Natural transformation is assumed to be the most likely mechanism by which DNA from transgenic plants could be horizontally transferred to bacteria. In order to determine the occurrence of naturally transformable bacteria amongst bulk and rhizosphere soil bacteria, different transformation strategies were employed using either plasmid DNA (IncQ plasmids pSM1890 and pSM1885, conferring GFP, Smr, Gmr and GFP, Smr, Tcr, respectively) or genomic DNA from rhizosphere isolates, which were chromosomally tagged with mini-Tn5~(GFP, Tcr, as transforming DNA. Transformation assays were done in microtiter plates (262 isolates and pSM1890 or pSM1885), on filters (i) with rhizosphere bacterial community mixed with pSM1890 or pSM1885, (ii) with 24 rhizosphere or soil bacterial isolates mixed with genomic DNA of the corresponding mini-Tn5-tagged strains, and in the rhizosphere of tobacco plants inoculated with rifampicin-resistant bacterial isolates and genomic DNA of the corresponding mini-Tn5-tagged strains added. One transformant colony was obtained when Brevundimonas vesicularis was transformed with genomic DNA of the corresponding mini-Tn5-tagged strain. Attempts to reproduce this result were unsuccessful. With this single exception, transformants were neither detected in the collection of isolates nor in the rhizosphere bacterial community. Acinetobacter baylyi BD413 used as a positive control showed drastically reduced transformation frequencies with plasmid pSM1890 as transforming DNA when mixed with the rhizosphere pellet. All transformants were characterized by BOX-PCR fingerprints, and three different BOX patterns were revealed. Sequencing the 16S rRNA gene showed that all transformants could be assigned to Acinetobacter sp. Since transformants were only observed in the positive control, the introduced BD413 either underwent genomic rearrangements, or competence of the Acinetobacter population present in the rhizosphere was stimulated by the introduction of BD413. The various transformation assays performed indicate that the proportion of rhizosphere or bulk soil bacteria which are naturally transformable is negligibly low.

RecA-mediated recombination requires regions of homology between donor and recipient DNA for successful integration. This paper investigates the effect of the relationship between the length of gene-sized inserts (434, 733, 2228 and 2400 bp) and flanking sequence homology (100 – ca. 11 000 bp) on transformation frequency in Acinetobacterbaylyi strain BD413. Both insert size and size of the homologous region were varied, which improves on previous studies that kept insert size constant and varied only the homologous flank size. Transfer frequency of a non-homologous single small gene for gentamicin resistance (aac(3)I; 773 bp) was increased 18-fold when flanking homology was changed from about 2000 bp to 8000 bp, but was reduced 234-fold when two genes were inserted (nptII-gfp; 2400 bp) between similar homologous regions. To investigate the effect of smaller regions of flanking homology (100 – 2000 bp), a partial nptII-gfp deletion (434 bp) was restored. This confirmed that a minimum of 500 bp on each flank was required for transformation to be affected by flanking homology. The data obtained allowed development of a multiple regression equation to predict transformation frequency from homology, insert size and total fragment size for gene insertions. We also show that the ratio of flanking homology to insert size and not the total size of donor DNA is the most important variable determining transformation frequency. The equation developed was consistent with results previously reported by others, and so will be useful when using A. baylyi as a model for gene transfer by transformation in the laboratory, environment and for biosafety.