On the 'forgotten" gene that could be harmful in transgenic plants: regulators did not miss it

Scientific assessment of Publication: Podevin & Du Jardin on possible new peptides derived from potential ORFs in the CMV promoter

GenPeace team

A paper by Podevin and du Jardin entitled “Possible consequences of the overlap between the CaMV 35S promoter regions in plant transformation vectors used and the viral gene VI in transgenic plants” was recently published in GM Crops and Food 3: 1-5 (http://www.es.landesbioscience.com/journals/gmcrops/2012GMC0020R.pdf). Some of the paper results were reinterpreted by groups against biotechnology as a proof that regulators all over the world did not see an important source of risk in transgenic plants.

The authors argument that the cauliflower mosaic virus 35S promoter has a long overlap to the P6 viral gene sequence and that could lead to the expression of segments of this gene or putative ORFs in other reading frames. As take from the paper, the authors “ more specifically … address whether potential expression of the ORFs contained by the P35S promoter overlapping with gene VI: (1) may affect the plant phenotype and (2) show similarity to known allergenic and toxic proteins”.

The first question we must ask is if there is any potential for such expression. The overlapping sequence from P6 is, indeed, part of the CMV promoter and no part of the gene lay downstream of it. This promoter drives the transcription of sequences lying downstream of it (starting from position +1). An inspection of Figure 1 (see scheme below) shows that position +1 is well below (3´) the end of P6 coding sequence . This means that essentially no part of the P6 gene could be transcribed, as IT IS NOT PRESENT in the transgenic construction as a DNA segment able to be transcribed from the CMV promoter. The only possibility for P6 to be transcribed would be from a plant promoter located upstream of the CMV promoter. There is no evidence in the literature or in regulatory documents suggesting such transcription exists.

Figure 1. (A) Relative position of P6 gene 3’ end and of the CMV promoter 3´end (8 bases upstream of position +1) and (B) possible small ORFs in frames +2 and +3 relative to the regular frame of expression of the P6 gene.

The second question is related to the possible (although never detected) transcription and further translation of any part of the P6 gene, either in its original frame or in one of the two remaining frames. If a plant promoter able to drive a significant number of copies of any P6 gene segment exists, there would be primary transcripts in the nucleus bearing a part of the p6 ORF or any one of the six small ORFs in the other two frames. However, to be translated, these RNAs have to be polyadenilated, what can only happens if a poliadenilation signal exists in the 3´end of the RNA. The only such signal is way down (3´) of the promoter, indeed after the transgene. Therefore, any such unlikely RNA would be longer than the usual mRNAs starting from the CMV promoter +1 position, but would contain the transgene and would, therefore, be detected in the regular Southern blots used to check for mRNA expression in all regulatory documents.

The unlikely existence of a convenient promoter just upstream of the insertion site and the fact that such long mRNA were never detected lead us to conclude that this is merely a theoretical speculation with very remote possibilities to occur.

Moreover, the peptides encoded in frames +2 and +3 (assuming +1 as the frame enabling the expression of P6) are very small (the largest is only 87 amino acids long) and therefore have a very remote chance of being toxic (see Figure 1B). As for allergies, they can´t elicit a response, due to their small size, except in association with haptens. Polypeptides representing the transcription and further translation of stretches of the P6 gene could be large enough as to elicit allergies or have any toxicity, but it was never reported that any CMV protein is allergenic, although they may be found in infected vegetables. Bioinfo data also suggests the inexistence of a potential for allergenicity or toxicity for any of the possible ORFs.

The route to damage (an essential step on risk assessment) therefore shows that if is very unlike that a new unexpected protein will be produced. Moreover, the damage (allergenicity or toxicity) is not defined, as there is no potential for either allergenicity or toxicity in the small peptides derived from the ectopic ORFs. Highly toxic proteins are very well known and the eventual toxicity of these peptides, if any, will be very low. We can than safely conclude that both the probability of damage and its magnitude are very small and the risk can surely be classified as negligible.  

As for the authors´ first concern (changing the phenotype), we should take into consideration that this can only happen if new unexpected proteins are produced in such amounts as to significantly disturb the plant metabolism. The probability of expression is, as we discussed above, very small. Moreover, there is no reason to expect that small peptides or even parts of the P6 protein could change the plant phenotype. Finally, any phenotypic change derived from this eventual expression would be readily detected by the plant growers and would very unlikely lead to any health or environmental harm. To conclude, the risk for this hypothesis is also negligible.

The risk assessment described above follows many R.A. guidelines. The conclusions are clear: negligible risks are to be expected. The conclusions are in contrast to those described by Genok (http://www.genok.com/news_cms/2013/january/vitenskapelig-vurdering-av-publikasjon-fra-podevin-og-du-jardin-2012/181), an institution clearly against modern biotechnology and prone to find risks were they do not exist; in the absence of similarity to know allergens or toxic peptides, Genok sticks to the idea that these databanks are still incomplete and that regulators must therefore be ultra cautious.

They also argument that “a potential change in the plant phenotype in new GM plants can be identified by transcriptomics, proteomics, or other profiling technology”. This is simply not true essentially  because the complexity of the data usually generated by these techniques preclude any useful conclusion on risk assessment .

Moreover, they argue that “applicants or producers of GM plants should provide the genetic information concerning new ORFs to the regulators”. Genok possible does not know, but applicants do discuss the existence of new ORFs, whenever relevant, what is seldom the case.

Finally, Genok also criticizes the methodology and says “This approach ignores the potential availability of protein domains of toxins and allergens in the linear translated sequences. Domains are the functional portions of proteins and consist of at least 25 amino acids”. As explained in the paper, the authors used Blastx to compare sequences and this software displays conserved domains in the first box of results, if they exist…