In Vitro Synergism of Azithromycin Combination with Antibiotics against OXA-48-Producing Klebsiella pneumoniae Clinical Isolates

Carbapenem-resistant Klebsiella pneumoniae has globally emerged as an urgent threat leading to the limitation for treatment. K. pneumoniae carrying blaOXA-48, which plays a broad magnitude of carbapenem susceptibility, is widely concerned. This study aimed to characterize related carbapenem resistance mechanisms and forage for new antibiotic combinations to combat blaOXA-48-carrying K. pneumoniae. Among nine isolates, there were two major clones and a singleton identified by ERIC-PCR. Most isolates were resistant to ertapenem (MIC range: 2–>256 mg/L), but two isolates were susceptible to imipenem and meropenem (MIC range: 0.5–1 mg/L). All blaOXA-48-carrying plasmids conferred carbapenem resistance in Escherichia coli transformants. Two ertapenem-susceptible isolates carried both outer membrane proteins (OMPs), OmpK35 and OmpK36. Lack of at least an OMP was present in imipenem-resistant isolates. We evaluated the in vitro activity of an overlooked antibiotic, azithromycin, in combination with other antibiotics. Remarkably, azithromycin exhibited synergism with colistin and fosfomycin by 88.89% and 77.78%, respectively. Bacterial regrowth occurred after exposure to colistin or azithromycin alone. Interestingly, most isolates were killed, reaching synergism by this combination. In conclusion, the combination of azithromycin and colistin may be an alternative strategy in dealing with blaOXA-48-carrying K. pneumoniae infection during a recent shortage of newly effective antibiotic development.


Introduction
Carbapenem resistance in Enterobacteriaceae is currently a global concern [1]. Among Enterobacteriaceae, Klebsiella pneumoniae plays a major role in resistance to carbapenems [2,3]. Most K. pneumoniae isolates are resistant to carbapenem by carbapenemase production [2,4]. Apart from metallo-carbapenemases (such as NDM and IMP) and an Ambler class A carbapenemase (KPC) that strongly hydrolyze carbapenemases, OXA-48 produces a weak carbapenemase activity yet is responsible for a broad range of carbapenem susceptibility [4]. Although bla OXA-48 -carrying K. pneumoniae isolates are endemic in Turkey, India, some countries in Europe, Africa, and the Middle East, it has been reported to be widespread worldwide [5][6][7][8].

OXA-48 Expression Level among Different Carbapenem-Resistant K. pneumoniae
All blaOXA-48-carrying K. pneumoniae isolates had comparatively different levels of carbapenem susceptibility. To investigate the correlation of blaOXA-48 with carbapenem resistance, blaOXA-48 expression among different clonal clusters was evaluated. In cluster A and a singleton (KP162), the relative expression of blaOXA-48 in each isolate was compared to that of imipenem-susceptible isolate, KP203. Overexpression of blaOXA-48 was observed in KP166 and KP197 isolates which exhibited imipenem MIC of 32 and 8 mg/L, respectively (Figure 2a), indicating that blaOXA-48 expression was probably related to carbapenem resistance level in K. pneumoniae cluster A. Of note, blaOXA-48 expression of KP162 isolate belonging to a singleton was significantly downregulated compared to that of KP203, indicating that other mechanisms played a role in carbapenem resistance of KP162 ( Figure  2a). Among isolates belonging to cluster B, overexpression of blaOXA-48 was significantly observed in KP241 and KP1184 isolates whose imipenem MICs were 128 and 64 mg/L, respectively ( Figure 2b). No significant expression of blaOXA-48 was observed in carbapenem-resistant KP260 and KP262 isolates compared to that of imipenem-susceptible KP221 isolate (Figure 2b). Therefore, among cluster B, expression levels were slightly related to carbapenem susceptibility in a few isolates. In conclusion, blaOXA-48 expression correlated to carbapenem resistance in some strains likely in a strain-specific manner, indicating the involvement of other mechanisms.

