ABSTRACT
Conjugation, mobilization, and retromobilization are three related mechanisms of horizontal gene transfer in bacteria. They have been extensively studied in gram-negative species, where retromobilization, the capture of DNA from a recipient by a donor cell, was shown to result from two successive steps: the transfer of the conjugative plasmid from the donor to the recipient followed by the retrotransfer of the mobilizable plasmid to the donor. This successive model was established for gram-negative bacteria but was lacking experimental data from the gram-positive counterparts. In the present work, the mobilization and retromobilization abilities of the conjugative plasmid pXO16 from Bacillus thuringiensis subsp. israelensis were studied using the mobilizable plasmids pUB110 and pE194 and the “nonmobilizable” element pC194 lacking the mob and oriT features (all from Staphylococcus aureus). Experimental data suggested a successive model, since different retromobilization frequencies were observed between the small plasmids. More importantly, retromobilization was shown to be delayed by 50 and 150 min for pUB110 and pE194, respectively, compared to pXO16 conjugation. Natural liquid foods (cow milk, soy milk, and rice milk) were used to evaluate the putative ecological impact of these transfers. In cow and soy milk, conjugation, mobilization, and retromobilization were shown to occur at frequencies of 8.0 × 10−1, 1.0 × 10−2, and 1.2 × 10−4 transconjugants per recipient, respectively. These data are comparable to those obtained with LB medium and about 10-fold lower than in the case of rice milk. Taken together, these results emphasize the potential role of plasmid capture played by B. thuringiensis in natural environments.
Horizontal gene transfer can occur by three main mechanisms (12): phage transduction, transformation of naked DNA, and conjugation, a process whereby DNA is transferred from a donor to a recipient bacterium. It involves an intimate association between the interacting cells that is not susceptible to DNases present in the mating medium (4). After transfer, the recipient becomes a transconjugant, possessing the ability to start new rounds of conjugation. With the exponential number of prokaryotic genomes available, the remarkable extent to which horizontal gene transfer, and especially conjugation, has played an important role in bacterial evolution becomes increasingly apparent. Indeed, conjugation allows access to a large gene reservoir, with a well-known example being the propagation of antibiotic resistance genes (12).
Any DNA can be transferred by conjugation, but some DNA structures are specifically designed for conjugative transfer: the conjugative plasmids, the conjugative transposons, and the integrating conjugative elements. They have the genetic determinants, the tra genes, to establish the mating pair and the conjugative apparatus necessary for DNA transfer. This mechanism has been extensively studied in gram-negative bacteria, where intimate contact is established by extracellular filaments, the sex pili, anchored on the donor cell surface (16). A relaxase catalyzes the cleavage at the nic site of the origin of transfer (oriT) and remains covalently bound to the single-stranded DNA to form, with auxiliary proteins, a nucleoprotein complex called the relaxosome (13). This complex is transferred to the recipient cell by a type four secretion system (T4SS) (10). A coupling protein is required to make the contact between the relaxosome and the secretion machinery (26). It has been hypothesized that it participates in the active secretion of the substrate by “pumping” the single-stranded DNA strand from the donor into the recipient cell (32).
Recent studies have suggested that several components of the gram-positive conjugative systems share homologies with the gram-negative T4SS (1, 11, 37). However, differences between conjugation in gram-negative and gram-positive bacteria lie in the mechanism that conjugative plasmids use to establish cell-cell contact to initiate transfer. Whereas mating pair formation in gram-negative bacteria always implies sex pili, diverse surface components have been suggested to be involved in gram-positive conjugation. Recently, chemical communication systems have also been characterized in gram-positive bacteria and were shown to be involved in mating pair formation of integrating conjugative elements in Bacillus subtilis (8).
Besides conjugation, mobilization involves plasmids unable to promote their own transfer but that take advantage of a coresident conjugative plasmid to move intercellularly. Mobilizable plasmids typically carry the genetic determinants for their own DNA processing (a relaxase and, optionally, a coupling protein, encoded in the mob region). They require a conjugative element to establish an efficient mating pair between parental cells and to provide the DNA transport apparatus (25, 29, 32).
