Scientific Narrative: Horizontal Gene Transfer and Transgenic Crops

Table of Contents

1. Introduction
2. Background - DNA
3. Background - Horizontal Gene Transfer
4. Background - Relative Frequency of HGT
5. Background - Barriers to HGT
6. Background - HGT and Natural Selection
7. Movement of DNA from biotech crops to soil microorganisms under natural conditions
8. Movement of transgenes from ingested food to enteric bacteria via horizontal transfer
9. Incorporation of transgenes from ingested food into mammalian cells
10. Likelihood of transgenes and other genes being involved in HGT
11. The effect of modular makeup of transgenes on the rate and impact of HGT
12. Antibiotic resistance marker genes and biotech crops

I. Introduction and Background Information

Crops developed through biotechnology, such as insect resistant and herbicide tolerant corn hybrids, have been on the market since in the mid-1990s. Farmers in the United States have adopted these crops rapidly because they provide better yields, less crop loss from damaging pests, lower input costs and better crop quality.

The adoption of these crops has been accompanied by considerable discussion about benefits and risks of biotechnology. Some of the discussion has focused on the issue of the horizontal transfer of genes from crops developed through biotechnology, or transgenes, to other organisms.

Horizontal gene transfer (HGT) is defined as the movement of genes between independent, co-existing organisms, in contrast to the inter-generational "vertical" transfer of genes from parent to offspring. In this sense, HGT does not include the transfer of genes through sexual reproduction methods, such as cross-pollination. HGT occurs regularly and includes the movement of genetic material between two closely related species, such as two similar bacterial species, or between very dissimilar species, such as from bacteria to plants. In theory, genetic material has the potential to move horizontally between any two species, but HGT involving eukaryotes is rare, while HGT among prokaryotes is well known.

But, considering the millions of species that co-exist in close physical contact and the universal presence of DNA in environments, the trivial amount of HGT that occurs is actually remarkable. Organisms are much more likely to reject foreign DNA than accept it, and evolution has provided them with a number of barriers to prevent the uptake and incorporation of foreign DNA.

Specifically, the questions about HGT focus on whether transgenes might move:

• from crops developed through biotechnology to soil microorganisms and have unexpected environmental impacts;
• from ingested material from plants developed through biotechnology to intestinal cells or intestinal microflora and have adverse impacts.

These questions also apply to a component of some of the transgenes found in certain biotech crops currently on the market: the antibiotic resistance marker used to identify transgenic cells at the earliest stages of crop development. Some have specifically questioned whether horizontal transfer of the antibiotic resistance marker from biotech crops could have negative impacts on the environment, animal agriculture and human health. However, the antibiotic resistant markers are now rarely used in the development of biotech crops.

This paper is a summary of the available scientific information that is relevant to assessing the potential impact of HGT on human health, animal agriculture and the environment.

Return to Table of Contents

Ever since Avery et al. (1944) proved that DNA is the genetic material, molecular biologists have amassed incontrovertible evidence that DNA is chemically the same, regardless of its source. DNA is contained in the cells of virtually all organisms, and its chemistry is conserved across life forms. In addition, the DNA in all organisms is subject to mutation, recombination, restriction and replication.

The universality of DNA among diverse species provides the basis for recombinant DNA technology. Genes are composed of the same four nucleotides, so distinguishing transgenic DNA from non-transgenic DNA based on chemical makeup is not possible. In addition, all of the components used to create transgenes, such as promoter sequences, the desired trait and marker sequences, were originally discovered in living organisms.

Because of DNA's universality, all organisms that are consumed or that decay in the environment, transgenic or not, contain DNA. DNA from all organisms and sources, including the component sequences that code for regulatory and housekeeping gene functions, is in the environment. Because of their identical chemical make-up, there is no reason to believe an organism is more likely to take-up a transgene than another gene.

While the overall chemistry of DNA is similar irrespective of its source, species uniqueness results from differences in the sequence of nucleotides that is then translated into a sequence of amino acids. Because the code for converting a chain of nucleotides into a chain of amino acids is universal, one organism, irrespective of the degree of their phylogenetic divergence, can translate a gene from another organism.

Return to Table of Contents

As previously noted, HGT is defined as the movement of genes between independent, co-existing organisms. The scientific literature clearly indicates that HGT occurs and has likely played an important role in speciation and evolution, especially in prokaryotes (Smith et al., 1992; Syvanen, 1994; Syvanen, 1999; Woese, 1998; Jain et al., 1999). Literature reviews and entire volumes on HGT have been published (Roberts et al., 1996; Schmidt and Hanklen, 1996; Wostemeyer et al., 1997; Nielsen et al., 1998; Syvanen and Kado, 1998).

3.1   Methods of HGT

There are three distinct mechanisms of HGT: conjugation, transduction and transformation (Salyers et al., 1995).

• Conjugation is the direct exchange of DNA from one cell to another. Conjugation requires physical contact between the two cells. This distinguishes it from the other two methods of HGT.
• Transduction is viral-mediated transfer of DNA between cells.
• Transformation is the direct uptake of DNA from the environment. Since DNA is found in all living things, it is found in soil, water, and digestive tracts, so transformation could occur in any of these environments (Lorenz and Wackernagel, 1994; Doerfler et al., 1998; Dröge et al., 1998; Nielsen et al., 1998).

3.2   HGT between prokaryotes

HGT is most common among prokaryotic species. All three mechanisms of HGT have played a key role in bacterial speciation and have led antibiotic resistance in bacteria (Davies, 1994). In prokaryotes, any gene has the potential to move horizontally between any two species (Sprague, 1991). Nonetheless, certain factors affect the propensity for and probability of a gene being transferred between two prokaryotes.

3.2a   Conjugation

In conjugation, one bacterial cell acts as the donor, the other as the recipient. The ability to transfer DNA by conjugation depends on a so-called fertility factor carried by the donor cell and absent in the recipient cell. The fertility factor may be located on a self-replicating plasmid or, occasionally, integrated into the donor's chromosome. Physical contact between the donor and recipient cells is required for HGT. Successful HGT via conjugation requires both contact (conjugative functions) and mobilization and transfer of DNA. Some plasmids do not have the genes required for these functions. In bacterial species in which the fertility factor is integrated into the chromosome, chromosomal genes can also be transferred during conjugation.

3.2b   Transduction

Transduction between prokaryotes occurs when bacterial viruses, or phages, acquire host DNA, then infect another bacterium and transfer DNA from the host bacterium to the infected bacterium. Some phages are able to mobilize bacterial genes and carry them between bacteria. Within that group, generalized transducing phages can carry any part of the chromosome, while specialized transducing phages can carry only limited parts of the host chromosome. Finally, HGT via transduction is further constrained by the specificity of phages for certain hosts.

