Single or in Combination Antimicrobial Resistance Mechanisms of Klebsiella pneumoniae Contribute to Varied Susceptibility to Different Carbapenems

Resistance to carbapenems has been documented by the production of carbapenemase or the loss of porins combined with extended-spectrum β-lactamases or AmpC β-lactamases. However, no complete comparisons have been made regarding the contributions of each resistance mechanism towards carbapenem resistance. In this study, we genetically engineered mutants of Klebsiella pneumoniae with individual and combined resistance mechanisms, and then compared each resistance mechanism in response to ertapenem, imipenem, meropenem, doripenem and other antibiotics. Among the four studied carbapenems, ertapenem was the least active against the loss of porins, cephalosporinases and carbapenemases. In addition to the production of KPC-2 or NDM-1 alone, resistance to all four carbapenems could also be conferred by the loss of two major porins, OmpK35 and OmpK36, combined with CTX-M-15 or DHA-1 with its regulator AmpR. Because the loss of OmpK35/36 alone or the loss of a single porin combined with bla CTX-M-15 or bla DHA-1-ampR expression was only sufficient for ertapenem resistance, our results suggest that carbapenems other than ertapenem should still be effective against these strains and laboratory testing for non-susceptibility to other carbapenems should improve the accurate identification of these isolates.


Introduction
The increasing prevalence of extended-spectrum βlactamases (ESBLs) and plasmid-mediated AmpC βlactamases in Enterobacteriaceae is a critical concern for scientists trying to develop treatments against bacterial infections. TEM-, SHV-and CTX-M-type ESBLs and CMY-and DHA-type AmpC β-lactamases are commonly found in multidrug-resistant Enterobacteriaceae [1][2][3][4]. Carbapenem has been recommended as an effective drug for these strains. However, carbapenem-resistant Enterobacteriaceae have recently been reported worldwide. In Klebsiella pneumoniae, numerous reports have confirmed carbapenem resistance by ESBLs or AmpC β-lactamases combined with the loss of outer membrane porins OmpK35 and/or OmpK36 [5][6][7], or by carbapenemases alone [8].
Currently, four carbapenems are used clinically (imipenem, meropenem, ertapenem and doripenem). The mechanisms of carbapenem resistance were mainly detected from clinical isolates with individual or combined genetic alterations. No comparisons have been made on all four carbapenems and on the contributions of each resistance mechanism towards carbapenem resistance. In the present study, we created mutants with individual or combined resistance mechanisms from a susceptible clinical K. pneumoniae isolate and studied each resistance mechanism in response to these four carbapenems.

Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Table 1. The K. pneumoniae clinical strains were isolated from different patients in Taiwan or America, and strain NVT2001 has been found to be resistant only to ampicillin [9]. The low-copy-number plasmid pACYC177 was used in the cloning experiments. Escherichia coli and K. pneumoniae strains were cultured at 37°C in Mueller-Hinton broth (MHB) or Luria-Bertani (LB) broth with the appropriate antibiotics.

Conjugation experiments and plasmid content
Direct transfer of the β-lactamase-carrying plasmids from K. pneumoniae clinical isolates to azide-resistant E. coli J53 was performed by filter conjugation [10]. Transconjugants were selected on LB agar plates containing sodium azide (100 μg/ml) for counterselection and cefotaxime (1 μg/ml) for selection of plasmid-encoded resistance. Because E. coli J53 does not retain the bla KPC-2 -carrying plasmids of clinical K. pneumoniae for any period of time [11], the bla KPC-2 -carrying plasmid was directly transferred to the K. pneumoniae strains NVT2001S, ΔompK35, ΔompK36 and ΔompK35/36 by filter conjugation [10]. Transconjugants were selected on LB agar plates containing streptomycin (500 μg/ml) for counterselection and cefotaxime (0.5-2 μg/ml) for selection of plasmid-encoded resistance. Resistance genes, including bla CTX-M , bla SHV , bla TEM , bla DHA , bla CMY , bla MOX , bla LAT , bla BIL , bla KPC and bla NDM , were detected among the transconjugant plasmids using PCR with specific primers (Table S1) and then further confirmed by sequencing. All the sequence analyses were conducted using the NCBI database (http://www.ncbi.nlm.nih.gov/). Isolates carrying resistance genes that had previously been sequenceconfirmed were used as positive controls in every PCR assay.