OXA-48 Expression Level among Different Carbapenem-Resistant K. pneumoniae
All bla OXA-48 -carrying K. pneumoniae isolates had comparatively different levels of carbapenem susceptibility. To investigate the correlation of bla OXA-48 with carbapenem resistance, bla OXA-48 expression among different clonal clusters was evaluated. In cluster A and a singleton (KP162), the relative expression of bla OXA-48 in each isolate was compared to that of imipenem-susceptible isolate, KP203. Overexpression of bla OXA-48 was observed in KP166 and KP197 isolates which exhibited imipenem MIC of 32 and 8 mg/L, respectively (Figure 2a), indicating that bla OXA-48 expression was probably related to carbapenem resistance level in K. pneumoniae cluster A. Of note, bla OXA-48 expression of KP162 isolate belonging to a singleton was significantly downregulated compared to that of KP203, indicating that other mechanisms played a role in carbapenem resistance of KP162 ( Figure 2a). Among isolates belonging to cluster B, overexpression of bla OXA-48 was significantly observed in KP241 and KP1184 isolates whose imipenem MICs were 128 and 64 mg/L, respectively (Figure 2b). No significant expression of bla OXA-48 was observed in carbapenem-resistant KP260 and KP262 isolates compared to that of imipenem-susceptible KP221 isolate (Figure 2b). Therefore, among cluster B, expression levels were slightly related to carbapenem susceptibility in a few isolates. In conclusion, bla OXA-48 expression correlated to carbapenem resistance in some strains likely in a strain-specific manner, indicating the involvement of other mechanisms.

Impact of bla OXA-48 -Carrying Plasmid in Carbapenem Susceptibility
Transformation of bla OXA-48 -carrying plasmids to E. coli DH5α was used to investigate the impact of bla OXA-48 -carrying plasmids on carbapenem susceptibility. The presence of bla OXA-48 -carrying plasmid evidently increased imipenem MIC from 32-fold to 512-fold, whereas the MICs of meropenem and ertapenem appeared to raise at least 1024-fold in E. coli transformants ( Table 2). Although KP162 isolate had a low level of bla OXA-48 expression, its plasmid showed remarkably increasing carbapenem MICs in the transformant ( Figure 2a and Table 2). This was similar to imipenem-susceptible KP203 isolate that its transformant was resistant to all tested carbapenems ( Table 2). Among cluster A, the transformants of KP166 and KP197 with high carbapenem resistance and bla OXA-48 expression had the same level of carbapenem resistance as that of imipenem-susceptible isolate KP203 (Table 2 and Figure 2a). Although isolates that belonged to cluster B (including KP221, KP260, and KP262) had near-identical expression level of bla OXA-48 , they differed in carbapenem susceptibility ( Figure 2b). Moreover, their transformants showed almost equal carbapenem MICs (Table 2). On the other hand, KP241 and KP1184 overexpressed bla OXA-48 , but its transformants had the lowest carbapenem MICs (Figure 2b and Table 2). The results of bla OXA-48 expression and transformation indicate that apart from bla OXA-48 , there may be other mechanisms involved in carbapenem resistance. Additionally, marked changes in the MICs of ceftriaxone, fosfomycin, and amikacin were also observed in all transformants ( Table 2). . The relative number of blaOXA-48 transcripts of K. pneumoniae isolates was normalized to 16S rRNA expression and calculated using the 2 −ΔΔct method compared to the expression of imipenem-susceptible K. pneumoniae isolates KP203 and KP221. p-values were calculated using unpaired t-test ( **, p-value < 0.01; ***, p-value < 0.001 and ns, non-significant).

