We obtained positive PCR amplification using primers specific for p-35S in five of the seven Mexican maize samples tested (Fig. 1). Four criollo samples showed weak albeit clear PCR amplification, whereas the Diconsa sample yielded very strong amplification comparable in intensity to transgenic-positive Bt1 and RR1 controls. The historical negative control (data not shown) and the contemporary sample from Cuzco, Peru, were both invariably negative. Low PCR amplification from landraces was due to low transgenic abundance (that is, a low percentage of kernels in each cob), not to differential efficiency in the reaction, as demonstrated by internal control amplification of the maize-specific alpha zein protein 1 gene (Fig. 1, zp1). During the review period of this manuscript, the Mexican Government (National Institute of Ecology, INE, and National Commission of Biodiversity, Conabio) established an independent research effort. Their results, published through official government press releases, confirm the presence of transgenic DNA in landrace genomes in two Mexican states, including Oaxaca. Samples obtained by the Mexican research initiative from sites located near our collection areas in the Sierra Norte de Oaxaca also confirm the relatively low abundance of transgenic DNA in these remote areas. The governmental research effort analysed individual kernels, making it possible for them to quantify abundances in the range of 3–10%. Because we pooled all kernels in each cob, we cannot make such a quantitative statement, although low PCR amplification signal from criollo samples is compatible with abundances in this percentage range.

 

Figure 1 PCR amplification of DNA from the maize-specific alpha zein protein gene (top panel) and the CMV p-35S promoter (centre and bottom panels).   Full legend   High resolution image and legend (21k)

Using a nested primer system, we were able to amplify the weak bands from all CMV-positive criollo samples (Fig. 1) sufficiently for nucleotide sequencing (GenBank accession numbers AF434747–AF434750), which always showed at least 98% homology with CMV p-35S constructs in commercially used vectors such as pMON273 (GenBank accession number X04879.1) and the K1 sample (accession number AF434746).

Further PCR testing of the same samples showed the presence of the nopaline synthase terminator sequence from Agrobacterium tumefasciens (T-NOS) in two of the six criollo samples (A3 and B2; GenBank accession numbers AF434752 and AF434751, respectively) and the Diconsa sample (K1; accession number AF434753). We detected the B. thuringiensis toxin gene cryIAb in one criollo sample (B3) (data not shown). We confirmed all of the PCR results through repeated testing.

We performed inverse PCR (iPCR) to reveal the various genomic contexts in which the CMV construct was embedded in the Oaxacan criollo maize. This method enabled us to sequence unknown DNA regions flanking the known p-35S sequence in each of the samples. For each sample, iPCR yielded 1–4 DNA fragments differing in size. We isolated these fragments from electrophoresis gels and attempted to sequence them individually, yielding sequences in eight cases (GenBank accession numbers AF434754–AF434761; Fig. 2). Sequences adjacent to the CMV p-35S DNA were diverse, suggesting that the promoter was inserted into the criollo genome at multiple loci. When compared with GenBank (BLAST, February 2001), two sequences were similar to synthetic constructs containing regions of the adh1 gene found in transgenic maize currently on the market, such as Novartis Bt11 (Fig. 2, samples A3 and K1). Notably, these two sequences had high homology with each other. Other sequences represented maize-native genomic DNA, including retrotransposon regions, whereas others showed no significant homology with any GenBank sequence (Fig. 2). The diversity of transgenic DNA constructs present in criollo samples suggests the occurrence of multiple introgression events, probably mediated by pollination. In some of these events, the introgressed DNA appeared to have retained its integrity as an unaltered construct (as with adh1 (ref. 10), whereas in others the transgenic DNA construct seemed to have become re-assorted and introduced into different genomic backgrounds, possibly during transformation or recombination¹³. The apparent predominance of re-assorted sequences obtained in our study might be due to PCR bias for amplification of short fragments, as intact functional constructs are expected to be much longer.

 

Figure 2 Inverse PCR sequences of flanking regions adjacent to the CMV p-35S in landraces (A2, A3 and B3) and in Diconsa seed (K1).   Full legend   High resolution image and legend (34k)

Our results demonstrate that there is a high level of gene flow from industrially produced maize towards populations of progenitor landraces. As our samples originated from remote areas, it is to be expected that more accessible regions will be exposed to higher rates of introgression. Our discovery of a high frequency of transgene insertion into a diversity of genomic contexts indicates that introgression events are relatively common, and that the transgenic DNA constructs are probably maintained in the population from one generation to the next. The diversity of introgressed DNA in landraces is particularly striking given the existence in Mexico of a moratorium on the planting of transgenic maize since 1998. Whether the presence of these transgenes in 2000 is due to loose implementation of this moratorium, or to introgression before 1998 followed by the survival of transgenes in the population, remains to be established. The intentional release of large amounts of commercial transgenic seed into the environment since the mid-1990s represents a unique opportunity to trace the flow of genetic material over biogeographical regions, as well as a major influence on the future genetics of the global food system.

