A key stage in determining the phenotype(s) conferred by a plasmid is its displacement, or ‘curing,’ to create a plasmid-free strain. However, many plasmids are very stable, not only because they contain multiple replicons, but also because they can encode post-segregational killing systems that reduce the viability of plasmid-free segregants. We have developed an efficient curing strategy that involves combining key regions of the replicons and the post-segregational killing loci into an unstable cloning vector carrying sacB, which confers sensitivity to sucrose. Targeting plasmids of both the F family of Escherichia coli and the broad-host-range IncP-1 family, we demonstrated displacement of susceptible resident plasmids from all clones tested. Growth on sucrose allowed the isolation of many clones without either plasmid. This strategy is highly efficient and avoids the stress of inducing and surviving the effects of post-segregational killing systems or other lethal gene products.

Introduction

Many bacteria carry a variety of mobile genetic elements that can contribute significantly to their diversity and adaptability (1). Although current DNA sequence databases allow predictions about function based on bioinformatics, it is still important to experimentally test the contribution of plasmids to the phenotype of their host. To do this, one must (i) obtain a plasmid-free segregant that has lost the plasmid and determine what properties have changed, and (ii) transfer the plasmid to a new strain and determine what new phenotype(s) is acquired (2). Important for this empirical work is the displacement of the plasmid from its host, a process called curing. While some plasmids are naturally unstable, many are very persistent and active curing strategies are needed to obtain plasmid-free bacteria.

The classic strategy to cure a strain of its resident plasmid is to stress the host bacteria in some way, such as by growth at high temperatures or in the presence of detergent, mutagens, or other DNA-modulating agents (2). Unfortunately this may also favor mutation of the chromosome irrespective of whether the plasmid has been displaced, thereby seriously undermining any conclusions (although this issue does not seem to have been systematically investigated). Alternatively, plasmid incompatibility—arising when two plasmids carry related replication or stable inheritance functions—can be used to displace the resident plasmid (3). Displacement can be unidirectional if one of the two plasmids contains a second unrelated replicon, especially if it has a naturally higher copy number (4). If plasmid elements (e.g., copy number regulators or replication origins) are cloned into a compatible vector, they can block replication of the endogenous plasmid. This is a powerful approach that is also the basis of a general plasmid classification system (5).

However, a major barrier to curing is that many plasmids encode what are called post-segregational killing systems (PSKs) (6). Killing of plasmid-free segregants happens because the plasmid leaves behind a toxin that becomes active after loss of the plasmid. Known systems encode either a slowly activated mRNA which encodes a toxin, or a protein toxin itself. Production of active toxin is prevented while the plasmid is present, either by regulation of translation of the toxin-encoding mRNA with antisense RNA or by antidote proteins. In both cases, the regulator or antidote is unstable and so rapidly disappears once the plasmid that encodes them is no longer present. If the bacteria survive loss of the plasmid, they may have an increased chance of carrying mutations because some PSKs target DNA gyrase (7,8)—inducing the SOS response—causing elevated error-prone repair and mutation rate. Therefore, to improve a plasmid displacement strategy based on incompatibility, we proposed that the curing vector should be supplemented with the plasmid's PSK system's antitoxins or anti-sense RNA genes. This point does not seem to have been generally recognized in the design of plasmid displacement strategies to date, although while this manuscript was being prepared, a paper that implicitly covers this point with respect to the Agrobacterium tumefaciens Ti plasmid was published (9).

Using this principle, we devised a general strategy to achieve efficient plasmid displacement and apply it to displace F-like and IncP-1 plasmids from their hosts. pO157 is a typical F-like plasmid, possessing multiple replicons and stable inheritance functions (10,11). Previously pO157-free strains have been obtained by incompatibility with a mini-replicon derived from just part of the pO157 genome, but curing was only successful in a proportion of the transformants (12). By contrast, our approach results in highly efficient displacement and can be extended to other F-like plasmids. We have also developed a similar strategy to cure IncP-1 plasmids (13,14) demonstrating the general applicability of this strategy to other groups of plasmids.

Materials and methods

Bacterial strain, plasmids, and growth conditions

E. coli K12 strains used were DH5α (15), S17–1 (16), NEM259 (17), and JM109 (18). E. coli O157:H7 Sakai strain stx (toxin deficient) (12) was used to reduce health risks. Plasmids used or constructed during this work are listed in Supplementary Table S1. The standard medium was Luria Bertani (LB) broth, or L-agar (LB solidified with 1.5% w/v agar) and growth was at 37°C. Antibiotics used were kanamycin, 50 µg/mL; penicillin, 150 µg/mL (broth) or 300 µg/mL (agar); streptomycin, 30 µg/mL; and trimethoprim, 100 µg/mL.