3.2c   Transformation

In prokaryotic transformation, a bacterium takes up a DNA molecule. Transformation does not require the DNA donor to be alive. Therefore, it is possible for the DNA donor and DNA recipient to be separated in time and space (Leff et al., 1992). Uptake of plasmid DNA also occurs without regard to sequence or source (Hanahan, 1983). Finally, DNA uptake is mediated through a single-stranded DNA intermediate (Palmen and Hellingwerf, 1997), and, therefore, host restriction may not play a significant role in "detoxifying" foreign DNA.

A number of factors affect the amount of transformation occurring in prokaryotes.

• Not all bacterial cells are "competent," or able to be transformed, all of the time. Bacteria must enter a highly regulated state of competence to participate in transformation (Havarstein, 1998). A repertoire of cell functions for binding and processing external DNA must be induced (Solomon and Grossman, 1996). This occurs only under certain conditions. Some bacterial species are competent throughout their growth cycle, some at certain stages in their growth cycle, while others can be made competent only under highly artificial laboratory conditions. For these species, transformation can occur only when unique nutrient and culture conditions are provided. However, even under optimal laboratory conditions, the frequency of successful transformation is low.
• The probability of HGT through transformation depends not only on the bacterial species involved but also on environmental variables. All DNA in an environment have a finite probability of being utilized in transformation. Therefore, the frequency of DNA uptake depends on its concentration in the environment (Hanahan, 1983). Transformation is possible in soil, water or other environments where DNA and competent bacteria are present (Lorenz and Wackernagel, 1994; Droge et al., 1998; Nielsen et al., 1998). Therefore, certain, naturally competent soil bacteria may be in a position to take up DNA, but scientists expect the frequency of occurrence to be extremely low. For some bacteria, proximity to plant tissue (and the nutrients being secreted by the plant cells) increases their capability to incorporate new DNA (Droge et al., 1998). Finally, certain soil types have been found to bind DNA, protecting it from degradation and, therefore, extending the duration during which bacteria are exposed to DNA (Lorenz and Wackernagel, 1994; Nielsen, 1998).
• Integration typically occurs by homologous recombination (Smith et al., 1981), and the precise mechanism is dependent on the percentage of sequence similarity between the new DNA and endogenous sequences of the host genome. Homologous recombination frequencies of 10-5 are common (Shen and Huang, 1986). Because, some bacteria have a well-developed mechanism specifically for transporting environmental DNA into the cell, and integrating it into their genome via recombination/repair, it has been speculated that natural transformation is an avenue for bacteria to produce and introduce variability (Syvanen, 1994; Havarstein, 1998).

3.3   HGT involving eukaryotes

Compared to HGT among prokaryotes, HGT involving eukaryotes is even more rare, especially when it involves the movement of genes between organisms in different kingdoms. HGT from eukaryotes to prokaryotes has never been shown experimentally under non-laboratory conditions (Bertolla and Simonet, 1999). However, evolutionary studies comparing genomic sequences obtained from prokaryotes and eukaryotes provide growing evidence that gene transfer has occurred (in both directions) over a geological time frame (Mazodier and Davies, 1991; Smith et al., 1992; Syvanen, 1994; Woese, 1998; Jain et al., 1999).

3.3a   HGT involving plants

DNA sequence homology between plants and bacteria has led to the speculation that HGT from bacteria to plants, and vice versa, has played a role in evolution (Prins and Zadoks, 1994).

Conjugation-like transfer from the bacteria to plants is well documented for the plant pathogen Agrobacterium (Sheng and Citovksy, 1996). As part of the infection process, Agrobacterium transfers and integrates some of its DNA into the genome of a variety of plant species (Lessl and Lanka, 1994; Buchanan-Wollaston et al., 1987). Some species in the plant genus Nicotiana contain additional genes, unrelated to the infection process, that biologists believe were obtained from Agrobacterium (Furner et al., 1986). Agrobacterium is the only genus of bacteria known to "conjugate" with plants, but the transfer of genes from bacteria to plants is not surprising. According to microbiologist Julian Davies, "Microbes are masters at genetic engineering, and heterologous expression vectors of broad host range in the form of integrons were present in bacteria long before they became the vogue for biotechnology companies in the 1980s" (Davies, 1994).

A special and ancient form of HGT from bacteria to plants has happened when plant chlorplasts arose from cyanobacteria through endosymbiosis. It has been estimated that about 4,500 of Arabidopsis protein encoded genes (about 18% of the total) were acquired from the cyanobacterial ancestor of plastids (Martin et al., 2002).

The importance of HGT in the reverse direction - from plants to bacteria - under natural conditions remains unclear but has been receiving increased attention. Of the three methods of HGT, transformation is recognized as the method with the greatest potential for movement of DNA sequences from plants to other organisms. However, experiments to induce HGT via transformation from plants to Agrobacterium under natural conditions have not been successful (Broer et al., 1996). Nonetheless, based on sequence homology, it appears likely that a number of genes have been transferred from a plant into a bacterium (Froman et al., 1989; Wakabayashi et al., 1986). For example, sequence analysis suggests that E. coli's phosphoglucose isomerase gene is of plant origin (Froman et al. 1989). Another 3-4 bacterial genes appear, based on sequence homologies, to have a eukaryotic origin (Carlson and Chelm, 1986; Doolittle et al., 1990; Lamour et al., 1994). More indications of HGT from plants to prokaryotes will likely be identified as prokaryotic and eukaryotic genome sequencing progresses.

3.3b   HGT involving animals

Recent analysis of the human genome (International Human Genome Sequencing Consortium, 2001) revealed that at least 100 genes share extensive homology between higher animals (vertebrata) and prokaryotes. For 25 of 35 of those genes tested, the bacterial gene could be detected in the human genome by PCR. The authors conclude that human genes appear likely to have resulted from HGT from bacteria at some point in the vertebrate lineage. However, subsequent evolutionary studies and database research suggest that the claim of HGT from bacteria to vertebrata was unfounded (Salzberg et al., 2001; Roelofs and Haastert, 2001; Stanhope et al., 2001). Despite the fact that HGT between prokaryotes and eukaryotes is controversially discussed, recent research results strongly suggest that more recent within the evolutionary timeframe, the adzuki bean beetle, Callosobruchus chinensis, has acquired a genome fragment from a bacterium, its natural endosymbiont Wolbachia (Kondo et al., 2002).

Gene transfer from bacteria to mammalian cell lines has been demonstrated in the laboratory from Agrobacterium to HeLa cells (Courvalin et al., 1995) and Escherichia coli to CHO K1 cells (Waters, 2001).

Uptake of ingested pieces of DNA by some host cells occurs naturally. This has been occurring since organisms began relying on each other as food sources. Interestingly, this phenomenon is not restricted to multicellular organisms; uptake and incorporation of DNA in the process of eating appears to have occurred in single-celled eukaryotes, such as protists (based on molecular evolutionary data from Smith et al. (1992).

