Robert Wager, Peter LaFayette and Wayne Parrott

This recently published paper on transgenic corn attempts to prove that transgenes are being incorporated into the local landraces, most likely via cross pollination. It is our assessment that the molecular data presented in the current paper fall far short of proof, and that the results have been over-interpreted. It is clear that all the results of inverse PCR are artifactual, and, based on the data provided, we cannot support the author's assertion that they have found evidence of transgenes, multiple integration events, or that recombination possibly may have caused transgenes to become "re-assorted and introduced into different genomic backgrounds". Furthermore, the claims of introgression made in this paper cannot be supported based on the data provided.

We wish to make 8 points as follows:

1. The authors claim to have amplified Cry1Ab using one primer internal to the coding sequence, and one external to it. Overall this is a sound strategy. Unfortunately the results fall into the category of "data not shown." There is no evidence in the paper that this particular band was ever sequenced to verify its identity.

2. The description of the upstream and downstream primer pairs is reversed in the materials and methods. Based on the orientations of the primers on the GenBank sequences, it is obvious that the nomenclature is reversed in Figure 2. What the authors call upstream sequences are really downstream, and vice versa.

3. Inverse PCR was used to clone the flanking sequences, using EcoRV, which occurs once in the 35S promoter. A consequence of iPCR is that 35S sequences should flank any amplified fragment. Furthermore, the primer sites should remain on the amplified fragments. Finally, the EcoRV site should be reconstituted when the DNA ends come together to constitute a circle. However, not a single one of the amplified fragments in Figure 2 has an EcoRV site. There should be 13 bp of 35S sequence, including the EcoRV site, immediately adjacent to the iCMV1 primer. This homology is absent in all the fragments listed as downstream in Figure 2 (A2#2 has 7bp and B3 10 bp homology, respectively). Some fragments (A3, A2#2, B3), are missing a primer site, while the K1 upstream sequence is missing both the primer sequences. Unexplainably, the primer site present in the B3 fragment is oriented outward rather than inward.

4. The BLAST results reported in Figure 2 have been further compromised by including the primer sequences. Omitting the primer sequences prior to the BLAST search provides different results, particularly for the first A2 and the second A3 fragment. Results are given at the end of this writeup.

5. The authors also claim that the adh1 sequences found in the K1 and A3 fragments are "similar to synthetic constructs containing regions of the adh1 gene found currently on the market, such as Novartis Bt11." This interpretation cannot be correct because the homologous region they found is not the part used in synthetic constructs. The part found by the authors is about 40 kb away from the adh1 introns used in the synthetic constructs. [The adh1 introns used to enhance gene expression are described in D. Mascarenhas, I.J. Mettler, D.A. Pierce, and H.W. Lowe. 1990. Intron-mediated enhancement of heterologous gene expression in maize. Plant Mol Biol 15:913-920.]

To summarise the results for Figure 2, fragments K1 (downstream), A3, and A2 appear to be a retrotransposon found near alpha zein, bz, and adh1. The second A3 fragment is more closely related to rice than maize. As reported by the authors, B3 does appear to be a retrotransponson, and K1 and B3 (the two upstream sequences) have no significant match to anything.

None of the sequences cloned have any of the expected vector sequences, the necessary primer sequences, or the EcoRV sites, from which we can only conclude these bands represent artifacts of amplification. The most likely explanation is a short stretch of similarity between the primers and the amplified regions. There is a significant match of the primers used for the inverse PCR shown in Figure 2 to known regions of the maize genome. Performing a BLAST search with the option "short nearly exact matches" of the inverse PCR primers against the maize database yields the following:

As these identities are at the 3' end of the primers, mis-priming would be expected, leading to amplification of sequences other than the intended sequences.

One main reason for the problems is that the iCMV2 primer is designed incorrectly. Based on the text provided in the article's introduction, it appears that they used the sequence of pMON273 to design their primers. The result is that only the first 12 bases from the 5' end or the primer are homologous to the 35S promoter used in other vectors-- the intact iCMV2 primer sequence (5'-agtgacagataggatcgggaat-3') is only found in pMON273 (a petunia vector, GenBank X04879.1). Performing a BLAST search (short, nearly exact matches option) with this primer yields:

gi|17352178|gb|AF434757.1|AF434757

Zea mays landrace sample

44

0.001

gi|17352177|gb|AF434756.1|AF434756

Zea mays landrace sample

44

0.001

gi|17352175|gb|AF434754.1|AF434754

Zea mays variety sample

44

0.001

gi|58071|emb|X04879.1|ARCAMVPR

CaMV promoter in pMON273 vec.

