Spread of Recombinant DNA by Roots and Pollen of Transgenic Potato Plants, Identified by Highly Specific Biomonitoring Using Natural Transformation of an Acinetobacter sp.

Abstract

Transgenic potato plants with the nptII gene coding for neomycin phosphotransferase (kanamycin resistance) as a selection marker were examined for the spread of recombinant DNA into the environment. We used the recombinant fusion of nptII with the tg4 terminator for a novel biomonitoring technique. This depended on natural transformation of Acinetobacter sp. strain BD413 cells having in their genomes a terminally truncated nptII gene (nptII′; kanamycin sensitivity) followed by the tg4 terminator. Integration of the recombinant fusion DNA by homologous recombination in nptII′ and tg4 restored nptII, leading to kanamycin-resistant transformants. DNA of the transgenic potato was detectable with high sensitivity, while no transformants were obtained with the DNA of other transgenic plants harboring nptII in different genetic contexts. The recombinant DNA was frequently found in rhizosphere extracts of transgenic potato plants from field plots. In a series of field plot and greenhouse experiments we identified two sources of this DNA: spread by roots during plant growth and by pollen during flowering. Both sources also contributed to the spread of the transgene into the rhizospheres of nontransgenic plants in the vicinity. The longest persistence of transforming DNA in field soil was observed with soil from a potato field in 1997 sampled in the following year in April and then stored moist at 4°C in the dark for 4 years prior to extract preparation and transformation. In this study natural transformation is used as a reliable laboratory technique to detect recombinant DNA but is not used for monitoring horizontal gene transfer in the environment.

Molecular techniques have been used now for about 2 decades to introduce new traits such as resistance to diseases, pests, and herbicides into plants of agronomical importance. Many of the transgenic plants contain antibiotic resistance genes, which were used as selection markers during their construction (9). The use of transgenic plants in agriculture leads to the presence of recombinant DNA in the environment on a large scale. The concern has been raised that an unintended transfer of the recombinant genetic material into the soil microbiota may occur and increase, e.g., antibiotic resistance in bacteria, including human pathogens (7, 24, 43).

At the same time, the recombinant and thus specific nucleotide sequences of the DNA of genetically modified organisms enabled the quantitative tracing of the fate of DNA from transgenic organisms in the environment by applying PCR amplification. DNA of high molecular weight has been found to be present in soil sites where free DNA (10) or plant material (10, 29, 41) had been deposited and to persist in nonsterile soils for several months (33, 34, 41, 42). It was suggested that DNA released from eukaryotic and prokaryotic cells constitutes an extracellular gene pool which can be used by naturally competent bacterial cells that take up DNA and integrate it into their genomes (natural transformation) (19, 40). In microcosm experiments transformation was found to occur in nonsterile soils (26, 28, 37). A transfer of recombinant DNA from transgenic plants to microbes in the soil has not been found (10, 25, 29). However, Kay et al. have demonstrated in planta gene transfer from transplastomic tobacco plants to Acinetobacter sp. strain, BD413 when the plants were experimentally coinfected by Acinetobacter and Ralstonia solanacearum (17).

The nptII gene, which is present as selection marker gene in the genomes of several transgenic plants (9), has previously been used to determine the prerequisites for a horizontal transfer of plant DNA into competent bacteria. It was found that recombinant plant DNA can transform competent cells to antibiotic resistance when the recipient cells provide DNA homology for transgene integration by homologous recombination (5, 11, 17). Integration was not detectable in the absence of homology (4, 17, 27).

To assess the level, frequency, and dynamics of DNA spread from plants during growth, we employed transgenic potato plants carrying nptII as selection marker and measured the presence of the recombinant DNA in their environment. For detecting the recombinant DNA we used a biomonitoring assay based on natural transformation of Acinetobacter sp. strain BD413. This species does not discriminate between its own DNA and foreign DNA during the DNA uptake process (3, 18, 30). The assay has previously been successfully applied to detect nptII genes in leaf DNA extracts from several transgenic plants including potato, rape, tobacco, tomato, and sugar beet plants (5). Recently, it was also applied to the detection of recombinant DNA from transgenic sugar beet plants in environmental samples (23). We have now modified the genetic system for biomonitoring in order to make it specific for a given recombinant construct, in our case the DNA of a transgenic potato having an nptII-tg4 terminator fusion (8, 31). By monitoring samples from soil and rhizospheres of field plot- and greenhouse-grown transgenic potato plants we found that DNA is spread during the growth of the plants and not only during the decay of plant litter deposited in soil.

MATERIALS AND METHODS

Construction of plasmids and bacterial strains.

