2.2. Susceptibility of E. coli and E. amylovora towards antimicrobial compounds
The susceptibility of the wild types and respective single mutants with deletions in
acrB or
tolC of
E. coli and
E. amylovora to a variety of antimicrobial compounds were examined in complex and minimal medium using the determination of minimal inhibitory concentrations (MIC) (
Tables 1–
4). Double mutants defective in
acrB and
tolC served as negative controls. Different plant-derived antimicrobial compounds, which were previously reported to be substrates for
E. amylovora AcrAB-TolC [
33–
34], bile salt as a reported substrate for AcrAB-TolC in
E. coli [
37], and various other antimicrobials were tested. Single deletion of
acrB or
tolC and simultaneous deletion of
acrB and
tolC resulted in increased susceptibility of
E. coli mutants toward all tested plant-borne antimicrobial compounds. The respective MICs for phloretin, (+)-catechin, naringenin, quercetin, and berberine decreased more than 4-fold, 4-fold, 8-fold, 8-fold, and 16-fold, respectively, in complex medium and about 8-fold in minimal medium except for berberine, for which the MICs were reduced about 64-fold, as compared to
E. coli wild type MICs (
Tables 1 and
3). Similarly,
E. amylovora mutants defective in
acrB or
tolC and the double mutant exhibited 4-fold, 8-fold, 8-fold, 16-fold, and 32-fold decreased MICs towards (+)-catechin, phloretin, naringenin, quercetin, and berberine, respectively, in both, complex and minimal medium (
Tables 2 and
4). Consequently and despite of the fact that
E. coli is neither a plant pathogen nor usually exposed to plant-borne chemical defense reactions, AcrAB-TolC of this human-associated bacterium can exclude phytoalexins. This interesting result further substantiated the broad substrate spectrum of this MDE system [
14,
38] and is in line with the idea that intestinal microbes are constantly challenged by toxic substances derived from plant-borne diet in mammals [
39]. Alternatively, our result might indicate a high degree of phylogenetically conserved functionality of the AcrAB-TolC complex among
enterobacteriaceae.
The three tested
E. amylovora mutants were about 8-fold more sensitive to bile salt as compared to the wild type regardless of the used medium. In contrast, the three
E. coli mutants exhibited a 64-fold increased sensitivity to bile salt in both media in comparison to the wild type (
Tables 1–
4). This result clearly demonstrated that
E. amylovora may possess an alternative MDE pump during the detoxification process for bile salts such as demonstrated for EefABC in
Enterobacter aerogenes [
40]. It remained obscure why the
E. coli mutants exhibited higher susceptibilities towards bile salt since one might have assumed a better adaptation of this bacterium to this toxic compound [
41].
In contrast to the mutants of
E. amylovora, growth medium-dependent susceptibility towards the following group of antimicrobials was observed for the three
E. coli mutants: acriflavine, novobiocin, ampicillin, cefoperazone, mitomycin, tetracycline, nalidixic acid, norfloxacin, ciprofloxacin, SDS, ethidium bromide, and crystal violet. For those substances the
E. coli wild type was about 5- to 20-fold more resistant as compared to the mutants in complex medium but only 4- to 10-fold more resistant in minimal medium (
Tables 1 and
3).
E. amylovora mutants were 5- to 40-fold more sensitive to those compounds as compared to their wild type regardless of the used growth medium (
Tables 2 and
4). These apparently conflicting results might be rather due to the auxotropic status of
E. coli TG1 and its general physiological fitness in minimal medium [
42] than due to substrate specificity alterations.
The MICs of the aminoglycosides amikacin and tobramycin for the
acrB mutants of
E. coli and
E. amylovora did not differ from those determined for the wild type strains suggesting that those substances did not act as substrates of AcrB. In contrast, the same substances caused 8-fold and 16-fold increased sensitivities for the
tolC mutants and the
acrB /
tolC double mutants of both species in complex and minimal medium, respectively. It was previously demonstrated that TolC additionally to its function with AcrB could specifically interact with another RND-type transporter of
E. coli termed AcrAD in extruding a variety of hydrophilic aminoglycosides from the periplasm and cytoplasm [
38,
43].
Additive effects in the reduction of MICs due to simultaneous disruption of
acrB and
tolC regardless of the used species and medium could be observed for erythromycin, rifampin, jasmone, and clotrimazole. The MIC values of the
acrB mutants for these compounds were reduced 4- to 10-fold. In contrast, respective MIC values declined in the
tolC mutants about 10- to 40-fold. Susceptibilities increased 64- to 200-fold in the
acrB /
tolC double mutant of
E. coli and about 80-fold in the respective
E. amylovora double mutant (
Tables 1–
4), suggesting that outer membrane-bound TolC might interact with additional partners during the extrusion of these compounds. Chollet
et al. [
44] demonstrated that telithromycin, a ketolide comparable to macrolides, is not a substrate for AcrAB-TolC but it is efficiently recognized by another PAβN-sensitive system pump. Elkins and Mullis [
14] showed that the MFS-type tripartite system, EmrAB-TolC, of
E. coli has the extraordinary capacity to transport mammalian steroid hormones. This transporter is ‘silent’ under normal laboratory conditions and its contribution to MDE resistance towards steroids and macrolides is masked by the overlapping substrate repertoire of the AcrAB-TolC system [
14]. Likewise, MacAB of
E. coli is cooperating with TolC in exporting macrolides [
45]. The herein observed additive effects in MIC reduction may be related to presence of EmrAB in both,