Several studies have addressed the fate of ingested DNA in mammalians, including attempts to detect transgenic DNA in chicken (Khumnirdpetch et al., 2001) or cows (Klotz and Einspanier, 1998) fed with glyphosate tolerant soybean and in pork (Weber and Richert, 2001), pigs (Klotz et al. 2002) or dairy cows, beef steers and broiler chicken (Einspanier et al., 2001; Flachowski et al., 2000) fed with Bt corn. In none of those studies was transgenic DNA detectable by PCR in various samples.

Experiments conducted by Doerfler's laboratory at the University of Cologne in Germany found that very small amounts of DNA fragments may be absorbed across the epithelium in the gastrointestinal tract (Schubbert et al., 1994; Schubbert et al., 1997; Doerfler and Schubbert, 1998; Hohlweg and Doerfler, 2001).

HGT requires not only that the DNA be taken up by cell, but also that the ingested fragments are maintained in the recipient cells by integration into the genome. The work of Schubbert et al. (1997, 1998) provides an ambiguous demonstration of integration of ingested DNA fragments into an animal genome. When mice were fed daily with GFP DNA for 8 generations, no GFP DNA was detectable in DNA isolated from tail tips and internal organs by PCR, which "argue against the germline transfer of orally administered DNA" (Hohlweg and Doerfler, 2001).

Return to Table of Contents

Complete HGT involves not only the uptake of new DNA by a cell but its incorporation into the cell's genome. The frequency of complete HGT under natural conditions is unknown. Based on genome sequencing, which is helping to refine phylogenetic relationships, the following inferences have been made about the relative frequencies of HGT within and between kingdoms (from highest to lowest frequency):

• bacterium to bacterium:
• bacterium to plant (very species-specific);
• plant to bacterium;
• eukaryote to eukaryote.

4.1   Experimental Estimates of Rates of HGT between Bacterial Species

Under optimal laboratory conditions, HGT via conjugation between two bacterial species that can cross conjugate, ranges from 10-1 to 10-3 (expressed as the total number of cells receiving DNA divided by the total number of potential recipients). However under natural conditions, HGT through conjugation between the same two species drops by many orders of magnitude to 10-8 to 10-9 (Frischer et al., 1994; Jiang and Paul, 1998). It can be assumed that conjugation provides a higher success rate of HGT than other mechanisms because the cells are in direct contact with each other. Therefore, bacterial species with the ability to conjugate will experience significantly more HGT than bacterial species that cannot cross conjugate and are limited to viral mediated HGT or uptake of DNA from the environment.

4.2   Phylogenetic Estimates of HGT between Bacterial Species

Comparing DNA sequencing data provides another method for estimating the rates of HGT between bacterial species. Escherischia coli and Salmonella spp. are very close relatives, having diverged from a common ancestor to form the two species approximately 100 million years ago, which is very recent in prokaryotic evolution. Since their speciation, the two bacteria have transferred a large number of genes back-and-forth. In fact, approximately 18% (about 750 genes) of the E. coli genome has been obtained through HGT (Lawrence and Ochman, 1998).

This percentage provides an estimate of the highest possible frequency of HGT because:

• HGT is much more common within the prokaryotes than either between prokaryotes and eukaryotes or within the eukaryotes, and
• the more closely related two species are, the higher the frequency of HGT between them.

In spite of these predisposition factors that encourage gene exchange between E. coli and Salmonella spp., the rate of HGT is about one gene every 50,000 years (Ochman et al., 2000)

Return to Table of Contents

When considering:

• the degree to which millions of species co-exist in close physical contact with each other,
• the ability of viruses to move freely between host organisms, and
• the amount of DNA in certain environments,

one realizes the trivial amount of HGT that occurs is actually remarkable. Organisms are much more likely to reject foreign DNA than accept it, and evolution has provided them with a number of barriers to prevent the uptake and incorporation of foreign DNA. Physical barriers to HGT include cell membranes and cell walls. Cell membranes are also studded with proteins that permit the movement of only certain molecules into a cell. Extracellular nucleases excreted by cells degrade DNA in the environment. Also, organisms have developed internal defense mechanisms, especially intracellular nucleases that degrade foreign DNA that slips through the external barriers.

If DNA integrates into the host chromosome, it can escape enzymatic degradation by nucleases. Integration is a low-frequency event and occurs either through homologous recombination or illegitimate recombination. Homologous recombination requires some sequence homology between a stretch of the host chromosome and the incoming DNA and the degree of homology determines the efficiency of integration (Vulic et al., 1997; Majewski and Cohan, 1998; Majewski et al., 2000). In bacteria, frequencies of homologous recombination of 10-5 are common. Illegitimate recombination involves random insertion and no sequence similarity; it is 1000 times less likely than homologous recombination (Rayssiguier et al., 1989).

Even if a sequence is integrated, it may not be expressed. In eukaryotes, foreign sequences may be silenced by methylation (Sutter and Doerfler, 1980; Doerfler, 1983; Orend et al., 1995).

Return to Table of Contents

The discussion above focuses on the probability of, and impediments to, HGT between two individual organisms. This information is essential, but not sufficient for assessing the potential for environmental or public health risks from HGT. The questions listed earlier in this paper require an analysis of what happens after successful transfer of an intact gene from one individual to another. Gene transfer does not provide the potential for an incipient problem unless a significant number of individuals in a population of the recipient species acquire the new gene. While all DNA fragments may participate equally in HGT, only those nucleotide sequences that provide a selective advantage and do not carry significant fitness costs will become widespread in a population (Nielsen, 1998; Malik and Saroha, 1999).

The clearest example of this is seen in the evolution of antibiotic resistance. When a bacterial population is exposed to a new antibiotic, the initial gene frequency for resistance to that antibiotic is expected to be low. However, when the population is subjected to selection pressure provided by the new antibiotic, those organisms with the resistance gene will increase in number. Eventually, the resistance gene is pervasive in this bacterial population and may become fixed in the presence of continuing selection pressure provided by the antibiotic.

It has been demonstrated that antibiotics can increase the frequency of HGT 10-1000 fold, because the selective pressure accelerates the spread of resistance genes (Salyers et al., 1995; Torres et al., 1991).

Return to Table of Contents

II. Discussion of Research Data

Nielsen et al. (1998, 2001) and Droge et al. (1998) have published reviews of the work on horizontal gene transfer from biotech plants to terrestrial bacterial.

A number of experiments have been undertaken to determine if, under non-laboratory conditions, DNA can move from biotech plants to soil bacteria and fungi (Paget et al., 1998; Gebhard and Smalla, 1999).

Experimental attempts to establish microbial transformation - the most likely mode of HGT from plants to prokaryotes - with transgenic DNA under non-laboratory conditions have been uniformly negative. To date, no experiment has provided evidence for the transfer of either transgenes or other genes from plants to soil microbes under non-lab conditions.

These experimental findings confirm results of investigations of HGT during and after field tests of biotech crops. Numerous field releases of a variety of biotech crops have been made over the past 14 years. Follow-up studies to assess HGT in the field have revealed no detectable transgene transfer to soil microbes under natural conditions (Smalla et al., 1994; Paget and Simonet, 1994; Badosa et al., 2004).