Escherichia coli DH5α (12) and XL10 Gold (Stratagene, La Jolla, Calif.) were the recipients for cloning experiments. Plasmid DNA was purified by alkaline lysis with plasmid purification kits (Qiagen, Hilden, Germany) or by rapid boiling (15). The nptII gene of pBlue-Km1 (located on a 1.8-kb BamHI-HindIII fragment of Tn5 [4]) was fused with the tg4 terminator by replacement of the 0.96-kb NcoI-XbaI fragment (containing a part of nptII and downstream nucleotides) with the 0.90-kb NcoI-SphI fragment from pSR8-36 (31), giving pBlue-Km-tg4 (Fig. 1A; the incompatible SphI and XbaI ends were fused as blunt ends produced by treatment with T4 DNA polymerase [MBI Fermentas, St. Leon-Rot, Germany]). A deletion of 233 nucleotides covering nptII codons for the C-terminal 16 amino acids (resulting in nptII′) and the spacer DNA in front of tg4 was introduced into pBlue-Km-tg4 by inverse PCR of the 5,180-bp plasmid with primers del-0 (AGCGGCGATACCGTAAAGCA), complementary to nucleotides 744 to 725 of the nptII open reading frame and del-3 (AGCCGCTTTCGACGGATTCG), complementary to nucleotides 9 to 28 of the tg4 terminator, and ligation of the product, yielding pMR13 (Fig. 1A). The deletion cassette (1,555 bp) was amplified by PCR with primers npt-Eco1 (ggaaTTCACGCTGCCGCAAGCACTCAG; EcoRI site underlined, noncomplementary nucleotides in lowercase) and npt-Eco2 (ggaattcGTTTACCCGCCAATATATCCTG), treated with EcoRI, and cloned into the EcoRI site of the broad-host-range IncQ vector pKT210 (1), yielding pMR30 (Fig. 1). This plasmid was introduced into Acinetobacter sp. strain BD413 by electroporation (5), and transformants were selected on medium with chloramphenicol (25 μg ml⁻¹). The plasmid pKm1 (5) was linearized with EcoO109I (cutting four times outside of nptII) to prevent cointegrate formation during transformation.

FIG. 1.

(A) Construction of the marker rescue plasmid pMR30. The nptII gene, tg4 terminator, the position and size of the nptII-inactivating deletion, and the nptII′-tg4 fusion are indicated. Small arrows, primer binding sites used for inverse PCR; dotted lines, cloning sites used for deletion formation and subcloning into the broad-host-range plasmid pKT210. (B) Recombinational repair of the nptII gene on pMR30. Shaded areas indicate homologous regions available for recombinational nptII completion. The product is the filled-up marker rescue cassette of pMR30 that confers kanamycin resistance.

For increased strain stability the marker rescue cassette of pMR30 was integrated into the chromosome of Acinetobacter sp. strain BD413 with the alkM gene (32) as the insertion site. The Acinetobacter alkM gene was amplified with primers alkM-f (ccaccggtaccATGAATGCACCTGTACATGTC; noncomplementary nucleotides in lowercase) and alkM-r2 (atcaactcgAGGTCTGATTACTTGCCG) and cloned into the EcoRV site of pBluescript II SK(+) (Stratagene), giving pBlue-alkM1. From pBlue-Km-tg4 (Fig. 1A) a 2.28-kb SalI-NotI fragment containing the nptII-tg4 fusion was excised, blunted, and cloned into the single BsgI site within alkM, giving plasmid pMR30R-cw. Natural transformation of strain BD413 Rif^r (a spontaneously rifampin-resistant BD413 mutant) with EcoRV-linearized pMR30R-cw and selection for kanamycin resistance resulted in strain JV28-Km^r, which carried the nptII-tg4 fusion in the chromosomal alkM gene. The nptII-tg4 cassette was replaced by the nptII′-tg4 cassette through allelic exchange by nonselective natural transformation with pMR13 DNA. At a concentration of 10 μg ml⁻¹ for the XmnI-linearized plasmid DNA in the transformation culture, kanamycin-sensitive transformants arose at a frequency of about 3.4%, as determined by replica plating colonies grown on Luria broth (LB) agar plates onto LB agar plates containing 50 μg of kanamycin ml⁻¹. Five of the transformants were characterized by PCR with primers specific for alkM, nptII, tg4, and the deleted region. The results showed the deletion in the marker rescue cassette in the chromosomes of all of them. One of the strains was termed JV28.

Plants, field plots, and preparation of extracts.

Samples were obtained from potato plants grown in a randomized block design in field plots at Groß Lüsewitz near Rostock, Germany, from 1996 until 2000. Different areas of the field were used for planting out the tubers every year. The parental potato line was Désirée; the transgenic control line (DC1) contained the nptII-tg4 fusion, and the transgenic lines DL4, DL5, DL10, and DL12 additionally contained a T4 lysozyme gene. The soil, field plot design, and sampling procedure have been described (14). In short, “rhizosphere extracts” were prepared from 5 g of freshly harvested root material with adhering soil (often combined from five plants per plot) by aqueous extraction with a stomacher blender in a total volume of 50 ml and purification from most soil particles by low-speed centrifugation (2 min; 500 × g; 20°C). Aliquots of the rhizosphere extracts were stored at −20°C. They were used for transformation immediately after thawing without further purification. Soil extracts were also prepared by the rhizosphere extraction protocol using soil material equivalent to 5 g dry weight and yielding 50 ml of extract (referred to as stomacher soil extracts). Alternatively, for the extraction of total DNA from soil samples the protocol of Widmer et al. (42) including hot sodium dodecyl sulfate (SDS) and ultrasonic treatments was applied with the modifications described previously (23). This method yields 100 μl of extract from 100 mg of soil (dry weight). These extracts are referred to as SDS soil extracts. Leaf DNA was extracted as described previously (5) from the potato plant lines listed above and from the following plant lines: Beta vulgaris subsp. vulgaris, L5 (parental) and L3 (transgenic; beet necrotic yellow vein virus resistance) (21); Lycopersicon esculentum, wild type (parental) and FLAVR SAVR (transgenic; antisense polygalacturonase gene; Calgene, Davis, Calif.); Nicotiana tabacum cv. Samsun, wild type (parental), XynZ-34, and XynZ-46 (transgenic; xylanase production) (13); Brassica napus cv. Drakkar, wild type (parental), B600, and B675 (transgenic; fatty acid production; R. Töpfer, Bundesanstalt für Züchtungsforschung, Groß Lüsewitz, Germany). Purified plant DNA was stored in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) (36) at 4°C.

