Missense Mutations in the CrrB Protein Mediate Odilorhabdin Derivative Resistance in Klebsiella pneumoniae

NOSO-502 is a preclinical antibiotic candidate of the odilorhabdin class. This compound exhibits activity against Enterobacteriaceae pathogens, including carbapenemase-producing bacteria and most of the colistin (CST)-resistant strains.

ABSTRACT

NOSO-502 is a preclinical antibiotic candidate of the odilorhabdin class. This compound exhibits activity against Enterobacteriaceae pathogens, including carbapenemase-producing bacteria and most of the colistin (CST)-resistant strains. Among a collection of CST-resistant Klebsiella pneumoniae strains harboring mutations in the genes pmrAB, mgrB, phoPQ, and crrB, only those bearing mutations in the gene crrB were found to be resistant to NOSO-502. CrrB is a histidine kinase which acts with the response regulator CrrA to modulate the PmrAB system, which induces the restructuring of lipopolysaccharide on the outer membrane and thus leads to CST resistance. Moreover, crrB mutations also enhance the transcription of neighboring genes, such as H239_3063, encoding an ABC transporter transmembrane region, H239_3064, encoding a putative efflux pump also known as KexD, and H239_3065, encoding an N-acetyltransferase. To elucidate the mechanism of resistance to NOSO-502 induced by CrrB missense mutations in K. pneumoniae, mutants of NCTC 13442 and ATCC BAA-2146 strains resistant to NOSO-502 and CST with single amino acid substitutions in CrrB (S8N, F33Y, Y34N, W140R, N141I, P151A, P151L, P151S, P151T, or F303Y) were selected. Full susceptibility to NOSO-502 was restored in crrA- or crrB-deleted K. pneumoniae NCTC 13442 CrrB (P151L) mutants, confirming the role of CrrAB in controlling this resistance pathway. Deletion of kexD (but not other neighboring genes) in the same mutant also restored NOSO-502-susceptibility. Upregulation of the kexD gene expression was observed for all CrrB mutants. Finally, plasmid expression of kexD in a K. pneumoniae strain missing the locus crrABC and kexD significantly increased resistance to NOSO-502.

INTRODUCTION

Klebsiella pneumoniae belongs to the Enterobacteriaceae family and is present in the gastrointestinal tract and the nasopharynx of healthy humans (1). This species is associated with community-acquired bacterial pneumonia and with hospital-acquired infections responsible for severe diseases such as pneumonia, septicemia, urinary tract infections, or soft tissue infections (2). Recently, the World Health Organization (WHO) has classified this species as one of the priority pathogens for research and development of new antibiotics, and carbapenem-resistant K. pneumoniae (CR-Kp) isolates are one of the principal threats (3). CR-Kp isolates have become resistant to all, or nearly all, antibiotics available and the incidence of infections due to these multidrug-resistant (MDR) pathogens has increased dramatically (4). Colistin (CST), belonging to the polymyxin antibiotic family, is one of the last-resort antibiotics used for the treatment of CR-Kp infections (5). Unfortunately, recent reports indicated that in certain countries, more than 15% of CR-Kp isolates are resistant to CST (6, 7).

In K. pneumoniae, resistance to polymyxins and to other cationic antimicrobial peptides is mainly mediated by the addition of 4-amino-4-deoxy-l-arabinose (LAra4N) or phosphoethanolamine (pEtN) to lipid A of the lipopolysaccharide (LPS) present in the outer membrane. Addition of cationic groups decreases the affinity of polymyxins for the LPS and strongly limit their antibacterial effect. Indeed, polymyxins must bind to the lipid A moiety of bacterial LPS to exert their bactericidal activity by permeabilizing the cell membranes (8).

LAra4N synthesis requires the products of the pmrHFIJKLM operon and pmrE gene, and pEtN synthesis is encoded by the pmrC gene. These operons and genes, which are highly conserved among pathogenic Enterobacteriaceae, are positively regulated by well-studied phosphorelay signaling systems, the PmrAB and PhoPQ two-component systems (TCSs). Furthermore, MgrB, a small transmembrane protein, was identified to exert negative control against the PhoPQ system (9).