Impact of blaOXA-48-Carrying Plasmid in Carbapenem Susceptibility
Transformation of blaOXA-48-carrying plasmids to E. coli DH5α was used to investigate the impact of blaOXA-48-carrying plasmids on carbapenem susceptibility. The presence of blaOXA-48-carrying plasmid evidently increased imipenem MIC from 32-fold to 512-fold, whereas the MICs of meropenem and ertapenem appeared to raise at least 1024-fold in E. coli transformants ( Table 2). Although KP162 isolate had a low level of blaOXA-48 expression, its plasmid showed remarkably increasing carbapenem MICs in the transformant ( Figure  2a and Table 2). This was similar to imipenem-susceptible KP203 isolate that its transformant was resistant to all tested carbapenems ( Table 2). Among cluster A, the transformants of KP166 and KP197 with high carbapenem resistance and blaOXA-48 expression had the same level of carbapenem resistance as that of imipenem-susceptible isolate KP203 (Table 2 and Figure 2a). Although isolates that belonged to cluster B (including Relative bla OXA-48 expression levels of K. pneumoniae. RT-qPCR assay of bla OXA-48 expression was performed in K. pneumoniae cluster A (a), a singleton isolate KP162 (a), and cluster B (b). The relative number of bla OXA-48 transcripts of K. pneumoniae isolates was normalized to 16S rRNA expression and calculated using the 2 −∆∆ct method compared to the expression of imipenemsusceptible K. pneumoniae isolates KP203 and KP221. p-values were calculated using unpaired t-test (**, p-value < 0.01; ***, p-value < 0.001 and ns, non-significant).  Figure 3). The presence of both OmpK35 and OmpK36 was only observed in imipenem-and meropenem-susceptible KP203 and KP221 belonging to clusters A and B, respectively. These results demonstrate that bla OXA-48 expression together with loss of OMPs, particularly OmpK36, have an affluential role on carbapenem susceptibility among bla OXA-48 -carrying K. pneumoniae clinical isolates.   Apart from Omp profile detection by SDS-PAGE, the expression levels of ompK genes and killing effects of imipenem were determined to correlate with the resistance mechanism. Among isolates in cluster A and a singleton, isolate KP162 showed significantly re-

Ompk35 and Ompk36 Expression and Killing Effect of Imipenem against bla OXA-48 -Carrying K. pneumoniae
Apart from Omp profile detection by SDS-PAGE, the expression levels of ompK genes and killing effects of imipenem were determined to correlate with the resistance mechanism. Among isolates in cluster A and a singleton, isolate KP162 showed significantly reduced expressions of ompK35 and ompK36 genes (Figure 4a,c). The most downregulation was of ompK35 was KP1184 (Figure 4b), but ompK36 showed a similar expression level in cluster B (Figure 4d).   RT-qPCR assay of ompK35 and ompK36 expression was performed in K. pneumoniae cluster A (a,c), a singleton isolate KP162 (a,c), cluster B (b,d). The relative number of ompK35 and ompK36 transcripts of K. pneumoniae isolates was normalized to 16S rRNA expression and calculated using the 2 −∆∆ct method compared to the expression of imipenem-susceptible K. pneumoniae isolates KP203 and KP221. p-values were calculated using unpaired t-test (*, p-value < 0.05; **, p-value < 0.01; ***, p-value < 0.001 and ns, non-significant). Killing effect of imipenem against K. pneumoniae cluster A and a singleton (e), and cluster B (f). Mean of bacterial viable cells at the treatment with 1 × MIC of imipenem was plotted at 0, 2, 4, 6, and 24 h of incubation. All experiments were performed in triplicate and the detection limit of the viable cells was 10 2 CFU/mL (dashed lines). The range of azithromycin MIC of all nine bla OXA-48 -carrying K. pneumoniae was 8-1024 mg/L. Most of the K. pneumoniae isolates were resistant to azithromycin (Table 1). However, azithromycin was still effective against isolates KP203 and KP1184 with MICs of 8 and 16 mg/L, respectively (Table 1). KP260 and KP262 were highly resistant with MIC of 1024 mg/L. The presence of erythromycin resistance methylase genes was determined by using PCR. All isolates carried ermC gene (Table 3). Moreover, high-level azithromycinresistant isolates (KP260 and KP262) carried not only ermC but also ermB, and these isolates belonged to cluster B (Figure 1 and Table 3).  In addition, the presence of fosfomycin-modifying enzyme genes was detected in all K. pneumoniae, of which five and four isolates were susceptible and resistant to fosfomycin, respectively. The most common gene was fosA5, found in all isolates ( Table 3). The coexistence of fosA3 with fosA5 was exhibited in five isolates (KP197, KP166, KP262, KP1184, and KP241).