A third mechanism related to DNA conjugative transfer is retromobilization, i.e., the capture of recipient DNA by a donor harboring the conjugative plasmid. Captured DNA can be either chromosomal markers, in which case it is named “retrotransfer,” or mobilizable plasmids, in which case it is called “retromobilization.” Retromobilization was long subject to controversy because it went against the principle of unidirectional transfer of DNA. Two models were proposed to explain the transfer of a mobilizable plasmid from a recipient to a donor strain (36). The bidirectional model (one-step model) (Fig. 1A) postulates that retromobilization is a single event during which DNA molecules move freely between both partners. The name “concurrent model” refers to the fact that transfer of both plasmids takes place simultaneously. The unidirectional model (two-step model) (Fig. 1B) involves two transfer events, the first step being the transfer of the conjugative plasmid to the recipient and the second step corresponding to the transfer of the mobilizable plasmid from the recipient to the donor. This “successive” model proposes that retromobilization corresponds to two successive transfer events.
Schematic representation of retromobilization models (36). (A) Concurrent model: transfer of conjugative plasmid to the recipient occurs simultaneously with transfer of mobilizable plasmid to the donor via one conjugative event. (B) Successive model: a first conjugative event allows transfer of the conjugative plasmid to the recipient, while a second event allows transfer of the mobilizable plasmid to the donor, which could be the same as in the first event or another donor cell present in the medium. Large oval, conjugative plasmid; small circle, mobilizable plasmid; light line, recipient cell; bold line, donor cell; white arrow, transfer of indicated plasmid; black arrow, results of transfer(s).
Several studies have demonstrated that gram-negative retromobilization is mechanistically identical to conjugation and mobilization and that it follows a successive model of transfer (18, 19, 34). However, since gram-negative and gram-positive conjugation systems differ somewhat, so could retromobilization. So far, a single report on retromobilization in gram-positive bacteria has briefly described retromobilization of pUB110 by the conjugative plasmid p19 of B. subtilis at frequencies similar to those of mobilization (27). It should be noted that no retromobilization was observed with the conjugative transposons Tn916 and Tn925 during matings between B. subtilis and Bacillus thuringiensis (33).
The Bacillus cereus sensu lato group of spore-forming gram-positive bacteria contains six closely related species (30). Among these, B. cereus sensu stricto is an opportunistic pathogen mainly associated with food poisoning (31), B. thuringiensis is widely used as a biopesticide because of its ability to produce entomopathogenic delta-endotoxins (7), and Bacillus anthracis is the causative agent of anthrax (28). The other members of this group are Bacillus mycoides, Bacillus pseudomycoides, and Bacillus weihenstephanensis. Conjugation in the B. cereus sensu lato group was mostly studied with B. thuringiensis since insecticidal toxin genes are often carried by large conjugative plasmids (17). B. thuringiensis strains frequently harbor multiple other conjugative plasmids not encoding toxin genes, such as pXO11, pXO13, pXO14, pXO15, pXO16, and pAW63 (9, 39). Among them, pXO16 (∼350 kb) displays a specific aggregation phenotype during conjugation which involves a proteinaceous molecule on the cell surface but apparently no aggregation-inducing pheromone (2, 22). The conjugative system of pXO16 is very efficient, with frequencies reaching 100% in B. thuringiensis subsp. israelensis matings (23). pXO16 was also shown to be able to mobilize small “nonmobilizable” plasmids lacking both a mob gene and an oriT site (3). From Michaelis-Menten-like kinetics of pXO16 transfer, it was shown that this system is not only efficient but also fast (5), since the conjugative transfer of one plasmid molecule takes 3.5 to 4 min. Having donated the plasmid, the donor needs a period of recovery of about 10 min before it can redonate the plasmid. It takes 40 min for a new transconjugant to become able to function as a donor. Finally, pXO16 can promote its own transfer and mobilize small plasmids in natural environments, such as river water and dipteran larvae (35), dairy products (38), and the intestine of gnotobiotic rats (40).
