The evidence for horizontal gene transfer is accumulating. Dr. Mae-Wan Ho reviews a selection of recent scientific papers.
Researchers find that horizontal gene transfer may be most likely to happen in the mouth of farm animals, and to a lesser extent in the rumen and in silage [1].
The insect-resistant maize line CG00526-176 contains three bacterial genes: the cry1A(b) specific to lepidopterans, the bar gene conferring tolerance to glufosinate, and a bla gene encoding TEM-1 b-lactamase (ampicillin resistance). The bla gene originates from the cloning vector PUC18 and is not expressed in maize, but has bacterial regulatory sequences that would allow it to become functional were it to be transferred back into bacteria. There are at least two copies of crylA(b) and bla genes integrated into the DNA of the maize line.
Researchers investigated the survival of DNA from the transgenic maize and the transfer of the antibiotic resistance bla gene to bacteria in the presence of saliva, rumen fluid and silage effluent, which are relevant to horizontal gene transfer in the oral cavity, the rumen, and in silage.
E. coli was the test micro-organism for horizontal gene transfer. Degradation of DNA was followed by gel electrophoresis as well as by polymerase chain reaction (PCR). Both pUC18 plasmid and transgenic maize DNA were used in the experiments.
On gel electrophoresis, plasmid DNA and maize DNA were shown to be degraded rapidly by rumen fluid or silage effluent within one minute, but both were incompletely degraded after at least l hour exposure to saliva.
On PCR, a much more sensitive analysis, large fragments of the bla gene (>350bp) were still found in rumen fluid up to 30 mins for the plasmid and up to1 min for maize DNA. Even larger fragments (>350 and >684 bp) from plasmid and maize DNA were found up to 30 min incubation in silage effluent, and up to 24h and 2 h respectively in saliva.
PCR analysis also showed that fragments of the cryl1A(b) (>1914bp) in maize DNA could be found up to 1 min with rumen fluid, 5 min with silage effluent, and 60 min with saliva.
Plasmid DNA exposed to saliva for 24h was still capable of transforming E. coli to ampicillin resistance, but at low efficiency: 20 cfu (colony forming units) per millilitre compared with 1.6 x103 cfu per millilitre after 24h in sterile water. Previous exposure to rumen fluid for 30s reduced transformation 5-fold. No transformants were obtained after the plasmid DNA was exposed to silage eflluent or rumen fluid for longer than 1 min.
However, when E. coli and plasmid were simultaneously added to filter-sterilized silage effluent or rumen fluid, 4.75x103 cfu per ml transformants were recovered after 4.5h in rumen fluid and 11cfu per ml were recovered after 3h in silage effluent.
Thus, horizontal gene transfer can occur before the DNA is completely broken down, even when the breakdown is rapid, as in the rumen or in silage. DNA breakdown is extremely slow in saliva, and hence the oral cavity may be a very important site for horizontal gene transfer.
Those who dismiss horizontal gene transfer in GM safety often retreat to the tired old argument that it happens only between similar or homologous gene sequences. Indeed, it has been found that homology increases the frequency of horizontal gene transfer 10 million to 100 million-fold [2]. The experiments were done with DNA from transgenic potato plants carrying the nptII gene for kanamycin resistance. It transformed naturally competent cells of the soil bacteria Pseudomonas stutzeri and Acinetobacter BD413, both harboring a plasmid with an nptII gene containing a small deletion with the same high efficiency as nptII genes on plasmid DNA (3x10-5 -1x10-4), despite the presence of more than 106 fold excess plant DNA.
However, in the absence of homologous sequences in the recipient cells, the transformation dropped by at least about 108 fold in P. stutzeri and 109 fold in Acinetobacter, below the detection limit.
More than 60 bacterial species have been shown to take up and incorporate DNA (undergo transformation). Many bacteria like Bacillus subtilis and Acinetobacter sp. strain BD413, apparently take up DNA of any source into the cytoplasm. Stable maintenance and expression depends on integration into the genome by genetic recombination.
