A New and Improved Host-Independent Plasmid System for RK2-Based Conjugal Transfer

Bacterial conjugation is a process that is mediated either by a direct cell-to-cell junction or by formation of a bridge between the cells. It is often used to transfer DNA constructs designed in Escherichia coli to recipient bacteria, yeast, plants and mammalian cells. Plasmids bearing the RK2/RP4 origin of transfer (oriT) are mostly mobilized using the E. coli S17-1/SM10 donor strains, in which transfer helper functions are provided from a chromosomally integrated RP4::Mu. We have observed that large plasmids were occasionally modified after conjugal transfer when using E. coli S17-1 as a donor. All modified plasmids had increased in size, which most probably was a result of co-transfer of DNA from the chromosomally located oriT. It has earlier also been demonstrated that the bacteriophage Mu is silently transferred to recipient cells by these donor strains, and both occurrences are very likely to lead to mutations within the recipient DNA. Here we report the construction of a new biological system addressing both the above mentioned problems in which the transfer helper functions are provided by a plasmid lacking a functional oriT. This system is compatible with all other replicons commonly used in conjugation experiments and further enables the use of diverse bacterial strains as donors. Plasmids containing large inserts were successfully conjugated and the plasmid modifications observed when E. coli S17-1 was used as donor were eliminated by the use of the new host-independent vector system.


Introduction
Due to its well established genetics and good transformation competence Escherichia coli is the most frequently used host for manipulation of DNA via a variety of recombinant DNA technologies. After the modifications have been made it might be necessary to transfer the constructs designed in E. coli to alternative hosts at high frequencies. This becomes relevant for example during construction of large numbers of transposon insertion mutants or for transfer of metagenomic libraries in functional screening studies across species barriers. Transformation of naked DNA is often inefficient, or sometimes even impossible, depending on the host of interest. The use of conjugation often solves these problems as the transfer system is mainly acting in a recipient-independent manner [1]. While the recipient-independency is an attractive feature, there also exist limitations due to the requirements of complex machinery and also due to protection systems in recipient cells, such as CRISPR and restriction-modification [2,3]. The E. coli strain S17-1 and its analogue SM10 are heavily used as donor strains in such transfer procedures, which is reflected by a very high citation frequency (nearly 5000 as of October 2013) of the paper in which these strains are described [4].
E. coli S17-1/SM10 contain a chromosomally integrated RP4 plasmid, which is essentially the same as the more studied broadhost-range self-transmissible IncP plasmid RK2 [5]. Conjugal transfer of plasmids based on this system requires the presence of an origin of transfer (oriT) in the plasmid to be transferred, as well as the gene products of two separate tra-clusters [6] which are provided in trans from the RP4 integrated in E. coli S17-1. A number of small and specialized oriT-containing vectors have been developed from the large RK2 plasmid (60 kb) [7], but other types of plasmids containing oriT may also be conjugated by E. coli S17-1/SM10 [8,9].
In spite of their extensive use there are several problems associated with the E. coli strains S17-1 and SM10: They both contain an active bacteriophage Mu genome (within the tetracycline resistance gene of RP4) which has been shown to mobilize itself into recipient strains [10,11]. This may cause problems as Mu DNA may randomly mutate the recipient genome and/or the transferred plasmid. Another demonstrated problem is that these strains not only mobilize oriT-carrying plasmids, but also their own chromosomal DNA to recipient strains at frequencies of 10 24 per donor cell [12]. Furthermore, there are also reports describing generation of plasmid modifications of unknown nature in conjugation experiments involving E. coli strains S17-1/SM10 [13,14].
In addition to these findings we here report that plasmids transferred from E. coli S17-1 to other bacterial species quite often become modified by insertion of DNA from the donor host chromosome, presumably as a result of mobilization of DNA from the active oriT within the inserted RP4. This represents a rather serious problem as it very likely can lead to inactivation of genes in such transferred plasmids. There have been established alternative conjugation systems which address some of the above mentioned problems separately, such as a modified E. coli S17-1 strain in which the Mu genome has been inactivated [11]. In this study we present a new and improved system for conjugal transfer of mobilizable plasmids which overcomes both the problems of bacteriophage Mu and chromosomal DNA mobilization from the donor. This system is constructed in a way that all the functions required for conjugal transfer are present on a broad-host-range (RK2-compatible) plasmid, a feature that allows the use of diverse bacterial hosts as donors for conjugation of oriT-containing constructs.