In this study, the conjugation, mobilization, and retromobilization abilities of the conjugative plasmid pXO16 of B. thuringiensis subsp. israelensis were compared. The mob and oriT regions of mobilizable plasmids were analyzed for their effects on mobilization and retromobilization. The influence of the mating medium was explored by the use of different food matrices. Kinetics experiments were also performed to analyze in detail the different types of transfer and to get more insights into their mechanisms.
MATERIALS AND METHODS
Strains and plasmids.The strains and plasmids used in this study are reported in Table 1. In the present work, strain 4Q7 refers to the rifampin spontaneous mutant of 4Q7 and pXO16 refers to the Tn5401-tagged version of pXO16, resistant to tetracycline. pC194 was transferred by electroporation into 4Q7 to give strain DF001 (38). This plasmid does not possess any mob gene, and no oriT has been identified yet. The mobilizable plasmid pE194 was introduced by electroporation into B. thuringiensis subsp. israelensis GBJ002. All plasmids were transferred to the different strains by matings during this work to obtain the right combinations of strains and plasmids. The presence of the different plasmids was verified by PCR (see below).
Strains and plasmids used in this study
Culture media.LB medium was used for bacterial growth and mating experiments unless otherwise indicated. Its composition (g/liter) is sodium chloride (5), yeast extract (5), and tryptone (10). Solid LB agar plates were prepared with 1.4% agar. Antibiotics (Sigma) were added to the culture medium, when appropriate, at the following concentrations (μg/ml): streptomycin, 100; nalidixic acid, 15; tetracycline, 4; kanamycin, 50; rifampin, 50; chloramphenicol, 15; and erythromycin, 25. The different types of milk were UHT-sterilized full-cream cow milk (Delhaize, pH 6.7), sterilized soy milk (Alpro Soja, pH 7.4), and sterilized rice milk (Lima, pH 7.2). The sterility of these foods was tested by spreading 100-μl samples on LB plates. The absence of microorganisms was verified after 48 h of incubation at 30°C. Bacterial growth in the different food products was assessed every hour for 9 h at 30°C by plating appropriate dilutions on LB medium. The plate counts showed no significant growth differences among the four media (data not shown).
Mating experiments.Mating partners were streaked separately onto LB plates containing the appropriate antibiotic(s) for strain and plasmid selection and incubated at 30°C overnight. Separate precultures from single colonies were grown in 14 ml of LB medium, whole milk, soy milk, or rice milk (depending on the mating medium) without antibiotics and incubated overnight at 30°C without shaking. Precultures were performed in hermetically closed small glass bottles in which the volume of medium (14 ml) corresponded to more than 90% of the recipient volume. Equal amounts of mating partner cells (500 μl per optical density unit at 600 nm for the LB preculture, since it was not possible to measure the optical density in the different milks) from precultures were mixed, and mating medium was added to reach a final volume of 14 ml. The mixture was incubated at 30°C, without shaking, during the mating time (4 h, unless otherwise indicated). Precultures were plated at time zero as controls. After mating, appropriate dilutions were plated on selective media for parental and transconjugant strains and incubated for 24 to 48 h at 30°C. The transfer frequencies were calculated as the ratio of transconjugants per recipient cell at the end of the conjugation time. All results presented in the graphs represent mean frequencies of a minimum of four independent repetitions. It should be noted that in some cases, no transfer was detected, although transfer could have occurred below the detection limit of the system, which was 10−7 transconjugants per recipient. It is important to note that the mean frequencies reported in the graphs do not take into account either the “undetected” transfers or the statistically eliminated frequencies (see “Statistical analysis” below). Kinetic experiments were performed in the same way, but the dilutions were plated on selective antibiotics every 10 min for the first 2 h and every 15 min for the next 2 h. The resulting curves came from the means of the results of at least two repetitions.