Synergistic Activity of Azithromycin with Other Antibiotics against bla OXA-48 -Carrying K. pneumoniae
The in vitro activities of azithromycin in combination with either imipenem, colistin, or fosfomycin against bla OXA-48 -carrying K. pneumoniae were performed by using the checkerboard assay. Despite azithromycin resistance, the synergism was revealed in combination with colistin (88.89%), fosfomycin (77.78%), and imipenem (11.11%), respectively (Table 3). Azithromycin with colistin, the most effective combination, exhibited synergism in all isolates, except in ermCand ermB-co-carrying KP262. Azithromycin and fosfomycin combination showed a synergistic effect in all isolates belonging to cluster A and a singleton that carried ermC. Remarkably, synergism of this combination was also observed against isolates in cluster B that carried ermC without ermB. The combination of azithromycin with imipenem had the least activity that was synergistic against only isolate KP1184 (Table 3). No antagonism was observed in our study.

Time-Kill Curves of Azithromycin Combination with Colistin against K. pneumoniae
According to checkerboard results, the most effective combination was azithromycin plus colistin. We, therefore, investigated the activity of this combination by time-killing assay. The time-killing curves of all isolates are shown in Figure 5. Although all isolates were susceptible to colistin, the regrowth was usually observed at 2-6 h after exposure to colistin alone (Figure 5a-i). This was similar to the presence of azithromycin alone in which the concentration of 1 × MIC could not eliminate the growth of both azithromycinsusceptible and azithromycin-resistant isolates. These results indicated that single use of either colistin or azithromycin may be inadequate for bla OXA-48 -carrying K. pneumoniae.
Antibiotics 2021, 10,1551 assay. The time-killing curves of all isolates are shown in Figure 5. Although all were susceptible to colistin, the regrowth was usually observed at 2-6 h after exp colistin alone (Figure 5a-i). This was similar to the presence of azithromycin a which the concentration of 1 × MIC could not eliminate the growth of both azithro susceptible and azithromycin-resistant isolates. These results indicated that singl either colistin or azithromycin may be inadequate for blaOXA-48-carrying K. pneumo

Discussion
Currently, carbapenem-resistant K. pneumoniae, an urgent threat, has spread wide, including Thailand [5,[18][19][20]. The predominance of carbapenem-resistant moniae in Thailand and any other country in Asia is NDM producers or NDM wit 48 co-producers, but OXA-48 alone producers are also reported [9,19]. OXA-48 pr exhibit a wide range of carbapenem susceptibility. The phenotype of blaOXA-48-car pneumoniae in our study is slightly in accordance with that of isolates from Taiw most isolates were resistant to carbapenems and all isolates were resistant to ert In combination, the synergism was observed in eight bla OXA-48 -carrying K. pneumoniae isolates (KP162, KP221, KP197, KP203, KP166, KP260, KP262, and KP1184) ( Figure 5). Notably, KP241 that showed synergy by checkerboard assay had no synergy result by time-killing assay and vice versa for KP262 isolate (Table 3, Figure 5c,h), indicating nonaccordance of checkerboard and time-killing assay.

