Pathogenic Escherichia coli cause a wide range of infectious diseases in various animal species and humans, including diarrhoea [1], meningitis, urinary tract and, soft tissue infections [2]. Many E. coli produce extended-spectrum beta-lactamase (ESBL)-enzymes, which- in addition to penicillins and others- hydrolyse newer, third-generation cephalosporins and monobactams [3], and limit antimicrobial therapy. Several ESBL-gene families (bla_CTX-M, bla_SHV, bla_TEM and bla_OXA) are encoded on plasmids [4]. Prior research has demonstrated that some of the ESBL-carrying plasmids influence factors other than resistance, like the ability of Klebsiella strains to invade epithelial cells [5]. In these ESBL-plasmids, addiction models such as toxin-antitoxin (TA) systems have been described [6], which prevent the cell from “post-segregational killing”, therefore ensuring the maintenance of the plasmid in the bacterial cell during replication [7]. Several TA systems have been discovered in Gram-negative and Gram-positive bacteria differing basically in functionality and type of the antitoxin. Type I system antitoxins are small antisense RNA molecules, which mostly inhibit toxin mRNA translation or degrade toxin mRNA while type II system protein antitoxins interact post-translationally with protein toxins. Other systems are type III to type V TA systems [8–10]. One of the first [11] and most studied is type I hok/sok TA system in E. coli. Following replication, in plasmid-free daughter cells type I unstable RNA antitoxin molecules (e.g. sok) degrade rapidly, while stable toxins (e.g. hok) induce cell membrane porins, therefore impairing ATP synthesis and subsequently causing bacterial cell death [12]. In ESBL-plasmids most frequently represented systems not only include hok/sok but also pemK/I and ccdA/B, which seem to be associated with CTX-M-15 and CTX-M-9 encoding plasmids of IncF replicon type [13]. To investigate, which important role ESBL-plasmids play concerning both resistance and factors not related to resistance it is necessary to construct ESBL-plasmid-“cured”-variants (PCVs) and compare those pheno- and genotypically to ESBL-wild-type strains. Different methods are known to “cure” bacterial plasmids, most of them use chemical treatment like ethidium bromide or acridine orange in different concentrations added to bacteria in Luria-Broth (LB) [14]. These methods often involve the problem that “curing” the ESBL-plasmid of the wild-type strain causes the death of the bacterium due to the loss of an operating TA system. In this study we “cured” TA systems-containing ESBL-plasmids from wild-type E. coli testing an established method [15] using enhanced temperatures, brain-heart-infusion (BHI) medium and several weeks of continuing sub-cultivation protocols.

As ESBL-plasmid-“curing” of the strains applied in this study did not work treating bacteria with ethidium bromide or acridine orange a previously described method [15] using a heat technique was performed. This method was adequate to construct viable PCVs and to our knowledge we are the first ones to describe a successful “curing” of ESBL-plasmids, which carry genes for toxin-antitoxin systems. The results are surprising because one might think that bacteria cannot survive without this plasmid. Strains used in this study all had at least one complete ESBL-plasmid-encoded toxin-antitoxin system, including hok/sok, srnB/C, vagC/D and pemI/K, which were encoded on plasmids of replicon types FIA or FIB carrying bla_CTX-M-9, bla_CTX-M-14 and bla_CTX-M-15 or combinations of these three ESBL-genes. The addiction systems and replicon types found have been characterized in ESBL-producing E. coli before and certain combinations of ESBL-enzyme, addiction systems and replicon type might contribute to the success and spread of multi-resistant E. coli strains [13]. An association of CTX-M-15 with FIA and CTX-M-9 with FIB plasmids has been described before and the same was true for the isolates used in this study (Table 1). Mnif et al. also observed that the occurrence of CTX-M-14 correlated with FII replicons, however, FIB replicon type plasmids carrying CTX-M-14 have also been found in this earlier study [13]. No correlation for CTX-M-14 and FII replicons was observed in this study, as this ESBL-type was also present on FIA/FIB plasmids (Table 1). This might be partly due to the fact that bla_CTX-M-14 was only detected in combination with other CTX-M enzymes (Table 1). So why did bacteria, whose TA system-containing plasmid was “cured” using a heat technique, survive? One might speculate that some toxins degenerate irreversibly above certain temperatures or that the TA genes are not expressed at 45°C. Or is it that continuing soft stimulation using high temperatures allows a slow adaption to changing environmental conditions and therefore “curing” of the plasmid, while chemicals, like ethidium bromide, together with the loss of a TA system mean too much stress for bacteria? Pulsed-field gel electrophoresis, plasmid-profile-analysis and bioinformatic analyses confirmed that the genetic change is restricted solely to the loss of an ESBL-plasmid and that genetic identity of PCV and wild-type strain remains given. Constructed PCV strains might be important tools to investigate the influence of an ESBL-plasmid on its bacterial host.