Scientists at JIC assure us that the CaMV 35S promoter is safe on grounds that people have been eating cabbages infected with the virus all the time. We have shown in detail in at least two further publications [2,3] why the CaMV 35S promoter in the viral genome and the intact virus is not the same as the cut-out promoter in GM constructs integrated into transgenic plants. But the JIC scientists have persistently failed to cite our papers or to reply to them. It would be tedious to repeat our arguments here. So, let s look at the new point raised by one of them, Paul Christou, who claims their recent publication [4] shows transgenic rice lines are stable. Their claim to stability is not at all supported by the data presented.

The publication in question is a report of studies on 40 independent rice lines representing 11 cultivars, each containing 2 or 3 different transgenes (not 3 or 4 as it says in the text) delivered by particle bombardment. The transgenes were delivered either by cotransformation, ie, the genes in two or more different vectors, or in the form of cointegrate vector, ie, all three genes linked together in one vector. The authors made the following claims.

1. "Approx. 75% of the lines (29/40) demonstrated Mendelian inheritance of all trangenes, suggesting integration at a single locus".

2. Levels of expression varied, but primary transformants showing high level expression of four of the transgenes, gna (snowdrop lectin), gusA (beta-glucuronidase), hpt (hygromycin resistance) and bar (tolerance to herbicide phosphinothricin) "faithfully transmitted these traits to progeny".

3. In the case of transgenes cry1Ac and cry2A (insecticidal endotoxins from Bacillus thuringiensis or bt), "transgene expression was stably inherited when primary transformants showed moderate or low-level expression".

4. The results show that "six transgenes (three markers and three insect-resistance genes) were stably expressed over four generations of transgenic rice plants".

5. The results showed that "transgene expression was stable in lines of all the rice genotypes" analysed.

Now, claims 4 and 5 are contradicted by claims 1 to 3 which indicate that at least 25% of the lines did not show Mendelian inheritance, a sign of instability, and that only some of the transformants gave stable expression in later generations. So we can dismiss claims 4 and 5 immediately, and concentrate on the first three. But even those are not borne out by the data presented. This seems to be a general problem with this publication. The text states what the data do not support, and I shall highlight some of the inconsistencies here.

The experimental details are very sketchy. No genetic maps were given on the constructs or the plasmids used, nor whether any plasmid sequences are integrated into the transgenic plants. No genetic map of any of the inserts in the plant genome is given. Going by the information in Table 1 (p.391) and "Transforming plasmids" (p.389), I believe the 40 transgenic lines consisted of the following categories.

1. Twelve lines from 5 cultivars transformed with the co-integrate vector containing three marker genes, all driven with the CaMV 35S promoter, but in which orientation, the paper does not say.
2. Eight lines from 4 further cultivars co-transformed with gna driven by the maize ubiquitin promoter in one plasmid and gusA and hpt both driven by the CaMV 35S promoter in a second plasmid.
3. Six lines from 2 of the 4 cultivars in (b) co-transformed with gna driven with the rice sucrose-synthase-1 promoter in the first plasmid and the same gusA and hpt second plasmid as in (b).
4. Eight lines from two cultivars different from all the above, cotransformed with a plasmid carrying a bt toxin gene, cry2A, driven by the CaMV 35S promoter, and a second plasmid as in (b).
5. Six lines from the same cultivars as (d), cotransformed with a plasmid carrying the bt toxin gene, cry1Ac, driven by the maize ubiquitin promoter and a second plasmid with hpt, driven by the CaMV 35S promoter.
6. There is little or no overlap between the transforming genes/plasmids and cultivars, so it is impossible to tell if all the transgenes behave the same in all cultivars. The claim in the title "multiple heterologous transgenes show similar behaviour in diverse genetic backgrounds" is unjustified, to say the least.

The transgenic lines were characterised as follow.

• enzyme assays to detect bar and gusA expression
• western blots to detect the protein products of crylAc, cry2A and gna
• polymerase chain reaction (PCR) to detect the DNA of all six transgenes
• Southern blots to detect the plant genomic fragments that have integrated the transgenes after the genomic DNA is cut with restriction enzymes with unique sites in each of the transforming plasmids used. This means that for single copy of the non-rearranged insert at a single site, only one, or at most two bands should be present. The presence of more than two bands is an indication of repeats and rearrangments.

There are two kinds of transgenic instability, functional instability as in gene silencing, and structural instability as in rearrangement and loss of transgenes. Of the analytical methods used, only Southern blot is really informative on the structural instability of the transgenic lines. Southern blot data were presented on only four R3 lines from category (a) involving the cointegrate vector carrying three marker genes. Instability is also indicated whenever the progeny of transgenic lines deviate significantly from Mendelian ratios. However, and this is important, failure to significantly depart from the Mendelian ratio is not a sign of stability, unless corroborated by quantitative molecular data.