Discussion
Currently, carbapenem-resistant K. pneumoniae, an urgent threat, has spread worldwide, including Thailand [5,[18][19][20]. The predominance of carbapenem-resistant K. pneumoniae in Thailand and any other country in Asia is NDM producers or NDM with OXA-48 co-producers, but OXA-48 alone producers are also reported [9,19]. OXA-48 producers exhibit a wide range of carbapenem susceptibility. The phenotype of bla OXA-48 -carrying K. pneumoniae in our study is slightly in accordance with that of isolates from Taiwan that most isolates were resistant to carbapenems and all isolates were resistant to ertapenem [21]. The silence of bla OXA-48 was revealed in imipenem-and/or meropenem-susceptible K. pneumoniae isolates, including two isolates (KP203 and KP221) in our study [20]. Imipenem alone had inadequate activity in vivo against imipenem-susceptible isolates carrying bla OXA-48 and led to treatment failure [22,23]. These results indicate an inappropriate treatment for imipenem-and/or meropenem-susceptible K. pneumoniae carrying bla OXA-48 by carbapenems. In contrast, imipenem or meropenem monotherapy has been reported to be effective against OXA-48-producers [24]. Not only carbapenems but also other antibiotics have limited activities to bla OXA-48 -carrying K. pneumoniae that are MDR strains. Fortunately, none of our isolates were resistant to the last line antibiotic, colistin. Colistin resistance became widespread among carbapenem-resistant K. pneumoniae [25]. According to the clonality in our study, this data indicates the clonal spread (ERIC-PCR cluster A and B) of bla OXA-48 -carrying K. pneumoniae in our hospital. OXA-48 strongly hydrolyzes ertapenem rather than hydrolyzing imipenem and meropenem [26,27]. It is in accordance with our results that all bla OXA-48 -carrying K. pneumoniae were resistant to ertapenem. However, phenotypes of imipenem and meropenem susceptibility were diverse among our isolates, indicating the involvement of other resistance determinants apart from bla OXA-48 expression level. E. coli transformants carrying bla OXA-48 -plasmids isolated from all K. pneumoniae isolates displayed a rising of carbapenem MICs. This experiment indicated that different plasmids displayed different levels of carbapenem resistance. Nevertheless, colistin with azithromycin remained effective against almost all of these isolates. In this study, we used E. coli DH5α as a recipient cell due to lacking K. pneumoniae competent cells. Making in-house competent cells is arduous, and K. pneumoniae reference strains are unavailable in our facility. Although they are different species, the spread of these plasmids generally occurs between K. pneumoniae and E. coli [28].
Porins, OMPs acting as pores that the specific substrates can diffuse into intracellular compartment, operating as carbapenem entries in K. pneumoniae are OmpK35 and OmpK36 [10]. Loss of OmpK35 results in increase imipenem and meropenem MICs in K. pneumoniae producing ESBLs [2,29]. The deletion of ompK35 results in 2-fold and 4-fold increase of imipenem and ertapenem MICs, respectively, but no change of these MICs is observed in ompK36 deletion strain indicating that OmpK35 plays a superior role to OmpK36 in carbapenem susceptibility [30]. This is similar to our results which showed that isolates lacking OmpK35 or lacking both OmpK35 and OmpK36 had no difference in carbapenem resistance levels (Table 3). However, loss of OmpK36 together with KPC production confers resistance to carbapenem in K. pneumoniae [10,31]. The mutations of OmpK36, particularly, insertion of Gly115-Asp116 into loop 3 with KPC or OXA-48 production resulted in the elevation of carbapenem MICs [32]. It is in agreement with our results which showed that higher resistance to carbapenems was found in isolates with loss of OmpK35 and OmpK36 compared to isolates with intact OMPs. Thereby, OXA-48 production with lacking OMPs affects the magnitude of carbapenem resistance in bla OXA-48 -carrying K. pneumoniae isolates. Isolate KP162, which was resistant to carbapenem with the lowermost expression of bla OXA-48 (Figure 2a), also showed the most downregulation of ompK35 and ompK36 (Figure 4a,c). Although all isolates were sharply killed in early exposure to imipenem (2-4 h), the regrowth usually occurred after 6 h (Figure 4e,f), indicating that imipenem alone is inadequate for treatment of bla OXA-48 -producing K. pneumoniae. Notably, the results of the transformation experiment demonstrated that the E. coli transformants were resistant to not only carbapenems but also other antibiotics, indicating multiple resistance genecarrying plasmids. This data supported the evidence that bla OXA-48 -carrying K. pneumoniae frequently were multidrug-resistant organisms.
Due to multidrug resistance in bla OXA-48 -carrying K. pneumoniae, the treatment options were limited. Antibiotic combination therapy is inevitably used during a shortage of novel effective antibiotics. The main aim of this study was to forage and obtain the combinations of antibiotics that may have a potential effect against bla OXA-48 -carrying K. pneumoniae. Azithromycin, a macrolide antibiotic that inhibits translocation and transpeptidation of protein synthesis by binding to 50S ribosomal subunit at 23S rRNA [15], was chosen to be combined with other commonly used antibiotics (imipenem, colistin, and fosfomycin). The majority of isolates in our study were resistant to azithromycin, indicating an ineffective effect of azithromycin in single use against bla OXA-48 -carrying K. pneumoniae. All isolates carried ermC alone or with ermB. These azithromycin resistance genes are generally found in Gram-positive cocci and spread to Gram-negative bacilli by plasmid-mediated horizontal gene transfer [33,34].
Remarkably, in our study, azithromycin with colistin was the most potent combination observed by checkerboard assay, and its synergism was confirmed by the time-killing assay against bla OXA-48 -carrying K. pneumoniae. Furthermore, we performed the in vitro activity of this combination against additionally 26 carbapenem-resistant K. pneumoniae clinical isolates producing various patterns of carbapenemase. The synergism was observed in 15 isolates (57.69%) (Supplementary Materials Table S1). Interestingly, only half of NDM producers (6 of 10 isolates) and NDM with OXA-48 producers (6 of 12 isolates) showed synergism of the antibiotic combination (Table S1). The in vitro synergism of this combination has been reported against MDR P. aeruginosa, MDR-Acinetobacter baumannii, colistin-resistant E. coli, and MDR-K. pneumoniae [16,35]. The synergistic mechanism of the combination is revealed in MDR-A. baumannii that colistin, like antimicrobial peptide, disrupts the bacterial cell membrane and enhances azithromycin uptake, resulting in bacterial death [16]. The limitation of our study was that the synergistic mechanism of azithromycin and colistin was not performed. Moreover, the purpose of using azithromycin, especially in cystic fibrosis caused by P. aeruginosa, is to reduce inflammation response and to inhibit biofilm formation, not to kill the bacteria [36]. Recently, azithromycin has been used as an immunomodulator to modulate immune response to respiratory tract infection [37].
Additionally, synergism was remarkably observed in the combination of azithromycin and fosfomycin. Our previous study demonstrates that fosfomycin resistance genes (fosA5 and fosA3) silence in carbapenem-resistant K. pneumoniae (including bla OXA-48 -carrying isolates) probably leading to insufficient activity [9]. Therefore, the activity of azithromycin with fosfomycin was not determined by time-killing assay. Moreover, amikacin seemed effective against OXA-48-producing strains. Unfortunately, amikacin-heteroresistant subpopulations have been reported among amikacin-susceptible populations, plausibly leading to treatment failure [38]. Therefore, amikacin was not included in the combination testing in this study.

