Progress-report RTD-Project GMOCARE

Executive summary (mid-term review: 18 months)

Genetically modified plants

Generally speaking, the project's focus is directed towards answering the question 'Is there a difference between the parent and the genetically modified cultivar?' Thereto, a wide range of transgenic (GM) plants has been produced to investigate the occurrence of an unintended effect due to the genetic modification. At the mid-term review, it was questioned whether the current production of transgenic plants including their relevant controls should be limited to one or two potato transformants e.g. steryl glucosyl transferase and SAM antisense lines including their field tested harvest. If there is a significant (unintended) difference, the difference should also be compared to the natural variation of that crop. Thereto, it can be foreseen that sampling and the range of controls will be an increasing problem. However, the consortium concluded that this latter testing was clearly not envisaged by the project. 

Potato

The Scottish Crop Research Institute produced potato lines that showed the following genetic modifications i.e. Mal1 (defective glycoprotein processing), SAM35S (polyamine metabolism), W2GBSS (starch composition), FK (modified sugar metabolism) and a steryl glycosyl transferase antisense potato lines (i.e. lower glycoalkaloid levels). The GM plants were continuously characterised on sizes of T-DNA, vector backbone integration, copy number and expression levels. Especially, phenotypes with metabolic and/or developmental disturbations were of interest.

The Free University of Brussels (VUB) continued their production of a molecularly well characterised potato transgenic line (DHDPS) showing increased lysine content. So far, there were no observations of phenotypic alterations (i.e. free lysine levels were below 20%).

 Tomato

A selected range of molecularly well characterised tomato transgenic lines was produced by the University of London (RHUL) i.e. 35S crtl (2-fold increase of b-carotene and lutein), crtB (increased carotenoids), anti-Psy1 (dramatically reduced carotenoids), beta-Lyc (2.5-fold increase of b-carotene) and HMGR (increased end-product phytosterols). There were no observations of deleterious effects on the physiology of plants, which were modified in their carotenoid biosynthesis of which the carotenoid (phytosterol) profiles were significantly changed. The transformant crtl showed increased seed dormancy and a slower growth.

Arabidopsis and Tobacco

For academic purposes Arabidopsis Wassilewskija was selected and characterised further i.e. one parental line, 3 CHS (modified flavonoid biosynthesis) antisense (single copy) and 3 CHS (multi copy) transformants. In addition, the construction of 10 different Arabidopsis lines with antisense/sense constructs of DFR under control of an ethanol inducible promoter was achieved and their propagation and growth from 1st generation is under way.

In case of Nicotiana Sylvestris mutants like RAEC-1 (mutated dhdps-r1), RLT-70 (mutated aklys 1) and hybrid RLT-70/RAEC-1 seeds were grown in the greenhouse. There were observations of effects on the shape of these plants (i.e. an atypical leaf morphology), which were constructed by means of classical mutagenesis.

Measuring the unintended effects

The initial step in the assessment of the safety will in most cases be a compositional analysis in order to determine whether a GM plant, as those described above, is substantial equivalent to its non-modified control. The traditional 'targeted' analytical approach is based on the measurement of single compounds in GM versus non-GM plants and envisaged too in the project.

General sampling strategy

SCRI presented their general sampling strategy for the compositional analysis of tubers, which was adopted by the project as follows: 1) start with field trial of 6 plots; 2) collect randomly tubers of 5 plants of each line per plot, 3) take from each plant a sample of tubers of average size and 4) prepare a sub-sample of 3 'sets' of tuber tissue of each of about 100 g (wet weight), which can be freeze-dried if required. After ample discussion, it was adopted that a 'non- and/or targeted' comparative analysis (see below) should be performed at the level of tuber to tuber, fruit to fruit, seed to seed or flower to flower, rather than based on a plant to plant comparison. Consequently, this design of sampling and comparative analysis will be the focus and strategy for the reminder of the project.

Traditional 'targeted'  analytical approach

Given time constraints it will be pertinent to prioritise which GM potato lines should be fully characterised from the field trial carried out in 2001 (SCRI). Thereto, the project will envisage the controls, MAL1, 35ScaMV and SAMPAT GM-lines as being most promising. This will give valuable information comparing their previous glasshouse performance with field performance.