Recently, the crrAB operon, a new TCS, was identified as responsible for CST resistance in K. pneumoniae (10). CrrB is the signal-transducing histidine kinase and CrrA is an adjacent response regulator. Around the crrAB locus are found, in the opposite orientation, the H239_3059 gene and an operon consisting of four genes, crrC, H239_3063, H239_3064, and H239_3065. H239_3059 encodes a glycosyltransferase, whereas crrC, H239_3063, H239_3064, and H239_3065 were predicted to encode a possible transporter protein, an ABC transporter transmembrane protein, a putative RND-type efflux pump, and a putative N-acetyltransferase, respectively (Fig. 1). Previous phylogenetic analysis has shown the crrAB genes are present almost exclusively in some Klebsiella strains (11). Mutation or inactivation of the crrB gene is responsible for the upregulation of expression of a set of genes: crrAB, crrC, H239_3059, H239_3063, H239_3064, H239_3065, pmrAB, pmrHFIJK, pagP, and phoPQ (12). These mutants display lipid A modifications with incorporation of 4-amino-4-deoxy-l-arabinose (LAra4N) and palmitoylation, resulting in an extremely high CST resistance level (MIC > 512 μg/ml). Previous studies hypothesized that CrrB mutations increase the transcription of the crrC gene, which in turns upregulates the expression of the pmrC and pmrE genes and the pmrHFIJKLM operon, through the overexpression of the PmrAB two-component system, thereby inducing CST resistance via LPS modification. Overexpression of the efflux pump H239_3064 also contributes to CST resistance (13, 14).

FIG 1.

Schematic diagram of genome organization in K. pneumoniae strains and the NOSO-502 resistance pathway.

The H239_3064 sequence is 100% identical to the KexD protein from Klebsiella pneumoniae MGH78578 (CP000647) (12). The KexD protein was described by Ogawa et al. as a component of an energy-dependent efflux pump and belongs to the resistance nodulation division (RND) family (15). The KexD protein was found to have about 50% identity to the AcrB protein of K. pneumoniae. The inner membrane protein AcrB, in association with the periplasmic protein AcrA and with the outer membrane protein TolC, is a major tripartite RND efflux system that confers resistance to tetracycline, chloramphenicol, ampicillin, nalidixic acid, and rifampin in Gram-negative bacteria (16). KexD could function with AcrA and with TolC or KocC of K. pneumoniae to efflux antibiotics such as novobiocin or erythromycin; however, these assemblies remain to be elucidated (15).

Recent studies have described the pharmacological properties of NOSO-502, the first preclinical candidate of a novel antibiotic class, the odilorhabdins (ODLs) (17, 18). ODLs are a family of cationic peptides produced by a bacterial symbiont (Xenorhabdus) of nematodes that specifically inhibit bacterial translation by interacting with the 30S subunit of the bacterial ribosome at a site not exploited by any known ribosome-targeting antibiotic (19, 20). NOSO-502 exhibits activity against Enterobacteriaceae pathogens, including carbapenemase-producing Enterobacteriaceae (KPC, NDM, OXA-48, and VIM) and CST-resistant strains due to multiple mechanisms of acquired or plasmid-mediated resistance (chromosome-encoded mutations or deletions of pmrA, pmrB, mgrB, or phoQ genes, or plasmid-encoded CST resistance genes mcr-1, mcr-2, and mcr-3). Among all the CST-resistant strains tested, only those bearing mutations in the crrB gene were found to be resistant to NOSO-502 (17). These results suggest that CrrB mutations could induce a mechanism of NOSO-502 resistance different from that of polymyxins and that would not involve the LPS modifications mediated by PmrAB or PhoPQ.

The objective of this study was the identification of the NOSO-502 mechanism of resistance induced by CrrB missense mutations in Klebsiella pneumoniae. To investigate that, we have selected spontaneous mutants with a high level of resistance to NOSO-502 and CST, confirmed the location of mutations in the crrB gene, evaluated the role of crrA, crrB, and neighboring genes in the resistance pathway to NOSO-502, and identified the efflux pump KexD as responsible for this phenotype.

RESULTS

Selection of K. pneumoniae mutants resistant to NOSO-502 and colistin.