Experimental designs to distinguish between DNA spread by roots and pollen.

For each of the experiments of Table 2, several nontransgenic (line Désirée) or transgenic potato plants (line DL10) or plants from both lines were grown in the field or greenhouse. Depending on the experiment the plants were grown separately (distance greater than 10 m) or side by side. For experiment 1, rhizosphere extracts were prepared from four plants grown in the field (summer 2002) and six plants grown in the greenhouse (two plants in 2001, four plants in 2002). Experiment 2 was carried out with eight plants (DL10) grown in 2002 in the greenhouse. For experiment 3, six rhizosphere extracts were prepared each from five combined Désirée plants of the 1999 field plots sampled in spring (before flowering). Due to the randomized block design, each plot of nontransgenic plants was neighbored by plots of transgenic lines on at least one side. The transgenic roots could thus invade the Désirée plots. For experiment 4, nine rhizosphere extracts were used; each of them was obtained from five combined DL10 plants from the 1999 field plots. The plants of experiments 5 to 9 were grown in summer 2002 in plastic containers to prevent invasion of their soil by roots of neighboring plants. In the field plot, the containers were installed in the field soil in order to provide the same growth conditions as those for plants growing without plastic containers. Plants for experiment 6 were grown in a separate greenhouse in which no transgenic flowering potato plants were present. Side to side with the four Désirée plants several DL10 plants were grown, but their flowers were removed immediately after appearance and before opening. Roots of the Désirée plants from this experiment were also used for experiment 7. The parental or transgenic genotypes of all individual plants of experiments 1, 2, and 5 to 9 were verified by PCR amplification of the nptII-tg4 fusion and/or by marker rescue transformation of strain JV28 with leaf-extracted DNA.

TABLE 2.

Analysis of the impact of transgenic root invasion and transgenic pollen dispersal on the occurrence of recombinant DNA in rhizosphere and surface soil samples from plants in field plot and greenhouse

Expt	Plant line	Sample type	Time of sampling and experimental setupa	Presence of transgenic DNA from:		No. of transformants/109 recipient cellsb in:
				Pollen	Roots	Field plot	Greenhouse
1	Désirée	Rhizosphere	Before flowering of and distant from transgenic plantsc	−	−	≤1.0 (4)	≤0.8 (6)
2	DL10 (transgenic)	Rhizosphere	Before flowering	−	+	NDd	59.6 ± 45.7 (8)
3	Désirée	Rhizosphere	Before flowering; grown close to transgenic plants	−	+	22.0 ± 19.6 (6)	ND
4	DL10 (transgenic)	Rhizosphere	During flowering	+	+	75.6 ± 50.7 (9)	ND
5	Désirée	Rhizosphere	During flowering of neighboring transgenic plants	+	−	9.1 ± 6.6 (8)	14.4 ± 14.4 (4)
6	Désirée	Rhizosphere	Flowers of neighboring transgenic plants removed	−	−	ND	≤1.1 (4)
7	Désirée	Rhizosphere	Roots powdered with transgenic pollene	+	−	ND	35.8 ± 24.8 (4)
8	Désirée	Surface soil	Soil powdered with transgenic pollen; SDS soil extracts	+	−	26.8 ± 19.2 (3)	59.8 ± 10.9 (2)
9	Désirée	Surface soil	Samples of expt 8; stomacher soil extracts	+	−	22.4 ± 17.1 (3)	11.8 ± 5.7 (2)

^aField plot plants of experiments 1 to 4 were grown directly in the field soil, while those of experiments 5 to 9 were grown in field soil in plastic containers (∼25 cm in diameter) to prevent invasion of roots from neighboring plants.

^bThe numbers of experiments are given in parentheses. Details on the sampled plants and the experimental design are provided in Materials and Methods. Transformation frequencies are given with standard deviations (n > 2) or deviations from the means (n = 2). When no transformants were found, the limit of detection obtained from four to six experiments is given.

^cSamples were taken at several time points from juvenile plants until the first buds of transgenic flowers were visible.

^dND, not determined.

^ePollen was introduced by gentle flicking of a fresh flower from a transgenic plant ∼10 cm above the freshly harvested roots.

Marker rescue transformation and PCR control of transformants.

