IS5 Element Integration, a Novel Mechanism for Rapid In Vivo Emergence of Tigecycline Nonsusceptibility in Klebsiella pneumoniae

Abstract

Tigecycline nonsusceptibility is concerning because tigecycline is increasingly relied upon to treat carbapenem- or colistin-resistant organisms. In Enterobacteriaceae, tigecycline nonsusceptibility is mediated by the AcrAB-TolC efflux pump, among others, and pump activity is often a downstream effect of mutations in their transcriptional regulators, cognate repressor genes, or noncoding regions, as demonstrated in Enterobacteriaceae and Acinetobacter isolates. Here, we report the emergence of tigecycline nonsusceptibility in a longitudinal series of multidrug-resistant (MDR) and extensively drug-resistant (XDR) Klebsiella pneumoniae isolates collected during tigecycline therapy and the elucidation of its resistance mechanisms. Clinical isolates were recovered prior to and during tigecycline therapy of a 2.5-month-old Honduran neonate. Antimicrobial susceptibility tests to tigecycline determined that the MIC increased from 1 to 4 μg/ml prior to the completion of tigecycline therapy. Unlike other studies, we did not find increased expression of ramA, ramR, oqxA, acrB, marA, or rarA genes by reverse transcription-quantitative PCR (qRT-PCR). Whole-genome sequencing revealed an IS5 insertion element in nonsusceptible isolates 85 bp upstream of a putative efflux pump operon, here named kpgABC, previously unknown to be involved in resistance. Introduction of the kpgABC genes in a non-kpgABC background increased the MIC of tigecycline 4-fold and is independent of a functional AcrAB-TolC pump. This is the first report to propose a function for kpgABC and identify an insertion element whose presence correlated with the in vivo development of tigecycline nonsusceptibility in K. pneumoniae.

INTRODUCTION

Tigecycline is an extended-spectrum glycylcycline. Tigecycline nonsusceptibility is concerning given the growing number of infections otherwise resistant to all other antibiotics, including carbapenems and colistin. Currently, tigecycline is not approved for use in pediatrics or neonates and is used only when there are no other alternatives. In the few pediatric case studies published, 1 to 2 mg/kg of body weight/day appeared to be an effective dose without further complications (1, 2). A reduction in mortality, compared to that for monotherapy alone, has been noted when treatment is combined with colistin (3). However, these case studies failed to report if isolates were collected throughout tigecycline therapy. It is therefore unknown whether tigecycline MIC values increase during treatment.

Previously reported mechanisms of tigecycline nonsusceptibility include mutations of efflux pumps or their regulator genes, especially ramR and S10 ribosomal protein mutations (4). Longitudinal data are uncommon, with most studies reporting data from adults or cultures from a single time point during patient treatment. Rarer still are data from pediatric infections, especially those occurring in the tropics. One report from India (2) described upregulated expression of the AcrAB-TolC efflux pump, but this was seen in a minority of tested isolates (2 of 57).

AcrAB-TolC efflux pump activity is regulated by several AraC-type regulators, including RamA, MarA, RarA, and SoxS (5–7). Whole-genome sequencing has identified point mutations within regulator genes or their noncoding regions responsible for constitutive acrAB expression (7–9). Recently, the RarA-regulated OqxAB efflux pump system has been identified in Klebsiella pneumoniae. However, this may be a supplementary nonsusceptibility mechanism, as its increased expression relies on the presence of a functional AcrAB efflux pump (9). In addition to point mutations, insertion sequence (IS) elements can upregulate expression of acrAB and adeABC efflux pump genes and produce tigecycline nonsusceptibility in Escherichia coli and Acinetobacter baumannii, respectively, but has not yet been described in K. pneumoniae (10, 11). The regulation of the OmpF porin by OmpR can also independently contribute to E. coli and K. pneumoniae tigecycline nonsusceptibility (12, 13). Here, we report a novel mechanism of tigecycline nonsusceptibility which evolved in vivo prior to completion of tigecycline therapy in a treatment-naive individual.

MATERIALS AND METHODS

This study was undertaken as a quality improvement infection control initiative, authorized by U.S. Army Medical Command policy number 13-016 and the Walter Reed Army Institute of Research Institutional Review Board work unit number 1812.

Patient and isolate data.

