The mechanism of T-DNA integration in plant genome is still largely unknown although the Agrobacterium-mediated transformation has become one of the most widely used techniques for creating genetic modification of crops and researches on plant genes. The difficulty in producing transgenic monocots by this technique has limited its application in many important food crops such as maize, wheat and rice until recently.

Early in a series of experiments, we obtained transgenic rice plants of five Chinese varieties including maintainer and restorer lines of hybrid rice via Agrobacterium-mediated transformation (Zhai et al. 2000). The wide spectrum resistance gene for resistance to bacterial blight, Xa21, was integrated and stably inherited through several generations. Most of the T₀ plants and their T₁ progenies showed high level of resistance to the disease. These transgenic rice plants are not only useful in breeding excellent rice varieties with bacterial blight resistance, but are also valuable for the study of the mechanism of T -DNA integration in a monocot.

In this study, TAIL-PCR (Thermal asymmetric interlaced PCR) technique was used to amplify T-DNA flanking sequences in 20 resistance positive transgenic rice plants, including 6 T₀ and 14 T₁ plants. Twenty-four fragments were amplified and sequenced, of which 14 were rice DNA, 9 contained vector backbone sequences of three types, and the remaining one was a segment of the exogenous gene Xa21.

Of the 14 flanking sequences whose nature was rice DNA, 11 were at the left border (LB) of T -DNA and 3 at the right border (RB). No homology was found between these sequences and the sequence of transfer vector pCXK 1301. Moreover, strong hybridization signals could be detected when they were hybridized with rice genomic DNA. The sequences of these T-DNA end-joining regions are compiled in Fig. 1 in a linear form and the schematic map of this type of junction is shown in Fig. 2A.

None of these sequences had an AT-content over 70% (Data not shown), of which six had even lower AT-content than the average of rice genomic DNA (61.5%). The length of these fragments did not have significant influence on amount of AT-content. All but two of them bore just a few motifs such as A-box, T-box, AT-tract, ATC-tract and the recognition sites of DNA topoisomerase which were characteristics of scaffold/matrixattachment region (S/MAR) (Takano et al. 1997, Sawasaki et al. 1998). The copy numbers of these sequences in rice genome was determined when hybridized with rice genome DNA, and most of them were single copy (Data not shown). However, exogenous genes introduced by direct gene transfer methods such as particle bombardment and calcium phosphate have been found to have a preference to integrate in the repeated sequences and AT -rich regions bearing motifs of S/MAR in plant genomes (Takano et al. 1997, Sawasaki et al. 1998).

Studies on T-DNA processing have shown that T-strand theoretically contains remnant

border sequences of 22nt (Nucleotide) and 3nt in its left and right ends, respectively, when it is produced from T-DNA. No 22nt LB remnants remained intact in the 14 fragments we obtained. In L608, L612, L593 and L590, the whole LB sequence and several nucleotides adjacent to LB were deleted. However, deletion of the whole RB was not found, and at least one (in R612 and R003) or three (in R010) nucleotides of the RB sequence were retained, correlating with the position of major and minor nick sites identified during T-DNA processing in Agrobacterium (Wang et al. 1987). This suggests that the right T-DNA end remained intact, to some extent, during transfer and integration while the left border junctions showed more variations. This is in agreement with that found in many dicots indicating that the T-DNA integration in monocots possibly involves the same mechanism as dicots (Guus et al. 1989, Reinhold et al. 1991, Maria et al. 1997). It has been known that VirD2 protein remains to covalently join the right end of the T-strand after nicking the T-DNA borders. It pilots and protects T-DNA during the transfer and integration while LB are often partially or all deleted for lacking the protection.

Transgenic plant No.585 was the progeny of No.45 and their flanking sequences were almost the same, indicating that the integrated T-DNA could inherit with no changes in the insertion site. A BLAST X search against GeneBank was performed for these rice DNA bearing sequences but no homology to any protein-coding genes could be found.

Of the 24 sequenced fragments, 9 contained vector backbone sequences of three types. The backbone sequence is referred to the bacterial replicon region of binary vector (Maria et al. 1997). Type I is the backbone sequence directly linked to the T-DNA across either LB or RB (Fig. 2B). In three amplified fragments at LB (L301, L591 and L592-4) and three at RB (R301, R591 and R597), sequences linked to either border were found to read through the LB or RB. In addition, Southern hybridization in transgenic plants No.301 and 591 showed - that both contained only one copy of the transgene Xa21, indicating that the whole transfer vector possibly is integrated into rice genome in both cases.