The following targeted analytes were measured: glycoalkaloids, carbohydrates, vitamin C, fatty acids, amino acids and protease (trypsin) inhibitors(s). It was concluded that the chemical analysis of the FK (modified sugar metabolism) GM-potato lines showed only marginal modifications. No higher levels of glycoalkaloid (GA) content were measured compared to the controls, but rather a decrease of glycoalkaloids (e.g. solanine, chacocine) in the GM-lines. Whether this effect might be due to the transformation event has been discussed, but it was concluded that first a repeat of the experiment will be needed, which is under way. Moreover, SCRI did not found a correlation between tuber size (cv. Desiree) and the total glycoalkaloid content. This in contrast to the observations published by Engel et al. (In: Novel Food Regulation in the EU - Integrity of the Process of Safety Evaluation. Berlin: Federal Institute of Consumer Health Protection and Veterinary Medicine, pp. 197-209). The question remains whether there is a cultivar specificity? It was SCRI's observation too that the vitamin C content varied even amongst the control tubers. The statistical analysis of results grouped as values of group/tubers is in progress.

RHUL studied in GM-lines of tomato fruit/potato tubers and tobacco leaves the following targeted metabolites: isoprenoids (caretonoids), sterols, tocopherol. It was RHUL's observation that the transgenic DHDPS (lysine content) potatoes including controls (VUB) did not differ in their phytosterols and isoprenoids. Also the Nicotiana Sylvestris mutants like RAEC-1 leaves (VUB) sent to RHUL, which showed an atypical leaf morphology, did not differ in their phytosterols and isoprenoids. It appeared that the phytosterol levels were quite variable between transgenic tubers received from SCRI. With respect to these latter analyses it was decided to repeat these measurement by transferring more material. 

The 'non-targeted' analytical approach

The drawback of the traditional analytical approach (see above) is that altered levels of unknown compounds cannot be detected and that it is restricted by the availability of analytical methods. Therefore, new 'holistic' approaches are needed which can monitor changes in composition at various integration levels by the generation of non-biased fingerprints of (extracts) of relevant plant parts using genomics, proteomics and metabolomics.

Genomics (analysis of the transcriptome (mRNA) or gene expression profiling). The Veterinary Food Administration (VFA) started the construction of representative cDNA libraries of Arabidopsis by subtractive hybridisations. In collaboration with RIKILT about 1500 cDNA's were cloned and spotted on an array. As reported earlier, two sets of tomato cDNA microarrays were constructed (RIKILT). These microarrays consisted of specific cDNAs derived from red and green tomato fruit as well as cDNAs corresponding to genes with known function. Hybridisation experiments were carried out in order to determine levels of gene expression in subsequent ripening stages of tomato (i.e. green-breaker-turning-light red-red). A significant differential gene expression has been defined as a 2-fold increase or decrease of the signal in both duplo's. Currently, the construction of a potato array by subtraction of potato high from low glycoalkaloid-levels as well as cDNAs correlated to genes with tuber germination is underway.

It was questioned whether it would be necessary to go through all the ripening stages for testing of the selected GM-tomato lines on unintended effects. Because, it can be foreseen that data management will be an increasing problem. Although, the 'background' expression profiles will be necessary, it was adopted to focus both the genomics and proteomics screening on the stages light-red and red and in case of metabolomics on the red stage only. It was also discussed to what extent the various stages of ripening should be calibrated/validated. It was adopted (RHUL) to count the days post breaker as the criterion for ripening.   

Proteomics (analysis of the proteome (proteins) or protein expression profiling). The evaluation of sample preparation and extraction procedures for the proteome analysis of plant material was completed and optimised to find a practical approach for sample delivery and compatibility with other analyses done within the project. Better quantification of gels was pursued by comparing different staining methods and by evaluating day-to-day variation. SYPRO Ruby was thus the most sensitive dye, but was also relatively expensive (University of Kuopio, UKU). 2D-gels were run from the samples delivered by  other partners. The identification of more landmark proteins from 2D-gels of tomato, potato, tobacco and Arabidopsis was continued from tryptic peptides using MALDI-TOF-MS and ESI-MS. The identification of protein expression including landmark proteins was performed as follows: 1) 2D-gel electrophoresis, 2) tryptic digestion of some spots, 3) MALDI-TOF or ESI-MS and 4) database searches for identification. It was reported that tomato protein 2D patterns changed upon ripening (27 increasing spots and 26 decreasing spots etc). HPLC -ESI-MS for final identification of proteins using, for example, SEQUEST algorithms of GM and non-GM material is under way.  Moreover, UKU's proteomics indicated that all tomato cultivars delivered by RHUL were infected with tomato mosaic virus (ToMV) i.e. the tomato proteome was contaminated with viral coat protein and, thus, possibly also with infection-related plant proteins. This material was not discarded, but kept in the database as a kind of background information for viral 'infected' proteins profiles.

