Analysis of the transgenic papaya genome sequence suggests that transgenes
generally stay put following integration and can achieve stable expression
level from generation to generation.

Enhanced understanding of the mechanisms of transgene insertion and
rearrangement in plant chromosomes is essential not only for the routine
production of transgenic loci, but also ultimately to spur the development
of directed transgene integration approaches. The availability of the
complete genome sequence of the SunUp variety of papaya recently reported in
Nature by Ming et al.1 represents a major step in this regard. This is the
third complete genome sequence of a multicellular plant to be published
(Arabidopsis and rice were the other two) and the first ever genome sequence
of a transgenic organism. From a biosafety perspective, the papaya
sequencing project also provides the first definitive molecular evidence
against in situ transgene rearrangements, one of the main suspected causes
of 'transgene instability'. Although no comparative sequencing over multiple
generations is available, the fact that the transgene remains structurally
and functionally intact in this distant descendent of the original
integration event is convincing proof that transgenes generally become fixed
elements of the plant genome and can achieve a consistent and predictable
expression level from generation to generation.

Papaya is a tropical fruit that was almost wiped out in Hawaii by the papaya
ringspot virus a decade ago. The onslaught of the virus on the island
prompted pioneering efforts in the early 1990s to create a transgenic
variety of papaya resistant to ringspot using ballistic methods. These
efforts resulted in the creation of two virus-resistant transgenic cultivars
'SunUp' and 'Rainbow'. The former produces a red-fleshed fruit that
expresses the coat protein gene of an attenuated mutant of the virus,
conferring resistance via post-transcriptional gene silencing.

Ming et al.1 used a whole genome shotgun approach to facilitate the
sequencing of >90% of the euchromatic papaya genome, including 92.1% of
previously identified expressed sequence tags, 92.4% of known genetic
markers and the genomic sequences surrounding the integrated transgene DNA.
The sequences flanking the transgene appear very similar to those around
'natural' DNA integration events, such as the occasional integration of
chloroplast DNA fragments into the papaya genome, as reported in the same
publication1, and tobacco genome, as reported previously2. This supports
research on both direct DNA transfer (mostly particle bombardment) and
Agrobacterium-mediated transformation, which has shown that transgene
integration is the consequence of a natural process carried out by enzymes
involved in DNA break and repair3.

The SunUp genome sequence lends support to many aspects of the current model
for transgene integration in plants (Fig. 1) and suggests that identical
methods are involved in artificial and 'natural' DNA integration events.
These include the tendency for exogenous DNA sequences to undergo
recombination, rearrangement and truncation before or during integration
(but rarely after integration), the tendency for exogenous DNA to integrate
at AT-rich sites resembling topoisomerase recognition sequences, and the
tendency for junction regions to show evidence of microhomology and
nontemplated DNA synthesis (filler DNA)3. All these processes occur
regardless of the transformation method because they rely on enzymes
endogenous to the nucleus of the host cell.

The initial stages of transgene integration are characterized by complex
rearrangements of input sequences resulting from a combination of
homology-dependent and homology-independent processes stimulated by the
presence of a high concentration of free DNA ends (Fig. 1a). This often
results in the formation of transgene concatemers that may comprise any
number of genes, from fewer than 10 to more than 100, some of which are
complete and some truncated (Fig. 1b). These structures compete with
individual genes for integration sites, which are thought to reflect the
positions of naturally occurring DNA nicks and breaks4. DNA repair complexes
in the vicinity of such breaks are thought to incorporate exogenous DNA into
the repair, generating complex structures that need to be resolved before
the DNA duplex can be sealed. The frequent occurrence of topoisomerase sites
near both integrated transgenes and natural integration sites suggests that
this enzyme plays an important role in the resolution process (Fig. 1c). The
fact that most transgenic plants have a single integration site suggests
that the efficiency of the integration process is low, but the presence of
transgene arrays interspersed with genomic DNA in some transgenic plants
indicates that the integration process might stimulate further integration
events nearby, perhaps through the recruitment of DNA repair complexes4
(Fig. 1d).