Type II has the backbone sequence not directly linked to a T-DNA border (Fig. 2C). The fragment L010 has a total length of 273nt with 15nt remnants at LB. The sequence between its 57^thnt and 162^thnt is a backbone fragment. The two end segments (i.e. the segment between l^stnt and 56^thnt and the segment between 163^thnt and 273^thnt) are homologous to vector pCXK1301. Moreover, the whole fragment showed strong hybridization signals in Southern analysis probed with rice genomic DNA, indicating that both of its end segments possibly originated from rice DNA.

Type III is rather complicated. As in transgenic plants R615 and R592-1 the flanking sequences outside RB were from the backbone sequence linked to LB in the vector (Fig. 2D), which might be attributed to integration of tandem T-DNAs into the rice genome. The other unusual flanking sequence was found in transgenic plant No.52, in which the flanking sequence outside LB is a segment of the exogenous gene Xa21 (Fig. 2E). Until recently, it has been assumed that during Agrobacterium-mediated plant transformation, only the sequence between the two T -DNA borders is transferred to the plant. Indeed, T -DNA is traditionally defined as the sequences that lie between these borders. However, evidences in the literature suggest that DNA sequences residing outside the T -DNA border may be occasionally transferred to the plant. Martineau et al.(1994) reported that binary vector sequence outside T -DNA left border could integrate into the genomes of transgenic plants by a rate of 20-30%. In addition, Maria et al.(1997) found backbone sequences in over 75% transgenic tobaccos. In this study, 37.5% (9/24) sequenced flanking fragments had backbone sequences, slightly higher than the statistic of Martineau et al. In agreement with that found by Maria et al., the integrated backbone sequences were either linked to the T -DNA across the border sequence or independent of the T -DNA.

The results from the present study provided information about the characteristics of flanking sequences of integrated T-DNA in rice genome. They will contribute to utilization of the Ti-plasmid transformation system in a more comprehensive way and get full use of the system in monocot transformation. The flanking sequences can be used to map the T -DNA insertion sites in rice genome and to facilitate the research into the positional effect and stability of exogenous gene in transgenic plants. Finally, we think that the evaluation of transgenic plants should include the possible effect caused by vector backbone sequences.

References

Guus, B., K. Zdena, G. Nigel and H. Barbara, 1989. Recovery of Agrobacterium tumefaciens T-DNA molecules from whole plants early after transfer. Cell 57: 847-857.

Enoki, H., T. Izawa, M. Kawahara, M. Komatsu, S. Koh, J. Kyozuka and K. Shimamoto, 1999. Ac as a tool for the functional genomics of rice. Plant J. 19(5): 605-613.

Takano, M., H. Egawa, J.-E. Ikeda and K. Wakasa, 1997. The structures of integration sites in transgenic rice. Plant J. 11(3): 353-361.

Maria, E., B. Burgund and B. Stanton, 1997. Integration of T-DNA binary vector 'backbone' sequences into the tobacco genome: evidence for multiple complex patterns of integration. Plant J. 11(5): 945-957.

Martineau, B., T. Voelker and R. Sanders, 1994. On defining T-DNA. Plant Cell 6: 1032-1033.

Reinhold, M., K. Zsuzsanna, N. Christiane, B. Guus, C. Andreas, A. Karel, P. George, S. Jeff, H. Barbara and K. Csaba, 1991. T-DNA integration: a mode of illegitimate recombination in plants. EMBO J. 10(3): 697-704.

Sawasaki, T., M. Takahashi, N. Goshima and H. Morikawa, 1998. Structures of transgene loci in transgenic Arabidopsis plants obtained by particle bombardment: Junction regions can bind to nuclear matrices. Gene 218: 27-35.

Wang, K., S.E. Stachel, B. Yimmerman, M. Van Montagu and P.C. Zambryski, 1987. Site-specific nick in the T -DNA border sequence as a result of Agrobacterium vir gene expression. Science 235: 587-591.

Zhai, W., X. Li, W. Tian, Y. Zhou, S. Pian, S. Cao, S. Zhao, B. Zhao, Q. Zhang and L. Zhu, 2000. Introduction of a blight resistance gene, Xa21, into Chinese rice varieties through an Agrobacterium-mediated system. Science in China (Series C) 43(4): 361-368.