A newborn, term female infant born with an omphalocele underwent multiple abdominal surgeries in Honduras and was transferred to a U.S. referral center 26 days after birth for more definitive care. Isolates were cultured throughout antibiotic treatment and hospitalization (Table 1; see also Table S2 in the supplemental material). While in Honduras, the infant received numerous antimicrobials, including carbapenems, aminoglycosides, and extended-spectrum penicillins, but not tigecycline. After transfer to the United States, surveillance and blood cultures grew extensively drug-resistant (XDR) K. pneumoniae susceptible only to colistin and tigecycline. Antibiotic therapy was modified to 0.25 mg/kg/day colistin and 2 mg/kg/day tigecycline. Bacteremia resolved after 3 days of therapy, but respiratory and peritoneal cultures remained positive. Both antibiotics were continued until decreased tigecycline susceptibility was noted prior to completion of a 3-week treatment plan, at which point only colistin therapy was continued and the infection resolved.

TABLE 1.

K. pneumoniae isolates selected for further study and tigecycline MICs

Isolate or strain	Collection day	Ongoing tigecycline treatment	MIC (mg/liter)
			Etest	Microbroth dilutiona	Vitek-2
KP40	0	No	1	1	1
KP47	7	Yesc	1	1	1
KP49	9	Yes	1	1	1
KP52-1	12	Yes	1	NDd	1
KP64	24	Yes	3	4	ND
KP66	26	No	4	4	4
VA6048(ramA) strainb	NAe	NA	4	8	ND
VA6048 (ramRVA12262-WT) strainb	NA	NA	1	1	ND

^aManual broth microdilution experiments ran in replicates of three with at least two subsamples per replicate. Medium was made immediately prior to use, and drug was a fresh suspension.

^bControl strains; see reference 8.

^cTigecycline therapy was initiated on day four, with the first posttherapy isolate collected on day 7.

^dND, not determined.

^eNA, not applicable.

Susceptibility testing.

Multiple methods were used to quantify drug susceptibilities, including Etest (bioMérieux, Durham, NC), manual broth microdilutions (MBD), and the Vitek-2 antimicrobial sensitivity testing platform. Except for testing on the Vitek-2, which was performed once after quality control confirmation, all susceptibility tests were repeated at least three times on select Klebsiella clinical isolates (Table 1).

Bacterial isolates.

In addition to the K. pneumoniae isolates listed in Table 1 and in Table S2 in the supplemental material, we used E. coli JM109, AG100, and AG100A ΔacrAB strains (14) as surrogates to our K. pneumoniae isolates, as transformation was unsuccessful. JM109, AG100, and AG100A ΔacrAB strains were transformed with pMQ300 (15) or pMQ300 containing the kpgABC genes amplified from K. pneumoniae KP64 (see below) to test whether overexpression alters antibiotic resistance profiles.

Efflux pump inhibition assays and growth studies.

Cation-adjusted Mueller-Hinton (CA-MH) agar plates with and without 200 μM phenylalanine-arginine β-naphthylamide (PAβN) were swabbed with 0.5 McFarland standard cell suspensions, and tigecycline Etests were performed (16). For growth studies, CA-MH medium was inoculated from freezer stocks to CA-MH agar before transfer to fresh broth and grown to an optical density at 600 nm (OD₆₀₀) of 0.5 prior to growth study inoculation. Growth studies were repeated three times, with two or three replicates per experiment.

Clonality and genomic sequencing.

Clonality of patient isolates was determined by pulsed-field gel electrophoresis and whole-genome mapping (WGM), according to methods previously described (17). Genomic DNA was isolated using a genomic DNA purification kit (Fermentas) and sequenced using the Illumina MiSeq system. Sequence reads were assembled with Newbler (Roche Diagnostics Corp., Branford, CT). Comparative genomics and qRT-PCR primer designs were performed using Geneious (Biomatters Ltd., Auckland, New Zealand).

kpgABC expression studies.

Samples were grown in CA-MH broth to an OD₆₀₀ of 0.6 before addition of RNAprotect bacteria reagent and RNA extraction using the mirVana RNA isolation kit (Ambion). A total of 500 ng of RNA was converted to cDNA using Superscript II (Life Technologies) followed by qRT-PCR using SsoFast Sybr green (Bio-Rad). Non-RT controls were performed to ensure no DNA contamination and that sample replicative efficiencies for standard curves were between 95 and 100%, depending on target and primer sets (see Table S1 in the supplemental material). Gene targets included marA, oqxA, acrA, kpgA, kpgB, the sensor histidine kinase gene, and ompR-like genes flanking kpgABC, rarA, and ramA, with mdh and rpoB genes as controls. Control strains included the tigecycline-resistant clinical isolate VA6048(ramR) and its complement VA6048(ramR_VA12262-WT) (8). Three technical replicates were performed per experiment, with a minimum of three experimental replicates. Data analysis used the cycle threshold (ΔΔC_T) method, reported values were averaged across all experiments, and the standard errors of the means were calculated.

kpgABC overexpression.