Concentrated competent-cell suspensions of Acinetobacter sp. strain BD413(pMR30) or Acinetobacter sp. strain JV28 were prepared as described previously (5) and stored at −80°C until use. The transformation cultures (20 ml each; recipient titer, 2.5 × 10⁸ ml⁻¹) were prepared by dilution of 0.5 ml of freshly thawed competent-cell suspension into LB (36). Extract volumes corresponding to 100 mg of fresh root material (1.0 ml of rhizosphere extract) or 200 mg of dry soil (2.0 ml of stomacher soil extract or 0.2 ml of SDS soil extract) were added to a transformation culture, followed by aeration for 90 min at 30°C. In transformation assays with purified plasmid or plant DNA the transformation culture volumes ranged from 0.2 to 20 ml and various DNA concentrations were used. Transformants were selected on LB agar containing kanamycin (10 μg ml⁻¹) and, when strain JV28 and nonsterile extracts were used, rifampin (100 μg ml⁻¹) and cycloheximide (75 μg ml⁻¹). Typically, the cells of a 20-ml assay mixture were concentrated by centrifugation and spread on four selective plates. The recipient titers were determined on LB agar. Transformants obtained from nonsterile soil or rhizosphere extracts were checked by PCR in two ways. First, the identity of the transformants was verified by randomly amplified polymorphic DNA PCR (RAPD-PCR) using the primer F1G (CGGATGGGTGATTTTTAGGA) and parameters as described previously, including four cycles with 40°C annealing temperature (38). All transformants yielded a pattern of six separate bands, which was undistinguishable from that produced by the recipient strain JV28. Other strains including E. coli K-12 and Pseudomonas stutzeri gave no detectable PCR products. Second, by using a primer specific for the correctly filled up deletion (R1b; AGCGCATCGCCTTCTATCGC) and a reverse primer binding within alkM (alkM-r1; AGCTATGCTCTGGCATGG) it was verified that the kanamycin-resistant clones indeed were transformants of JV28 with a correctly filled up marker rescue cassette. Transformants obtained with strains containing pMR30 were verified by PCR with primer R1b and primer L5 (CAACCATATCGGTGCGCTCT; binding within the strA gene of pMR30).

Magnetic capture hybridization.

For the detection of recombinant DNA by PCR, the target DNA in rhizosphere extracts was isolated by magnetic capture hybridization. The method was adapted from that of Jacobsen (16) with the following specifications. The target sequence was the T4 lysozyme gene present in the transgenic potato lines except for DC1. A magnetic hybridization probe was prepared by coupling a 5′-biotinylated 98-mer oligonucleotide binding to the center of the T4 lysozyme gene (nucleotides 373 to 470 of the sense strand of the T4 lysozyme open reading frame) to paramagnetic M-280 streptavidin beads (Dynal, Skøyen, Norway). For capturing recombinant DNA, 200 μl of rhizosphere extract was boiled for 10 min and centrifuged for 10 min at 16,000 × g and 4°C and 50 μl of the supernatant was added to 330 μl of hybridization solution and 20 μl of a 10-mg ml⁻¹ suspension of the probe-carrying paramagnetic particles. The tubes were incubated in a rotating hybridization oven for 4 h at 62°C. After magnetic separation and the washing of the beads, the beads were resuspended in 25 μl of sterile water. The presence of the target sequence was determined by PCR with primers complementary to recombinant fusion sites at both ends of the T4 lysozyme gene, namely, the fusion of the signal peptide and T4 lysozyme coding sequence (CCGGGTTGGCGTCCATGAAT) and of the 35S terminator and vector sequences (CATGCCTGCAGGTCACTGGA). The PCR mixtures contained 5 μl of the sample in a final volume of 30 μl. Amplification was carried out with PCR Ampliwax Gems (Perkin-Elmer, Weiterstadt, Germany) and 0.6 U of AmpliTaq (Perkin-Elmer) as recommended by the supplier. PCR conditions were 5 min at 94°C; 40 cycles of 94°C for 30 s, 65°C for 45 s, and 72°C for 90 s; and finally 12 min at 72°C in a DNA Thermal Cycler 480 (Perkin-Elmer). The detection limit of the magnetic capture step plus PCR was determined with pSR8-36 (31) added to the rhizosphere extract prior to denaturation and was between 10 and 50 molecules.

RESULTS

The transgene-specific marker rescue system.

The transgenic potato lines used in this study were constructed by Agrobacterium tumefaciens-mediated transformation of potato leaf disks with the plasmid pSR8-36 (31). The transferred DNA (T-DNA) of this plasmid contains the marker gene nptII, which is expressed from a nos promoter and which is followed by the eukaryotic tg4 terminator. We have exploited the nptII-tg4 region of the T-DNA for a specific biomonitoring of DNA from the transgenic potato plants. For this, inverse PCR was used to truncate the nptII gene by 51 bp at the downstream end and by the deletion of a further 182 bp directly fused to the tg4 terminator (Fig. 1A; see Materials and Methods). The deletion cassette was cloned into a broad-host-range plasmid, giving plasmid pMR30 (Fig. 1A). The truncation of nptII caused kanamycin sensitivity.

When cells of Acinetobacter sp. strain BD413 containing pMR30 take up DNA with the full-length nptII fused to tg4 (i.e., the construct present in chromosomal DNA from the transgenic potato plants or in pSR8-36), this DNA can lead to the fill up of the deletion by recombination events in the two homologous regions upstream and downstream of the deletion (i.e., in the nptII′ and tg4 sequences; Fig. 1B), which were 775 and 456 bp, respectively. This results in a restoration of nptII, measurable by the formation of kanamycin-resistant transformants.

Determination of sensitivity and transgene specificity.