SCRI analysed the 2D-gels (i.e. pictures) prepared by DCHGR (Danish Cancer Society) of non- and GM-potato lines (cv. Desiree and Record), but a visual inspection of the gels did not indicate consistent differences. At least, there were a lot of differences/changes between proteins of the GM-potato lines, but less differences were found in the patatin area. There were also variations between controls (i.e. SAMPAT vector control, DES TC control). It was concluded that the alignment of spots and the subsequent comparative analysis of GM-potato proteomes should be repeated by using imaging software. In addition, it was discussed whether a (sub-) fractionation of tubers, e.g. an aqueous and membrane extract, would be a step forward to look more closely at differences in the proteome, because a fractionation might increase the sensitivity. After ample discussions, it was decided that a sub-fractionation of plant tissue is not an issue that should be covered by the project. However, it was agreed that the Food Standard Agency (FSA-UK)-project on unintended effects (SCRI, IFR, RIKILT) will cover this latter item.

To improve the 2D-PAGE/MALDI-TOF analysis it was suggested by DCHGR to enter antibodies raised against plant proteins like patatin. Because, the availability of antibodies will allow western blotting and therefore, it will be a very efficient approach to build up a preliminary database on plant proteins. This approach will also provide the unique possibility to calibrate and pre-screen protein expression profiles. At least, it will certainly strengthen the interpretation of the experiments i.e. necessary turnover in data handling. However, this approach was found to be too optimistic and not feasible at the moment as collections only existed with individual scientists and not centrally. Also a production de novo was found to be too expensive regarding the project's total budget.

Proteomics and the analysis of unintended post-translation modifications. A strategy of glycome analysis of minor amounts of glycoprotein post-PAGE and in whole plant tissue was developed and its application to the selected non- and transgenic plant materials was demonstrated using, for example, the 2D-gels of UKU with a focus on the patatin area (i.e. the major potato storage protein). BOKU (University of Vienna) demonstrated that the patatin isoforms differed in glycosylation. However, the minor changes in the glycosylation patterns may not be responsible for the differences observed upon 2D-gel electrophoresis as performed by UKU. In addition, it was reported that the proteome of leaves versus tubers of the potato cv. Desiree showed different glycosylation patterns.

Related to the glycomics of Arabidopsis, it was demonstrated that Arabidopsis is not capable of synthesising high complex N-glycans, because this plant cannot generate the Lewis A epitope on the outer arms. However, other forms of glycosylation e.g. arabinogalactans (AGP) are present in Arabidopsis. It is envisaged to analyse these sugars by using the beta-glycosyl-Yariv reagent, which synthesis is underway (BOKU). Also the analysis of plant-type protein O-glycans is under development.

Proteomics and the analysis of unintended allergen expression. The collection of pollen from (wild type) Arabidopsis and subsequent selection of Arabidopsis-specific human IgE was partly finalised. For the development of tools to profile (possible) alterations in endogenous allergens, the collection of human sera containing specific IgE antibodies directed to rapeseed was continued and analysed for specific subclasses between allergic patients. It appeared that 140 sera were screened of which only 2-8 were positive for Arabidopsis. It was reported that 4 proteins in rapeseed reacted with one of the selected human serum: one protein corresponded to cobalt-dependent methionine synthase. However, extracts of Arabidopsis flowers using the same human serum sample showed a complete other proteome pattern. It was questioned whether these were different proteins? The examination of potato 2-DE with sera from patients allergic to potato was discussed too, but such sera were not available at the Pasteur Institute.

Generally speaking, it was recognised that in case of the 'rare' allergens it will always be the case that only very limited amounts of suitable human sera will be available, if any. Therefore, it is proposed (INRA-CEA) to apply the strategy of 'pre-selected' transgenic (crop) plants as follows: 1) establish 2D-gels of the proteome of non- and GM-Brassica napus or Arabidopsis lines (e.g. using seeds/pollen/flowers), 2) investigate the 2D-gels by means of a comparative analysis in order to identify significant differences in those 'gel-regions' shown to be of interest in case of allergenicity (i.e. IgE-binding), 3) transfer only those proteomes of GM-lines versus controls with an unintended/unexpected difference to a specialised institute like INRA-CEA and Pasteur and 4) prepare 2D-immunoblots using selected IgE-specific human sera (i.e. pre-selected).

Metabolomics (analysis of the metabolome or metabolite chemical fingerprinting). The utilisation of CEA's special growth chambers (i.e. controlled conditions) for potato, tomato, Arabidopsis and tobacco pot plants has been evaluated for the metabolite profiling (metabolomics) experiments in vivo i.e. testing during growth). At present biomass of two potato lines (i.e. Désiree, Record of SCRI) and two GM-tomato lines (i.e. Crtl and anti-Psy1 of RHUL) have been fixed for 13C-NMR analysis at about 2 weeks after the beginning of flowering. NMR analysis of potato tubers (i.e. labelled 10%), Arabidopsis thaliana (i.e. 7 transformant lines of VFA) and tomato fruit (seeds) is under way. Fortunately, the UK authority had certified SCRI's potatoes as being virus (TSWV) free.