The kpgABC genes were PCR amplified from KP64 genomic DNA using primers kpgABC-HindIII-F (5′-TCAAAGCTTATGATTTATAAAGGCAACGACAAA-3′) and kpgABC-BamHI-R (5′ TAAGGATCCCTACTTCGTCCATTCTTTATGCA-3′). The resulting amplicon and pMQ300 plasmid (15) were digested using HindIII and BamHI restriction enzymes (Thermo Fisher, Pittsburgh, PA) and ligated to create pMQ300-kpgABC with a selectable hph resistance marker. pMQ300-kpgABC was transformed into E. coli AG100 and AG100A ΔacrAB strains and selected on tryptic soy agar (TSA) plates containing 100 μg/ml hygromycin B. Colonies containing pMQ300-kpgABC were tested for tigecycline and colistin susceptibilities using Etest strips (bioMérieux, Durham, NC) in triplicate per the manufacturer's protocol. All other antibiotics were tested using the Vitek-2 antibiotic susceptibility testing platform (bioMérieux).

Nucleotide sequence accession numbers.

Sequence contigs of isolates KP52-3 (accession number JPGS00000000) and KP66 (number JPGT00000000) are deposited at DDBJ/EMBL/GenBank.

RESULTS

A total of 13 extended-spectrum-beta-lactamase (ESBL)-positive K. pneumoniae isolates were collected during treatment. All isolates were XDR, according to the recent proposed international classification scheme (18) (see Table S2 in the supplemental material), and tested positive for bla_KPC by RT-PCR (19). Isolates KP40 through KP52-2 were susceptible to colistin and tigecycline based on the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines. Colistin and tigecycline antibiotic coverage was initiated 4 days after the initial culture was obtained, with the second culture taken 3 days later (7 days after initial culture). Multiple antimicrobial testing methods reported an MIC of 2 μg/ml or greater to tigecycline in isolates KP64 and KP66, while previous isolates were considered susceptible. Some variations between testing methods were noted, but this did not alter interpretation of results (Table 1). Isolates KP52-3, KP64, and KP66 became nonsusceptible (MIC = 16 μg/ml) to colistin, while all previous isolates were susceptible (see Table S2); however, colistin therapy after dosing with a combination of tigecycline and colistin resolved the infection in vivo, while isolates maintained in vitro nonsusceptible MIC values. Isolates KP40, KP47, KP49, KP64, and KP66 were chosen for further analysis based on their antibiotic profiles and collection dates (Table 1).

Efflux pump inhibition and growth assays.

Mutations leading to increased efflux pump activity have been reported in tigecycline-nonsusceptible isolates (8, 11, 20, 21). To determine if efflux pumps are involved in reduced tigecycline susceptibility, we conducted efflux pump inhibition assays on isolates KP47, KP52-1, KP64, and KP66 and the VA6048(ramR) control strain. We found that exposure of nonsusceptible strains to tigecycline in the presence of PAβN resulted in an average 4-fold MIC decrease, while susceptible isolate MICs declined slightly from 1 to 0.75 μg/ml, suggesting a role for efflux pumps in nonsusceptible isolates. Growth curves were performed to assess if efflux pump activity was associated with a fitness cost in nonsusceptible isolates. Similar to the findings of De Majumdar et al. (5), tigecycline-nonsusceptible isolates did not differ from susceptible isolates in their rate of growth or final cell concentration, suggesting no fitness cost (data not shown). We were unable, however, to conduct comparative growth studies for nonsusceptible and susceptible isolates in the presence of tigecycline due to the sensitivity of susceptible isolates and the inherent instability of the antibiotic (22).

Comparative genomics.