Cells of Acinetobacter sp. strain BD413(pMR30) were transformed with DNA of the plasmid pSR8-36 containing the nptII-tg4 fusion. As shown in Fig. 2, the transformation frequencies obtained with pSR8-36 DNA increased linearly up to 0.1 μg ml⁻¹ with an ascent of 1.0 (single-hit curve). In the linear range, one transformant was obtained per 1.5 × 10⁴ nptII-tg4 fusions present in the transformation culture. At higher DNA concentrations the increase of the transformation frequency leveled off. At the highest DNA concentration tested (100 μg ml⁻¹), about 3% of the recipient cells were transformed. Similarly high transformation frequencies were previously obtained by Palmen et al. (30) using Acinetobacter sp. strain BD413 and a plasmid carrying an nptII gene embedded within chromosomal DNA. Linearization of pSR8-36 by ClaI treatment prior to transformation did not change the transformation frequency (e.g., transformation frequencies of 1.04 × 10⁻² and 1.05 × 10⁻² were obtained at a DNA concentration of 1 μg ml⁻¹ with circular and linearized pSR8-36 DNA, respectively). This suggests that transformation by linear DNA (e.g., as recovered from plants) is as efficient as that by circular DNA.

FIG. 2.

Transformation of Acinetobacter sp. strain BD413(pMR30) by genomic DNA from the transgenic potato line DC1 (•), plasmid DNA of pSR8-36 containing the nptII-tg4 fusion of DC1 (▴), and pKm1 containing the nptII gene without the tg4 terminator (▾). The numbers of nptII genes per 1 μg of DNA are 1.36 × 10⁵ (DC1), 1.02 × 10¹¹ (pSR8-36), and 1.58 × 10¹¹ (pKm1). Data are from three determinations; error bars, standard deviations.

DNA of the transgenic potato plants contains the nptII-tg4 fusion at about 10⁶-fold-lower concentration per mole of nucleotide than pSR8-36 DNA due to the large excess of potato DNA. In fact, the transformation frequencies with DNA from the transgenic potato (DC1) were nearly 10⁶-fold lower than those with corresponding concentrations of pSR8-36 DNA (Fig. 2). The limit of detection of transgenic potato DNA was 20 ng ml⁻¹ (200 ng total in 10 ml of transformation culture), corresponding to 2.7 × 10⁴ nptII-tg4 fusion molecules, which yielded an average of 1.3 transformants. As observed with pSR8-36 DNA, at concentrations above 0.1 μg ml⁻¹ the increase of the transformation frequency leveled off, indicating the beginning of saturation of the system. The results suggest that the transformation frequency depends primarily on the number of target sequences in the assay, as was observed previously (5).

Compared to transformation by pSR8-36, transformation by pKm1 DNA was about 6,000-fold less efficient at low DNA concentration and at least 1,000-fold less efficient at high DNA concentration (Fig. 2). This is explained by the lack of the tg4 terminator next to nptII in pKm1, which is required for the efficient repair of nptII′ by homologous recombination. The strong preference of the recipient cells to integrate the nptII-tg4 fusion DNA indicated the high transgene specificity of the pMR30 marker rescue system.

Specific detection of transgenic potato DNA.

We examined whether the novel system could be used to discriminate between plant DNA with the nptII-tg4 fusion (i.e., the transgenic potato DNA) and the DNA of other transgenic plants having nptII but different downstream nucleotide sequences. As shown in Table 1, the DNA from transgenic potato plants (DC1, DL4, and DL5) consistently gave Km^r transformants of Acinetobacter cells with pMR30 (average transformation frequency, 8 × 10⁻⁹). As expected, DNA of parental plants without nptII did not yield any transformants. DNA from five transgenic plants containing nptII, including tomato, sugar beet, tobacco, and rape plants, and five nontransgenic parental lines also did not produce Km^r transformants (Table 1). This result is consistent with the data obtained with pKm1 DNA in Fig. 2 showing that the absence of the second homologous recombination site decreases the transformation frequency strongly. In the experiments of Table 1 the transformation frequency fell below the detection limit (2 × 10⁻¹⁰). It is concluded that the novel marker rescue system is as specific for its cognate transgenic fusion with plant DNA as with plasmid DNA.

TABLE 1.

Specific detection of the potato transgene by natural transformation of Acinetobacter sp. strain BD413(pMR30) by using leaf-extracted DNA from various nontransgenic and transgenic plants (with nptII)

Plant line	No. of Kmr transformantsa	n
Solanum tuberosum
Désirée (parental)	0	8
DC1 (transgenic)	45.8 ± 10.3	4
DL4 (transgenic)	37.7 ± 16.9	3
DL5 (transgenic)	40.9 ± 16.2	7
Beta vulgaris subsp. vulgaris
L5 (parental)	0	2
L3 (transgenic)	0	3
Lycopersicon esculentum
Parental	0	1
FLAVR SAVR (transgenic)	0	3
Nicotiana tabacum
Samsun (parental)	0	4
XynZ-34 (transgenic)	0	2
XynZ-46 (transgenic)	0	1
Brassica napus
Drakkar (parental)	0	4
B600 (transgenic)	0	1

^aDetermined with 3 μg of purified leaf DNA per 20-ml transformation culture; the data are means of n experiments ± standard deviations.

Chromosomal integration of the marker rescue cassette in Acinetobacter.