Bacterial strains, plasmids and growth media
The bacterial strains and plasmids used in this study are described in Table 1. The fosmids used in the conjugation experiments are from a previous study and were constructed using the broad-host-range cloning vector pRS44 [15]. This vector harbours two replicons: ori2, leading to single plasmid copy in E. coli; and oriV, leading to high copy-number in strain EPI300 when induced. This strain harbours a gene encoding a high copynumber replication protein for oriV, TrfA, which is under a tightly regulated inducible promoter on its chromosome. Upon induction the vector copy-number increases from a single copy-to high copy-number as replication then occurs via oriV [16].
The growth media used were Lysogeny Broth (LB, 5 g yeast extract, 5 g NaCl and 10 g tryptone per litre) and Lysogeny Agar (LA, LB supplemented with 20 g agar per litre) for E. coli strains, LB and Difco Pseudomonas Isolation agar (PIA) for Pseudomonas fluorescens, and Yeasy Mold (YM) broth and YM agar for Xanthomonas campestris. Antibiotics were used at the following concentrations when relevant: ampicillin, 100 mg mL 21 (E. coli); chloramphenicol, 12.5 mg mL 21 (E. coli), 30 mg mL 21 (X. campestris); kanamycin, 50 mg mL 21 (E. coli and P. fluorescens); tetracycline 10 mg mL 21 (E. coli), 15 mg mL 21 (X. campestris) or 25 mg mL 21 (P. fluorescens). Expression of trfA (wt, i.e. low copynumber) from PmG5 in the recipient P. fluorescens and X. campestris strains was induced by addition of m-toluate at 0.5 mM, enabling replication of the transferred fosmids from oriV. Counter selection of sacB containing strains was done using sucrose at 5% (w/v). Clones in E. coli EPI300 were switched from single copy-to high copy-number by L-arabinose induction at 0.01% (w/v).

Standard DNA manipulations and conjugative matings
Routine DNA manipulations and agarose gel electrophoresis were performed according to the methods of Sambrook and Russel [17], or by using commercially available kits. DNA sequencing was performed using the Big Dye Terminator version 1.1 Cycle Sequencing Kit (Applied Biosystems). Transformations of E.coli DH5a, ER2566 and EPI300 were performed according to the RbCl transformation protocol (New England BioLabs) or through electroporation according to Sambrook and Russel [17] (13 V cm 21 , 200V, 25 mF).
Conjugative matings were performed as follows: cells from 2 mL exponential phase growing cultures (OD ,0.4) of donorand recipient strains were mixed, concentrated after centrifugation and deposited onto LA without antibiotic selection (30uC, overnight). The mixtures were then plated on appropriate selective media and incubated at 30uC for 48 h, for P. fluorescens or 72 h for X. campestris.
Plasmids were isolated from cultures of P. fluorescens and X. campestris using commercial plasmid isolation kits and isopropanol precipitation, and retransformed into E. coli EPI300 by electroporation.

Homologous recombination for targeted deletions within RK2
For inactivation of oriT, E. coli ER2566 cells (recA + ) containing plasmid RK2 were first transformed with pTA16, followed by selection of chloramphenicol-resistant transformants. As pTA16 cannot replicate extra-chromosomally in this strain, these represented cells in which recombination had happened between either the oriT-1 or the oriT-2 fragment. The cells were first cultivated in the presence of chloramphenicol overnight, then, after reinoculation (0.5% overnight-culture to fresh media), for at least 6 hours in the absence of selection. Different dilutions were next plated on LA plates containing sucrose, selecting for cells not containing sacB, i.e. where also the second cross-over had occurred (at the remaining oriT fragment). Correct alteration was confirmed through PCR reactions and sequencing, and in addition the new RK2 derivative (pTA17) was confirmed being conjugation deficient (due to inactivation of oriT).
Construction of the kanamycin sensitive derivative of pTA17 (pTA19) was performed similarly as described for pTA17, but with the use of pTA15 instead of pTA16.