Bacterial Strains
Thirty-five OXA-48-carrying K. pneumoniae isolated from nonduplicate patients were collected from the routine laboratory's stocks at the King Chulalongkorn Memorial Hospital, Bangkok, Thailand, from 2017-2018. Our study was approved by the Institutional Review Board of Faculty of Medicine, Chulalongkorn University (IRB 221/62). Neither human nor animal was involved in this study. The need for consent was waived by the ethics committee.

Detection of Antibiotic Resistance Genes
Multiplex PCR was used for the detection of carbapenemase genes including bla NDM-like , bla OXA-48-like , and bla KPC-like as described by Poirel et al. [40]. The presence of metallocarbapenemase genes including bla IMP-like and bla VIM-like was performed by using multiplex PCR as described by Ellington et al. [41]. Fosfomycin-modifying enzyme genes including fosA5 and fosA3 were detected as in a previous study [9]. Erythromycin resistance methylase genes including ermA, ermB, ermC, and ermF were detected by PCR [42][43][44]. The primers used in this study are listed in Table S3.

Clonal Study
The genetic relatedness of nine OXA-48-producing K. pneumoniae isolates was characterized by the Enterobacterial Repetitive Intergenic Consensus (ERIC) PCR (ERIC-PCR) [45]. The dendrogram of ERIC-PCR was generated by BioNumerics software, version 8.0, using the UPGMA. Clonal relatedness was defined as >90% similarity.

Expression Level of bla OXA-48 , ompK35 and ompK36
The expression level of bla OXA-48 , ompK35, and ompK36 mRNA was studied by RT-qPCR. Total RNA of nine bla OXA-48 -carrying K. pneumoniae was extracted by using TRIzol ® Reagent (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized by using RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, Vilnius, Lithuania). The number of bla OXA-48 transcripts was determined by using the Luna ® Universal qPCR master and the QuantStudio5 (Applied Biosystems, Foster City, CA USA). The relative number of bla OXA-48 transcripts was normalized with 16S rRNA and determined by using the 2 −∆∆ct method. This experiment was performed in triplicate.