To achieve high genetic stability for the monitoring strain without the need for selective pressure to maintain pMR30, we inserted the marker rescue cassette from pMR30 into the alkM gene of the chromosome of Acinetobacter sp. strain BD413 Rif^r, yielding strain JV28 (see Materials and Methods). The alkM gene is involved in alkane degradation, which is of no relevance for growth on LB media. The transformation frequency with 0.1 μg of ClaI-linearized pSR8-36 DNA ml⁻¹ with strain BD413(pMR30) Rif^r ([3.7 ± 1.3] × 10⁻³) was about equal to that with strain JV28 ([3.1 ± 0.8] × 10⁻³). This indicated that recombinant DNA monitoring by JV28 was as sensitive as that by strain BD413(pMR30). The results suggest that (i) the recombination frequency of the recipient cells was not influenced by the chromosomal or extrachromosomal location of the target sequence and (ii) multiple marker rescue cassettes provided by the low-copy-number plasmid (copy number about 3) did not increase the transformation frequency in comparison to that for a single cassette per chromosome. For the following experiments JV28 was used as the recipient strain.

Detection of recombinant DNA in the plant rhizosphere.

The transgenic potato lines DL4 and DL5, expressing the T4 lysozyme gene, the transgenic control line DC1, and the parental line Désirée were studied in field release experiments for their performance from 1996 to 1998 (14). In 1999 and 2000 the studies were continued with two further T4 lysozyme gene-expressing potato lines (DL10 and DL12) and involved planting the tubers in a different area of the field every year. The field design was a randomized block design with six to nine plots per line, each containing 15 plants, in which each plot of the parental line was neighbored by plots of transgenic lines on at least one side. During the field releases rhizosphere extracts were sampled for bacterial community analyses (14, 20). We assayed these nonsterile extracts for their content of recombinant DNA by the marker rescue assay. Typically, 20 ml of transformation culture (JV28) was mixed with 1.0 ml of rhizosphere extract. The Km^r transformants were identified as JV28 by RAPD fingerprinting and PCR amplification of the filled-up nptII-tg4 region (see below). With rhizosphere extracts from juvenile plants, transformants were obtained from six out of nine plots with transgenic plants (Fig. 3). At the stages of flowering and senescence transformants were obtained with all of the extracts (18 of 18). The transformation frequencies increased from juvenile to flowering plants and remained at the high level until senescence. These results indicated the frequent presence of free transforming recombinant DNA in the rhizosphere extracts at any growth stage of the plants. The DNA may have been released from the roots into the rhizosphere or set free by cell disruption during extract preparation (see Discussion).

FIG. 3.

Transformation of the biomonitoring strain Acinetobacter sp. strain JV28 with rhizosphere extracts from a field release experiment with potato plants. The fractions of samples yielding transformants and the average transformation frequencies of the positive samples for the transgenic potato plants (DC1, DL10, and DL12) and parental plants (Désirée) are given separately. Tubers were planted on 19 May 1999. Sampling dates were 30 June (juvenile), 11 August (flowering), and 20 September (senescent) 1999.

Surprisingly, transformants were also obtained with 13 out of 18 rhizosphere extracts obtained from parental plant plots at early and late stages of development (Fig. 3). The average transformation frequencies were, however, lower than those for the transgenic plants but also increased at the time of flowering (Fig. 3). These data clearly indicated the presence of nptII-tg4 fusion DNA in rhizospheres of parental plants. A corresponding result was obtained by a different DNA-monitoring approach in which the presence of the recombinant T4 lysozyme gene in potato rhizosphere extracts was determined by PCR amplification of DNA recovered by magnetic capture hybridization (16). A survey of the field plots from 1996 and 1997 by this method gave four positive scores out of 116 extracts from plots of plant lines without the T4 lysozyme gene (Désirée and DC1). The PCR signals were obtained with samples from flowering plants (1 out of 44 samples [2.3%]) and senescent plants (3 out of 38 samples [7.9%]), but not from juvenile plants (34 samples). From the transgenic plant lines with the T4 lysozyme gene (DL4 and DL5) a much higher proportion of the rhizosphere extracts gave PCR products (98 out of 116 samples).

We have carefully addressed the possibility of false-positive transformed JV28 clones by PCR analysis of 190 clones from transformations with rhizosphere extracts from the field plots (143 transformants) and the greenhouse (32 transformants) and with SDS soil extracts from stored soil samples (15 transformants; see “Persistence of recombinant DNA from pollen in soil” below). Of these, 95 transformants were checked with primers binding on both sides of the deletion and 135 were checked with primers binding on one side and within the deletion (primers specified in Materials and Methods). All 190 transformants yielded the expected products. In addition, 20 transformants composed from these three groups were characterized by RAPD-PCR and were not distinguishable from the parental strain Acinetobacter sp. strain JV28 (see Materials and Methods).

DNA spread by roots and pollen.

Since the rhizosphere samples from transgenic and parental plants studied in Fig. 3 came from a randomized block design field experiment in which plants grew in close proximity (leaves touching), it was suspected that the recombinant DNA detected in nontransgenic samples was spread from the plots of transgenic plants. Two possible routes of DNA spread into rhizosphere extracts from other plants were considered: (i) roots of transgenic plants may have grown into the areas of the parental plants, causing the presence of recombinant DNA there, and (ii) pollen of nearby flowering transgenic plants that was deposited on the soil surface may have been transported into the rhizosphere by, e.g., rain, or introduced into the rhizosphere during sampling of the roots.