Construction of pTA-Mob
The mobilization plasmid pTA-Mob was constructed from pTA19 by first removing the AseI-AvrII fragment (9.4 kb) containing the replication origin, oriV, and the ampicillin and tetracycline resistance genes. An AseI-AvrII fragment containing the broad-host-range replicon pBBR1, as well as a gentamycin resistance gene was next ligated into the same restriction sites. This pBBR1rep-Gm r fragment was amplified from plasmid pBBR1MCS-5 [18] in two PCR reactions using the following primer pairs (restriction sites are underlined): Gent-Fwd:

Results and Discussions
Detection of plasmid-modifications after conjugal transfer from E. coli strain S17-1 to Pseudomonas fluorescens and Xanthomonas campestris In an extension of the previously reported inter-species transfer experiments with fosmids from a metagenomic library [15], we discovered that some of the fosmids had increased in size after being conjugally transferred to P. fluorescens and X. campestris (see Materials and Methods for information about the fosmids). For analyses of the plasmids within transconjugants of P. fluorescens and X. campestris, the plasmids needed first to be retransformed into E.
coli EPI300 in order to obtain high quality plasmid DNA. This step also ensures that the analysis is performed on only one fosmid. Fosmid preparations of the transferred/retransformed fosmids were then digested with restriction endonuclease HindIII and the resulting fragments were separated by agarose gel electrophoresis. The digestion patterns were then compared to the corresponding results obtained from the original plasmid preparations from the E. coli donor strains (Figure 1). Lanes 1-3 show a case where the restriction fragment band patterns for a randomly selected plasmid, designated 62, remain the same after conjugal transfer both to P. fluorescens and X. campestris. However, for another plasmid, designated 83, the restriction fragment band patterns were altered after being conjugated to P. fluorescens (lanes 5-7) (original fosmid band pattern in lane 4). Of the plasmids obtained after conjugation of fosmid 83 to X. campestris, one appeared very similar to the original fosmid in the E. coli donor (lane 8), while the band pattern originating from another X. campestris transconjugant was clearly different (lane 9). A final example is illustrated by  S17-1(lpir) lpir lysogen of strain S17-1 [21] Psedomonas. fluorescens

NCIMB 10525
Pseudomonas fluorescens wild type NCIMB transfer of a fosmid designated 37 (lane 10 shows the original fosmid). No obvious differences could be observed for the fosmid after transfer to P. fluorescens (lane 11), but the fosmids in lanes 12 (from P. fluorescens) and 13 (from X. campestris) were clearly different from the one in the donor. Notably, these two lanes display indistinguishable restriction band patterns even though the plasmids had been transferred to two different species, indicating that the modifications were originating from the donor. The experiments described above revealed that fosmid modifications occasionally occur for a given fosmid, that each individual transfer event can lead to a different outcome for a given fosmid, that passage through two different hosts can apparently lead to the same modification, and finally that the modifications always involve increase in the fosmid size.

Analysis of the inserted DNA
To characterize the nature of the apparently inserted DNA, ten such altered DNA bands were excised from the agarose gel, purified, ligated into pLitmus28 and end-sequenced. Three of the obtained ten sequences did not give any significant hits against the public databases when BLAST analysis was performed, indicating that these were originating from the metagenomic insert DNA. Analyses of the remaining seven sequences revealed that four of these originated from the E. coli donor strain chromosome, all within a region of ca. 35 kb, whereas the last three fragments were from a gene within the Tn7 transposon. Historically the Tn7 was used to inactivate the kanamycin resistance gene within RP4 during construction of the conjugation E. coli donor strain S17-1 [4].
Based on these findings it appeared that the fosmid modifications were results of insertions of chromosomal DNA from the donor into the fosmids. We suspected that the inserted chromosomal DNA was located near the integrated RP4, and that the insertions do occur as a consequence of co-activation of the chromosomally located oriT element (within RP4) during conju-gation. As the fosmid vector used (pRS44) contains several elements from the RK2 plasmid, including oriV, oriT and parDE, it seemed possible that homologous recombination may occur between fosmid-and RP4 sequences. Correspondingly, the Tn7sequences could possibly originate from the Tn7-transposon inserted into the kanamycin resistance gene of the integrated RP4, alternatively partly due to Tn7 transposition. As E. coli S17-1 is recA 2 , homologous recombination should not occur within this donor host. However, it appeared fully feasible that the fosmid and DNA mobilized from the chromosomally located oriT could be cotransferred to the same recipient cell, and that homologous recombination could occur between these single-stranded molecules, given that the recipient was recA + . Such a hypothesis may explain how DNA fragments originating from the E. coli chromosome are integrated into the fosmid clones.