To evaluate the propensity of K. pneumoniae to generate in vitro mutants resistant to NOSO-502 and CST, two carbapenemase-producing K. pneumoniae strains (NCTC 13442 and ATCC BAA-2146) were selected. The presence of the TCS crrAB in the genomes of these strains was first confirmed by PCR, and then MIC values of NOSO-502 and CST against K. pneumoniae ATCC BAA-2146 and K. pneumoniae NCTC 13442 were determined to be 1 μg/ml.

The selection of spontaneous K. pneumoniae ATCC BAA-2146 mutants was carried out by plating 8.3 × 10⁸ CFU onto nutritive agar medium containing 4× and 10× the MIC for NOSO-502. Twenty mutants were selected at 4× the MIC (frequency of 2.4 × 10⁻⁸) and 6 mutants at 10× the MIC (frequency of 7.2 × 10⁻⁹). All 26 mutants obtained exhibited high-level resistance to NOSO-502 and CST. The MIC values of NOSO-502 against these mutants increased by a 128 to >128-fold factor compared to a wild-type strain, whereas the MIC values of CST increased by more than 256-fold.

The same protocol was applied with K. pneumoniae NCTC 13442 by plating 5.7 × 10⁹ CFU. Fifty-three mutants were selected at 4× the MIC (frequency of 9.3 × 10⁻⁹) and 1 mutant at 10× the MIC (frequency of 1.8 × 10⁻¹⁰). Out of the 53 mutants of K. pneumoniae NCTC 13442 selected, only 7 of them were coresistant to NOSO-502 (MIC values increased by 128 to >128-fold) and to CST (MIC values increased by >128-fold).

Antibiotic susceptibility profiles for these mutants are shown in Table 1. Moreover, a slight increase in MIC values for tigecycline (TIG) was observed in some mutant strains (4-fold increase maximum). No significant changes in susceptibility to imipenem (IPM), ciprofloxacin (CIP), or gentamicin (GEN) were seen.

TABLE 1.

MICs of NOSO-502, CST, and comparators against selected mutants coresistant to NOSO-502 and CST

K. pneumoniae strain	Mutant	NOSO-502 MIC-fold selection	CrrB substitutionb	MIC (μg/ml)a
				NOS	CST	IPM	GEN	TIG
ATCC BAA-2146 WT	NA	NA	NA	1	1	128	>256	32
ATCC BAA-2146	AT_P151S_1	4×	P151S	128	>256	128	>256	64
ATCC BAA-2146	AT_P151S_2	4×	P151S	256	>256	128	>256	64
ATCC BAA-2146	AT_N141I_1	4×	N141I	128	>256	128	>256	64
ATCC BAA-2146	AT_P151S_3	4×	P151S	128	>256	128	>256	64
ATCC BAA-2146	AT_P151S_4	4×	P151S	256	>256	128	>256	64
ATCC BAA-2146	AT_P151S_5	4×	P151S	128	>256	128	>256	64
ATCC BAA-2146	AT_P151S_6	4×	P151S	256	>256	256	>256	32
ATCC BAA-2146	AT_N141I_2	4×	N141I	128	>256	128	>256	32
ATCC BAA-2146	AT_P151S_7	4×	P151S	128	>256	128	>256	64
ATCC BAA-2146	AT_P151S_8	4×	P151S	128	>256	128	>256	64
ATCC BAA-2146	AT_P151S_9	4×	P151S	256	>256	256	>256	64
ATCC BAA-2146	AT_P151S_10	4×	P151S	256	>256	256	>256	64
ATCC BAA-2146	AT_P151S_11	4×	P151S	256	>256	256	>256	64
ATCC BAA-2146	AT_P151S_13	4×	P151S	>256	>256	128	>256	32
ATCC BAA-2146	AT_F303Y_1	4×	F303Y	256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_14	4×	P151S	256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_15	4×	P151S	>256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_16	4×	P151S	256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_17	4×	P151S	256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_18	4×	P151S	256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_19	10×	P151S	256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_20	10×	P151S	256	>256	128	>256	32
ATCC BAA-2146	AT_ F303Y_2	10×	F303Y	256	>256	256	>256	32
ATCC BAA-2146	AT_P151S_21	10×	P151S	256	>256	128	>256	32
ATCC BAA-2146	AT_P151S_22	10×	P151S	>256	>256	256	>256	32
ATCC BAA-2146	AT_ F303Y_3	10×	F303Y	256	>256	128	>256	32
NCTC 13442 WT	NA	NA	NA	1	1	8	≤0.5	2
NCTC 13442	NC_P151L	4×	P151L	256	>256	4	≤0.5	8
NCTC 13442	NC_W140R	4×	W140R	128	>256	16	≤0.5	4
NCTC 13442	NC_P151T	4×	P151T	256	>256	8	≤0.5	8
NCTC 13442	NC_P151A	4×	P151A	256	>256	8	≤0.5	4
NCTC 13442	NC_F33Y	4×	F33Y	64	>256	4	≤0.5	4
NCTC 13442	NC_S8N	4×	S8N	64	>256	8	≤0.5	1
NCTC 13442	NC_Y34N	10×	Y34N	16	>256	8	≤0.5	2