In the laboratory, HGT from biotech plants to microorganisms has been demonstrated under optimized conditions (Hoffmann et al., 1994; Schluter et al., 1995; Gebhard and Smalla, 1998; De Vries and Wackernagel, 1998; Nielsen et al., 2000; Meier and Wackernagel, 2003). Using a model system based on biotech Brassica plants and Aspergillus niger, a pathogenic fungus that infects brassicas, Hoffman et al. (1994) demonstrated successful transformation of A. niger with an antibiotic resistance marker gene. The rate of successful transfer was too low to calculate a frequency of HGT, however. Interestingly, the fungi that acquired the antibiotic resistance gene during co-cultivation with biotech plant tissue lost the resistance during further strain cultivation, even under continuous selective pressure. In addition, an intact antibiotic resistance gene was never identified in the resistant colonies.

Utilizing a different model system, Schluter et al. (1995) attempted to stimulate HGT between biotech potatoes and Erwinia chrysanthemum, a pathogen tightly associated with potatoes, in order to calculate rates of HGT under a wide variety of conditions. Under conditions they describe as "idealized" natural conditions, they calculated a HGT frequency of 2 X 10 -17, a rate they describe as "so rare as to be essentially irrelevant in any realistic risk assessment of biotech crops." These idealized natural conditions included using a bacterial marker gene linked to a functional origin of replication. It is important to note that most transformed plants do not contain origins of replication in their transgenes. Therefore, one might expect the frequency of transformation with transgenes to be even lower than the 2 X 10 -17 calculated by Schluter et al. (1995).

Several studies were performed in which transgenic plant DNA with a functional antibiotic resistance gene (nptII) was incubated with natural transformable bacteria (Acinetobacter sp. or Pseudomonas stutzeri) with a non-functional copy of nptII (317 or 10 bp deletion, respectively) (Gebhard and Smalla, 1998; De Vries and Wackernagel, 1998; Nielsen et al., 2000; Meier and Wackernagel, 2003). The degree of HGT was determined by the restoration of antibiotic resistance and transfer frequencies of 1.4 x 10 -8 to 5.4 x 10 -9 were detected. The laboratory conditions were set up to produce maximum competence of bacteria for DNA uptake and restoration of antibiotic resistance required homologous integration of the biotech plant derived nptII into the bacterial DNA. Even under optimized conditions, homologous integration could not be demonstrated in the laboratory in the absence of homologous sequences (De Vries et al., 2001).

De Vries et al. (2004) used a similar approach to study the probability of HGT from transgenic plant DNA to Acinetobacter sp. DNA extracted from tissues of transplastomic (transgenic DNA integrated in the plastid genome) tobacco containing the aadA gene (resistance against spectinomycin and streptomycin) was exposed to Acinetobacter cells with a non-functional aadA copy. Depending on the degree of homology of the aadA bordering sequences between the plant and Acinetobacter DNA, transformation frequencies of "not detectable" to 1.4 x 10 -4 were observed. This is a higher transformation frequency compared to the experiments described above. The authors suggest that the higher copy number of plastid genes compared with nuclear genes is the reason for this. Previous observations that sequence homology is essential for transgenic plant DNA integration were confirmed in this study.

Some laboratory studies have failed to demonstrate HGT under optimized, laboratory conditions. Broer et al. (1996) infected transgenic tobacco plants with Agrobacterium tumefaciens but could find no instance of HGT from plant to bacterium at a detection level of 6 x 10 -12. Nielsen et al. (1997) examined the frequency of HGT between transgenic plants and Acinetobacter under laboratory conditions and found no transformation with plant DNA.

However, as mentioned above, evolutionary evidence indicates the probable horizontal transfer of a few genes from plants to bacteria via transformation. Therefore, genes can move from plants to bacteria under natural conditions. However, altering the genetics of soil microorganisms via transformation from transgenic plants would appear to be ecologically insignificant since the elements in transgenes already occur and are available for transformation from other organisms.

The more important question is not whether transgenes in plants can move, at some low frequency, to soil organisms but, rather, whether there is sufficient selective advantage to maintain these transgenes over other endogenous genes found in the environment. Concerns about environmental impacts of HGT are germane only in those instances where the transgene spreads through the population.

(Note: Some of the above experiments are discussed in greater detail later in the document in the section on HGT of antibiotic resistance marker genes.)

Return to Table of Contents

In an effort to assess the potential for enteric bacteria to be transformed with ingested DNA, Scott et al. (2000) conducted a series of experiments that utilized the Green Fluorescent Protein (GFP) experimental system. They first constructed a bacterial plasmid integration vector that included a GFP expression cassette and a 450 base pair region of homology to genomic sequences in a variety of bacteria. This construct should increase transformation frequencies to artificially high levels when compared to transgenic DNA sequences because the extensive amount of DNA sequence homology greatly favors integration through homologous recombination. Not surprisingly, this plasmid could transform strains of Lactococcus, Enterococcus and Streptococcus when the researchers used optimized in vitro methods to transform the bacteria. However, when the GFP-expressing Lactococcus was cultured in a simulated human gut environment inoculated with human fecal flora, the transformed Lactococcus exhibited impaired survival relative to non-transformed enteric bacteria. Therefore, HGT was not successful because the gene did not persist in the population even though it was taken up and integrated into the bacterial host genome.

The possibility of plasmid DNA being incorporated via a normal biological process into endogenous gut bacteria is minimized due to the non-conjugative nature of typical plasmids used in recombinant DNA laboratories (Hamer, 1977) and the low frequency with which unaided transformation (uptake of DNA) occurs. Furthermore, beyond the difficulty of unaided transformation is the general lack of stable incorporation of exogenous DNA into host genomes (Behr et al., 1989). Maniatis et al. (1982) calculated the probability of transferring plasmids into natural bacteria in the gut environment to be less than one in one million. However, the transfer of biotech plant genes should be significantly less. The Maniatis et al. (1982) calculation assumes the plasmid is free rather than incorporated into the plant genome. Movement of transgenes from ingested plant material to gut bacteria would require the transgene construct to be precisely removed from the plant genome and incorporated into the bacterial genome.

Mitten et al. (1996) evaluated the potential transfer of a transgene from ingested transgenic plant material to gut bacteria. They estimate a transformation frequency of 1 in 750 billion bacterial cells. More details on this work are discussed later in the section that focuses on HGT and antibiotic resistance markers.