44

0.001

gi|11079401|gb|AC007335.6|AC007335

Homo sapiens chromosome ...

32

4.7

gi|15624875|gb|AC024607.3|

Mus musculus chromosome 5 clone

30

19

Note that the first 3 hits are to the sequences deposited by Quist and Chapela.

Furthermore, a BLAST search of the 354 bp pMON273 sequence against Genbank yields many hits to CaMV promoters, but in each instance starting at bp 13 of pMON273 (iCMV2 is bp 24-3 of pMON273).

The conclusion is that is iCMV2 was doomed from the start and would only yield non-artifactual results with plants transformed with pMON273. This also explains why there is no CaMV sequence adjacent to the iCMV2 primer in the data shown in Figure 2.

6. The entire iPCR work appears to have been done with the assumption that the iCMV2 primer site would be joined to plant DNA- however, the authors are ignoring that there almost certainly are additional vector sequences upstream of this primer. That none were present should have served as a signal that something was wrong.

7. The results in Figure 1 do suggest the presence of the 35S promoter sequences. The sequences amplified match the expected sequence, but in the absence of additional data to support genomic integration, these can only be considered as preliminary data. Likewise, the authors claim to have amplified the NOS terminator from 3 samples. The sequences provided are a perfect match to the NOS terminator. However, the primer sequences are not provided in the paper, and again, in the absence of additional data to support genomic integration, these also remain as preliminary data. The possibility of contamination cannot be ruled out. Such is the nature of PCR data-- on their own, they fall short of proof.

8. It is also necessary to examine the claims of introgression made in this paper. It helps to begin with a definition of introgression:

"Initial hybridization is followed by recrossing with the parental species in such a way that certain features of one species become transferred to the other species without impairment of taxonomic integrity." R.W. Allard 1960. Principles of Plant Breeding. John Wiley, New York.

"Method involves the transfer of a small amount of germplasm from one species to another. It follows hybridization of the two species and backcrossing of the recombination products over several generations to one parental species. This may result in the transfer of certain features of one species to another without impairing the latter's taxonomic integrity". F.N. Briggs and P.S. Knowles, 1967. Introduction to Plant Breeding. Reinhold Publishing.

Assuming that one was to give the authors the benefit of the doubt and agree that the local landraces have crossed with transgenic corns, the authors present zero evidence that the transgenes have survived and been maintained over repeated cycles of crossing. As is, absolutely nothing in this paper can support the authors' speculations that "the transgenic DNA constructs are probably maintained in the population from one generation to the next."

Whether introgression will eventually occur in the absence of a deliberate breeding program is debatable in this particular case. One has to consider that the various landraces have been cultivated next to each other for centuries, and the various land races have not lost their integrity. Furthermore, US hybrid corn has been imported for decades, and yet the land races have not lost their identity. Yet, there is the probability that cross-pollination occurs between the various land races, and between North American hybrids and local landraces.

There are probably biological reasons why the various land races have not succumbed to centuries of gene flow. There are cultural reasons as well. As a key part of their local culture, maize farmers select for desirable plants and seeds each harvest season, thus ensuring that each land race maintains its integrity from generation to generation. Their incentive is that the various land races have their own set of uses, as determined by colour, taste, dough-making ability, use of stover as animal feed, etc. Farmers select those plants in the field whose ears have the morphology associated with the desired traits (see for example, D. Louette and M. Smale. 2000. Farmer's seed selection practices and traditional maize varieties in Cuzalapa, Mexico. Euphytica 113:25-41).

Regardless of the current Nature paper, crossing between transgenic hybrids and local landraces of corn will probably occur sooner or later, if it hasn't happened already. Gene flow - by chance or intent - has given rise to a large amount of biodiversity, which is balanced as farmers select for specific characteristics that make each local maize variety unique. To imply that this age-old system will now be disrupted and that sustainable food production will be adversely affected is indefensible, unduly alarmist, and irresponsible.

In the end, the results in this paper are not sufficiently reliable to prove anything. We call upon the authors to retract the manuscript, and to annotate the GenBank sequences accordingly.