It is observed that the Nicotiana mutant RLT-70 accumulated threonine if compared to the wildtype, whereas the mutant RAEC accumulated both lysine and threonine (see above). However, the hybrid RAECxRTL-70 accumulated lysine only (i.e. threonine was not visible upon NMR analysis). It was speculated whether lysine blocked threonine or whether other unexpected metabolic mechanisms might be involved. Obviously, there is an unintended effect. Does lysine block threonine synthesis? It was agreed that CEA should continue this work (i.e. repeats). Moreover, it was discussed whether it could be demonstrated that mutation breeding might inhibit a higher risk of unintended effects if compared to genetic (recDNA) breeding. Thereto, it was questioned whether VUB could generate a GM-line instead of a mutant, and that would show a comparable phenotype as the mutant RAECxRTL-70 i.e. an atypic leaf morphology. Because, the availability of a comparable GM-line would allow the project to rank the breeding techniques with respect to their risk of potential unintended effects. Moreover, this could also generate highly relevant information in view of the attitude of the public and even more the EU legislation (e.g. new directive 2001/18/EC). To make a good comparison between mutants RAEC, RLT and their hybrid with similar transgenic plants, it will be necessary to generate Nicotiana sylvestris transgenics with feedback insensitive AK and DHDPS genes. However, these transformations would take about one year. Notwithstanding these time constraints, it was adopted that VUB will start these transformations, and it was also agreed to analyse these GM lines by proteomics and metabolomics. Which means, that these analyses will have to be done near the very end of the project.

With respect to Nicotiana sylvestris wt and GM-lines of VUB, CEA reported that there was no evidence for compositional differences between the wildtype and transformed lines.

For the metabolite profiling experiments in vitro (post harvest) sample preparation and extraction procedures for proton-NMR etc. were established by IFR and RIKILT. The fractionation procedure for tomato and potato was tested and modified to reduce running time.

At present the following techniques are under investigation: 1H-NMR/2D-NMR; GC-FID/GC-MS; HPLC/LC-MS. Moreover, the development of extraction and derivatisation procedures (starting from lyophilised potato and tomato powders) to provide samples for metabolic profiling by gas chromatography was completed. The assignment of gas chromatograms of representative samples using reference standards and GC/MS is in progress including the acquisition of gas chromatograms of derivatised potato and tomato samples. Work is also foreseen regarding the development of automated procedures for data extraction and to assure repeatability of whole procedure (extraction, derivatisation, measurement) and consistency between different batches of samples.

GM potato lines i.e. Mal1 (defective glycoprotein processing), SAM35S (polyamine metabolism), W2GBSS (starch composition) and FK (modified sugar metabolism) were analysed by metabolomics. No significant differences were observed. However, the SAM35S antisense potato-line(s) demonstrated an unexpected atypical variation in the amino acid proline, whereas the phenotypes looked alike (parent cv. Desiree). It was discussed whether this effect might be due to the various positions of the insert. It was also questioned whether information has been gathered on the flanking sequences? It was accepted to continue this work including an investigation by DNA-microarrays and proteomics in due time. It is envisaged that in terms of number of samples analysed  it should be expected to be more like 80% potato, 20% tomato and 0% Arabidopsis by the end of the project.

Pyrolysis GC-MS for membrane-structures did not operate as it should (RIKILT). It was accepted to leave this methodology out of the workplan until further notice. Evaluation of the possibility of using 2D heteronuclear experiments for compound identification stranded on lack of sensitivity. The option LC-NMR for identification is possible, but too much work if identification of all major compounds is required. However, LC-NMR would be an option for identification for compounds, which have been empirically targeted by statistics as relevant in the context of substantial equivalence.  

Conclusion

To summarise, one of the main topics of the 30 months progress-meeting will be a summarising collection of all results of the profiling techniques. Thereto, an 'independent' will be convened where all data will be put on a row including a throughout evaluation of the various workpackages. The aim will be to integrate all the work on molecular/phenotypic and targeted analysis, metabolomics, proteomics and genomics etc. and secondly, to rank and validate the data related to the main objective of the project (i.e. potential to identify unintended effects). It is anticipated that the outcome of this exercise should be a pre-recommendation of those factors and tools of importance that showed promise in defining a (new) strategy and methodology with the potential to identify unintended effects.