Conjugal gene transfer between bacteria in soil and rhizosphere

The extent of possible conjugal transfer of recombinant DNA present in genetically engineered microorganisms (GEMs) was studied. Occurrence of transfer of recombinant DNA is only one of the concerns regarding the use of GEMs (Chapter 2). Other potential hazards preventing the application of GEMs for agricultural purposes are (1) possible pathogenicity, (2) disturbance of ecological balance, (3) unwanted biochemical reactions, and (4) negative effects on diversity and specific populations (Levin and Strauss, 1991). The fate of GEMs introduced into soil is not very predictable, since knowledge on most of the aforementioned aspects is scarce. In most countries strict regulations limit large scale field studies with GEMs, which slows down research.Research was focused on (1) development of sensitive detection techniques to study GEMs in soil, (2) detection of conjugal plasmid transfer to indigenous bacteria in soil, (2) assessing the fate of recombinant DNA present in different genomic locations. The soil isolate Pseudomonas fluorescens R2f was chosen as model microorganism. All studies were done in vitro and in soil microcosms which were planted whith wheat.Studying conjugal transfer between homologous donor and recipient strains introduced into soil we found that matings could occur on the transconjugant selective plates, thus obscuring the real number of transconjugants in soil (Chapter 3). The use of nalidixic acid instead of streptomycin (in conjunction with rifampicin) to select for recipients which received the plasmid and to counterselect the donor, was shown to prevent these plate matings. The use of nalidixic acid to prevent plate mating was later confirmed by Walter et al., (1991). Employing this donor counterselection, it was shown that the number of transconjugants decreased with decreasing numbers of introduced donor and recipient cells. However, transconjugants could only be detected when the soil was amended with nutrients or bentonite clay or when plant roots were present. The fact that plasmid transfer between P.fluorescens R2f strains was enhanced in the rhizosphere was not entirely surprising since this organism was originally isolated from the rhizosphere. Stimulation of plasmid transfer by addition of nutrients or the presence of plant roots was also found by others (Stotzky, 1989; Edwards, 1993).Experiments in which both donor and recipient are introduced, might be indicative for factors affecting conjugal gene transfer, however they do not necessarily predict transfer to indigenous bacteria. Most potential indigenous recipients are generally in a different physiological state than the freshly cultured, metabolically active, introduced recipients. To detect transfer to indigenous bacteria, another donor counterselection method was developed (Chapter 4). A phage specific for the donor strain, ΦR2f, was isolated, which could be used to lyse the donor prior to plating on transconjugant-selective plates. The selftransmissible plasmid RP4p which containes suitable antibiotic resistance genes for selection, had been marked previously with an eukaryotic DNA fragment for hybridization purposes, giving RP4p. RP4p transfer to indigenous bacteria was observed in the rhizosphere of wheat (Chapter 5). The number of indigenous transconjugants detected was around 10 3cfu/g of soil, while transconjugant numbers in the corresponding bulk soil were just below 10 2cfu/g of soil. All indigenous transconjugants analysed contained the plasmid, and all were able to transfer RP4p to a P.fluorescens recipient strain in control filter matings. The transconjugants were identified as belonging to the genera Pseudomonas, Enterobacter, Comamonas and Alcaligenes. These genera fit very well into the known host range of RP4 (Thomas, 1989).Transfer of RP4p to indigenous bacteria was also studied in four different soils (Chapter 6). Highest numbers of transconjugants per g of dry soil were found in Montrond silt loam and Flevo silt loam (10 3-10 4), whereas in Ede loamy sand and Loss silt loam transconjugant numbers were around 10 2. The presence of plant roots affected transconjugant numbers to a significant extent in Ede loamy sand and Loss silt loam but not in the other soils. High clay and organic matter contents as present in Flevo silt loam and Montrond silt loam might be favourable for conjugal transfer and obscure any stimulatory effect of the rhizosphere.Since selftransmissible plasmids are not likely to be used as vectors for inserting recombinant DNA in GEMs, possible transfer of a marked IncQ plasmid was studied (Chapter 7). A marker cassette was constructed based on two antibiotic resistance genes, npt II and aad B conferring resistance against kanamycin and gentamycin, and part of a Bacillus thuringiensis endotoxin gene, cry IVB was used as molecular marker. This marker cassette was cloned an IncQ plasmid (a RSF1010 derivative) wich resulted in plasmid pSKTG. Mobilization of the non-selftransmissible plasmid pSKTG was studied in filter matings, in sterile soil and in natural soil. In filter matings, pSKTG was only mobilized in the presence of RP4p. Transfer frequencies in bi- and tri-parental matings were in the same range, indicating that apparently cell-to-cell contact was not a limiting factor. In sterile soil, mobilization frequencies of pSKTG with RP4p present in the same strain were 100-fold lower then on filters (10 -4). When RP4p was present in a separately introduced strain, (triparental) transfer was found to occur with a frequency of 10 -6. In microcosms with non-sterile soil planted with wheat, mobilization of pSKTG to indigenous bacteria was detected only in the presence of RP4p. The results obtained in filter matings using using the total soil bacterial community suggested the occurence of genetic elements capable of plasmid mobilization. Such elements have been found by Hill et al. (1992) and Top (1993), elements with mobilizing capacity are probably present at low levels in natural soil, since pSKTG could not be shown to be mobilized in situ in the absence of RP4p.The influence of the location of heterologous DNA in Pseudomonas fluorescens R2f on gene stability, expression and transfer following introduction into Ede loamy sand and Flevo silt loam was studied (Chapter 8). Three strains were used with markers on different genetic elements, i.e a selftransmissible plasmid (RP4p), a mobilizable plasmid (pSKTG), and a chromosomally inserted marker gene cassette (KTG). In vitro filter mating experiments showed that the selftransmissible plasmid was transferred with high frequencies (about 10 -2) and that this plasmid could mobilize pSKTG with similar frequencies. The chromosomally inserted marker cassette could be mobilized by RP4p to a recipient strain with low frequency (10 -8). In sterile soil, transfer of the chromosome could not be detected in the presence of RP4p. The three strains showed poor survival in Ede loamy sand and good survival in Flevo silt loam. Moreover, a partial, but significant loss of expression of the gentamycin resistance gene aad B was observed in Ede loamy sand, but not in Flevo silt loam.RP4p was found to be transferred from an introduced donor to indigenous bacteria in both non- sterile soils planted to wheat, whereas transconjugants harbouring the mobilizable plasmid pSKTG or the chromosomally inserted marker cassette were not detected when using a donor strain without IncP1 plasmid.A lot of data concering gene transfer in soil have been published since this project started in november 1988. At that time however, most knowledge on conjugal gene transfer had been obtained using selftransmissible plasmids and introduced donor and recipient strains. In this work we showed that care should be taken to avoid matings on selective plates when such experiments are performed. The experiment which showed that RP4p could transfer to indigenous bacteria in soil and rhizosphere was one of the first reports on this phenomenon. Later, we confirmed that RP4p could also transfer to indigenous bacteria in soils of different type and texture. This indicated that the soil environment is not a barrier for gene transfer. The observation that an IncQ plasmid could be mobilized in situ to indigenous bacteria by RP4p present in a different strain, showed that tri- parental matings could take place. Transfer of an IncQ plasmid, or markers present on the chromosome could not be detected when RP4p was not added. However, indications were found for the presence of selftransmissible genetic elements in soil bacteria which could mobilize an IncQ plasmid.Transfer frequencies of recombinant DNA present on either a selftransmissible, a mobilizable plasmid, or on the chromosome can be compared (some are based on estimations) using the data obtained (See Table 1). Multiplication of the figures with the number of introduced donor cells (per gram of soil) gives, depending on the conditions in soil, the number of transconjugants per gram of soil. Since IncQ plasmids and chromosomal inserts cannot propagate their own transfer, we assumed the presence of selftransmissible plasmids in 10 4indigenous bacteria per g of soil (total 10 8cfu/g of soil) in soil, with the capability of plasmid and chromosome mobilization. Frequencies clearly indicate that transfer of both the IncQ plasmid and the insert by naturally occurring genetic elements is below the detection limit (which is between 10 and 100 bacteria per gram of soil) and that transfer of those elements is a very rare event. It is difficult to indicate up to what level transfer of heterologous genes is acceptable. We therefore propose that if bacteria containing heterologous genes are to be applied for agricultural purposes, the recombinant DNA will be inserted into the chromosome to prevent unnecessary transfer.