Isolates KP47, KP49, KP52-1, and KP66 underwent whole-genome mapping to assess for gene inversions or structural changes greater than 2 kb between tigecycline-susceptible and -nonsusceptible strains. A 3- or 4-kb phage insertion was located in isolate KP49 but thought unlikely to be associated with tigecycline nonsusceptibility, given that the insertion was not maintained in the subsequent isolates. To detect differences less than 2 kb and base pair mutations, we sequenced the genomes of isolates KP47, KP49, KP52-3, KP64, and KP66 and focused on comparing previously published genes and their noncoding regions, indicated in tigecycline nonsusceptibility. We found no sequence differences between susceptible and nonsusceptible isolates in the rarA, marA, ramA, ramR, acrA, ompR-like, and oqxA genes previously indicated in tigecycline resistance in other E. coli, A. baumannii, or K. pneumoniae isolates (6, 9, 21, 23). These genes were evaluated further for elevated expression (see below). Of exception, however, was one base pair mutation arising in a noncoding region of KP47 and maintained in later isolates. This single-nucleotide polymorphism (SNP) is located closest to an operon predicted to have phosphonate transport activity but is unlikely to be associated with tigecycline nonsusceptibility, as isolates KP47–KP52-3 remained susceptible. Genome sequencing did not reveal other point mutations that would alter the amino acid sequence in the S10 ribosomal protein or other putative antibiotic efflux pumps (4, 9).

To determine if other previously unidentified mutations contribute to the development of tigecycline nonsusceptibility, we compared the entire genomic sequences of our isolates to each other. This analysis located an IS5 insertion site containing target repeat sequences CTAAGTG and CTAAGG and inverted repeat sequence AAGGTGCGAAYAAG (24) within a promoter region of the KP64 and KP66 isolates. The IS5 element placement was the result of internal transposition events, as the element is found within the genome of earlier isolates and the total genome size of all isolates is the same. This promoter region is predicted to encode three proteins with homology to an ATP-binding cassette (ABC) transporter, a resistance-nodulation-cell division (RND) superfamily multidrug efflux pump transporter, and an N-acetyltransferase with similarity to other known efflux pump genomic architectures (Fig. 1a). The operon and its arrangement, including the two-gene operon containing an OmpR-like response regulator sensor and downstream histidine kinase, are conserved in the K. pneumoniae CG43, MGH78578, JM45, HS11286, and KCTC2242 genomes. These operons are not found in the K. pneumoniae 1084, NTUH-K2044, SB3432, 342, or Kp13 genomes. The putative RND multidrug efflux transporter protein does not have significant homology with other known efflux pump transporter proteins (25–27). AcrB, which is involved in tigecycline resistance, is the closest related protein in the K. pneumoniae proteome at 48.5% protein identity. Other efflux pump proteins are no more than ∼45% similar (Fig. 1b). Therefore, we refer to this arrangement as the K. pneumoniae glycylcycline (kpgABC) operon.

FIG 1.

(a) KP64 and KP66 genome arrangement surrounding the IS5 integration site. Target repeat of IS5 integration occurs directly upstream of the kpgABC. (b) Phylogenic protein identification relatedness of the KpgB efflux protein to other K. pneumoniae efflux pump proteins. Percentages of protein identity to KpgB relative to other proteins are shown after protein names. Proteins without annotation are identified by arbitrary whole-genome sequencing scaffold number. (c) Relative fold gene expression of tigecycline-nonsusceptible KP64 relative to susceptible KP49 using either the mdh or rpoB control genes. Similar results were found for KP47 (data not shown).

Expression analysis.

Expression analysis was performed on isolates KP47, KP49, KP64, and KP66 using primers with at least 97% efficiency and a low detection limit of 100 copies. Control genes mdh and rpoB were validated to be expressed at the same levels in all isolates under our tested conditions. Quantitative RT-PCR analysis showed significant upregulation of the kpgA and kpgB genes in isolate KP64 relative to that in KP47, by 132-fold and 48-fold, respectively (Fig. 1c). Expression of acrA, marA, rarA, oqxA, and ramA was not significantly altered. Integration of the IS5 element in KP64 effected transcription of kpgA and kpgB but not its opposing operon encoding an OmpR-like response regulator, and histidine kinase as the first gene in the operon was expressed similarly in KP47, KP49, KP64, and KP66. To verify if increased expression of acrA and ramA could be detected, the expression of these genes in the VA6048(ramR) control strain relative to the VA6048(ramR_VA12262-WT) control strain was tested (8). Elevation of gene expression of both was seen, suggesting our assay could detect increased ramA expression due to RamR (data not shown). The expression of structural efflux pump gene oqxA, also regulated by RamR, was also not increased in nonsusceptible isolates relative to that in susceptible isolates.

kpgABC overexpression and antibiotic profiles.

