New evidence may pull the plug on GMOs. Dr. Mae-Wan HoTransgenes unstable in more ways than one
Transgene instability has been known at least since 1994  (reviewed in Genetic Engineering Dream or Nightmare, p.140), though it is seldom, if ever, reported in the popular media. Transgenes (the synthetic foreign genes transferred into the genetically modified organism (GMO)) can become silent or inactive during growth and development of the GMO, or in its progeny. This has been attributed to defence mechanisms that silence genome invaders such as viruses. But transgenes can also stop working on account of structural factors intrinsic to the transgenic DNA inserted into the genome of the GMO  (reviewed in Living with the Fluid Genome, pp. 128-135). Transgenic DNA has been artificially constructed by stitching together synthetic copies of DNA from different sources, and often contain additional weak points that tend to break and rejoin (recombination hotspots). The most widely used cauliflower mosaic virus (CaMV) 35S promoter is associated with such a recombination hotspot , as we have warned [4-6]. Transgenic constructs are also designed with ends that can break into genomes such as the repeated sequences of viral vectors, and the left and right borders of the T-DNA of Agrobacterium, widely used as a vector. These ends too, are recombination sequences, and facilitate movement of the transgenic DNA within as well as between genomes. For more details see  (Horizontal Gene Transfer from GMOs Does Happen, SiS 38)
Transgene instability makes transgenic varieties illegal and ineligible for patent protection
During the transformation process that creates the GMO, the transgenic construct tends to undergo deletions, duplications, and rearrangements, integrating at unpredictable sites in the host cell genome, causing widespread damage both at and away from the site(s) of integration. The precise configuration of the DNA integrated, the site(s) of insertion, and the particular collateral damages done to the host genome are therefore specific for each transgenic ‘event’. Each ‘event’ is a single cell that has integrated transgenic DNA and from which a transgenic plant is generated, which is then bred through a number of generations to give a transgenic line.
However, the particular cell may have integrated one to hundreds of copies of the transgenic construct in a variety of different rearranged, deleted, and duplicated configurations, and at more than one site (locus) in its genome. Complex transgenic loci (containing multiple rearranged or partial copies of the transgenic construct) are very unstable and tend to rearrange further or become lost in subsequent generations. Proponents claim that unstable events will be eliminated through stringent selection, and only those lines that have single stable inserts will reach the market.
Unfortunately, that does not seem to be the case, evidence of transgene instability has emerged in transgenic varieties that has been commercialised and grown in more than one countries for years  (MON810 Genome Rearranged Again, SiS 38)
To qualify for commercial release in Europe, for patent protection in Europe and the United States, and other protection under the UPOV (International Union for the Protection of New Varieties of Plants) Convention , a transgenic line must be distinct, uniform and stable (the DUS test). It is likely that none of the transgenic varieties that have been commercially released passes the DUS test, which makes them both illegal and ineligible for patent protection. But our regulators have been bending, if not breaking the law so far in failing to withdraw commercial approval .
Transgene instability is a serious safety issue
Transgene instability is a serious safety issue, as it not only changes the very nature of the transgenic plant, but also increases the likelihood that the transgenic DNA could spread horizontally to the genomes of cells of unrelated species by direct uptake of the DNA .
According to a review published in 2004 , the loss of transgenes during reproduction occurs at a frequency of 10 to 50 percent of transgenic plants, regardless of whether they are produced by Agrobacterium-mediated or particle bombardment transformation. The transgene may be lost or deleted in part, or else rearranged or moved to another location in the genome. Transgene instability appears to depend on the nature of the transgene, the host genome, and the site of integration, and not on the transformation method. There may be integration hotspots in the genome that are inevitably also disintegration hotspots, as revealed by experiments in ‘gene therapy’  (Gene Therapy Risks Exposed, SiS 19), which creates transgenic human cells, and confirmed in large scale analysis of transgenic loci in plants  and in the common carp .
In plants, transgene loci resulting from all transformation systems (except for homologous recombination) exhibit short sequence homologies between the integrated transgenic DNA and flanking genomic sequences of 1 to 8 bp, and between the rearranged transgene fragments . Transgenes tend to be integrated into gene-rich regions, and reduced in the centromeric regions of chromosomes. They also show propensity for AT-rich regions and at transitions between normal base composition to a poly-T or A-rich region. These ‘ hotspots’ for integration may be sites that tend to be exposed and break more often, and hence also hotspots for disintegration. Another reason for transgenic instability is the transgenic process itself, which may destabilise the genome by causing genome scrambling and chromosomal abnormalities.
Transgene instability is now widely reported in the scientific literature, and some examples are given below.