PCR.Transconjugants were regularly tested by PCR for the presence of the plasmids. Each PCR sample of 20 μl contained 7.7 μl of distilled sterile water, 4 μl of buffer specific for Taq polymerase, 2 μl of deoxynucleoside triphosphate (2 mM), 2 μl of each primer (10 μM), 1.2 μl of the cofactor MgCl2 (25 mM), 0.1 μl of Taq polymerase, and 1 μl of the DNA to be amplified. The primer sequences used during this work are listed in Table 2. They correspond to specific sequences of the different plasmids being tested for. The primers for pXO16 amplify the Tn5401-borne tetracycline resistance gene.
Primer pairs used in this study
Statistical analysis.The frequencies obtained during the different repetitions were statistically analyzed by using Dixon's test (14). This test for extreme values is a convenient method to determine outliers from continuous data.
RESULTS
pXO16 mobilizes more efficiently pUB110 and pE194 than pC194.Preliminary mating experiments using a donor harboring the conjugative plasmid pXO16 and a cured recipient were performed in liquid LB at 30°C for 4 h without shaking (Fig. 2A). All the strains used in the course of this work were cured B. thuringiensis subsp. israelensis strains harboring different plasmid combinations. Under these conditions, biparental pXO16 conjugation was shown to occur at a mean frequency of 8.4 × 10−1 transconjugants per recipient (T/R), similar to what was reported by Jensen and collaborators. (23).
Schematic representations of biparental and triparental matings. Conjugation refers to the transfer of the conjugative plasmid, while mobilization and retromobilization refer to the transfer of the mobilizable plasmid. (A) Biparental conjugation: conjugation to the recipient (1). (B) Biparental mobilization: conjugation to the recipient (1), mobilization to the recipient (2), and transfer of both plasmids to the recipient (3). (C) Biparental retromobilization: conjugation to the recipient (1) and retromobilization to the donor (2). (D) Triparental matings: conjugation to the recipient (1) and to the helper strain (2), mobilization to the recipient (3), transfer of both plasmids to the recipient (4), and retromobilization to the donor (5). Large oval, conjugative plasmid; small circle, mobilizable plasmid; light line, recipient cell; bold line, donor cell; white arrow, transfer of indicated plasmid; “T”-base white arrow, transfer of both types of plasmid; black arrow, results of transfer(s).
It was also previously shown that pXO16 is able to mobilize nonmobilizable plasmids, i.e., plasmids defective for their mob and/or oriT regions (3, 38). Biparental mobilization experiments were then performed using two mob+ plasmids, pUB110 and pE194, and the mob-lacking plasmid pC194, all originating from S. aureus. At the end of the mating, three types of transfer can be distinguished, as shown in Fig. 2B. pXO16 transfer to the recipient occurred at mean frequencies of 7.1 × 10−1 T/R (Fig. 3A), similar to the frequencies obtained for biparental conjugation with pXO16 alone. This indicated that there was no effect of the presence of the mobilizable plasmid on the transfer of pXO16. Mobilization of the mob + plasmids pUB110 and pE194 occurred at mean frequencies of 5.0 × 10−2 and 2.5 × 10−2 T/R, respectively, while those obtained with the mob-lacking plasmid pC194 were 104times lower, at 1.6 × 10−5 T/R. Similar trends and frequencies were obtained for the transfer of these plasmids selected concomitantly with the conjugation of pXO16. The fact that the mobilization of pUB110 and pE194 was more efficient than that of pC194 indicated that the presence of typical features of a mobilizable plasmid, i.e., the mob and oriT regions, favors mobilization. However, it is important to note that pC194 mobilization still represented a significant amount of transfer.