On p. 390, it states that,

"Genetic analysis confirmed that all lines carried a single transgenic locus with one to seven copies of the three-gene construct." And

"R2 and R3 plants of lines K496-4 and K496-1 consistently lacked detectable GUS activity although the primary transformants showed moderate GUS activity, indicating that the gusA transgene had undergone silencing. Plants from C549-1 consistently lacked detectable PAT activity; however in this case the primary transformants also lacked PAT activity. Southern blot hybridisation of K496-4, K496-1 and C549-1 genomic DNA revealed banding patterns identical to those of the primary transformants showing that there had been no loss or rearrangement of the genes during transmission from Ro to R1. Stable inheritance of the gusA, hpt and bar transgenes was also found for the remaining (non-silenced) lines, from Ro through to R3 (Fig. 3)."

Figure 3 (p.392), the only Southern blot data presented, is on just four lines in the R3 generation. The first line, K496-4 represented by seven plants, indeed showed identical banding patterns in all lines probed with hpt, gusA and bar, but four or five bands are present and the banding pattern for each enzyme is different. These are signs of multiple repeats and rearrangements. More seriously, the intensity of the bands varied more than ten-fold, suggesting that some plants may have had deletions of multiple copies, or else other plants have had amplified copies. The second line, K496-3 represented also by 7 plants did not have any hybridisation signal for three of the plants, while one had at least 20 times the intensity of the banding pattern. Of the remaining three, one had twice the intensity of the other two. These plants cannot be said to follow Mendelian inheritance. The third line K496-2 is represented by 2 plants only, one with multiple banding pattern and the other without any bands, again, not quite Mendelian . That leaves only the fourth line, K496-1, represented by two plants, which showed identical multiple banding patterns in both plants which are different for all the enzymes, again indicating repeats and rearrangments. On the most generous interpretation of the data, there is evidence that one out of the 12 transgenic lines in category (a) may be stable to the R3 generation.

On p. 395, it claims that,

"With the exception of the line K495-1, PCR analysis confirmed a 3:1 Mendelian segregation in all lines at the DNA level. Many lines also showed Mendelian segregation for protein expression, but the ratio was distorted for gusA in lines K496-1 and K496-3 due to silencing& Line C549-1 carried a single integrated copy of the construct, but bar gene expression was undetectable even in the Ro plant. Further analysis showed that the CaMV 35S promoter driving bar in this line was rearranged.."

However, no data were presented to show Mendelian segregation for any of the lines.

With regard to the transgenic lines with gna driven by two different promoters in categories (b) and (c) , it states on p391,

"In most of the lines (11/14), the three transgenes segregated together as a single Mendelian trait (3:1 ratio), indicating cointegration of the two cotransforming plasmids at a single locus. Three lines showed an aberrant 1:1 ratio."

However, Table 2 (pp.394-5) contains data only on 6 lines in the R1 generation, of which one gave the aberrant 1:1 ratio. At R2 and R3, only 4 lines are left. So, at most, there is evidence to support the stability of 4 out of 14 lines. Even these are questionable, if the standards of Fig. 3 are to go by. Quantitative PCR or Southern blot should distinguish between homozygous and heterozygous positives, so the proper Mendelian ratio to use is 1:2:1, and not 3:1.

The same criticisms apply to the data in Table 3 (p.396) on inheritance of the transgenes in categories (d) and (e) above. Of the 13 (not 14 as claimed in the text) lines analysed in R1, 5 failed to give Mendelian 3:1 ratio. Four lines remained in R2 and by R3, only two lines were left. So again, there is evidence of stability for at most 2 out of 14 lines. And yet it states on p.391,

"We analysed 6 cry1Ac-transgenic lines. All lines showed 3:1 segregation for the transgenes in the R1 and R2 generations." According to Table 3, two of the lines did not give 3:1 ratio in R1 and none were analysed in R2.

On p. 392,

"Most seedlings from lines expressing Cry2a at high levels (M7-10, M7-12, M7-13 and M7-14) died within two weeks of emergence, probably due to the toxic effects of high-level endotoxin expression."

The western blots of Cry1Ac and Cry2A (Fig. 4 C-D, p.393) showed multiple extra band, differing between plants in the same line, hardly an indication of stability.

Another factor that already biases the data towards apparent stability is that the seeds were germinated in the presence of hygromycin, so all plants that had lost the hpt gene or in which the hpt gene was silenced would have been eliminated.

A generous interpretation of the data presented would suggest that 7 out of 40 (18%) of transgenic rice lines may be stable to the R3 generation.

1. "Top research centre admits GM failure" ISIS News7/8, Feb. 2001 www.i-sis.org/i-sisnews7-4.shtml
2. Ho, M.W., Ryan, A. and Cummins, J. (2000). Hazards of transgenic plants with the cauliflower mosaic viral promoter. Microbial Ecology in Health and Disease 12, 6-11.
3. Ho, M.W., Ryan, A. and Cummins, J. (2000) CaMV 35S promoter fragmentation hotspot confirmed, and it is active in animals. Microbial Ecology in Health and Disease (in press).
4. Gahakwa D, Maqbool SB, Fu X, Sudhakar D, Christou P and Kohli A. Transgenic rice as a system to study the stability of transgene expression: multiple heterologous transgenes show similar behaviour in diverse genetic backgrounds. Theor Appl Genet 2000: 101: 388-99.