Mae-Wan Ho - Institute
of Science in Society and Department of Biological Sciences, Open
University, Walton Hall, Milton Keynes, MK7 6AA, UK
The following reply to Trewavas and Leaver's long correspondence in
Nature was rejected on ground that "no new point" was raised. I
am posting it again in case T&L's fallacious argument is taken
seriously by anyone.
Sir - Trewavas and Leaver (1) claim that genetic engineering is no
different from the natural processes that have been occurring throughout
evolution. Moreover, they argue that this is supported by what I wrote in
my book (2).
Perhaps they have not read the whole book, for I explain, especially in
the second edition, that while artificial genetic engineering is
uncontrollable, random and unpredictable, natural genetic engineering is
quite precise and repeatable because it is regulated by the organism as a
whole (3). This regulatory system has evolved over hundreds of millions of
years. Under steady state conditions, proof-reading, DNA-repair and other
mechanisms ensure that the DNA remains constant and stable. But
predictable rearrangements, amplifications, deletions and mutations may
also take place during normal development and in response to certain
In contrast, DNA polymerase reactions in the test-tube are prone to
error. Artificial transformation gives unpredictable, unrepeatable results
and the transgenic lines obtained are often unstable. Trewavas and Leaver
assume that lethal insertions [for the plant] are selected out, while
potentially innocuous insertions [for consumers] are detected by substantial
equivalence. But current risk assessment hardly addresses the
unpredictability of the process or the stability of transgenic lines (4).
There are two main reasons for the unpredictability and instability of
artificial genetic engineering. First, foreign DNA is broken down and
inactivated by restriction mechanisms in the host cells. Second, the
transgenic constructs -made up of DNA originating from many different
sources - are structurally unstable (5).
Several expression-cassettes are usually stacked in series, each
cassette consisting of a gene with a promoter (as well as terminator). The
promoter is often taken from a virus in order to make the gene
over-express. The stacked cassettes are, in turn, spliced into a vector,
the most widely used being the T-DNA of Agrobacterium; and it is
this whole construct which is intended for integration. But unpredictable
deletions, rearrangements and repeats are invariably present in the actual
insert. Transgenic instability often persists after the insertion event
(6). The constitutive over-expression of transgenes placed under viral
promoters may be one cause of gene-silencing (7), and recombination
hotspots may be another.
The borders of T-DNA are recombination hotspots, as is the 3 end
of the 35S promoter from the cauliflower mosaic virus (CaMV) (8), which is
used in practically all transgenic plants currently released.
Recombination hotspots are expected to increase the likelihood of
secondary mobility and horizontal gene transfer (9). Secondary mobility
within the host genome may result in rearrangements and other effects that
could drastically alter the agronomic and other properties of the
That plant genomes, like animal genomes, harbour retrotransposons,
relict retroviruses and pararetroviruses (10) is no reason for
complacency. These sequences have existed for millions of years in the
genome, and are probably no longer harmful, either to the plant itself or
to other organisms interacting with it. However, should they recombine
with the CaMV promoter (or modules thereof) in the transgenic DNA, live
viruses may well be regenerated. The transgenic DNA may also be mobilized
by these relict elements.
Horizontal gene transfer could spread transgenic DNA to unrelated
species, in principle, to all species that interact with the transgenic
plant, including bacteria and viruses in all environments, and animals
that feed on the plant. Chief among the potential dangers are the spread
of antibiotic resistance marker genes, the creation of new viruses and
bacteria that cause diseases, and insertion mutagenesis, including cancer
in mammalian cells. These potentials should be addressed by empirical
investigations instead of bland reassurances of substantial
1. Trewavas, A. and Leaver, C. Nature403, 12 (2000).
3. See Shapiro, J. Trends in Genetics13, 98-104 (1997).
4. See Ho, M.W. and Steinbrecher, R. Environmental and Nutritional
Interactions2, 51-84 (1998).
5. The structural instability of artificial vectors made by joining DNA
from widely different sources is a text-book topic. See Old, R.W. and
Primrose, S.B. Principles of Gene Manipulation (fifth edition),
Blackwell, Oxford (1994); also Prazeres, D.M.F. et al. Tibtech17, 169174 (1999).
6. See Srivastava, V., Anderson, O.D. and Ow, D.W. Proc. Nat. Acad.
Sci. USA96, 11117-11121 (1999).
7. See Finnegan, J. and McElroy, D. Bio/Technology12,
8. See Kohli, A., et al. The Plant Journal17,
9. See Ho, M.W., Ryan, A. and Cummins, J. Microbial Ecology in
Health and Disease (in press).
10. Jakowitsch, J., et al. Proc. Nat. Acad. Sci. USA96, 13241-13246 (1999).