Construction of a new mobilization system for transfer of any plasmid containing oriT
Given that the observed modification problems occur as a result of DNA-mobilization from oriT in the E. coli S17-1 chromosome, inactivation of oriT within the donor strain should eliminate the problem as this would block mobilization of the chromosomal DNA (a view also described by Babic and co-authors [12]). However, we also sought to eliminate the possibility of Mu genome mobilization as well as other reported E. coli S17-1/SM10-related problems, and in addition we wanted to circumvent the restriction of using one particular strain as donor. A plasmid-based mobilization system that provides the tra-gene products for mobilization of oriT RK2 -containing vectors, which itself does not contain an intact oriT, could solve these issues. The most obvious strategy to achieve this appeared to be the substitution of the replication functions in an oriT-inactivated version of RK2 with another replication system. To avoid incompatibility problems with the plasmids to be transferred, the replicon of this new plasmid should satisfy three criteria: (i) it should not belong to the Figure 1. Agarose gel electrophoresis analysis of HindIII-digested fosmid clones before and after conjugation from E. coli S17-1 to P. fluorescens::TnRS48 or X. campestris::TnRS48. Lane 1, plasmid 62 before transfer, and lanes 2 and 3 after transformation to E. coli from P. fluorescens and X. campestris, respectively. Lane 4, plasmid 83 before conjugal transfer, and after transformation to E. coli from P. fluorescens (lanes 5-7) and X. campestris (lanes 8 and 9). Lane 10, plasmid 37 before transfer, and after transformation to E. coli from P. fluorescens. (lanes 11 and 12) [18,19].
In order to make precise modifications of the large-sized RK2 plasmid (60 kb), a system was developed that allows for homologous recombination between ligated fragments within a suicide plasmid (pTA10) and the target region of RK2, with subsequent selection of the altered RK2-derivatives (see Materials and Methods). This system was used to make two deletions within the RK2 plasmid, resulting in inactivation of the oriT site as well as the kanamycin resistance gene, generating the plasmid pTA19. Further, a fragment in pTA19 containing the replication origin, oriV, and the ampicillin and tetracycline resistance genes was replaced with a fragment containing the pBBR1 replicon and the gentamycin resistance gene (giving pTA-Mob, Figure 2).
Plasmids up to 220 kb in size were successfully transformed into E. coli DH10B/pTA-Mob cells, demonstrating that very large plasmids are maintained together with the relatively large-sized pTA-Mob. Next, oriT-containing plasmids were conjugated from E. coli DH10B/pTA-Mob cells using P. fluorescens as recipient, and the relatively easy attainment of transconjugants confirmed that the RK2 Tra functions were still intact. To test the system with respect to the described problem concerning modifications of conjugatively transferred plasmids, fosmid clones 37 and 83 were analysed after mobilization from E. coli DH10B/pTA-Mob. Plasmids from 11 different P. fluorescens transconjugants were digested with HindIII as described above (Figure 1), and the DNA fragments generated were analysed by agarose gel electrophoresis analysis ( Figure 3). As it can be seen, the restriction band patterns of plasmids originating from transfer experiments with fosmids 37 (three examples are given in lanes 2-4) and 83 (lanes [6][7][8] were all similar to that of the original fosmids (lanes 1 and 5).
Based on these findings we conclude that the established broadhost-range mobilization plasmid pTA-Mob is suitable for supporting in trans the necessary tra-genes for conjugation of oriT-containing plasmids from presumably a large number of bacterial species, and that its use also leads to elimination of all known problems associated with the heavily used E. coli S17-1/ SM10 strains. Since pTA-Mob is a broad-host-range plasmid, non-E. coli hosts may also potentially be used as donors for conjugation, given that the plasmid can replicate in these host(s) [18]. We have for instance previously shown that conjugation from E. coli is inefficient at low temperatures [20], indicating that it might be difficult to conjugate plasmids from E. coli to strictly psychrophilic species. With this new mobilization system this can potentially be solved through the use of a pTA-Mob containing ''intermediate host'' which handles the temperature-requirements of both strains. Such a strategy could also be used in any case where mixing of the host of interest with E. coli cells causes some form of problem. We also envision that the new system might be used as a model in environmental microbiology studies involving gene transfer between various genus/species.