Cloning and strain construction
DNA fragments of the bla CTX-M-15 , bla SHV-12 , bla DHA-1 -ampR, bla DHA-1 , ampR, bla KPC-2 and bla NDM-1 genes with their flanking regions were amplified from transconjugants by PCR with specific primers (Table S1). The generated PCR fragments were digested with PstI or BamHI and then cloned into pACYC177. The resulting plasmids were then transformed into the K. pneumoniae strains NVT2001S, ΔompK35, ΔompK36 and ΔompK35/36 by electroporation. The recombinant bacteria were plated onto LB agar plates containing kanamycin (25 μg/ ml), and the presence of the β-lactamase genes was confirmed  [10]. Transconjugants were selected on brilliant green containing inositol-nitrate-deoxycholate (BIND) plates containing cefotaxime (1 μg/ml) for the selection of plasmid-encoded resistance, and the growth of non-K. pneumoniae strains was effectively suppressed on BIND plates [12]. The presence of the β-lactamase genes was verified by PCR and sequencing.

Antimicrobial susceptibility test
Minimal inhibitory concentrations (MICs) of ertapenem, imipenem, meropenem and doripenem were determined using the E-test (Biodisk AB, Sweden). MICs of the other 15 antimicrobial agents were determined using a broth microdilution test according to the recommendations of the clinical and laboratory standards institute (CLSI) [13]. The following antimicrobial agents were used: aztreonam, ampicillin, piperacillin/tazobactam, cefazolin, cephalothin, cefoxitin, ceftriaxone, cefpodoxime, cefotaxime, cefotaxime/ clavulanic acid, ceftazidime, ceftazidime/clavulanic acid, cefepime, ciprofloxacin and gentamicin. All the results were interpreted according to the breakpoints established by the CLSI in 2013 [14].

Strain construction and porin loss in antibiotic resistance
The β-lactamases detected in the clinical plasmids are shown in Table 1, while bla CTX-M-15 , bla SHV-12 , bla DHA-1 -ampR, bla KPC-2 or bla NDM-1 was found in each of the five clinical plasmids. To evaluate the effect of these β-lactamases alone and in combination with porin loss on antimicrobial resistance, the clinical and recombinant plasmids were transferred into the NVT2001S strain and its porin-loss mutants. The loss of OmpK35 alone did not significantly influence the antibiotic resistance of the NVT2001S strain (Table 2). Conversely, the loss of OmpK36 alone conferred resistance to cefazolin, cephalothin and cefoxitin, and the loss of OmpK35/36 caused highly resistance against these cephalosporins ( Table 2).

ESBLs with or without porin loss in antibiotic resistance
The production of CTX-M-15 or SHV-12 alone conferred resistance to aztreonam and many of the cephalosporins tested (Table 3). With the loss of OmpK36, the CTX-M-15 and SHV-12 strains showed resistance to aztreonam, piperacillin/ tazobactam and all the cephalosporins tested, while some βlactam/β-lactamase inhibitor combinations were still effective against these strains (Table 3). AmpC β-lactamases with or without porin loss in antibiotic resistance NVT2001S harboring the clinical plasmid with bla DHA-1 -ampR showed significantly (≥4-fold) higher MICs for several antibiotics tested compared to that harboring the recombinant plasmid with bla DHA-1 -ampR. This result indicates that other resistant genes should exist on the clinical plasmid (Table 4). For the strains harboring the recombinant plasmid, the expression of bla DHA-1 -ampR could confer significantly (≥4-fold) higher MICs of many antibiotics tested compared to the expression of bla DHA-1 alone. No significant (≥4-fold) differences in MICs were found between the AmpR strain and its parental strain (Tables 4). With the loss of OmpK35/36, the DHA-1-AmpR strains were highly resistant to aztreonam, piperacillin/ tazobactam and all the cephalosporins tested except cefepime (Table 4).

Carbapenemases with or without porin loss in antibiotic resistance
In all the antibiotics tested, the KPC-2 strains showed susceptibility to ciprofloxacin and gentamicin. The NDM-1 strains showed susceptibility to aztreonam, ciprofloxacin and gentamicin, except gentamicin for the strains harboring the clinical plasmid with bla NDM-1 ( Table 5). The loss of porins did not increase the resistance of these strains to these antibiotics (Table 5).