Mating experiments were performed in LB medium with the conjugative plasmid pXO16. Results obtained with the mob + plasmids pUB110 and pE194 are shown in black and gray, respectively, while those obtained with the mob-lacking plasmid pC194 are shown in white. Each column corresponds to the mean of the results of a minimum of four independent repetitions and is expressed as the transfer frequencies (log T/R) after 4 h of mating. Error bars have a length of two standard deviations. Dotted columns mean that transfer was not detected in all repetitions. Transfer numbers on x axes refer to those used in Fig. 2. (A) Biparental mobilization, as depicted in Fig. 2B. (B) Biparental retromobilization, as depicted in Fig. 2C. (C) Triparental mating, as depicted in Fig. 2D.
pXO16 retromobilization, or plasmid capture, in biparental and triparental matings.Retromobilization can be seen as the capture of a mobilizable plasmid by a strain carrying a conjugative plasmid. Similar to the results for mobilization, the presence of the mob and oriT regions could affect frequencies. Therefore, retromobilization experiments were performed with mob + and mob-lacking plasmids in parallel. This is of particular interest since retromobilization was shown to be dependent on the presence of the oriT of the mobilizable plasmid in gram-negative species (19). In this type of mating, the conjugative plasmid pXO16 was present in the donor, while the recipient harbored the mob + plasmid pUB110 or pE194 or the mob-lacking plasmid pC194 (Fig. 2C). The results obtained after 4 h of matings at 30°C are illustrated in Fig. 3B. pXO16 conjugation to the recipient was not affected by the presence of the plasmid in the recipient, with mean frequencies of 5.0 × 10−1 T/R. pUB110 retromobilization was shown to occur at a rate of 1.7 × 10−4 T/R, while pE194 retromobilization displayed a 10-fold-lower frequency of 7.0 × 10−6 T/R. Interestingly, the mob-lacking pC194 plasmid could be retromobilized into the donor strain, albeit in only one of six experiments, at a frequency of 5.0 × 10−7 T/R (Fig. 3B). This single detection indicated that pC194 retromobilization occurred in other repetitions but that it generally took place at frequencies below the detection threshold of the system, which is 10−7 transconjugants per recipient.
In the environment, bacteria live in complex association, most often in biofilms, where they encounter several partners at a time. It was interesting to test more complex strain combinations, such as triparental matings. At the outcome of these matings, five different strain/plasmid combinations could be distinguished (Fig. 2D). Experiments were performed in parallel with pUB110 and pC194 in a “helper” strain, while the recipient strain was devoid of plasmid. In triparental matings, pXO16 conjugation to the recipient and to the helper strain displayed mean frequencies of 5.6 × 10−1 T/R, similar to the frequencies in biparental matings (Fig. 3C). Transfer of the mob + plasmid pUB110 to the recipient, simultaneously or not with pXO16, gave comparable frequencies of 8.0 × 10−4 T/R. These mobilization frequencies were ca. 102 times lower than the frequency obtained for biparental mobilization. Mobilization of pC194 alone to the recipient was detected only twice out of four repetitions, at a mean frequency of 3.2 × 10−6 T/R, while transfer of pC194 with pXO16 was shown to occur only once out of four repetitions, at a frequency of 4.3 × 10−6 T/R. pUB110 retromobilization was detected in each repetition at 3.5 × 10−5 T/R, which represented a 10-fold decrease compared to its frequency in biparental retromobilization. In contrast, pC194 retromobilization was not detected in any of the four repetitions. This does not mean that none occurred but rather that any transfer that did take place was below the detection limit of the system.
pUB110 capture by pXO16 in cow milk, soy milk, and rice milk.The ability of pXO16 to mobilize the mob-lacking plasmid pC194 was previously tested in milk and rice pudding (38), where it was shown that frequencies increased in milk, while they decreased in rice pudding. Moreover, retromobilization was only detected twice out of five experiments in milk and never in LB or in rice pudding. It was therefore interesting to test mobilization and, especially, retromobilization using a “genuine” mobilizable plasmid such as pUB110. The conjugation and mobilization capabilities of the conjugative plasmid pXO16 were then assessed in different types of “milk” used as mating media. The frequencies obtained in these foodstuffs were compared to those obtained in LB, a standard culture medium. Cow milk, soy milk, and rice milk were chosen for their relevance as potential contaminated foodstuffs, as well as for practical reasons, such as availability and ease of use.
