Introduction

The emergence and dissemination of extensively drug-resistant (XDR) Gram-negatives (e. g. colistin-only susceptible Pseudomonas aeruginosa, carbapenem-resistant Acinetobacter baumannii, and carbapenem-resistant Enterobacteriaceae) in the clinical setting has led to the renewed use of colistin for treatment of infections caused by these pathogens1. Consequently, colistin resistance has been increasingly reported2,3,4. Thus far, the problem has mostly involved Klebsiella pneumoniae and Acinetobacter, while acquired colistin resistance has remained uncommon among clinical isolates of Escherichia coli 5,6,7,8. Conversely, colistin resistance has been reported more frequently among E. coli isolates of animal origin, likely due to the extensive use of colistin in veterinary medicine9.

In E. coli, resistance to colistin has mostly been associated with the acquisition of transferable mcr-type genes, which encode enzymes able to modify the lipopolysaccharide (LPS) colistin target by the addition of phosphoethanolamine10,11,12. Chromosomal mutations in the pmrAB genes, encoding a two-component signal transduction system which regulates the endogenous LPS modification systems, have also been described in colistin-resistant (Col-R) E. coli isolates from swine and poultry, but without a formal experimental confirmation of their role in conferring resistance to colistin13.

In this study we characterized a Col-R E. coli strain of clinical origin carrying a novel pmrB allelic variant encoding a PmrB protein with a leucine to proline amino acid substitution at position 10, and demonstrate that this mutation is associated with acquisition of colistin resistance in E. coli.

Results

Resistance and clonal profiles of E. coli strain LC711/14

E. coli LC711/14 was isolated in 2014 from the urine of an outpatient suffering from an urinary tract infection. Routine susceptibility testing, using the Vitek2 automated system, revealed that the strain was susceptible to all tested antibiotics (amoxicillin-clavulanic acid, cefotaxime, ceftazidime, cefepime, piperacillin-tazobactam, ertapenem, imipenem, meropenem, gentamicin, ciprofloxacin, trimethoprim-sulfamethoxazole) except colistin. Resistance to colistin was confirmed by reference broth microdilution, which yielded an MIC of 4 mg/L (Table 1). The patient had not been previously treated with colistin.

Table 1 Colistin MICs and expression levels of pmrK of different E. coli strains and transformants complemented with the pACYC-pmrB, pACYC-pmrB t29c, pACYC-pmrBc28g/t29g and pACYC-pmrB t29a plasmids.
Full size table

Multilocus sequence typing (MLST) analysis revealed that strain LC711/14 belonged to ST59, a lineage previously described to include strains with high virulence potential14.

Colistin resistance mechanism in E. coli strain LC711/14: identification of a PmrB amino acid substitution involved in colistin resistance

LC711/14 was part of a collection of nine Col-R E. coli strains of clinical origin that were previously investigated to verify the presence of the mcr-1 gene, and resulted the only strain negative for mcr-1 15.

Transcriptional analysis by quantitative real-time PCR (qRT-PCR) showed that pmrK expression in LC711/14 was increased compared to what observed in the reference wild-type E. coli strain MG1655, which is colistin-susceptible (Col-S) (Table 1). These results suggested that the Col-R phenotype of LC711/14 was consequent to upregulation of the pmrHFIJKLM operon, encoding the endogenous LPS modification system16.

Molecular characterization of the chromosomal pmrAB genes, encoding a two-component system known to be involved in the regulation of the pmrHFIJKLM operon17, revealed a t29c transition in pmrB, resulting in a non-synonymous leucine to proline amino acid substitution at position 10 (Leu10Pro) of the protein compared with other E. coli PmrB proteins present in the non-redundant nucleotide collection Blast database (including PmrB of MG1655) (Fig. 1). While some polymorphisms of the PmrB sequence exist among different E. coli strains, some of which have previously been putatively associated with a Col-R phenotype (e. g. Val161Gly)13 (Fig. 1), the Leu10Pro substitution was not previously reported. In fact, a leucine residue appeared to be strictly conserved at this position in 256 records of full-length E. coli PmrB present in the non-redundant protein sequences Blast database (accessed on April 18, 2017), and also in the PmrB sequences of the 8 mcr-1-positive Col-R strains of the collection from which LC711/14 was taken15. The closest PmrB variant present in the database was that of E. coli ECOR35 (Accession number AEL97557), which was identical to that of LC711/14 except for the Leu10Pro substitution (Fig. 1). Data regarding the colistin susceptibility of ECOR35 strain are not available.