^aNOS, NOSO-502; CST, colistin; IPM, imipenem; GEN, gentamicin; TGC, tigecycline; NA, not applicable.

^bAmino acid positions where mutations in CrrB have been detected.

Amino acid substitutions in CrrB associated with NOSO-502 and colistin resistance.

In a previous study, no mutation or inactivation of the pmrA, pmrB, mgrB, or phoQ genes was associated with NOSO-502, whereas strains possessing mutations in CrrB exhibited this resistance profile (17).

All seven K. pneumoniae NCTC 13442 mutants coresistant to NOSO-502 and to CST were submitted to whole-genome sequencing (WGS) and screened for mutations in pmrA, pmrB, phoQ, phoP, mgrB, crrA, and crrB genes. Different substitutions in CrrB were detected in all isolates and included S8N, F33Y, Y34N, W140R, P151A, P151L, and P151T. All seven mutants and the parental strains also exhibited the same single 1-nt deletion after the 23rd amino acid of PhoQ, inducing a frameshift in the first 5% of the protein sequence and therefore most probably a complete functional impairment (Table 2). Further, the CrrAB system in the 26 coresistant K. pneumoniae ATCC BAA-2146 isolates was analyzed by amplification of the crrB gene. Subsequent sequencing identified mutations leading to amino acid changes (P151S, P141I, or F303Y) in the CrrB protein of all mutants (Table 1).

TABLE 2.

Alterations associated with colistin and NOSO-502 resistance in K. pneumoniae NCTC13442 mutants

K. pneumoniae strain	Mutant	Amino acid alteration
		CrrA	CrrB	MgrB	PhoP	PhoQ	PmrA	PmrB
NCTC13442 (wt)	NA	1-nt dela
NCTC13442	NC_P151L	P151L	1-nt dela
NCTC13442	NC_W140R	W140R	1-nt dela
NCTC13442	NC_P151T	P151T	1-nt dela
NCTC13442	NC_P151A	P151A	1-nt dela
NCTC13442	NC_F33Y	F33Y	1-nt dela
NCTC13442	NC_S8N	S8N	1-nt dela
NCTC13442	NC_Y34N	Y34N	1-nt dela

^aSimilar single 1-nt deletion (gene position 66) of phoQ inducing a frameshift in the first 5% of the protein sequence and therefore most probably a complete functional impairment.

Two-component regulatory system CrrAB controls NOSO-502 resistance.

To evaluate the role of the two components CrrB and CrrA in the resistance to NOSO-502, the genes were individually deleted in the mutant derived from strain NC_P151L and showing increased resistance to NOSO-502 and CST. Wild-type susceptibility to NOSO-502 and CST was restored in both cases. Indeed, MICs of NOSO-502 and CST against NC_P151L ΔcrrB or ΔcrrA were 1 μg/ml, similar to the ones against the parental strain NCTC 13442. Complementation of the ΔcrrB mutant was assessed by introducing the inducible plasmid pJN105 carrying either the wild-type crrB gene or the mutated gene crrB P151L. Susceptibility to NOSO-502 and CST was conserved with wild-type crrB (MICs = 1 μg/ml and 2 μg/ml, respectively), whereas the deletion mutant with plasmid expression of CrrB P151L displayed a high level of NOSO-502 (MIC = 64 μg/ml) and CST (MIC > 2,048 μg/ml) resistance. These results confirm that the CrrAB system controls the resistance pathway to these antibiotics (Table 3).

TABLE 3.