Biparental conjugation (Fig. 2A) was the first to be tested in the different milks (Fig. 4A). The transfer of pXO16 to the recipient took place at similar frequencies of about 6.0 × 10−1 T/R in cow and soy milks, similar to what was observed in LB, in which the frequency was 8.0 × 10−1 T/R. A twofold reduction for pXO16 conjugation was observed in rice milk, with frequencies of 3.0 × 10−1 T/R. This is consistent with the previous observation that pXO16 conjugation was barely affected by the mating medium (38).
Mating experiments performed in LB (black), in cow milk (dark gray), in soy milk (light gray), and in rice milk (white) with the conjugative plasmid pXO16 and the mobilizable plasmid pUB110. Each bar corresponds to the mean of the results of a minimum of four independent repetitions and is expressed as the transfer frequencies (log T/R) after 4 h of mating. Error bars have a length of two standard deviations. Transfer numbers on x axes refer to those used in Fig. 2. (A) Biparental conjugation. (B) Biparental mobilization. (C) Biparental retromobilization. (D) Triparental mating.
The transfer of pXO16 alone to the recipient in biparental mobilization (Fig. 2B) displayed frequencies similar to those obtained in biparental conjugation experiments, with again a threefold decrease in rice milk (Fig. 4B). Similar to matings in LB medium, the frequencies obtained for pUB110 mobilization were similar with or without the concurrent selection of pXO16 conjugation. Indeed, the frequencies were about 1.0 × 10−2 T/R in cow and soy milk and 4.0 × 10−3 T/R in rice milk. However, mobilization seemed to be more affected than conjugation by the medium, since the frequencies were approximately four times lower in cow milk and soy milk than in LB, where the frequencies were about 4.0 × 10−2 T/R. More significantly, pUB110 mobilization displayed a 10-fold decrease in rice milk compared to its mobilization in LB medium.
Biparental retromobilization of pUB110 was also tested in foodstuffs (Fig. 2C). No differences were noticed for pXO16 conjugation compared to the results of biparental conjugation experiments in the different types of milks (Fig. 4C). pUB110 retromobilization frequencies were similar in LB, cow milk, and soy milk, with a mean of 1.2 × 10−4 T/R, but displayed a 20-fold decrease in rice milk compared to the results for LB, with frequencies of 5.8 × 10−6 T/R.
To evaluate what could happen with complex combinations of strains in foodstuffs, triparental matings (Fig. 2D) were also carried out in the different types of milk. Similar to the other matings, pXO16 conjugation to the recipient or to the helper strain in rice milk was observed at frequencies 5 to 10 times less than those in LB medium and cow and soy milks (Fig. 4D). Mobilization of pUB110, selected or not with pXO16 conjugation, displayed similar frequencies in LB and cow and soy milks, with frequencies of around 8.0 × 10−4 T/R. Interestingly, pUB110 mobilization in rice milk provided 50-fold decreased frequency compared to the frequencies in the other media, with a frequency of 2.0 × 10−5 T/R. pUB110 retromobilization to the donor strain in cow milk and in soy milk was shown to occur at frequencies similar to those in LB medium, with a mean of 3.0 × 10−5 T/R. In contrast, retromobilization in rice milk was 10-fold decreased, with a mean of 5.0 × 10−6 T/R.
Kinetics study of mobilization and retromobilization.Four hours can be considered a long time of mating. Therefore, kinetics experiments were performed to compare the processes of conjugation, mobilization, and retromobilization. Transfer of pXO16 was already detected after 10 min and reached a maximum after 40 to 50 min. Frequencies remained around 6.5 × 10−1 T/R until 4 h (data not shown). This confirmed previous kinetics studies performed with this plasmid where it was shown that pXO16 conjugation is efficient and fast (23).