A series of research projects, commissioned by the British Food Standards Agency (FSA), also concluded that HGT of DNA from food derived through biotechnology to human gut bacteria is extremely unlikely (FSA; project codes FSG01007, G010008G01010 and G01011). Study G010008, later published in the journal Nature Biotechnology (Netherwood et al., 2004), included human volunteers. Seven ileostomists were given a single meal containing soy flower from glyphosate resistant soy and the presence of the transgene was monitored over time in the intestine (small bowel). In another experiment of the same study, healthy volunteers, with an intact intestinal tract, were given the same meal and the fate of the transgene was monitored in the feces. After having passed the complete intestinal tract, no transgene was detectable by PCR. However, the transgene was detected in the stoma from the ileostomists. When intestinal microflora of the ileostomists, sampled before the start of the experiment, was cultivated, a small portion of the transgene was detectable at very low levels by PCR in samples from three of the seven volunteers. The authors concluded that gene transfer from transgenic soybean appears to have occurred before the experiment. The fact that attempts to isolate the bacteria supposedly harboring the transgene fragment failed, is explained with the uncultivable nature of those bacteria. In their conclusions, Netherwood et al. (2004) state the gene transfer events in their study would be highly unlikely to pose a risk to human health. This is because a non-functional gene fragment was transferred and because it was transferred to just a few of the millions of intestinal bacteria.

Again, it must be emphasized that enteric bacteria are exposed continuously to an extremely wide spectrum of DNA fragments. The component-sequences of transgenes are all sequences to which enteric bacteria have been exposed through non-biotech foods or in accompanying microflora. Given the high numbers of foreign bacteria and viruses in this intestinal mixture at any one time, and the combinatorial genetic "mixing-and-matching" that normally occurs among prokaryotes and between prokaryotes and viruses, there is likely little new that is being added to this DNA mix from biotechnology. This led Schubbert et al. (1994) to state:

"This barrage of linear DNA fragments, i.e. of recombinationally highly active DNA fragments, in nature should mitigate any concerns that one might have had in the past about biological consequences of experiments carried out with recombinant DNA over the course of the past two decennia."

Return to Table of Contents

Studies from Doerfler's laboratory at the University of Cologne in Germany have addressed the question of the ultimate fate of ingested DNA in mammals. In the first set of experiments the Doerfler team fed mice pure, circular DNA extracted from the phage virus M13, and, 1 to 7 hours after feeding, they identified small amounts of absorbed DNA fragments (200-500bp) in the blood, epithelial cells of the intestinal wall, peripheral leukocytes, spleen, and liver cells. In follow-up studies, they screened 109 clones of splenocyte DNA extracted from the mice and found one M13-positive 1.3 kilobase fragment clone. The M13 DNA fragment was covalently linked to an 80-nucleotide fragment with 70% homology for the IgE receptor gene (Schubbert et al., 1997). Additional chimeric molecules of M13 DNA fragments were found associated with bacterial DNA (the authors suggest from normal gut bacteria) and with rearranged lambda phage DNA. These later findings raise serious questions about the technical conduct of these studies and suggest that the results are possibly common cloning artifacts as cautioned when creating genomic libraries (Maniatis, 1982). That concern is independent of the question of why an 'altered' IgE sequence (or even how the IgE sequence was altered) was found in the same clone.

These studies have been reviewed and criticized (Beever and Kemp, 2000). In general, they raise the possibility of technical methodological errors given the infrequent occurrence of incorporated M13 DNA fragments identifiable only after powerful amplification techniques were used. More specific issues related to experimental design include the

• excessive amount of DNA fed to the mice compared to normal DNA intake, and
• use of circular DNA from the M13 virus

Unlike plant or animal DNA, M13 DNA has short specific sequences that are not methylated. Human and mouse white blood cells, specifically the macrophages, respond to M13 DNA as if it were an infectious agent.

While DNA fragments from digested food may move from the mammalian digestive tract into other cells of the organism, there is no conclusive evidence that intact genes from foods, whether biotech or non-biotech, integrate into the genome of animal cells (Schubbert et al. 1997, 1998; Hohlweg and Doerfler, 2001; Einspanier et al., 2001).

Khumnirdpetch et al. (2001) attempted to detect transgenic DNA in broiler chickens fed meal from glyphosate tolerant soybeans from birth to seven weeks of age. Samples (meat, skin, duodenum and liver) were isolated from the birds at 1, 3, 5 and 7 weeks. PCR results of all samples taken over the seven-week feeding period were negative. The authors speculated that the negative detection results suggest that the transgenic DNA in glyphosate tolerant soybean meal had been fully degraded in the digestive tract of the broilers.

Weber and Richert (2001) attempted to detect the Bt gene and an endogenous corn gene in DNA extracted from 24 pork loin samples (12 fed Bt corn and 12 fed a control conventional corn). PCR followed by Southern blot analysis for ~200 bp fragments of the Bt transgene and endogenous shrunken-2 (sh-2) gene were uniformly negative. By comparison, adding corn DNA to the sample of swine DNA produced positive results, indicating that the DNA quality and PCR conditions were both favorable for detection of DNA fragments, had they been present in the original samples.

Using PCR followed by Southern blot, Klotz and Einspanier (1998) could not detect the transgene from glyphosate tolerance soybeans in either the blood or milk of cows. However, this highly sensitive technique detected a small fragment of a soybean chloroplast gene in white blood cells but not milk.

Einspanier's laboratory has published data from a study in which dairy cows, beef steers and broiler chickens were fed either grain from biotech maize or grain from non-biotech maize (Einspanier et al., 2001; Flachowski et al., 2000). The investigators used two DNA detection technologies [standard Polymerase Chain Reaction (PCR) and Light Cycler "real time" PCR]. No portion of the Bt gene could be detected by either method in any samples from the cows, steers or chickens fed Bt corn. Similar to the previous Klotz and Einspanier (1998) report, however, they could detect a small fragment of a chloroplast gene (tRNAleu) in lymphocytes of dairy cows and in muscle, liver, spleen and kidneys cells of chickens, but not in dairy milk, or any tissue samples from steers.

In a similar study, where pigs were fed Bt maize, short chloroplast DNA fragments were detectable by PCR in intestinal juices, but not in any pig organ investigated. Bt maize specific transgenic DNA fragments were never detected in any pig sample (Klotz et al., 2002).

Therefore, in light of the Einspanier et al. (2001) and Schubbert et al. (1997, 1998) studies, a primary determinant in detecting fragments of ingested DNA in host animal tissues may be the copy number of the sequence ingested. The copy number of the chloroplast genome per cell ranges from ~500 to 10,000 copies in roots and leaves, respectively (Bendich, 1987). Therefore, the copy number of the detected chloroplast gene is orders of magnitude higher than a transgene in a transgenic crop, which typically has only one copy per haploid genome. In addition, chloroplast gene sequences are also present in high numbers in the plant's nuclear genome, with >100 copies of some sequences being recorded (Ayliffe et al., 1998).

Based on these studies, it would be expected that fragments of both native genes and transgenes, as a consequence of food ingestion, would be distributed in a variety of animal tissues, probably in amounts that cannot be detected with current methods. It is important to reiterate that if uptake of small amounts of transgenic DNA fragments does occur, the rate will be trivial when compared to the uptake of other DNA fragments from ingested food. In the gut or in the environment, the amount of non-transgenic DNA exceeds transgenic DNA by many orders of magnitude.