Further study of the impact of the gene flow from commercial hybrids to traditional landraces in the centres of origin and diversity of crop plants needs to be carefully considered with respect to the future of sustainable food production. Long-term studies should establish whether, or for how long, the integrity of the transgenic construct is retained, and whether the relatively low abundance of transgene introgression detected in the 2000 harvest cycle in Oaxaca will increase, decrease, or remain stable over time.

Methods
Extraction and purification of genomic DNA For each cob (sample), all kernels (152–384 kernels per cob) were ground to a fine powder using a steel miller to obtain a pooled sample. Three hundred seeds were also ground from each of the bulk samples, except for the historical negative control, which consisted of 20 seeds. Before use with each sample, millers were thoroughly washed, soaked in 10% sodium perchlorate for 30 min, rinsed and then autoclaved. Genomic DNA was extracted from 100 mg of the powder as described elsewhere¹⁰, with an added purification step using a Geneclean I Kit (Bio 101).

Polymerase chain reaction For protocol I, amplification reactions using 50–100 ng of extracted genomic DNA were carried out in 25 µl containing 1 PCR buffer (Promega), 2.5 mM MgCl₂, 0.2 mM of each dNTP, 0.5 µM of each primer and 0.625 U of Platinum Taq Polymerase (GibcoBRL). We used a water negative control to verify that reactions were free of contamination. Amplifications were performed on a PTC-100 thermal cycler (MJ Research) with the following parameters: initial denaturation at 95 °C for 2 min; 40 cycles each with denaturing at 95 °C for 45 s, annealing for 1 min at 60 °C/56 °C for CMV/NOS, respectively, extension at 72 °C for 1 min; and a final extension for 5 min at 72 °C. For protocol II, where low amplicon yields were obtained, amplification was repeated as in protocol I but with only 20–25 cycles, followed by a re-amplification or nested amplification of a 1:250 dilution of the PCR products in a new reaction mix with partially or wholly nested primers for 10–15 cycles. Primers cm01 (5'-CACTACAAATGCCATCAT TGCGATA-3') and cm02 (5'-CTTATATAGAGGAAGGGTCTTGCGA-3')¹⁰ were used to detect CMV 35S promoters with protocol I. With protocol II we used nested primer pairs mp3 (5'-TCATCCCTTAC GTCAGTGGAGATAT-3') and mp4 (5'-GATAAAGGAAAGGCCATCGTTGAAG-3')¹², and cm02 and mp4. Primers cm01 and cr02 (5'-CTCTCGGCGTAGATTTGGTACA-3')¹⁰ were used to detect the cryIAb synthetic gene. Primers zp01 (5'-TGCTTGCATTGTTCGC TCTCCTAG-3') and zp02 (5'-GTCGCAGT GACATTGTGGCAT-3')¹⁰ were used to amplify the maize-specific alpha zein protein 1 gene (zp1) as an external control for the presence of maize DNA and the efficiency of the reaction.

Inverse PCR Inverse PCR reactions were modified from previously described protocols^{14, 15}. Genomic DNA for iPCR was digested with EcoRV (Promega), which targets a single digestion site internal to the p-35S. Restriction fragments were self-ligated with T4 DNA Ligase (Promega) (14 °C, 18 h), followed by heat inactivation (75 °C, 15 min) and phenol extraction. The purified, circularized products were resuspended in 10 µl 1:10 TE (10 mmol l^-1 Tris, pH 8.5; 1 mmol l^-1 EDTA). PCR amplification was performed using primers designed specifically for the CMV 35S promoter iCMV1 (5'-ACGTCTTCAAA GCAAGTGGA-3'), iCMV2 (5'-AGTGACAGATAGGATCGGGAAT-3') or iCMV3 (5'-GGAGAGGACACGCTGAAATC-3'), iCMV4 (5'-TAGTGGGATTGTGCGTCATC-3'). These primer pairs were designed to amplify outwards of the 35S promoter, downstream and upstream, respectively.

Nucleotide sequencing All nucleotide sequencing was carried out at the University of California at San Francisco Comprehensive Cancer Center, Genome Analysis Core Facility. All sequences mentioned above are available on the NCBI GenBank server (http://www.ncbi.nlm.nih.gov).