The putative kpgABC operon was cloned into plasmid pMQ300 for kpgABC expression studies and transformed into E. coli JM109, AG100, and AG100A ΔacrAB strains, as transformation of K. pneumoniae clinical isolates used in this study was repeatedly unsuccessful. Tigecycline MICs increased by 4-fold from 0.19 μg/ml in E. coli JM109(pMQ300)-expressing cells to 0.75 μg/ml in E. coli JM109(pMQ300-kpgABC)-expressing cells, demonstrating the involvement of kpgABC overexpression in increasing tigecycline resistance (Table 2). To test the involvement of the AcrAB-TolC pump, we employed E. coli AG100 and AG100A ΔacrAB strains, since we lacked a JM109 ΔacrAB mutant. Similar to JM109, the MIC of E. coli AG100 cells increased in those carrying pMQ300 and pMQ300-kpgABC. Susceptibility was further decreased in AG100A(pMQ300) ΔacrAB mutants compared to that in wild-type AG100, as expected. MIC levels increased with the introduction of pMQ300-kpgABC into AG100A ΔacrAB cells, although to a lesser extent than those in AG100 wild-type cells, suggesting that the involvement of KpgABC is independent of the AcrAB efflux pump. No variance in MIC values within triplicate replicates of the same isolates was noted. Antibiotic sensitivities were the same in cells with and without pMQ300, suggesting pMQ300 did not cause deleterious effects. MICs to other antibiotics were tested in kpgABC-expressing isolates, including colistin, but all resulting MICs were equal to that of its non-kpgABC counterpart (data not shown). Together, these data suggest that increased expression of kpgABC leads to a modest, but potentially clinically relevant, increase in tigecycline resistance that is independent of a functional AcrAB efflux pump.

TABLE 2.

E. coli kpgABC overexpression and resulting tigecycline MICs

Strain	MIC (μg/ml)		Fold MIC increaseb
	pMQ300a	pMQ300-kpgABC
JM109	0.19	0.75	4
AG100	0.125	1.0	8
AG100 ΔacrAB mutant	0.064	0.25	4

^aEmpty expression vector.

^bFold increase as a ratio of pMQ300-kpgABC to pMQ300.

DISCUSSION

Our findings are notable for several reasons. First, nonsusceptibility emerged relatively rapidly, before a full course of initial tigecycline therapy for bacteremia could be completed in a patient who was not previously exposed to tigecycline. Second, in vivo tigecycline data from longitudinal series are uncommon, and data from XDR bacterial infections in neonates are especially sparse. Third, nonsusceptibility in this case occurred via a mechanism that has not previously been reported in K. pneumoniae tigecycline nonsusceptibility, namely, an increase of kpgABC expression due to an IS5 insertion element.

In neonates, tigecycline is a drug reserved for use as a last resort. Based upon previous literature, a dose of 2 mg/kg/day tigecycline was administered with colistin. Serial cultures were obtained to monitor for resolution of bacteremia or development of resistance. Although colistin nonsusceptibility later developed during in vitro laboratory testing, we speculate these events are unlikely to be linked to the IS5 element, as the element was absent in the colistin-resistant isolate (KP52-3) and there are no reports in the literature that have directly linked colistin exposure and transposition events. Furthermore, colistin and tigecycline have substantially different mechanisms of action (28, 29).

Resolution of the infection should not diminish the significance of nonsusceptibility emergence for several reasons. First, early and complete source control was achieved. Studies have shown that source control and device removal was just as important as, if not more important than, antibiotics for treating carbapenem-resistant Klebsiella infections (30–32). Second, subsequent downtrending of inflammation and bacterial infection markers (e.g., erythrocyte sedimentation, C-reactive protein, and procalcitonin) was noted shortly after the discovery of nonsusceptibility. Finally, in vitro antibiotic sensitivity assays are not always reliable predictors of the concentration of drugs required for in vivo response. This phenomenon is generally attributed to many factors, including the host response, that impact the outcome of infection.

Many resistance mechanisms have been described in several clinically relevant species. The most common mechanisms for tigecycline resistance are the RND-type efflux pumps in A. baumannii and K. pneumoniae (33). In A. baumannii, resistance is commonly attributed to the constitutive overexpression of the adeABC efflux pump-encoding genes regulated by the adeR and adeS two-component system (10). In one study, overexpression was caused by the insertion of an ISAba1 element near the adeS gene, while others show that mutations in the adeR or adeS genes are directly responsible (10, 34). Still, overexpression of adeB is found independent of insertion or mutations in adeR or adeS genes, suggesting a larger regulatory network elevates adeB transcription (35). More recently, AdeABC-independent mechanisms, such as a mutation in the trm-encoded S-adenosylmethionine (SAM) methyltransferase, also contributed to tigecycline nonsusceptibility (36).