The World Health Organization (1993) and the U.S. Food and Drug Administration (1992) concluded there is no risk in consuming the DNA of biotech crops. The basis of this decision was that mammals have always ingested a large quantity of DNA from plants, animals, bacteria, parasites and viruses, and the proportion of transgenic DNA within native DNA of a biotech crop, for example, is less than 0.00042% (Beever and Kemp, 2000).

In a review on novel DNA in mammalian systems, Doerfler et al. (1998) conclude with the following perspective:

"Of course, we and our ancient history ancestors have all exposed our gastrointestinal surfaces to large amounts of foreign DNA of most variable origins for millions of years. Whatever the evolutionary consequences of this mass experiment in natural gene technology may have been, we just begin to understand in a modest way these consequences by applying concepts of gene technology ourselves. In the light of the magnitude and millenia duration of these natural processes, concerned scenarios about the application of gene technology and its consequences in our age belong to the realm of fairy tales."

Return to Table of Contents

The complete process of HGT depends upon following three separate events:

• gene transfer,
• gene perpetuation in the recipient, and
• gene persistence in the population.

Therefore, a number of factors determine whether the rates of fully effective HGT differ between transgenes or other naturally occurring genes. These factors are relevant whether one is considering HGT from transgenic crops to soil microbes, enteric bacteria or intestinal cells.

10.1   Uptake Factor - Relative Amounts of Genes and Transgenes

The most likely mode of HGT from biotech crops to other organisms is transformation. Only certain microbes are naturally competent and, therefore, "transformable" under natural conditions. Therefore, the physiological state (competent or non-competent) of the recipient microbe affects the rate of transformation, in general, but should not have an impact on "preferences" for transgenes or non-transgenes. In laboratory experiments, the success of bacterial transformation with transgenic DNA drops quickly when non-transgenic DNA molecules are added to the solution; so there is no greater tendency for transgenes or non-transgenes to be taken up by microbes. The probability of uptake of a given DNA sequence depends on the concentration of the sequence in the immediate environment of the recipient cell (Hanahan, 1983), whether the environment is the soil around biotech crops or the intestines of organisms that have ingested transgenic material. Therefore, an important variable in determining whether a cell is more likely to be transformed with transgenic or naturally occurring DNA is their relative amounts in the cell's immediate environment.

A transgene represents only a miniscule portion of the total DNA in the plant cell. Considering the genome size of most plants, the amount of natural DNA in a biotech plant is one million times greater than the amount of transgenic DNA (Beever and Kemp, 2000). So, if one focuses only on the biotech plant, the probability of one of the natural plant genes being taken up is one million times greater than the transgene. When one adds to that the amount of DNA organisms are exposed to, whether from other soil organisms or other ingested food, the amount of transgenic DNA available for uptake becomes even more miniscule in comparison to the amount of other naturally occurring DNA. Therefore, because DNA uptake is concentration dependent, the probability of transformation via a transgene is many orders of magnitude less than that of non-transgenes.

Finally, soil type may affect transformation frequencies because some soils bind DNA more tightly than others. DNA is very stable when bound to mineral clay, and DNA in this soil type is available for transformation (Lorenz and Wackernagel, 1994). Gebhard and Smalla (1999) demonstrated that DNA from biotech plants persists in the soil for months, and in one sample for up to two years. Because transgenes and non-transgenes are composed of the same building block molecules and are, therefore, chemically identical, no difference in the binding of transgenic DNA and non-transgenic DNA to specific soil types would be expected.

10.2   Host Perpetuation Factor - Transgene Composition

While the uptake of non-transgenic DNA is expected to far exceed the uptake of transgenic DNA based solely on their relative amounts, certain aspects of transgene composition may increase the likelihood that a transgene is integrated into some host genomes when compared to other genes. Most foreign DNA taken up by cells is rapidly degraded by nucleases. However, if the foreign DNA integrates into the host chromosome, it is not susceptible to the nucleases and will be perpetuated during cell division. Integration into the host genome occurs primarily through homologous recombination (Smith et al., 1981), at frequencies of approximately 10-5.

Some of the nucleases that bacteria use to degrade foreign DNA, i.e., restriction endonucleases, depend upon sequence-specific recognition sites along the DNA molecule. Some transgenes have been modified to reduce or eliminate some common recognition sites. Therefore, the intracellular stability of the transgenic sequence in some bacteria may be better than that of naturally occurring genes if the host's restriction endonucleases are targeted to the eliminated recognition sites. However, the absence of recognition sites is irrelevant in the case of transformation of mammalian intestinal cells. Transgenic DNA and non-transgenic DNA are equally likely to be degraded by mammalian nucleases as well as microbial nucleases that are not specific for the eliminated recognition sites.

The more bacterial sequences in the transgene, the higher the likelihood of successful integration into the recipient bacterial genome. The trend in plant transformation research for a number of years has been to continually reduce the amount of bacterial DNA being introduced into the plant. Nonetheless, some transgenes in currently marketed biotech-derived crops contain bacterial sequences. In terms of integration of ingested transgenic material into intestinal cells, little sequence similarity is expected between the transgene and mammalian genomes. Researchers in the lab that has demonstrated uptake of foreign DNA (non-transgenic) by mammalian cells have not conclusively been able to show that integration occurred (Schubbert et al., 1997; Schubbert et al., 1998). Equally important, the foreign DNA was not expressed by the host cells (Hohlweg and Doerfler, 2001).

Another aspect of transgene structure that can alter the frequency of perpetuation is whether or not the plant transformation was mediated by a transgenic plasmid containing an origin of replication. Plasmids were only used in first generation of biotech products, and most biotech crops on the market today do not contain origins of replications as part of their transgenes.

A final aspect of transgene construction that affects HGT is the nature of the promoter. If a eukaryotic promoter drives the transgene, even if bacteria are transformed, the transgene will be expressed poorly. This relates to the aspect of HGT discussed below: selective advantage.

10.3   Persistence in the Population Factor - Natural Selection and Fitness

While the focus of the discussion of HGT is often on events in single cells, the possibility that a HGT event could pose an environmental or health risk depends upon the gene persisting at some level in a population of cells. This occurs only if there is no fitness cost associated with the newly acquired gene. Furthermore, the probability that a HGT event will have environmental or health effects, depends upon the frequency of the new gene in the population. An increase in gene frequency depends upon selective pressure that favors the spread of that gene throughout the population. Therefore, implicit in the issues described above are questions about differences in selective advantages and disadvantages provided by transgenes and other genes.

The majority of the transgenes in biotech crops being currently grown by farmers would not be expected to impart a selective advantage to soil microbes and might, instead, confer a detrimental "growth drag" on them. This fitness cost would decrease the probability the new gene would survive in the host. However, because scientists do not understand all the selective pressures acting on soil microbes, it cannot be said with absolute certainty that a crop transgene would not confer some unexpected selective advantage. However, naturally occurring plants, animals and microbes release significantly more transforming DNA into the soil than biotech crops. Many of the microbial genes have proven survival value for soil microbes, so one would expect the persistence of these naturally occurring genes to exceed transgene persistence.