To identify and roughly quantify the contribution of both possible routes to DNA spread in the field, we conducted a series of experiments, the results of which are summarized in Table 2. First, rhizosphere extracts were prepared from parental plants (Désirée) grown separately (distance larger than 10 m) from transgenic plants either in a greenhouse or in field plots so that genetic cross contamination by roots and pollen was excluded. No transformants were obtained in these cases (Table 2, experiment 1). The absence of any transformants in these assays indicated that nptII genes from other sources such as soil microorganisms did not contribute to transformant formation with JV28 recipient bacteria. Further, the absence of transformants in these control experiments suggested that the recombinant DNA found in the plots of the parental plants before flowering (Fig. 3, juvenile parental plants) derived from invading roots of neighboring transgenic plants. To assess whether DNA is actually present in rhizosphere extracts from growing plants, rhizosphere extracts were prepared from juvenile transgenic plants (line DL10) grown in a greenhouse and sampled before flowering. Transformants arose at high frequency from all of the extracts (Table 2, experiment 2), indicating that transgenic roots can contribute to the transforming activity of rhizosphere extracts. When nontransgenic plants were grown in a field plot side by side with transgenic plants and not protected from the invasion by transgenic roots, transformants arose from rhizosphere extracts from nontransgenic plants before the flowering of the transgenic plants, although at low frequency (Table 2, experiment 3). This is in accordance with the data of Fig. 3 and supports the conclusion that roots of transgenic plants were a source of recombinant DNA in the block design field plots of parental plants. The combined effects of transgenic roots and transgenic pollen on the transformation activity in rhizosphere extracts were seen with extracts from flowering transgenic plants, which gave higher transformation frequencies than those observed before flowering (Table 2, experiment 4 versus 2). This is also in accord with the increasing transformation frequency obtained in Fig. 3 (transgenic plants, juvenile versus flowering and senescent).

To identify the effect of pollen production alone, the invasion of the root area of parental plants by roots of transgenic plants was prevented by the growth of both types of plants in plastic containers in the field plots. Containers were filled with the field soil and installed in the soil surface of the plot to provide the same growth conditions as those for the other plants in the plot. Greenhouse plants were grown in plastic containers without installation in soil. After the transgenic plants started flowering, transformants were obtained with rhizosphere extracts from the neighboring parental plants in both cases, i.e., when the growth occurred side by side in the field plot and in the greenhouse (Table 2, experiment 5). No transformants appeared at the time of flowering when all flowers from the transgenic plants were removed before pollen spread (greenhouse experiment; Table 2, experiment 6). As a control, transformants were always obtained when pollen from flowers of transgenic plants was directly powdered over parental root material (experiment 7). These results imply that pollen was spread from the transgenic plants onto the soil and contributed to the presence of recombinant DNA in the rhizosphere extracts of transgenic and parental plants.

To determine the efficiency of the rhizosphere extraction procedure for the recovery of DNA from pollen, we introduced tiny amounts of pollen from transgenic plants into surface soil samples collected from parental plants, divided the samples into two fractions, and extracted these either by the rhizosphere extraction protocol (stomacher soil extracts) or by a protocol for total DNA extraction using hot SDS and ultrasonication (42) (SDS soil extracts). Transformants were obtained with each of the extracts (Table 2, experiments 8 and 9). With the stomacher soil extracts (experiment 9) the average transformation frequency ranged from 20 to 80% of that obtained with the SDS soil extracts, indicating only partial release and/or recovery of DNA by the stomacher method.

Persistence of recombinant DNA from pollen in soil.

The fact that the transformation frequencies in the biomonitoring assay increased three- to fourfold at the time of flowering but did not decrease for several weeks after pollen production (Fig. 3) suggested that potato pollen or the DNA from it had persisted during this period. To test for DNA persistence, we assayed surface soil samples that had been taken in 1998 at a distance of 2 m from the field plots with transgenic and parental potato plants at the stage of senescence and since then had been stored moist in closed plastic bags at 4°C in the dark. SDS soil extracts were prepared from two such samples and gave transformation frequencies of 8 × 10⁻⁹ and 28 × 10⁻⁹ with 200 μl of extract added per 20-ml transformation culture. Similarly, transformants were obtained with 6 out of 10 SDS soil extracts prepared from surface soil samples devoid of discernible plant tissue litter, which were collected from the field plot in 1998 at a distance of 10 to 30 cm from transgenic potato plants during flowering and since then had been stored at 4°C in the dark. The transformation frequencies ranged from 6 × 10⁻⁹ to 25 × 10⁻⁹. These results show that DNA presumably spread by pollen can persist for at least 4 years in stored soil. We also examined soil samples which were taken in April 1998 from field plots in which transgenic potatoes had been grown in 1997. From these, 6 out of 10 gave transformation frequencies of 3 × 10⁻⁹ to 9 × 10⁻⁹. It is concluded that plant DNA either enclosed in plant material or as free DNA had persisted for 8 months during the winter period in the field site and had retained its transforming potential also during the following 4 years of storage.

