Increased Resistance to Carbapenems in Proteus mirabilis Mediated by Amplification of the bla VIM-1 -Carrying and IS 26 -Associated Class 1 Integron

Bacterial strains, antibiotic susceptibility testing, plasmid extraction, and conjugation assays

Proteus mirabilis isolates were identified at the species level using the API 20E system (bioMérieux, La-Balme-les-Grottes, France). Escherichia coli TOP10 (Life Technologies, Cergy-Pontoise, France) and E. coli J53 reference strain were used in cloning and conjugation experiments, respectively. ¹⁰ Minimal inhibitory concentrations (MICs) of carbapenems were determined by Etest (AB bioMérieux, Solna, Sweden), and were interpreted according to Clinical and Laboratory Standards Institute breakpoints. ¹¹ Mating-out assays were attempted using the VIM-1-producing P. mirabilis isolate PM5 as donor (representing the clonal strain) and E. coli J53 (azide resistant) as recipient. Selection was made using trypticase-soy plates supplemented with amoxillin 100 μg/mL and sodium azide (100 μg/mL). In addition, extraction of natural plasmids was attempted by using the Kieser method. ¹²

PCR, cloning, and sequencing

PCR approach was used to detect carbapenemase encoding genes, as previously described. ¹⁰ Cloning experiments were performed using plasmid pTOPO as vector, and expression in E. coli TOP10, and selection on trypticase-soy agar supplemented with ticarcillin (50 mg/L) and kanamycin (30 mg/L). Plasmid DNAs were extracted by using Qiagen columns (Qiagen, Courtaboeuf, France). Sequencing of the amplicons and recombinant plasmids was performed by the Microsynth company (Balgach, Switzerland). PCR mapping of the bla_VIM-1-surrounding sequences was performed with primers localized in the class 1 integron. ⁹ PCR fragments generated with primers on which the XbaI restriction site was added were cloned in pTOPO-PCR Blunt vector (Invitrogen, Thermo Fisher).

Quantification of carbapenemase activity

The production of carbapenemase was first evaluated by using the Carba NP test, as previously described. ¹³ In addition, specific activities with imipenem as substrate were measured. In brief, a 10-mL broth culture of each tested isolate was centrifuged and submitted to sonication. Then, the pellets were, respectively, resuspended in 500 μL of 100 mM sodium phosphate buffer (pH 7) and submitted to sonication. Supernatants were then used for kinetic measurements. ¹⁴ Hydrolysis was measured by UV spectrophotometry using 100 μM imipenem as substrate and 10 μL of the bacterial crude extract. The protein content was measured using the Bio-Rad DC protein assay.

Quantification of the bla_VIM-1 gene by real-time PCR amplification

Real-time experiments were performed at least three times in different PCR runs, using a Rotor-Gene Q (Qiagen, Hilden, Germany) with a KAPA SYBR FAST qPCR Kit (Kapabiosystems, Wilmington, MA). ¹⁵ The hcaT gene was chosen as the normalization control. The qPCR assays included primers VIM-1-qF1 (5′-GAG GTC CGA CTT TAC CAG ATT G-3′) and VIM-1-qR1 (5′-ATC ACC ATC ACG GAC AAT GAG-3′) for the bla_VIM-1 gene, and included primers hcaT-Pm-qF1 (5′-CAC TGT GGC TTG CCT TAG AT-3′) and hcaT-Pm-qR1 (5′-CTC GCC CTT TAA CCA GAT AGA C-3′) for the hcaT gene.

The quantity of each target in each isolate was expressed as the difference in Ct values between the hcaT and the bla_VIM-1 genes. PCR efficiency was ∼90% for both targets. Real-time PCR was performed by an initial denaturation at 95°C for 60 sec, then rounds were repeated at 95°C for 3 sec, 60°C for 20 sec, and 72°C for 10 sec using primers VIM-1-qF1 and VIM-1-rR1 on one hand, and hcaT-Pm-qF1 and hcaT-Pm-qR1 on the other hand. The Real-Time PCR reaction mix was prepared in a volume of 20 μL containing 20 ng of purified genomic DNA, using the KAPA SYBR FAST qPCR Kit Master Mix (2 × ) Universal, and 0.25 μM of each primer. Analysis was performed using the Rotor-Gene Q machine (Qiagen).

Clonal relationship

Genotyping was performed by pulsed-field gel electrophoresis (PFGE), using restriction enzyme NotI, as previously described. ¹⁶