Figure 1
figure 1

Protein sequence alignment of PmrB of E. coli MG1655 (accession no. NC_000913.3), ECOR35 (accession no. JN032071.1), LC711/14 (this work), and ZTA11/01748 (Col-R strain isolate from swine faeces, carrying a PmrB mutation putatively associated with colistin resistance13. Asterisks indicate conserved amino acid residues. Amino acids polymorphisms not associated with colistin resistance are underlined. The L10P substitution found in LC711/14 and demonstrated to be involved in colistin resistance in this work, and the V161G substitution described in a veterinary E. coli strain and putatively associated with colistin resistance are boldfaced. Provean (Protein Variation Effect Analyzer: http://provean.jcvi.org/seq_submit.php) analysis was carried out for the Leu10Pro (−2,67) and Val161Gly (−5.66) mutations, and indicated that both mutations had an impact on the PmrB topology. Structural and functional domains of the protein by Phobius prediction (http://phobius.sbc.su.se/) are also indicated: TM1 and TM2: transmembrane 1 and transmembrane 2 domains, are shaded in dark grey, while the periplasmic domain is shaded in light grey. The sub-domains (DHp: histidine phosphotransfer; CA: catalytic and ATP-binding) of the cytoplasmic portion of PmrB are overlined by black bars32. The number of residues of each domain are indicated in brackets.

Full size image

To investigate the role of the PmrB Leu10Pro amino acid substitution in colistin resistance, the effect of this substitution was studied in the reference Col-S E. coli strain MG1655. The PmrB sequence of this strain shares 360/364 identical amino acid residues with that of LC711/14, including the Leu10 residue (Fig. 1). A pmrB deletion mutant of MG1655 (MG1655_ΔpmrB), complemented with the pmrB allele from MG1655 containing the t29c substitution (PmrBMG1655 Leu10Pro) exhibited a Col-R phenotype (MIC, 4 mg/L), while the same strain complemented with the wild-type pmrB allele (PmrBMG1655WT) remained colistin susceptible (MIC, 0.125 mg/L) (Table 1). Analysis of the expression of pmrK revealed increased levels in MG1655_ΔpmrB expressing PmrBMG1655 Leu10Pro compared with MG1655_ΔpmrB expressing PmrBMG1655WT (Table 1).

Expression of PmrBMG1655WT in LC711/14 was also able to reduce the colistin MIC of this strain from 4 to 1 mg/L, restoring colistin susceptibility and decreasing pmrK expression to a nearly basal level (Table 1).

Taken together, these results supported the hypothesis that the Leu10Pro amino acid substitution in PmrB is able to confer a Col-R phenotype in E. coli by upregulation of the pmrHFIJKLM LPS modification system.

Investigation of the role of the amino acid at position 10 of PmrB

To further investigate the role of the amino acid residue at position 10 of PmrB of E. coli, additional mutants of PmrBMG1655WT were generated, including PmrBMG1655 Leu10Gly and PmrBMG1655 Leu10Gln.

Complementation of MG1655_ΔpmrB with these mutants did not alter susceptibility to colistin, and the same mutants, expressed in LC711/14, did not modify the Col-R phenotype of this strain (Table 1). Expression experiments performed by qRT-PCR confirmed that the pmrB mutant derivatives (PmrBMG1655 Leu10Gly and PmrBMG1655 Leu10Gln) were expressed in E. coli MG1655_ΔpmrB, while pmrB expression was not detectable, as expected, in MG1655_ΔpmrB (data not shown).