DISCUSSION

The recombinant DNA of genetically modified organisms can be specifically detected by PCR amplification using primers targeted to a recombinant fusion, i.e., binding to two sequences that are normally not contiguous. We have applied this principle in the novel biomonitoring of recombinant DNA from transgenic plants. The specificity of the marker rescue transformation with the nptII-tg4 fusion is based on the requirement of sequence identity for homologous recombination during transformation and on the presence of the two normally not contiguous sequences on the sides of the selective marker. The nptII gene itself lends most of its sequence as one recombination side, the fill up of the terminal nptII deletion constitutes the selective marker, and the tg4 terminator is the second recombination side. The precision of homologous recombination in the fill up of the terminal nptII deletion was underlined by the molecular analysis of 190 transformants from various experiments by PCR which revealed not a single case of irregular recombination. While 14 transformation tests with potato DNA containing the nptII-tg4 fusion always yielded about 40 transformants, no false positives were scored in 10 tests with DNA from five other transgenic plants having nptII without tg4 (Table 2). These DNA samples had previously given hundreds of transformants with a transformation-based bioassay (Acinetobacter sp. strain BD413 with pMR7) measuring the presence of the nptII gene alone (5). False positives were also not obtained with total DNA recovered from soil, indicating that nptII genes that might have been present in the soil microbiota were discriminated by the specificity of our marker rescue system. The rare false positives were found only when the transformation assay was swamped with nptII-containing plasmid DNA. These transformants probably arose from illegitimate recombination events, which are strongly facilitated by nearby homologous recombination (6). However, in the absence of any homology, the integration of nptII into transformable bacteria is extremely rare and remained undetectable (<10⁻¹³ per nptII [4]). Note that the principle of our assay also works with combinations of other antibiotic resistance genes and downstream nucleotide sequences (T. Herzfeld, J. de Vries, and W. Wackernagel, unpublished data).

The effectiveness of the biomonitoring of recombinant potato DNA by Acinetobacter sp. strain JV28 carrying a single marker rescue cassette in the chromosome was equivalent to that by a strain having the cassette on plasmid pMR30 and was similar to that by the previously described strain Acinetobacter sp. strain BD413 with pMR7 (5). Apparently, the presence of the marker rescue cassette in the chromosome or on the plasmid does not affect the efficiency of transformation; this may be explained by the rather low copy number of the plasmid. The sensitivity of our biomonitoring approach, giving about one transformant per 10⁴ target molecules, irrespective of the presence of a large excess of, e.g., plant DNA, is not much less than that of routine PCR applications, which generally also require >10³ target molecules (e.g., 10 ng of template DNA are usually required for the amplification of single-copy genes from eukaryotic genomes [PCR applications manual, 2nd ed., Roche Diagnostics GmbH, Mannheim, Germany, 1999]). Moreover, the DNA which is introduced into the assay does not have to be highly purified but can be present in aqueous extracts from soil without further removal of PCR-inhibiting substances such as humic acids (39).

Do transgenic plants spread recombinant DNA into the environment? Recombinant DNA has previously been detected in soil samples containing litter from transgenic tobacco (29, 41), potato (41), and sugar beet (10) plants. Since the methods used in this study for the extraction of total DNA included harsh steps such as ultrasonic and hot-SDS treatment, it is not clear whether the plant DNA was extracellular or released from plant cells during the extraction procedure. The specific detection of extracellular DNA in soil can be achieved by a mild aqueous elution technique not leading to cell disruption (2). When this method was recently applied in parallel to the procedure for total DNA extraction, it was demonstrated that a fraction of soil samples from field release experiments with transgenic sugar beets contained free recombinant DNA (23). Here we found that recombinant nptII DNA is present in aqueous extracts from rhizospheres of transgenic potato plants and from soil samples taken from field plots with transgenic plants or from plots without transgenic plants close by. These extracts were prepared without ultrasonic or hot-SDS treatment. However, the possibility that DNA was set free from tissue cells or pollen present in the soil material by mechanical forces during aqueous-extract preparation was not excluded. In fact, the presence of transforming DNA in extracts prepared from samples into which transgenic pollen was introduced directly from the potato flowers argues for a pollen breaking effect. Importantly, DNA was detected by our test from juvenile to senescent growth stages and not only during the decay of plant litter, as was found in previous studies. The data suggest that roots can spread DNA in the soil during plant growth, either as free molecules or within plant tissue material. This may occur by in situ destruction of rhizodermis or calyptra cells or the deposition of dead root tissue. The recovered DNA was of high molecular weight and was able to transform competent recipient cells. With respect to horizontal gene transfer, the potential for natural transformation is a more relevant measure than the potential for amplification by PCR, because it directly demonstrates that the material is still biologically functional. Detection by PCR and biological function may not always coincide (34).

The results of Table 2 suggest that, besides plant root material, pollen was a source of recombinant DNA. Pollen is probably particularly important in terms of long-distance gene spread, because as part of the reproductive system the function of pollen is gene movement. The distances covered by pollen have been determined with transgenic plants carrying a nuclear transgene by assaying the formation of transgenic seeds by nontransgenic bait plants. While potato pollen is transported by wind less than 10 m (22), the pollen of the sugar beet, which is also a wind pollinator, is transported over distances of at least 200 m (35). The fact that soil containing potato pollen and perhaps small tissue fragments retained much of its transforming activity over the winter period and during a subsequent storage period of four years suggests that DNA in pollen or released from it may be particularly stable. It was recently inferred that pollen of transgenic sugar beet plants was an important source for recombinant-DNA spread, because soil samples taken at distances up to 50 m from the plants gave positive results in PCR or transformation tests only when pollination had occurred (23). In those studies an experimental stop of the DNA dispersal through pollen by removal of flowers was not provided, as was conducted in this study.