip is coated with a large number of different, short DNA sequences at distinct positions, which bind the complementary sequences of transgenic sequences. A single PCR is then performed on the whole chip using primers labelled with fluorescent markers. The fluorescence is then measured at the different positions on the chip, allowing the transgenic sequences to be detected. The development of this chip is expected to result in a product on the market in early 2000.
In a few years time there may even be a dip-stick assay with which consumers can test the food they buy directly in the supermarket. But will GM food still be an issue then? This is likely to depend on whether adverse or beneficial (health) effects from consuming GM food can be shown. However, the current rejection of GM products by consumers and now retailers poses a more immediate question. If genetically modified material is not used in the food production - where does it go? Will tests on animal feeding stuffs be the next step?

References

1. An up-to-date list of applications for 'placing on the market' of GM plants can be found at www.environment.detr.gov.uk/acre/marketing.htm
2. Braunschweiger, G. & Concelmann, C., 'Gentechnisch veränderte Rohstoffe für die Lebensmittelwirtschaft: Was ist schon auf dem Markt', Bundesinstitut für den gesundheitlichen Verbraucherschutz und Veterinärmedizin, May 1997
3. http://www.transgen.de/ 'Gen-Soja: Anbauflächen in den USA nehmen weiter zu', March 1999
4. Jany, K.D. & Greiner, R., Gentechnik und Lebensmittel, Berichte der Bundesforschungsanstalt für Ernährung, 1998
5. Council Regulation No. 90/220 of the European Parliament and the Council of 23 April 1990 on the deliberate release into the environment of genetically modified organisms
6. Council Regulation No 258/97 of the European Parliament and the Council of 27 January 1997 concerning novel foods and novel food ingredients, Official Journal No. L 043, 14/02/1997, 0001-0007
7. Council Regulation 1139/98/EC, Official Journal of the European Communities, L 159/4, 1998; all other GM products are covered under the Novel Foods Regulation 258/97/EC
8. Strategic Diagnostics, Newport Beach, CA 92660, USA
9. 'Food test shows GM content', The Sunday Times, 4 April 1999
10. BATS Report 2/97, BATS, Switzerland.
11. Press release from JRC (http://www.jrc.it), 'Validation of a test to specifically detect genetically modified soybean (Roundup Ready) in dried powder samples (GMOs)', 3/26/99
12. Taqman 7700 System: Perkin Elmer Applied Biosystems (http://www.perkin-elmer.com/)
13. Studer, E., Rhyner, C., Luthy, J. & Hubner, P., Zeitschrift Lebensm. Unters. Forsch., 1998, 207, 207-13

Dr Pöpping is company molecular biologist at Eurofins Scientific, 69a Kilnwick Road, Pocklington, Yorkshire YO42 2JY, UK.

5 July 1999