In the work described above, Scott et al. (2000) transformed Lactococcus with a GFP expression cassette that was linked to 450 base pair sequence that is homologous to sequences in many bacteria (to promote homologous recombination). Transformed Lactococcus cultured in a simulated human gut environment inoculated with human fecal flora had impaired survival relative to non-transformed enteric bacteria. Therefore, HGT was not completely successful because the gene did not persist in the population even though it was taken up and integrated into the bacterial host genome.

10.3a   Natural Selection, Fitness and Antibiotic Resistance Markers

It should be noted here that a possible selective advantage could be conferred on some species of soil microorganisms or enteric bacteria if an antibiotic resistance marker gene that is used in some biotech crop varieties (see discussion below) is involved in a HGT event. In nature, microbes produce antibiotics to inhibit growth of other microbial species. Should one of these target species obtain a gene that provides resistance to an antibiotic secreted by a microbe with which it competes, the acquisition of this gene may provide a clear selective advantage to the recipient species. However, as described above, a competitive advantage is not guaranteed. Hoffman et al. (1994) demonstrated successful transformation of a soil fungus with an antibiotic resistance gene. However, the fungi that acquired the antibiotic resistance gene during co-cultivation with biotech plant tissue lost the resistance during further strain cultivation, even under continuous selective pressure. Similarly if enteric bacteria acquire an antibiotic resistance gene, this could be a selective advantage if the intestinal flora is ever exposed to that antibiotic.

The antibiotic resistance marker genes are typically derived from microorganisms and, therefore, are already found in nature. So, a transgene does not provide a novel antibiotic resistance gene; nor does it provide a significant number of that antibiotic resistance gene, when compared to the numbers of naturally occurring genes that provide the same trait. Also, antibiotic resistance genes used in biotech plants as markers have been altered so that they are expressed in plant cells but generally function poorly in bacterial cells.

It is widely believed that antibiotic resistance genes from other bacteria participate in HGT with much greater frequency than similar genes from plants because bacterial DNA would possess prokaryotic promoters as well as sequences that promote homologous pairing and recombination (Smith et al., 1981; Nielsen, 1998).

Return to Table of Contents

Transgenes that are added to plant genomes are comprised of novel combinations of components. However, each of the components derive from those found in nature. For example, commercially available Bt maize includes a common soil bacterium called Bacillus thuringiensis (often referred to as Bt) produces proteins that kill specific insect larvae. The independent elements of transgenes maintain their identity and individuality; that is, the sequences of the elements are not randomly mixed and integrated to produce wholly new, mosaic sequences.

Some have expressed concern that this modular makeup leads to an inherent structural instability, and such instability would lead to an increase in the potential for HGT of the transgene's component parts.

Others have expressed concern that fusion of different components that do not occur together in nature produces transgenic constructs that are inherently unsafe because the transgene appears to be unnatural and unfamiliar to transformed microbes. Data would not support this contention (Hanahan, 1983). DNA sequences from organisms that are native to a particular geography are not handled differently by soil microbes compared to sequences derived from foreign organisms. An exception to this might be those sequences that have significant homology with resident bacteria. Other DNA is recognized as, "unfamiliar" or foreign, and it is degraded by nucleases and not integrated.

DNA and microorganisms are ubiquitous in the environment, and most bacterial species that integrate environmental DNA into their genomes do so regardless of source and composition. It is possible that recombination between and among host and foreign DNA sequences during transformation has produced any number of heretofore "non-naturally" occurring sequences and novel combinations of genetic elements (Syvanen, 1994). New modular combinations may have resulted in new functions that may offer adaptive advantages (Spratt, 1988; Lujan et al., 1991).

Return to Table of Contents

Generally speaking, an antibiotic is any substance produced by an organism that has a toxic effect on another living organism. Many microorganisms produce antibiotics to kill other microorganisms.

Plants derived through biotechnology typically contain a herbicide resistance or antibiotic resistance gene linked to the gene of interest. During the transformation process, DNA containing an antibiotic or herbicide resistance gene is co-delivered into plant cells along with the gene of interest. Selective pressure is then used to identify those cells that were successfully transformed. To accomplish this, the cells are usually placed on medium containing an appropriate antibiotic or a herbicide. Only the cells containing the antibiotic resistance gene or the herbicide resistant gene (linked to the gene of interest) are capable of growth. Therefore, the plant cells that contain the target transgene gene can be identified. These cells are then regenerated into whole, transformed plants.

The antibiotic resistance genes or herbicide resistance genes used to recover transformed plant cells are commonly called "plant selectable markers." Plant selectable markers are expressed by the plant and are regulated by eukaryotic expression signals. The DNA used for plant transformations is typically amplified through bacterial culture. To allow growth of only the bacteria containing the desired DNA, an additional bacterial selectable marker is often included.

For selection in bacteria, the amp gene conferring resistance to ampicillin, the aada gene conferring resistance to spectinomycin or the nptII gene conferring resistance to kanamycin or neomycin have been used. These genes are expressed under prokaryotic regulatory sequences. For simplicity, these prokaryotic selectable markers are sometimes included in the DNA used to transform plants.

The nptII gene is the most commonly used antibiotic plant selectable marker. The safety risks of antibiotic resistant plants carrying this gene have been carefully reviewed (See Nap et al., 1992; Redenbaugh et al., 1994; U.S. FDA, 1998). These authors concluded that there is no risk associated with use of this antibiotic resistance gene as a selectable marker.

Furthermore, if HGT from crops developed through biotechnology to bacteria occurs, it would not remarkably increase the pool of natural antibiotic resistance because nptII, as well as aada and amp, are widespread in the bacterial domain.

12.1   Prevalence of antibiotic resistance genes in the natural environments

Because HGT of a specific piece of DNA depends on the relative concentrations of DNA fragments in the environment, the frequency of antibiotic resistance genes in a population is germane to any assessment of the potential of HGT of antibiotic resistance marker genes.

The prevalence of antibiotic resistant microorganisms in natural populations is high (Fontain and Hoadley, 1976; Talbot et al., 1980; Smalla et al., 1993). For example, Armstrong et al. (1981) sampled municipal water supplies in Oregon and found, on average, 34% of the bacteria were resistant to multiple antibiotics. Because antibiotic resistance is often found in non-antibiotic producing soil microorganisms, this suggests that these genes were selected for because they allowed certain microorganisms to resist antibiotics produced by competitors in their environment (Trieu-Cuot et al., 1987).

Malik and Saroha (1999) have concluded that the presence of antibiotic resistance genes is so ubiquitous, it is highly unlikely that the introduction of biotech plants with antibiotic resistance marker genes will alter the frequency or distribution of these genes.