Transgene complexity and instability in the common carp
Researchers analysed two individuals of a carp transgenic line with a human growth hormone gene. The line had been selected and bred to the fourth generation after transformation, when all the fish showed the transgenic trait . Each individual fish was found to contain about 200 copies of the transgene integrated at 4-5 sites, generally with repeats in a head-to-tail arrangement. A total of 400 copies of transgenes recovered from the two individuals fall into 6 classes, which differed somewhat between the two fish, indicating that the transgenic line was by no means uniform. The copies were either complete or partial transgene sequences. The major class 1, which comprised about 73 percent of clones from the two fish showed the original configuration. The other five classes were different from the original configuration in both molecular weight and restriction map, indicating that a proportion of the transgenes had undergone mutation, rearrangement or deletion during integration and reproduction. In three of the five types of aberrant transgenes in fish A, the flanking sequence of the host genome were identified as the carp b-actin gene, and carp DNA sequences homologous to mouse phosphoglycerate kinase-1 and human epidermal keratin 14 respectively.
Due to the limitation of the analytical method, those transgenes that had lost the plasmid replicon (replicating signal) or ampicilin resistance region could not be recovered, and this resulted in the underestimation of transgene classes.
Transgene instability in apple trees during vegetative propagation
Apple cultivars were transformed using Agrobacterium as vector to increase resistance to diseases like powdery mildew, apple scab and fire blight . A total of 64 plants of 15 different transgenic apple lines were transferred to the greenhouse, half of them grown as own rooted trees, and half grafted in different non-transgenic scion-rootstock. When tested after an unspecified time, 22 of the plants (34 percent) lost one or both genes. In the rest, four plants did not express the antibiotic marker gene, one had lost its promoter and in other three, the promoter was silenced by methylation.
However, plants that appear to have retained the transgene(s) may have only done so in part, as demonstrated in another experiment. Twenty-six lines carrying the attacin E gene from Hyalophora cecropia, the b-glucuronidase (gus) gene and the nptII gene were propagated vegetatively in vitro without selective agents for 4 years (50 generations) and then analysed . Neither expression nor integration remained stable in some lines, differences were found between plants of a single line and several plants were chimaeras of expressing and non-expressing cell clones. For example, twenty-three lines kept all three genes (at least in some of the plants). One line lost gusA and two lines lost all genes. Low levels of nptII expression were found in 12 lines, increased expression in 10 lines and only two had the same level of protein expression. Stable expression of gus was found in eight lines, though some plants were mosaics of cells that expressed the gene and cells that did not, Two lines had no activity at all, even though one had the gene. In three further lines, isolated blue spots of cells with gene expression were found against an overall white background of non-expressing cells.
Systematic and repeatable transgene elimination
Researchers in Brazil identified a remarkable systematic elimination of transgenes in a transgenic dry bean and a transgenic soybean at reproduction . The dry bean (Phaseolus vulgaris L.) line was obtained by particle bombardment with plasmid pMD4 containing the gus gene and the rep-trap-ren genes from bean golden mosaic geminiviru, both under the control of the CaMV 35S promoter, to make it immune to the virus. The soybean line was transformed with another plasmid pAG1 that contains a different combination of genes: the gus gene under the control of the act2 promoter and the ahas (acetohydroxyacid synthase) gene under the control of its promoter from Arabidopsis thaliana. In both, the transgenes were stable during the vegetative phase, but were eliminated during meiosis, the cell division that makes germ cells.
The transgenic bean line contains at least 3 copies of the transgenes integrated at three separate loci (sites). None of the copies were transferred to the progeny by self-crossing or reciprocal crosses to untransformed plants. Not a single progeny plant inherited any transgene locus. This phenomenon was systematically repeated for over two years in plants propagated by grafting (20 progenies of more than 300 plants from self-pollination, and 10 progenies of more than 100 plants from reciprocal crosses to untransformed plants).
Analysis of the host genome flanking the transgene inserts revealed that one integrated plasmid disrupted a ribosomal RNA gene while another was integrated into a sequence with no significant homology to known sequences. The third integrated sequence could not be isolated because it lacked the necessary plasmid sequence.
The same phenomenon occurred in the soybean transgenic line.
Several mechanisms have been suggested for the systematic elimination of transgenes, including intrachromosomal recombination, genetic instability resulting from tissue culture, and elimination of transgenes triggered by a process of genome defence against invading viruses.
The outstanding question is what became of the eliminated transgenes? Is it possible that the transgenic DNA could be transferred horizontally to bacteria in and on the plant, or to insects, or via insects to other plants?
While attention has focussed on horizontal transfer of transgenic DNA to bacteria, it may be that eukaryotic genomes are more promiscuous in accepting foreign DNA [7, 12] and hence better recipients for horizontal gene transfer, particular for transgenic DNA designed to invade genomes.