Agrobacterium is a marvellous organism for plant genetic engineering, but we would do well to remember that it has evolved to serve its own ends and not ours. In last month's issue of Trends in Plant Science, Natasha Smith[1xMore T-DNA than meets the eye. Smith, N. Trends Plant Sci. 1998; 3: 85Abstract | Full Text PDFSee all References][1]made clear that the processing of T-DNA plays a key role in defining the type of T-DNA molecule being integrated, and is of critical importance in plant biotechnology.A practical definition of the left border is that it is the point beyond which there is no T-DNA transfer, because it is the point where most of the T-DNA ends on the Ti plasmid — at least in simple systems such as the nopaline Ti plasmid pTi C58 and its derivatives. Nevertheless, it has been known for some time that several species of Agrobacterium carry more than one T-DNA on their Ti plasmid. For example, the octopine type of Ti plasmid, pTiA6, contains four T-DNA borders delimiting, in theory, three T-DNAs: TR (right), TC (central) and TL (left). However, the system is even more complex, because this type of plasmid is able to generate six different types of T-strands (composed of TR, TR+TC, TR+TC+TL, TC, TC+TL and TL (Refs [2xActivation of Agrobacterium tumefaciens vir gene expression generates multiple single-stranded T-strand molecules from the pTiA6 T-region: requirement for 5′ virD gene products. Stachel, S., Timmerman, B., and Zambryski, P. EMBO J. 1987; 6: 857–863PubMedSee all References][2]and [3xVirulence genes, borders, and overdrive generate single-stranded T-DNA molecules from the A6 Ti plasmid of Agrobacterium tumefaciens. Veluthambi, K., Ream, W., and Gelvin, S.B. J. Bacteriol. 1988; 170: 1523–1532PubMedSee all References][3]). The `left border' of TR can, alternatively, act as a left border; not be recognized and allow the transfer of TR+TC; or even play the role of a right border to initiate the transfer of TC+TL. Similar observations have been made with limited-host-range strains of Agrobacterium, where it has been postulated that the different lengths of T-DNA (encoding different numbers of oncogenes) could play a role in the host-range specificity[4xLimited-host-range plasmid of Agrobacterium tumefaciens: molecular and genetic analysis of transferred DNA. Yanofsky, M. et al. J. Bacteriol. 1985; 163: 341–348PubMedSee all References, 5xRole of T-region border in Agrobacterium host range. Paulus, F. et al. Mol. Plant–Microbe Interact. 1991; 4: 163–172Crossref | PubMedSee all References]. Thus, unprocessed left borders are not merely an invention of binary plasmid designers.Are there ways in which to improve processing at the left border and to transfer T-DNA in a more precise manner? In order to answer this question, it would be interesting to investigate the DNA region surrounding the left border, because an `attenuation activity' of sequences flanking the left border has been reported[6xSequence context of the T-DNA border repeat element determines its relative activity during T-DNA transfer to plant cells. Wang, K. et al. Mol. Gen. Genet. 1987; 210: 338–346Crossref | Scopus (27)See all References][6]. Termination of T-strand production probably involves nicking of the left border, and although this has not been fully investigated, the nicking probably has to be performed at a high rate, without initiation of a new T-strand to the left of the left border. Some as yet undescribed activity of virulence proteins might be responsible for this process; these proteins could recognize the region surrounding the left borders (e.g. the `attenuation domain') and block T-strand initiation. Certainly, a better understanding of the role of the left border during T-strand formation would allow the design of better vectors for Agrobacterium-mediated transformation: a combination of attenuation regions in the binary plasmid and of the corresponding virulence region on the helper plasmid might lead to the transfer of well-delimited T-DNA.In assessing the efficacy of T-DNA transfer for biotechnology, it is important to consider the small number of Ti plasmids that naturally occurs in Agrobacterium cells. Processing of the borders from a Ti plasmid involves a stoichiometric reaction between the endonuclease VirD2 and the border sequences. As a consequence, there might be a risk in using binary plasmids present in high copy number (especially in suboptimal induction of the virulence region), because the excess of borders could titrate-out VirD2, resulting in incomplete processing.Despite these considerations, in most cases correct processing is still observed. Although it is possible to improve on our present tools, it might be more convenient simply to eliminate those transgenic plants that contain DNA from outside the T-DNA integrated in their genome. As far as I am aware, colleagues who have undertaken field trials with plants transformed using Agrobacterium had to prove that the integrated foreign DNA was restricted to the region between the left and the right border.