12.2   The effect of using antibiotic resistance marker genes on development of resistance to antibiotics by bacteria

Antibiotic resistance in pathogenic bacteria is a serious public health issue. The number of pathogens resistant to one or more antibiotics has increased over time (Neu, 1992). It is universally acknowledged by physicians, public health officials and medical scientists that the evolution of antibiotic resistance can be traced to the widespread use of antibiotics (Davies, 1994). Antibiotics select for pathogens that are resistant to that antibiotic, while those that are susceptible to the antibiotic are killed. Therefore, the gene for resistance is perpetuated in populations treated repeatedly with the same, or a very similar, antibiotic, while the gene for susceptibility effectively - or actually - disappears from those same populations.

The rapidity with which antibiotic genes have spread throughout bacterial populations is due, in part, to HGT between bacteria. It has been demonstrated that antibiotics can increase the frequency of HGT 10-1000 fold, thus accelerating the spread of resistance genes while providing the selective pressure for their acquisition (Salyers et al., 1995). In clinical isolates of resistant strains, the origin of antibiotic resistance can be traced to genes that reside on plasmids or transposable elements that can readily spread, via conjugation, between bacterial genera (Trieu-Cuot et al., 1987). Other factors that promote rapid increases in gene frequencies in bacterial populations are short generation time and asexual reproduction.

12.3   Potential for transfer of antibiotic resistance genes from biotech plants to soil microorganisms

A number of experiments have been undertaken to determine if DNA can move from biotech plants to soil microorganisms (Droge et al., 1998). There are no cases where DNA transfer has been observed between biotech plants and soil microbes under natural conditions (Prins and Zadoks, 1994). Badosa et al. (2004) examined whether the bla ampicillin resistance gene present in transgenic corn could be detected in soil bacteria collected from commercial transgenic corn fields in Spain. Although many bacteria sampled in transgenic and conventional corn fields were naturally resistant to ampicillin, none of the 108 bacteria analyzed contained the plant transgene.

Using species known to be competent for transformation a number of laboratory experiments have attempted to determine if the DNA from biotech plants could move by HGT into bacteria under artificial conditions.

Gebhard and Smalla (1998) transformed Acinetobacter with a plasmid containing a plant nptII expression cassette. No transformants were obtained because the expression cassette contained eukaryotic promoter sequences. However, when they used biotech plant DNA to transform an Acinetobacter strain that harbored a defective nptII gene driven by a prokaryotic promoter, they were able to obtain transformants due to homologous recombination between a nptII gene under eukaryotic regulatory sequences and the defective nptII gene under prokaryotic regulatory sequences.

In another study, biotech plants expressing a hpt gene (conferring hygromycin resistance), driven by a promoter that works in both eukaryotic and prokaryotic cells, were co-cultivated with Aspergillus niger under sterile conditions (Hoffmann et al., 1994). Following co-cultivation, the fungi were placed under hygromycin selection. Hygromycin resistant colonies were obtained from both control transformations (co-cultivated with wild-type plants) and the biotech plant co-cultivation. The recovery of antibiotic resistant colonies was higher in the biotech plant DNA treatment, and some of these transformants contained fragments of the plant hpt expression cassette. Despite colony recovery under selective conditions, the acquired foreign DNA sequences were unstable in all but one case, and an intact hpt gene was never identified in any of the colonies.

In another example, HGT between biotech potato plants and a bacterial pathogen Erwinia was examined (Schluter et al., 1995). In this work the authors examined the incidence of HGT using a potato line containing DNA with a functional origin of replication linked to a bacterial selectable marker. Therefore, the transgenic DNA was optimized for transfer to bacteria. (Most biotech plants do not contain origin of replication sequences as part of their transgenes). Transfer events were readily identified because successful transformation led to an easily selected phenotype. Transformations were conducted under a wide range of conditions described by the authors as "idealized natural conditions" to "completely artificial conditions." Transfer was only detected under ideal artificial conditions.

Nielsen et al. (1997) examined the frequency of HGT between biotech plants and the easily transformed soil bacterium Acinetobacter. Again, under laboratory conditions, no transformation was observed using biotech plant DNA, and no transformation was observed when the origin of replication was removed from the recipient DNA.

In summary, no one has demonstrated the movement of an antibiotic resistance gene from a plant into bacteria under natural conditions. Furthermore, plant selectable markers in biotech plants are under eukaryotic regulatory sequences and, if transferred, would not be expected to function in bacteria.

The most important question is not, "can antibiotic genes move from biotech plants to soil microorganisms?" but rather:

• What is the likelihood that transgenic DNA containing an antibiotic resistance marker will be taken up by soil microbes in comparison to other antibiotic resistance conferring genes, and
• Is there sufficient selective advantage to maintain these transgenes over other endogenous genes found in the environment?

12.4   The effect of antibiotic resistance genes in transgenic plants on human health

The nptII marker is the most commonly used antibiotic plant selectable marker. Kanamycin and neomycin are rather toxic antibiotics and are of no clinical importance. The safety of biotech plants carrying this gene has been carefully reviewed and determined to be safe (Nap et al., 1992; Redenbaugh et al., 1994; U.S. FDA, 1998)

In an evaluation of potential transfer of a kanamycin resistance gene from biotech plants into enteric bacteria, Mitten et al. (1996) established one new kanamycin-resistant cell would be produced for every 750 billion already present in the gastro intestinal tract. They point out that 75-86% of human enteric bacteria, such as Streptococcus are already resistant to this antibiotic. Therefore, they estimate the transformation frequency of enteric bacteria would be approximately five orders of magnitude less than the frequency of natural mutation to antibiotic resistance for these organisms.

Some have argued that HGT from plants to bacteria is expected to be higher in biotech plants containing integrated plasmid, as demonstrated experimentally under artificial conditions using marker-rescue transformation (De Vries and Wackernagel, 1998). However, scientists believe the probability of an antibiotic resistant gene in plants moving via horizontal transfer to microorganisms is negligible. Agrobacterium and the particle gun are the most frequently used vectors to transform plants. With particle gun transformation, only the expression cassette (no plasmid) is integrated, while in Agrobacterium, only the expression cassette and T-DNA border sequences are integrated. In both cases, the integrated DNA is isolated from the original plasmid transformation vector and does not have the sequences required for autonomous replication or "shuttling".

Also, as described above, if HGT of antibiotic resistance marker genes were to occur, the consequence of such an event would be negligible inasmuch as these genes are already prevalent in populations of enteric bacteria. Therefore, the presence of antibiotic resistance markers in plant are unlikely to increase the prevalence of antibiotic resistant pathogens, and the ultimate impact of the highly improbable occurrence of HGT of antibiotic resistance marker genes from plants to other organisms is virtually zero.

It is important to reiterate that humans and animals are continuously being bombarded with foreign DNA in the form of bacteria and viruses. The average human comes into daily contact with a million species of bacteria and about 5,000 viruses (Jenkins, 1999).

Return to Table of Contents