The authors state, "This indicates a very low probability of non-homologous DNA fragments to be integrated by illegitimate recom-bination events during transformation". Should we be reassured? Not at all.
The high frequencies of homologous recombination obtained are relevant to GM constructs released in large concentrations into the environment in GM crops and transgenic wastes. GM constructs are highly chimerical, and contain homologies to many different species of bacteria and viruses, and hence capable of engaging in high frequencies of recombination with a wide variety of bacteria and viruses. I have pointed this out for years. It is gratifying to find a Harvard molecular geneticist who is saying the same thing [3]:
"Transgenes often contain DNA sequence homology to prokaryotes thereby increasing their likelihood of integration in bacteria significantly. Many studies have shown that DNA homology is the main barrier to HGT of chromosomal DNA (such as transgenes) in bacteria."
Still, 'illegitimate' recombination events should not be ignored. They may occur at lower frequencies, but will become substantial as GM constructs are released on massive scales. In particular, recombination hotspots associated with many GM constructs (such as those containing CaMV 35S promoter) may increase the frequency of illegitimate recombination.
The rhizosphere - surfaces around the plant roots - and the spermosphere - surfaces around the germinated seeds - are recognised hotspots for horizontal gene transfer between bacteria. But the frequency also depends on the location of the genes. It matters whether they are in the bacterial chromosome, or in a mobilizable plasmid, ie, a plasmid that can be transferred with helper functions supplied by other plasmids, or in a conjugative plasmid, ie, a plasmid that has its own functions for transfer during conjugation (mating between bacterial cells). Not surprisingly, researchers found the highest frequencies of horizontal gene transfer in both the rhizosphere and spermosphere when the GM cassette was in a conjugative plasmid, somewhat lower when it was in a mobilizable plasmid, but could not be detected when it was inserted into the bacterial chromosome [4].
However, that does not mean GM constructs located on bacterial chromosomes do not transfer. The authors were careful to point out that the main mode of horizontal gene transfer in both the rhizosphere and the spermosphere is conjugation, a process that require cell to cell contact between bacteria. Elsewhere, transformation (by direct uptake of DNA) will be more important, and there is evidence that chromosomal constructs are more efficient in transformation.
The authors warn: "On the basis of these experiments, we cannot rule out the possibility that horizontal gene transfer by transformation occurs at low frequencies in soil and that this process might have significant effect at field scale, which is an especially important point as regards risk assessment. Such rare events cannot be studied in microcosm experiments, but must be addressed in retrospective field studies."
The only retrospective study carried out has indeed found evidence of horizontal gene transfer from transgenic plants to soil bacteria (see "Horizontal gene transfer happens", ISIS News 5 ). An investigation yet to be done is horizontal transfer from GM plants to bacteria in the rhizosphere.
Transposable elements are genetic units that can move from one chromosome to another, with or without multiplying themselves in the process. Tol2 is a 4.7 kbp element found in the genome of the medaka fish Oryzias latipes. It has terminal inverted repeats and contains four genes similar to a group of transposable elements found in fruitflly, maize and snapdragon. There are some 10 to 30 copies of Tol2 in the medaka genome that are highly homogeneous in structure, and no variation in base sequence was found when 5 random clones were examined [5]. This is unusual, as transposable elements are typically heterogeneous, with many defective copies being present in the same genome.
The genus Oryzias contains more than 10 species. The researchers examined 10 species but Tol2 was found in only 2 of them: O. curvinotus and O. latipes. The structure of the Tol2 is homogenous and identical both within each species and between the two species, which are not closely related and do not crossbreed in nature.
These results suggest very recent horizontal gene transfer. The two species overlap in distribution probably somewhere in Southern China. Tol2 could have transferred from one species to another, or both species could have acquired it from the same source. They also illustrate the dangers of using transposable elements as gene transfer vectors (see "Stop release of GM insects", this issue).