Biparental mobilization (Fig. 2B) was also evaluated over time. The transfer of pXO16 to the recipient followed the same trend as in biparental conjugation, with a maximum frequency of transfer of 6.0 × 10−1 T/R after 40 min (Fig. 5A). Mobilization of pUB110 to the recipient, in association or not with pXO16 conjugation, followed an exponential curve, with a plateau reached after approximately 135 min at a frequency of about 5.0 × 10−2 T/R. Mobilization, like conjugation, was already detected after 10 min of matings.
Kinetics experiments performed with the conjugative plasmid pXO16 and the mob + plasmid pUB110. Each mark corresponds to the mean of the results of a minimum of two independent repetitions and is expressed as the transfer frequency (log T/R) at the corresponding time. Type(s) of transfer is indicated as follows: conjugation to the recipient (○) or to the helper strain (•), mobilization to the recipient (▪), transfer of both plasmids to the recipient (▴), and retromobilization to the donor (□). (A) Biparental mobilization with pUB110. (B) Biparental retromobilization with pUB110. (C) Biparental retromobilization with pE194. (D) Triparental mating with pUB110.
During biparental retromobilization with pUB110 (Fig. 2C), transfer of pXO16 to the recipient was also detected after 10 min of mating and reached a maximum after 40 min, with a frequency of 7.0 × 10−1 T/R (Fig. 5B). In contrast to what was observed in biparental mobilization, pUB110 retromobilization was only detected after 60 min of matings. However, retromobilization also displayed an exponential curve, a plateau being reached after 150 min of mating with 1.6 × 10−4 T/R. With pE194 (Fig. 5C), the delay between conjugation and retromobilization was increased up to 150 min and pE194 retromobilization reached a maximal frequency of 4.0 × 10−5 T/R.
Finally, triparental matings (Fig. 2D) were also assessed by kinetics experiments (Fig. 5D). pXO16 conjugation to the recipient and to the helper strain was shown to follow the same trend as the biparental conjugation kinetics. The plateau was reached after 50 min, with a frequency of 5.4 × 10−1 T/R, remaining stable until the end of the experiment. Mobilization of pUB110 alone to the recipient appeared after 70 min of mating, and its mobilization was detected after 80 min when associated with pXO16 transfer. After 4 h of mating, mobilization of pUB110 displayed frequencies of about 3.0 × 10−4 T/R that remained stable until 24 h (data not shown). Retromobilization of pUB110 to the donor followed a curve and frequencies similar to those observed for its mobilization to the recipient.
DISCUSSION
Conjugation, mobilization, and retromobilization participate in the evolution of bacteria. The fact that bacteria can not only spread genes by conjugation or mobilization but can also capture DNA from other bacteria gives force to the ecological significance of gene transfer. Retromobilization was first reported in gram-negative bacteria (for a review, see reference 6), where it was long subject to controversy concerning its mechanism. A molecular study of retromobilization in gram-negative bacteria leads to evidence that retromobilization follows the successive model (18, 19, 34). In the course of the present study, experiments were performed to identify the retromobilization mechanism in gram-positive species by using the conjugative plasmid pXO16, isolated from B. thuringiensis subsp. israelensis. This plasmid displays interesting features, such as fast and efficient transfer frequencies, a specific aggregation phenotype, and the ability to promote the transfer of nonmobilizable plasmids (2, 3, 4, 22, 23).