MICs of NOSO-502 and CST against K. pneumoniae NCTC 13442 CrrB (P151L) strain with deletion of the crrA, crrB, crrC, H239_3063, kexD, and H236_ 3065 loci and with complementation of the crrB (wt) or crrB (P151L) locus

K. pneumoniae strain	MIC (μg/ml)
	NOS	CST
NCTC 13442 wild type	1	1
NC_P151L	256	>2,048
NC_P151L ΔcrrA	1	1
NC_P151L ΔcrrB	1	1
NC_P151L ΔcrrC	256	2
NC_P151L ΔH239_3063	128	>2,048
NC_P151L ΔkexD	2	256
NC_P151L ΔH239_3065	256	>2,048
NC_P151L ΔcrrB/pJN105 (crrB [P151L])	64	>2,048
NC_P151L ΔcrrB/pJN105 (crrB [WT])	1	2

CrrB mediates NOSO-502 resistance through KexD.

The role of CrrC (a possible transporter protein) and KexD (H239_3064) in the resistance to CST of K. pneumoniae bearing CrrB mutations was previously demonstrated (13, 14). To examine whether crrC, H239_3063, kexD, or H239_3065 loci are involved in resistance to NOSO-502, these genes were deleted one by one in the mutant NC_P151L. As expected, the MIC of CST against the NC_P151L ΔcrrC strain was decreased by >1,024-fold compared to the parent strain, restoring a full CST susceptibility (Table 3). Interestingly, the MIC of NOSO-502 was unchanged (256 μg/ml), suggesting that the pathways controlled by the CrrAB regulatory system and conferring resistance to NOSO-502 and/or CST are different (Table 3).

On the other hand, deletion of the efflux pump KexD restored full wild-type sensitivity to NOSO-502 and decreased by a >8-fold factor the MIC of CST. A mutant bearing a deletion of the H239_3063 locus displayed a nonsignificant effect in susceptibility to CST (MIC >512 μg/ml) but 2-fold decreases in the MIC of NOSO-502 (MIC = 128 μg/ml), while deletion of H239_3065 did not affect the levels of NOSO-502 nor CST antibacterial activity (MICs = 256 and >2,048, respectively) (Table 3).

The crrAB operon (crrA and crrB) and the crrC operon (crrC, H239_3063, kexD, and H239_3065) are variably present in the K. pneumoniae population. In this context, the strain ATCC 43816 lacks these two operons (genome accession number NZ_CP009208.1) and was chosen for complementation with plasmids carrying KexD alone, the full-length operon (crrC_H239_3063-kexD-H239_3065), or a shorter version lacking crrC. As expected, the MICs of NOSO-502 and CST against the wild-type strain ATCC 43816 were 1 μg/ml. Complementation of this strain with KexD increases the strain resistance to NOSO-502 by 64-fold (MIC = 64 μg/ml) but have no effect on CST susceptibility (MIC = 1 μg/ml). Moreover, expression of the full-length crrC operon increased the MIC value of NOSO-502 by 128-fold and the MIC of CST by >2,048-fold, while the short-version had no effect on CST susceptibility (Table 4). These data clearly indicate that TCS CrrAB mediates NOSO-502 resistance through the KexD efflux pump, while the resistance pathway to CST involves CrrC.

TABLE 4.

MICs of NOSO-502 and CST against K. pneumoniae ATCC 43816 strain with complementation of the kexD locus, the full-length operon (crrC H239_3063- kexD- H239_3065), and a shorter version (H239_3063- kexD- H239_3065)

K. pneumoniae strain	MIC (μg/ml)
	NOS	CST
ATCC 43816 wild type	1	1
ATCC 43816/pJN105	1	2
ATCC 43816/pJN105 (kexD)	64	1
ATCC 43816/pJN105 (H239_3063- kexD- H239_3065)	128	1
ATCC 43816/pJN105 (crrC- H239_3063- kexD- H239_3065)	128	>2,048

Amino acid substitutions in CrrB associated with overexpression of KexD.

A previous publication showed a strong link between mutations in CrrB and overexpression of KexD (also known as KPN_02064) (14). The kexD transcript levels were measured to evaluate the impact of crrB mutations. Quantitative reverse transcriptase PCR (qRT-PCR) assays were performed with isolates bearing CrrB substitutions (P151L, P151T, P151S, W140R, and F33Y) and their corresponding parental strains (K. pneumoniae NCTC 13442 and K. pneumoniae ATCC BAA-2146). Upregulation of the expression levels of kexD was observed for all CrrB mutants compared to those of the parental NOSO-502-susceptible strains. A significant increase from 2.8 to 6-fold was measured (Fig. 2).