In K. pneumoniae, the AcrAB-TolC multidrug efflux pump is homologous to the AdeABC efflux pump, causing us to investigate this system as a likely mechanism of resistance. To determine if efflux pumps contributed to tigecycline nonsusceptibility, PAβN was used. PAβN competes with antibiotic extrusion from the cell, causing the internal antibiotic concentration to increase, thus requiring less drug supplementation to kill the cell and lowered MIC values (37). Previous reports found a 2- to ≥256-fold decrease in tigecycline nonsusceptibility, while growth of the isolate in the presence of tigecycline and PAβN compared to that of antibiotic alone (2, 38). Our study found a 4-fold decrease in MIC values in nonsusceptible isolates only, suggesting efflux pump-mediated tigecycline resistance.

K. pneumoniae tigecycline resistance is often associated with a mutation in the ramR response regulator causing increased expression of the ramA transcriptional activator and subsequent constitutive activation of acrAB (8). Expression studies correlated elevated ramA expression to an increase in tigecycline MIC from 2 to 4 μg/ml (39). Although we were unable to detect differences in the expression levels of ramA or acrA in susceptible and nonsusceptible isolates, we were able to detect differences in the VA6048(ramR) and VA6048(ramR_VA12262-WT) control strains, suggesting our assays were sufficient for detection of significant changes in ramA expression among isolates. The regulatory circuits of AcrAB activity can overlap several others, including marA- and soxS-containing networks, but we were also unable to find qRT-PCR- or sequencing-based evidence for the involvement of these regulatory systems (6). The RarA-mediated activity of the OqxAB multidrug efflux pump system is an alternative to the AcrAB system, but, again, we detected no increase in expression of the genes that encode these proteins.

Whole-genome maps were generated for KP47 and KP64, but we found no discernible difference between the chromosomes of these isolates; however, the resolution of WGM cannot detect insertions less than 2,000 bp. Whole-genome sequencing of KP47 and KP66 demonstrated that the only genetic difference between isolates is an IS5 insertion element upstream of an operon containing uncharacterized putative transporters, here termed kpgABC, with AcrA and AcrB being the closest K. pneumoniae homologs to the KpgA and KpgB amino acid sequence, respectively. Transposable elements have several evolutionary effects, including novel promoter creation causing antibiotic resistance in a broad range of organisms (10, 40–42). Here, we find a significant increase in expression of kpgABC due to an IS5-type insertion element in the promoter region of this operon.

The action of the KpgABC efflux pump can increase tigecycline resistance independent of a functional AcrAB-TolC, although the KpgABC pump cannot alone restore MIC values to the same wild-type level in E. coli ΔacrAB cells. Expression of kpgABC in E. coli did not change its antibiotic profile to any other drug, suggesting the KpgABC efflux pump exports only tigecycline given the detection limits of our methods. Furthermore, the MIC levels of E. coli strains expressing kpgABC increased but did not increase to the EUCAST clinical breakpoints to be considered truly “resistant,” as noted in the K. pneumoniae clinical isolates cultured in this paper. Since no kpgABC homologues are found within the E. coli strains used in this study, it is possible that these strains also lack any additional regulatory networks or other accessory proteins that could further enhance the activity of the KpgABC efflux pump. We further speculate based on the inability to fully restore wild-type MIC levels in E. coli that the KpgABC efflux pump has either a weaker affinity or lower rate of export of tigecycline than AcrAB-TolC, thereby allowing a greater buildup of drug concentration within the bacterial cells. Further research is needed to elucidate these possibilities. Nonetheless, with even a modest increase in tigecycline resistance levels due to overexpression of the kpgABC genes, the results suggest increased expression of kpgABC can have clinically relevant effects, especially in K. pneumoniae XDR cells, where treatment options are limited.

In conclusion, tigecycline nonsusceptibility rapidly arose in XDR Klebsiella pneumoniae isolates containing an IS5 insertion element upstream of an operon containing uncharacterized putative transporters, here termed kpgABC. This is the first report highlighting kpgABC and to identify an insertion element whose presence correlated with the in vivo development of tigecycline nonsusceptibility in K. pneumoniae clinical samples.