Interestingly, the SunUp genome contains truncated versions of the
tetracycline efflux pump (tetA) and neomycin phosphotransferase II (nptII)
marker genes, along with significant amounts of vector backbone DNA1. Such
events are commonly seen with both direct DNA transfer and
Agrobacterium-mediated transformation3, even though in the latter case only
the T-DNA is meant to be transferred. Vector DNA integrated during
Agrobacterium-mediated transformation can either be linked to the T-DNA,
indicating co-linear transfer, or integrate independently, suggesting that
aberrant vector DNA transferred to the plant cell in a co-linear fashion can
be fragmented and then integrated like any other exogenous sequence. The
presence of topoisomerase sites at transgene/genomic junctions after direct
DNA transfer4 and Agrobacterium-mediated transformation5 as well as flanking
integrated chloroplast DNA sequences in the papaya genome1 may even imply
that nuclease activity trims back linear DNA until such sites are exposed,
and the attachment of topoisomerase initiates the joining process. Although
the SunUp papaya variety demonstrates transgene stability over several
generations, the fact remains that some nonessential sequences are present
in the genome. Whereas counter-selective markers can limit the co-transfer
of vector sequences by Agrobacterium-mediated transformation, unlinked
fragments of the vector can still integrate. With direct transfer, however,
vector and/or marker integration can be completely eliminated through the
use of linear cassettes that contain only the promoter, open reading frame
and terminator6.

No endogenous genes were interrupted by the exogenous DNA introduced into
the SunUp papaya genome, so the transgenic plants are functionally similar
to their nontransformed counterparts in all but the virus-resistance
phenotype. Even so, comparative sequencing of the equivalent genomic region
in nontransformed papaya has not been reported so it is not clear whether
endogenous DNA has been deleted at the integration site or whether any
filler DNA has been added at the junctions. It would be very useful to carry
out additional sequencing experiments, which could be completed rapidly
using PCR to amplify across the unused integration site in nontransformed
plants using primers flanking the transgene in the SunUp variety.

Despite the stability of SunUp papaya, it is clear that current strategies
for gene transfer to plants still rely on chance events and the selection of
very rare transformants with appropriate properties from a large number of
primary events. An alternative transformation process that may become more
applicable in the future is transgene integration mediated by homologous
recombination, which would allow transgenes to be targeted to specific
genomic loci known to be favorable for stable, high-level expression.
Although achieved at a low efficiency in model species, the recent reports
of homologous recombination in rice7 and maize8 show how it could be applied
to food crops in the future. Another strategy that may facilitate
predictable transgene expression is the control of gene expression through
transcription factories. This has been elegantly demonstrated in experiments
showing the role of inter-chromatin interactions between two genes on two
different chromosomes that come together at a single focus for
transcription, facilitated by the estrogen receptor9. Designing constructs
that include sequences responsive to transcription factories may ensure
stable expression in future transgenic plants. With such advanced
technologies at our disposal, one day the creation of functional transgenic
plants may cease to be a numbers game and become a more precise science.

References
Ming, R. et al. Nature 452, 991?997 (2008).

Huang, C.Y., Ayliffe, M.A. & Timmis, J.N. Proc. Natl. Acad. Sci. USA 101,
9710?9715 (2004).

Kohli, A. et al. Plant Mol. Biol. 52, 247?258 (2003).

Kohli, A. et al. Proc. Natl. Acad. Sci. USA 95, 7203?7208 (1998).

Makarevitch, I. & Somers, D.A. Plant. J. 48, 697?709 (2006).

Fu, X.D. et al. Transgenic Res. 9, 11?19 (2000).

Terada, R. et al. Nat. Biotechnol. 20, 1030?1034 (2002).

D'Halluin, K. et al. Plant Biotechnol. J. 6, 93?102 (2008).

Nunez, E. et al. Cell 132, 996?1010 (2008).

www.checkbiotech.org