FIG 2.

Relative expression levels measured by RT-qPCR of kexD from K. pneumoniae ATCC BAA-2146 and K. pneumoniae NCTC 13442 crrB mutants. Each plot represents mutant gene expression relative to the wild-type strain, the latter set at equal to 1. The plot was generated by REST analysis (36). All genes are validated as overexpressed in K. pneumoniae crrB mutants (random test, REST, p _0.05).

Distribution of KexD in other species and in diverse sequence types of K. pneumoniae.

KexD amino acid sequence analysis was carried out by use of the Basic local alignment search tool (http://www.ncbi.nlm.nih.gov/BLAST). With respect to Gram-negative species with great clinical importance in hospitals, the KexD sequence presented the highest similarity (∼95% residue identity) with homologous proteins of Citrobacter freundii, and displayed identities of ∼80% (Escherichia coli, Enterobacter cloacae), ∼60%, (Acinetobacter baumannii, Pseudomonas aeruginosa), and ∼50% (Enterobacter aerogenes, Proteus mirabilis) to other multidrug efflux RND transporter permease subunits.

The presence of kexD was assessed in 200 K. pneumoniae genomes available in GenBank (http://www.ncbi.nlm.nih.gov/Genbank). The sequence types (STs) of these strains were determined by the Institute Pasteur Klebsiella MLST database (http://bigsdb.web.pasteur.fr/klebsiella/klebsiella.html). There are kexD orthologs found in K. pneumoniae sequence type 258 (ST258), ST11, ST16, ST29, ST34, ST86, ST111, ST147, ST244, ST258, ST270, ST273, ST307, and ST340, while it is absent in K. pneumoniae ST14, ST15, ST23, ST101, ST1518, and ST1536.

DISCUSSION

NOSO-502 belongs to a novel class of antibiotics, the Odilorhabdins (ODL) produced by the entomopathogenic bacterium Xenorhabdus (17). This compound exhibits activity against most Gram-negative and Gram-positive bacteria, including multidrug resistant (MDR) bacteria. (17). NOSO-502 is currently under preclinical development to treat infections caused by Enterobacteriaceae as Escherichia coli or Klebsiella pneumoniae. Understanding of how bacteria become resistant and how resistance genes are spreading is crucial for the development of antibiotics.

In a previous study, we observed that clinical K. pneumoniae isolates from a patient who had been treated with colistin (CST), and bearing CrrB mutations exhibited coresistance to NOSO-502 and CST (17, 21). CrrB is a signal-transducing histidine kinase that acts jointly with response-regulating transcription factor CrrA to regulate both LPS modifications via CrrC and the pmrAB pathway, but also the expression of the efflux pump KexD. We observed that, unlike the case for the polymyxin family, addition of LAra4N or pEtN to LPS does not seem to have any effect on the NOSO-502 antibacterial activity.

To evaluate the propensity of K. pneumoniae to generate in vitro mutants resistant to NOSO-502 and CST, two carbapenemase-producing K. pneumoniae strains were chosen in this study. CST and NOSO-502 coresistant mutants were obtained under NOSO-502 selection pressure at frequencies between 2.4 × 10⁻⁸ and 1.8 × 10⁻¹⁰ depending on the strain and NOSO-502 concentrations tested. Ten individual amino acid substitutions in CrrB (S8N, F33Y, Y34N, W140R, N141I, P151A, P151L, P151S, P151T, and F303Y) responsible for NOSO-502 and CST resistance were identified in the two CR-Kp strains studied. Five of them have previously been associated with CST resistance (12, 21). The colistin-resistant isolate Af44b harboring a P151L CrrB amino acid change was recovered from a South African patient who has been treated with CST (21). In this case, a clonal colistin susceptible isolate was recovered before treatment without any change in the CrrB amino acid sequence, confirming the close link between CrrB mutation and the in vivo emergence of colistin-resistant K. pneumoniae under colistin pressure. Colistin-resistant K. pneumoniae isolates with W140R (Col36), N141I (Col4, Col20), and P151S (Col7) substitutions in CrrB were collected from patients in the Taipei Veterans General Hospital (22). Three of them were isolated from biological matrices such as blood or sputum of patients treated with CST. For the last case, the patient does not appear to have been treated with CST prior to strain isolation, but authors could not trace the colistin usage for this patient in other hospitals. The CrrB S8N mutation in K. pneumoniae was identified in a controlled directed-evolution in vitro study (11). Among the 10 CrrB mutations identified in our study, three were at the same position as the mutations observed in other studies; however, the amino acid changes were different (amino acids 151 and 303). The F303S CrrB mutation was identified in a CST-resistant K. pneumoniae isolate from Iranian poultry (23). Because an insertion in MgrB and a mutation in PmrB were also characterized in this isolate, it is not clear whether the F303S CrrB mutation is solely responsible for the resistance. We therefore characterized two new CrrB mutations (F33Y and Y34N) conferring NOSO-502 and CST resistance. They are localized close to the CrrB Y31H mutation identified in a CST-resistant K. pneumoniae isolate recovered from a Taiwanese patient (22).