E. coli and
E. amylovora. A BLAST search in the genome sequence of
E. amylovora 237 revealed presence of an EmrB homolog with 83% identity and 92% similarity but no MacB homologs.
2.3. Allelic exchange analysis of AcrAB-TolC in E. amylovora and E. coli
As expected and in line with our previously published data [
33–
34], the MIC phenotypes of
acrB and
tolC single mutants of
E. amylovora could be fully restored to wild type levels with clones containing
E. amylovora acrAB and
tolC, respectively. Without any exceptions, the same results were obtained when respective
E. coli mutants were complemented with cloned
E. coli acrAB and
tolC genes (
Tables 1–
4). Next, single mutants of
E. amylovora were transformed to carry
acrAB or
tolC, respectively, derived from
E. coli and vice versa. Thus it was tested whether or not the individual components of the AcrAB-TolC complex from the two enterobacterial species can interact without interferences. Interestingly, all thus generated transformants showed fully restored wild type MIC phenotypes (
Tables 1–
4), suggesting that the AcrAB-TolC efflux systems of
E. coli and
E. amylovora are truly interchangeable despite of the variable ecological niches both organisms occupy. The results furthermore allowed the conclusion that the divergence in micro-ecological adaptation had not led to specialized MDE pumps with respect to AcrAB-TolC but that this system is multifunctional and ancestral. This confirmed an earlier study, in which MDE pumps from
Ralstonia solanacearum were used to complement respective
E. coli mutants [
32]. Our data are also in agreement with those from Tikhonova
et al. [
46] and Bokma
et al. [
47], who reported on the functionality of AcrB-MexB hybrid proteins and the directed evolution-based adaptation of
E. coli TolC to the MexAB translocase of
Pseudomonas aeruginosa. The later authors generated active hybrid pumps by challenging a library of mutated and shuffled TolC variants to adapt to the non-cognate
P. aeruginosa MexAB system. The obtained analysis of amino acid substitutions in TolC revealed that adaptation to the heterogenous efflux pump was conferred by substitutions of amino acyl residues located in the lower α-helical barrel in the periplasmic equatorial domain and the entrance coiled coils of TolC [
47]. Protein sequence alignment of
E. coli TolC versus that of
E. amylovora TolC revealed that all the substituted amino acid residues reported by Bokma
et al. [
47] are conserved in both enterobacterial species (data not shown) thus by further substantiating a tight phylogenetic relatedness of AcrAB-TolC in
E. coli and
E. amylovora.
Interestingly, any complementation led to MIC wild type level of the species, into which the heterologous alleles were transferred. These results underscore the principle that antibiotics resistance is determined by both, efflux and outer membrane permeability. In this context, it is remarkable that permeability for many of the tested compounds differed between E. amylovora and E. coli.
2.4. In planta virulence assays
Previously it was demonstrated that
E. amylovora mutants with defects in
acrB and
tolC were not causing fire blight symptoms on apple plants and showed significantly reduced
in planta survival [
33–
34]. These mutants could be complemented by plasmid-borne homogenous alleles. Since heterogenous alleles of
acrB and
tolC led to wild type MIC values in the respective
E. amylovora mutants
in vitro, it was tested whether
acrB and
tolC of
E. coli could restore virulence of
E. amylovora acrB and
tolC single mutants. Respective
E. amylovora transformants were inoculated to apple plants using the mutants and the wild type as controls. Plant shoot tips were inoculated with defined numbers of bacterial cells by the so-called prick technique [
48], which mimics the natural infection process. Typical fire blight symptoms in form of shepherd’s crook-like bending of the shoot tip after one week post inoculation as well as ooze formation and necrosis after three weeks post inoculation were induced by
E. amylovora wild type and the
E. amylovora mutants carrying
acrB or
tolC from
E. coli in all 15 plants inoculated per strain without exceptions (
Figure 1).
In contrast, the non-complemented mutants did not cause any disease symptoms thus confirming their previously reported
in planta phenotypes [
33–
34]. All inoculated plants showed the same lack of symptom development. Interestingly, for the first time these results led to the conclusion that enterobacterial AcrAB-TolC hybrids consisting of mixed components from a plant pathogen and an intestinal species are sufficient to effectively combat plant defense reactions. These results furthermore substantiated our
in vitro findings and showed that AcrB and TolC of
E. coli and
E. amylovora are apparently fully interchangeable. The herein obtained data are in line with recent findings of Krishnamoorthy
et al. [
49], who reported that the replacement of
E. coli AcrB with its close homolog, MexB, from
Pseudomonas aeruginosa formed a partially functional MDE system, AcrA-MexB-TolC,
in vitro and that certain single amino acid substitutions in AcrA and MexB, respectively, were sufficient to render this MDE hybrid fully functional. The data obtained herein demonstrated that complementation of
acrAB or
tolC of a plant pathogen by respective alleles derived from
E. coli can restore full virulence. In this respect, our results are in advancement towards those of Brown
et al. [
32], who had demonstrated that MDE components of the plant pathogen
R. solanacearum were successfully complementing
E. coli mutants
in vitro.