Overall, these results suggested that the two mutated PmrB proteins were not functional, and that the residue at position 10 plays a crucial role in the structure and function of the PmrB sensor kinase.

Prediction of the effect of PmrB mutations at position 10 at the protein level

According to current knowledge on the PmrB protein structure18 position 10 is located into the amino-terminal protein portion including the cytoplasmic secretion signal (aa 1-13), which is a domain putatively involved in the delivery process of the protein into the cell membrane19 (Fig. 1). Comparison of the secondary structure of PmrBMG1655WT with those of mutant derivatives analyzed in this work, using the YASPIN PREDICTION prediction tool, revealed that the mutations at position 10 could have a significant impact on the structure of the region encompassing the first 60 aa of the protein (Fig. 2), which includes the cytoplasmic secretion signal, the TM1 trans-membrane domain, and most of the periplasmic domain (Fig. 1).

Figure 2
figure 2

Secondary structure prediction of PmrB of MG1655 E. coli and its mutant derivatives. Analysis of the region encompassing the first 60 aa of the protein. In white are highlighted the coil regions. The helix regions are shaded in dark grey while the strand regions are boxed. The mutated aa in position 10 are boldfaced.

Full size image

The impact appeared to be higher with the Leu10Pro mutation, since the structure of the cytoplasmic domain and downstream regions were remarkably altered (Fig. 2), possibly due to the characteristic that proline is able to introduce kinks into alpha helices.

Discussion

The emergence of colistin resistance in the clinical setting has become a matter of major concern given the primary role that colistin has regained in the treatment of infections caused by XDR Gram-negatives.

Knowledge concerning colistin resistance mechanisms among E. coli isolates of clinical origin remains limited. Recent studies, prompted by the discovery of mcr-type transferable colistin resistance determinants, have revealed that these determinants, which are highly prevalent among E. coli isolates from animals, can also be found among Col-R isolates from humans10. In animal isolates, mutations of the PmrAB proteins have also been associated with colistin resistance, although their role has not been formally demonstrated13.

This work provides for the first time a formal demonstration that a mutation in the PmrB sensor kinase can be responsible for colistin resistance in E. coli, and underlines the ability of E. coli clinical strains to evolve a Col-R phenotype through a number of different mechanisms. The strain was susceptible to other antibiotics and, in this case, colistin resistance did not pose any challenge to antimicrobial treatment. The Leu10Pro PmrB mutation has not been reported in other clinical strains, and its potential epidemiological impact in the clinical setting remains to be clarified. However, the strain belonged to a clonal lineage described to have pathogenic potential in humans and to be capable of clonal expansion14.

The Leu10Pro PmrB amino acid change is apparently responsible for constitutive activation of the PmrB sensor kinase, leading to downstream up-regulation of the pmrHFIJKLM operon encoding the major endogenous LPS modification system, which is under the control of the PmrAB signal transduction system17. The mechanism by which the Leu10Pro substitution alters the function of PmrB remains unknown. The mutation is close to the N-terminus of the protein, in the first cytoplasmic domain. Comparison of the predicted secondary structure of the amino-terminal moiety of wild-type PmrB with its mutant derivatives revealed a significant impact of mutations inserted at this position (Fig. 2), which could alter the conformation of downstream domains. These alterations could lead either to a constitutive activation of the sensor kinase, due to an unbalance of kinase and phosphatase functions (in the case of Leu10Pro), or to a loss of function (in the case of Leu10Gly and Leu10Gln).

Interestingly, mutations in the PmrB sensor kinase have previously been demonstrated to be involved in acquisition of a Col-R phenotype by activation of the endogenous LPS modification systems in other pathogenic species including P. aeruginosa 18, 20, A. baumannii 21, 22 and K. pneumoniae 23, 24, underscoring the role of PmrB as an important mutational target in evolution of colistin resistance in Gram-negative pathogens. Among the various PmrB mutations reported to be associated with this phenotype, a Leu14Pro mutation has been reported in P. aeruginosa 18. Since the Leu14Pro in P. aeruginosa and the Leu10Pro of E. coli are mutations located in the same cytoplasmic secretion signal domain of PmrB, this further supports the role that similar mutations may have in evolution toward colistin resistance.