Downstream (as labeled in figure 2)

(bit scores and E values provided)

K1 AF434754

gi|17082476|gb|AF391808.2|AF391808

Zea mays chromosome 9S b...

101

1e-19

gi|7262818|gb|AF123535.1|AF123535

Zea mays alcohol dehydrog...

82

1e-13

gi|13606087|gb|AF090447.2|AF090447

Zea mays 22 kDa alpha ze...

80

4e-13

A3 AF434755

gi|13606087|gb|AF090447.2|AF090447

Zea mays 22 kDa alpha ze...

113

4e-23

gi|7262818|gb|AF123535.1|AF123535

Zea mays alcohol dehydrog...

113

4e-23

gi|17082476|gb|AF391808.2|AF391808

Zea mays chromosome 9S b...

70

5e-10

A2 AF434756

gi|13606087|gb|AF090447.2|AF090447

Zea mays 22 kDa alpha ze...

442

e-121

gi|17082476|gb|AF391808.2|AF391808

Zea mays chromosome 9S b...

387

e-105

gi|4185305|gb|AF090446.1|AF090446

Zea mays cosmid IV.1E1 22...

387

e-105

gi|7262818|gb|AF123535.1|AF123535

Zea mays alcohol dehydrog...

359

1e-96

gi|1314260|gb|U29136.1|ZMU29136

Zea mays putative matrix at...

281

3e-73

gi|1657766|gb|U68408.1|ZMU68408

Zea mays retrotransposon Op...

276

2e-71

A3 AF434757

gi|17149367|gb|AC091735.3|

Genomic sequence for Oryza sativ.

48

0.003

gi|13810565|dbj|AP003275.2|AP003275

Oryza sativa genomic DN..

48

0.003

gi|5257255|dbj|AP000364.1|AP000364

Oryza sativa genomic DNA...

48

0.003

gi|16418166|gb|AC091680.7|

Oryza sativa chromosome 10 BAC c.

44

0.043

A2 AF434758

gi|13606087|gb|AF090447.2|AF090447

Zea mays 22 kDa alpha ze...

170

2e-40

gi|2832242|gb|AF031569.1|AF031569

Zea mays 22-kDa alpha zei...

170

2e-40

gi|13677167|gb|AC015977.9|AC015977

Homo sapiens clone RP11-...

40

0.54

B3 AF434759

gi|3420038|gb|AF050455.1|AF050455

Zea mays gypsy/Ty3-type r...

363

7e-98

gi|433039|gb|U03680.1|ZMU03680

Zea mays W-22 clone PREM-1 r…

70

1e-09

gi|433043|gb|U03684.1|ZMU03684

Zea mays W-22 clone PREM-1E ...

62

4e-07

gi|4539654|gb|AF061282.1|AF061282

Sorghum bicolor 22 kDa ka...

58

6e-06

gi|13606087|gb|AF090447.2|AF090447

Zea mays 22 kDa alpha ze...

54

9e-05

gi|17082476|gb|AF391808.2|AF391808

Zea mays chromosome 9S b...

52

3e-04

gi|9626938|ref|NC_001497.1|

Cauliflower mosaic virus, compl...

50

0.001

Upstream (as labeled in Figure 2)

K1 AF434760

gi|20524|emb|X14591.1|PHCHSA

P.hybrida chsA gene for chalco...

50

0.002

gi|11128439|gb|AC006305.2|AC006305

Homo sapiens chromosome ...

46

0.036

gi|15217430|ref|NC_003070.1|

Arabidopsis thaliana chromosom...

42

0.56

gi|9022518|gb|AC074025.1|AC074025

Arabidopsis thaliana chro...

42

0.56

gi|6226515|ref|NC_001224.1|

Saccharomyces cerevisiae mitoch...

40

2.2

gi|16152267|gb|AC036102.8|

Homo sapiens chromosome 15

40

2.2

B3 AF434761

Note: This accession is labeled B2 in GenBank and A2 in the supplemental information.

gi|16116509|emb|AL583893.17|AL583893

Mouse DNA sequence fro...

40

1.1

gi|15623922|dbj|AP003331.3|AP003331

Oryza sativa genomic DN...

40

1.1

gi|15224037|ref|NC_003071.1|

Arabidopsis thaliana chromosom...

38

4.2

gi|14530833|gb|AC079037.3|

Oryza sativa chromosome 10 clone…

38

4.2