Susceptibility testing and carbapenemase identification

During the March–June 2013 period, a total of 23 P. mirabilis isolates were recovered at the Military hospital of Sofia from urine of patients hospitalized in three different units, namely two intensive care units and a neurosurgical ward. They exhibited variable susceptibility patterns to carbapenems, with four isolates remaining susceptible, and the other showing resistance at variable levels. While MICs of imipenem ranged from 0.25 to >32 mg/L, those of ertapenem ranged from 0.004 to 4 mg/L, and those of meropenem from 0.047 to >32 mg/L (Table 1). This prompted us to evaluate whether these isolates might be corresponding to different clones. PFGE analysis showed that the four imipenem-susceptible isolates corresponded to a single clone, and the 19 other isolates to a second clone (data not shown). Results of the Rapid Carba NP test showed that all the 19 imipenem-resistant isolates produced a carbapenemase, but no carbapenemase activity was detected in the susceptible isolates. PCR and sequencing identified the carbapenemase gene bla_VIM-1 in all imipenem-resistant isolates. Five isolates were retained for further analysis, with the bla_VIM-1-negative isolate being PM1, and four bla_VIM-1-positive isolates PM2 to PM5 exhibiting distinct resistance levels to carbapenems (Table 1). Those isolates all showed an identical susceptibility pattern to non-β-lactam antibiotics, being susceptible to fluoroquinolones, tetracycline, chloramphenicol, amikacin, tobramycin, gentamicin, and fosfomycin, but showing resistance to kanamycin and rifampicin.

Carbapenemase activities of the P. mirabilis isolates

To precisely determine the capacity of the different isolates to compromise the efficacy of carbapenems upon hydrolysis, β-lactamase activities were measured using imipenem as substrate. While no hydrolysis could be detected with a crude extract of isolate PM1, enzymatic activities of 43 ± 3, 46 ± 5, 82 ± 3, and 196 ± 8 U/mg of protein (1 U of enzyme activity being defined as the activity that hydrolyzed 1 μmol of imipenem per min) were obtained for isolates PM2, PM3, PM4, and PM5, respectively.

Copy number of the bla_VIM-1 gene

To assess whether the increased expression of the bla_VIM-1 gene could be related to an increased copy number, quantitative PCR was performed. Isolate PM1 used as control gave a negative result, as expected. For isolates PM2 and PM3, a single copy of the bla_VIM-1 gene was detected. Two copies were detected in isolate PM4 and finally, four copies were detected in isolate PM5. Interestingly, the bla_VIM-1 gene copy number very well correlated with the MIC data (Fig. 1; Table 1).

OPEN IN VIEWER

FIG. 1.

Copy numbers of the bla_VIM-1 genes measured by quantitative PCR. Isolate PM1 (bla_VIM-1 negative) was used as negative control. Isolates PM2, PM3, PM4, and PM5 (bla_VIM-1 positive) were used for testing. Vertical axis; bla_VIM-1 copy number normalized on hcaT. Horizontal axis; distribution of the different isolates. Error bars represent the standard deviation between the three experiments.

Genetic structures surrounding the bla_VIM-1 gene

Recombinant plasmid pPM2, recovered after XbaI cloning experiments using DNA of P. mirabilis PM2, possessed an ∼8-kb insert and harbored the bla_VIM-1 gene into a class 1 integron structure. In fact, the bla_VIM-1 gene was the first cassette in the integron structure, followed by aacA7 encoding resistance to aminoglycosides, dfrA1 (resistance to trimethoprim), and aacA1 (resistance to aminoglycosides) (Fig. 2). The 3′-extremity of this class 1 integron was truncated by an IS1 element, and consequently lacked the usual qacEΔ1 and sul1 genes. The whole class 1 integron was bracketed by two copies of insertion sequences IS26, as previously identified in some bla_VIM-1-positive strains. ⁹ Here, a likely IS26-mediated homologous recombination was at the origin of this integron acquisition. In fact, IS26 is very often identified in association with class 1 integron structures. It has been previously shown to be a key element in homologous recombination processes, leading to acquisition of some antibiotic resistance genes, such as bla_BES-1. ¹⁷ In fact, such integron structure is very similar to the one (named In-e541) reported from a bla_VIM-1-positive E. coli isolate recovered in Greece, but differed in its 3′-extremity with the insertion of IS1. ¹⁸

OPEN IN VIEWER

FIG. 2.

Schematic representation of the different bla_VIM-1 integron structures identified in isolates PM2 and PM3 (A), PM4 (B), and PM5 (C). Insertion sequences IS26 and IS1 are represented. The int1 integrase gene is followed by the following gene cassettes, namely bla_VIM-1, aacA7, dfrA1, and aadA1. Dashed arrows indicate the boundaries of the repeated fragments. Respective structures are not to scale between each other, to better represent the evolution of the whole content.

By PCR mapping, we showed that isolate PM2 possessed the same unique structure. A second copy of this whole bla_VIM-1-positive integron was identified in isolates PM3 and PM4, bracketed by another IS26 element at its 3′-end extremity. Finally, four copies of the same structure were identified in isolate PM5, with IS26 elements always bracketing those different copies (Fig. 2). This structure was likely the result of an IS26-mediated recombination process, leading to multicopies of the bla_VIM-1-positive integron.

In all isolates, sequencing showed that the promoter sequences located in the 5′-extremity of the class 1 integrons identified that are known to drive the expression of the right-end located gene cassettes were conserved. Therefore, the overall variability in bla_VIM-1 expression level could not be attributed to heterogeneity of the corresponding promoter sequences.

Mating-out assays remained unsuccessful despite repeated attempts, further correlating with a chromosomal location of the bla_VIM-1 gene in all positive isolates. In addition, repeated plasmid extractions did not reveal any plasmid band on gel electrophoresis, which is another correlating feature.