Transformation of the bla OXA-48 -Carrying Plasmids into E. coli DH5α
To investigate the role of bla OXA-48 -carrying plasmids isolated from K. pneumoniae on carbapenem susceptibility, the plasmid was extracted by using HiYield Plasmid Mini Kit (RBC, New Taipei City, Taiwan) and transformed into E. coli DH5α by using the heat shock method. E. coli DH5α transformants were selected on MHA supplemented with imipenem. The transformants were confirmed the presence of bla OXA-48 plasmid by PCR and tested for antimicrobial susceptibility by broth microdilution method.

Outer Membrane Protein (OMP) Study
To study OMP profiles of bla OXA-48 -carrying K. pneumoniae, OMPs were extracted by ultracentrifugation with N-Lauroylsarcosine (Merck Millipore, Kenilworth, NJ, USA) extraction as previously described [46]. Briefly, log-phase growth of K. pneumoniae in Nutrient Broth (Becton Dickenson Difco, Sparks, MD, USA) was broken by sonication (Sonics and Materials, Inc., Newtown, CT, USA). OMPs were extracted with N-Lauroylsarcosine solution and collected by ultracentrifugation. OMP profile was determined by SDS-PAGE.

Checkerboard Assay
The activity of azithromycin in combination with other antimicrobial agents including imipenem, colistin, and fosfomycin, was determined by checkerboard assay in 96-well culture plates. Each well of each column contained CAMHB supplemented with two-fold serial dilution of azithromycin, whereas the serial dilution of either imipenem, colistin, or fosfomycin was added in each well of each row. In the case of fosfomycin, 25 mg/L of G6P was also added. K. pneumoniae was inoculated into the checkerboard plates, and the plates were incubated for 18-24 h at 37 • C. The fractional inhibitory concentration index (FICI) was calculated using the following Equation: FICI = MIC of drug A in combination MIC of drug A alone + MIC of drug B in combination MIC of drug B alone The interpretation was defined as following, synergism (FICI ≤ 0.5), no interaction (FICI > 0. , and antagonism (FICI > 4).

Time-Killing Assay
The killing effect of imipenem alone was performed by the time-killing assay. Briefly, viable cells of K. pneumoniae exposure to 1 × MIC of imipenem were determined after incubation time for 0, 2, 4, 6, and 24 h at 37 • C with shaking were counted on MHA plates. The synergism of azithromycin in combination with colistin was performed by the time-killing assay. Briefly, viable cells of K. pneumoniae in different conditions including growth control (no antibiotic), 1 × MIC of azithromycin, 1 × MIC of colistin, and 1 × MIC of azithromycin plus 1 × MIC of colistin after incubation for 0, 2, 4, 6, and 24 h at 37 • C with shaking were counted on MHA plates. This experiment was performed in triplicate. The synergism was defined as the reduction of the viable cell at least 2log10-fold compared to the most effective single antibiotic at 24 h of incubation. The bactericidal activity was defined as the reduction of the viable cell at least 3log10-fold compared to the initial viable cell after 24 h of incubation.

Statistical Analysis
The statistical analysis was performed by with GraphPad Prism 5 (unpaired t-test) (GraphPad Software, San Diego, CA, USA).

Conclusions
Among two major clonal spreading events, bla OXA-48 -carrying K. pneumoniae was responsible for a wide range of carbapenem MICs, especially imipenem MIC. Imipenemsusceptible isolates had intact OmpK35 and OmpK36. OXA-48 production with lacking OMPs resulted in high resistance to carbapenems. The most effective combination was azithromycin with colistin. Although azithromycin is not currently used to treat K. pneumoniae, its combination with colistin may provide a potential activity for bla OXA-48 -carrying K. pneumoniae. Further in vivo study is needed to assess the application of this antibiotic combination. Our results highlighted azithromycin for bla OXA-48 -carrying K. pneumoniae treatment at least in part of novel combination therapy and knowledge for novel antibiotic development.