Compelling experimental evidence strongly suggests that pXO16 retromobilization follows a successive model of transfer, as was previously demonstrated in gram-negative species (18, 19, 34). The fact that pXO16 retromobilization frequencies, with the three small plasmids, were lower than pXO16 conjugation frequencies gave a first indication for a successive model of retromobilization. Indeed, if the concurrent model, which predicts a bidirectional flux of DNA between parental strains (36), was the case, conjugation and retromobilization frequencies would have been similar. Another interesting observation leading to the successive model of retromobilization is the fact that retromobilization frequencies were different depending on the small plasmid used. Indeed, retromobilization of mob-lacking pC194 was detected, not in all repetitions, at frequencies about 10 and 102 lower than those obtained with the mob + pE194 and pUB110, respectively. These results indicated that the presence of mob and oriT regions has a positive effect on retromobilization, although this effect varies depending on the plasmid. Results previously obtained for gram-negative bacteria have shown that retromobilization could not be detected with plasmids lacking mob (19). The high transfer frequencies displayed by pXO16 were determinant in demonstrating that even mob-lacking plasmids, such as pC194, could be captured, at least in biparental matings.
In gram-negative species, the best evidence that supported a two-step model of retromobilization was the observation of a 30- to 90-min delay between conjugation and retromobilization (19). In mobilization kinetics, transfer of the mob + plasmid pUB110 was detected immediately after the beginning of the mix of the parental cells. In contrast, a delay of approximately 50 min was observed between the detection of pXO16 conjugation and pUB110 retromobilization. This 50-min delay can be seen as the sum of several steps: (i) 4 min for the transfer of pXO16, (ii) 40 min for this new transconjugant to become a donor, and (iii) several minutes for the transfer of pUB110 to the donor. However, while this perfectly matches pUB110 kinetics, it is interesting to note that the delay was increased up to 150 min for pE194 retromobilization, indicating that plasmid capture is influenced by the type of mobilizable plasmid used.
In triparental experiments with pUB110, both mobilization and retromobilization were delayed by ca. 50 to 60 min compared to the time for conjugation. This can be explained by the fact that the conjugative plasmid, pXO16, is not coresident with the mobilizable plasmid at the beginning of the experiment. This is therefore good evidence for the successive model, since two steps are required to get mobilization and retromobilization: (i) transfer of the conjugative plasmid to the recipient and (ii) mobilization and retromobilization to the recipient and to the donor, respectively.
Contrary to pXO16 conjugation, which reached nearly 100%, mobilization and retromobilization were shown to be more controlled processes. Indeed, in kinetics experiments, their transfer frequencies were shown to reach a maximum after 135 to 150 min, no matter the transfer type or the type of mating (bi- or triparental). Moreover, pUB110 mobilization starts earlier in biparental experiments, leading to increased transfer frequencies compared to those of triparental matings and retromobilization in general. It is therefore possible that mobilization and retromobilization and possibly the aggregation phenotype are regulated by cellular density through a quorum sensing-like process.
The characteristics of the mating medium in pXO16 and pUB110 transfer were explored by the use of different food matrices, namely cow milk, soy milk, and rice milk, and compared to those of standard LB broth. Conjugation, mobilization, and retromobilization were detected in all food types, emphasizing their possible ecological impact from contributing to the transfer and accumulation of harmful genetic traits, such as virulence or antibiotic resistance genes, among foodborne bacteria. It is important to keep in mind that bacteria such as B. thuringiensis are used as biopesticides on crops and could therefore be recovered in the food chain (15).
Transfer frequencies were shown to be affected by the food medium used, with similar frequencies in cow milk and soy milk and lower transfer frequencies in rice milk. The reason for these differences between the matrices could be related to their physical properties (texture and fluidity) or to their composition, which could interfere with either the cell contact (aggregation) or cell communication (e.g., quorum sensing). A detailed study of the influence of the matrix type on plasmid conjugation is under way (P. Modrie et al., unpublished data).
ACKNOWLEDGMENTS
We acknowledge Lars Andrup for providing the pE194 plasmid.
This work was supported by the European Space Agency (MISSEX, AO-2004, Prodex C90255), the FRIA (Fonds pour la Formation à la Recherche dans l'Industrie et dans l'Agriculture) (Grant to P.M.), the National Fund for Scientific Research, and the Université Catholique de Louvain.
FOOTNOTES
- Received 5 December 2008.
- Accepted 19 January 2009.
- Copyright © 2009 American Society for Microbiology