Single or in combination antimicrobial resistance mechanisms in carbapenem resistance
The loss of OmpK35 or OmpK36 alone did not significantly (≥4-fold) influence carbapenem resistance of the NVT2001S strain. The loss of OmpK35/36 conferred 31-, 8-and 4-fold increases in the MICs of ertapenem, meropenem and doripenem, respectively, and led to ertapenem resistance (Table 6). Without porin loss, the CTX-M-15 and DHA-1-AmpR strains only showed a significant (≥4-fold) increase in the ertapenem MIC, while the KPC-2 and NDM-1 strains exhibited resistance to all four carbapenems (Table 6). With the loss of OmpK36, the CTX-M-15 and DHA-1-AmpR strains became resistant to ertapenem. With the loss of OmpK35/36, the CTX-M-15, SHV-12 and DHA-1-AmpR strains became resistant to all four carbapenems, except imipenem and doripenem for the strain harboring the recombinant plasmid with bla SHV-12 (Table  6).

Discussion
Of the four carbapenems in this study, the loss of OmpK35/36 alone could confer ertapenem resistance, and the expression of bla CTX-M-15 or bla DHA-1 -ampR alone was only sufficient to significantly (≥4-fold) increase the ertapenem MIC. Previous studies have shown that CTX-M β-lactamase activity against ertapenem is very low. CTX-M likely contributes to the decreased ertapenem susceptibility by binding with a high affinity to this molecule [15,16]. Conversely, the production of KPC-2 or NDM-1 alone could render K. pneumoniae strains resistant to all four carbapenems, and the ertapenem MIC was the highest. Our results suggest that among the four carbapenems, ertapenem is the least active against the loss of porins, cephalosporinases and carbapenemases. Previous studies have found that carbapenem resistance in clinical isolates can be conferred by porin loss combined with the production of ESBLs or AmpC β-lactamases. In particular, ertapenem resistance can be caused by porin loss with the CTX-M variants [17,18]. Our results further revealed that resistance to all four carbapenems could be rendered by the loss of OmpK35/36 combined with the expression of bla CTX-M-15 or bla DHA-1 -ampR in K. pneumoniae. Because carbapenems are frequently utilized as drugs of last resort for the treatment of a variety of infections caused by multidrug-resistant bacteria, this finding is notable when using carbapenems for treating infections due to ESBL-or AmpC β-lactamase-producing Enterobacteriaceae with the loss of porins.
DHA-1 is a plasmid-encoded AmpC β-lactamase and bla DHA-1 expression is transcriptionally regulated by the divergently read ampR gene [19]. The mechanism of AmpC induction is intimately linked to a cell wall recycling system [20][21][22]. AmpR can both activate and repress ampC expression according to its interaction with specific murein degradation products, and the murein synthesis was interfered with β-lactams [3,21]. Previous study showed that AmpR represses the synthesis of AmpC β-lactamase by 2.5-fold in the absence of an inducer, while its expression is induced more than 10-fold in the presence of a β-lactam [23]. The β-lactams also differ in their inducing abilities [3]. Of the strains harboring recombinant plasmids in this study, the DHA-1-AmpR strains showed higher MICs for many antibiotics tested compared with those of the DHA-1 strains. Whether this result is due to the expression of bla DHA-1 induced by AmpR in the presence of these antibiotics requires further studies. The DHA-1-AmpR strains with the loss of OmpK35/36 became highly resistant to multiple drugs, including the four carbapenems. However, cefepime should be an effective β-lactam against these strains.
Carbapenemase can effectively inactivate most β-lactam antibiotics, including carbapenems, and most carbapenemase genes are transferable. The rapid identification of carbapenemase producers is needed to prevent the development of outbreaks. Ertapenem is a sensitive indicator for detecting most of the carbapenemase producers [24,25]. However, ertapenem showed lower specificity for detecting carbapenemases in clinical isolates compared with imipenem and meropenem [26]. Of the four carbapenems in this study, our results also showed that the loss of OmpK35/36 or the loss of a single porin combined with the expression of bla CTX-M-15 or bla DHA-1 -ampR was only sufficient to lead to ertapenem resistance. Because resistance to all four carbapenems could be conferred by the loss of OmpK35/36 combined with the expression of bla CTX-M-15 or bla DHA-1 -ampR, special tests for Table 4. MICs of antibiotics against K. pneumoniae NVT2001S and its porin-loss mutants with AmpC β-lactamase and/or its regulator. carbapenemase detection should be applied, especially in regions where these β-lactamase-producing strains are endemic.  a Boldface numbers indicate a significant (≥4-fold) difference in the MICs of K. pneumoniae NVT2001S and its derived strains, while the underlined numbers were above the breakpoint of susceptibility established by CLSI in 2013 [14]. b The β-lactamase and/or its regulator on the plasmid were shown, and the plasmid was transferred into K. pneumoniae NVT2001S and its porin-loss mutants.