Here, we established the resistance pathway to NOSO-502 in K. pneumoniae isolates bearing mutations in CrrB. Upregulation of the efflux pump KexD observed in these mutants was responsible for the resistance profile. In this context, KexD might directly pump NOSO-502 out of the cell or efflux some other element essential to the antibacterial activity of NOSO-502. While kexD is predicted to encode an RND efflux pump component, in order to achieve an efflux activity it must interact with other RND efflux components, such as the periplasmic protein AcrA and the porin TolC, to form a functional system. This assembly was not evaluated in this study and the characterization of this novel efflux system could open interesting lines of research. Our results confirm that the CST-resistance pathway involving CrrC and PmrAB and resulting in LPS modification is different from that leading to NOSO-502 resistance, which is specifically dependent on the activity of KexD (Fig. 1). The role of KexD in the CST resistance of CrrB mutants remains unclear. Indeed, deletion of the gene encoding the efflux pump KexD in the high-level CST-resistant K. pneumoniae NC_P151L strain (MIC > 2,048 μg/ml) decreases the MIC of CST by >8-fold but does not restore full susceptibility (MIC = 256 μg/ml). On the other hand, plasmid expression of kexD in a K. pneumoniae strain lacking crrAB, crrC, H239_3063, kexD, and H239_3065 genes had no effect on the CST susceptibility of this strain, while a 64-fold increase of resistance to NOSO-502 was observed.

The kexD gene is present in several types of K. pneumoniae, including the clonal group CG258 (ST11 and ST258). In the United States and China, KPC was the most common carbapenemase identified in CR K. pneumoniae isolates. Most of those in China belong to ST11 (24), which is different from what is observed in the United States (predominance of ST258) (25, 26). In Europe, the highest incidence of carbapenem-resistant Enterobacteriaceae (CRE) is found in Italy and Greece. While other carbapenemases are present in these countries, KPC remains the most common etiology of carbapenem resistance, with the majority of isolates belonging to the clonal group 258 (27).

The increase in the use of CST to treat carbapenem-resistant or multidrug-resistant Gram-negative infections generates a concomitant increase in CST-resistant K. pneumoniae isolates. In a European Centre for Disease Prevention and Control (ECDC) report from 2013, the levels of K. pneumoniae resistant to CST were 8.8% of the total isolates analyzed and 32% of the carbapenem-resistant isolates (7). Chromosomal mutations or amino acid deletions of the mgrB, phoP, phoQ, pmrA, or pmrB genes, in addition to the plasmid-mediated mcr-1 and mcr-2 genes, are the main reported mechanisms of acquisition of CST resistance (8). CrrB mutations conferring CST resistance are rare. To date, only 14 clinical K. pneumoniae isolates have been described in the literature (2 isolates from the United States, 8 from Taiwan, 1 from France, 1 from Colombia, 1 from South Africa, and 1 from Greece) (10, 13, 21). Recently, a first polymyxin-resistant Citrobacter freundii strain with an A91T substitution in CrrB was identified from clinical isolates in a tertiary hospital from Recife, Brazil (28). The emergence on an Iranian poultry farm of CST-resistant Enterobacteriaceae strains, including K. pneumoniae with CrrB mutations, confirms the need to reassess the use of last-resort human treatment in livestock (23). The potent activity of NOSO-502 against all classes of carbapenemase-producing Enterobacteriaceae (CRE) strains and against almost all CST-resistant strains due to multiple mechanisms of acquired or plasmid-mediated resistance (chromosome-encoded mutations or deletions of pmrA, pmrB, mgrB, or phoQ genes, or plasmid-encoded CST resistance genes mcr-1, mcr-2, and mcr-3) makes this compound a potent candidate for treating CRE or CRE associated with CST-resistant infections.