Methods

Bacterial strains

The Col-R E. coli LC711/14 strain was isolated in 2014 from the urine of a patient admitted to the A. Manzoni Hospital in Lecco, Northern Italy. LC711/14 was previously reported as a Col-R mcr-1-negative clinical strain15. Identification of the strain at the species level was carried out using MALDI-TOF mass spectrometry (Vitek MS, bioMérieux, Marcy l’Etoile, France). The E. coli K12 strain MG1655 and its MG1655_ΔpmrB derivative (with the pmrB gene deleted) were from the Keio collection25. MLST analysis was carried out as previously described26, and ST was assigned in accordance with the E. coli MLST database (http://mlst.warwick.ac.uk/mlst/dbs/Ecoli/documents/primersColi_html).

Antimicrobial susceptibility testing

Routine susceptibility testing of LC711/14 was performed using the Vitek 2 system (bioMérieux). MICs of colistin were determined by reference broth microdilution27. Results were interpreted according to the EUCAST breakpoints, version 7.1 (www.eucast.org). MICs of the complemented strains were carried out in microdilution broth adding 64 mg/L chloramphenicol to the culture.

Recombinant DNA methodology, sequencing, site-directed mutagenesis and transcriptional analysis

Whole-cell DNA was extracted using the DNeasy Blood & Tissue kit (Qiagen GmbH, Hilden, Germany). The complete sequences of the pmrAB coding regions were amplified using primers and conditions reported in Supplementary Table S1. Primers were designed on the genome sequence of E. coli MG1655 (accession number: NC_000913.3). DNA sequences were determined for both strands at an external sequencing facility (GATC Biotech AG, Germany). The nucleotide and protein sequences were analyzed at the National Center for Biotechnology Information website (www.ncbi.nlm.nih.gov) using the Basic Local Alignment Search Tool (BLAST).

Plasmid pACYC-pmrB, used for complementation experiments and for site-directed mutagenesis experiments, is a pACYC18428 derivative carrying a cloned copy of the wild-type pmrB gene, amplified by PCR using primers pmrA-Ecoli_F and pmrB-ext-Ecoli_R (Table S1) from the genomic DNA of E. coli MG1655. The amplicon (1320 bp), covering the complete pmrB gene with its putative terminator and part of the upstream pmrA gene, was blunt-end cloned into pACYC184 digested with EcoRV. The authenticity of the cloned fragment was confirmed by sequencing. In the recombinant plasmid, the cloned pmrB gene was in opposite orientation of the tetracycline resistance cassette.

Site-directed mutagenesis of pmrB was carried out using the pACYC184-pmrB plasmid as a template. The primers (Table S1) were designed as described by Zheng et al.29. The authenticity of pmrB mutants was confirmed by sequencing on both DNA strands.

Recombinant plasmids were introduced into E. coli strains by electroporation, as previously described30. Transformants were selected on Mueller-Hinton agar (MHA) plates supplemented with 85 mg/L of chloramphenicol.

Expression of the cloned pmrB gene and mutant derivatives in E. coli MG1655_ΔpmrB transformed with the recombinant plasmids was verified by qRT-PCR using primers and conditions described in Table S1. Transcriptional analysis by qRT-PCR to measure expression of the pmrK gene was carried out as previously described30, using the pmrK_F and pmrK_R primers reported in Table S1. Expression of the gapA gene, evaluated with primers gapA_F and gapA_R (Table S1) was used as an internal standard. Normalization was performed against the gapA gene using the 2-ΔΔC T method (relative)31, and the values obtained were normalized against the value obtained with E. coli MG1655.

Protein structure analysis

Prediction of the secondary structure of the PmrB protein was carried out with the YASPIN PREDICTION prediction tool, using non-redundant (NR) and DSSP-trainer databases (http://www.ibi.vu.nl/programs/yaspinwww/).