MATERIALS AND METHODS

Bacterial strains and antimicrobial agents.

Reference strains are from the American Type Culture Collection (ATCC) and the National Collection of Type Cultures (NCTC). NOSO-502 was synthesized at Nosopharm, Nîmes, France. Ciprofloxacin (Sigma-Aldrich, ref: 1134335), gentamicin (Sigma-Aldrich, ref: G1397), imipenem (Sigma-Aldrich, ref IO160), polymyxin B (Sigma-Aldrich, ref: 92283), and tigecycline (Sigma-Aldrich, ref: PZ0021) were provided by the manufacturers as standard powders except for gentamicin and polymyxin B, which were in aqueous solution at 50 and 20 mg/ml, respectively.

MIC values.

MIC values were determined using the Clinical and Laboratory Standards Institute (CLSI) broth microdilution methodology, colony direct suspension, as described in the CLSI document M07-A10 (29). When necessary, streaking for inoculum preparation was performed on Mueller-Hinton agar supplemented with gentamicin at 10 μg/ml and with arabinose at 2% wt/vol. In this case, the reaction medium was equally supplemented with 10% of arabinose wt/vol. l-(+)-Arabinose (Sigma-Aldrich, ref: A3256), was purchased from Merck Life Science.

Determination of mutation frequency.

Bacterial strains were grown in antibiotic-free Luria Bertani broth at 35°C for 18 h. Approximately 10⁹ CFU of each strain were plated in duplicate onto cation-adjusted Mueller-Hinton agar plates containing NOSO-502 concentrations at 4× and 10× the MIC values. The plates were read at 24 h and 48 h of incubation at 35°C. The frequency of selected resistant mutants was calculated as the ratio of the number of bacteria growing divided by the number of bacteria in the original inoculum, which was calculated by plating several dilutions of the original inoculum (30).

Genomic analyses.

An Evotec-developed pipeline was used to find the genomic variants in the resistant clone(s). The reads were mapped to the reference genome and the variants were detected by identifying where the aligned reads differ from the reference genome. If a parental clone was provided, the variants also found in this parental clone were ignored. The detailed pipeline tools and parameters are given bellow. First, the Nextera sequencing adapters were removed from the reads and the reads were trimmed according to their base quality using the Trimmomatic software version 0.39 (31). Trimmomatic parameters were as follows: minimum base Phred score for 5′ ends of reads = 25; minimum base Phred score for 3′ ends of reads = 25; minimum base Phred score for 4 bp sliding windows = 30; and minimum read length = 40 bp.

Afterward, preprocessed high-quality reads were mapped to the reference genome, Klebsiella pneumoniae strain NCTC13442, whole-genome shotgun sequence (NZ_UGKW01000006, last update 9 October 2019). This mapping step was performed by Bowtie2 version 2.2.5 (32). The duplicated reads were then removed using the MarkDuplicates module of the Picard tools version 2.6.0-SNAPSHOT (Broad Institute, 2019). Samtools version 1.10 (33) was used to convert, sort by coordinates, index the aligned data files, and compute average coverage per contig for each sample. Mapping results quality was assessed by Qualimap v.2.2 (34). Then, short-range variants were identified using the FreeBayes software v1.3.2-dirty (35). The identified variants were then filtered using a custom R script. Only variants with an alternative allele proportion of >80%, without strain bias and not found in the parental clone, were conserved. Finally, those variants were annotated using a custom script and the GenBank file “Kpneumoniae_NCTC13442.20200623.gbk” (GenBank annotation file of K. pneumoniae NCTC13442 downloaded from the Refseq ftp server).

Sequence analysis of amino acid substitutions in CrrAB.

The crrB gene was amplified by PCR using the primers crrB Forward 1 (5′-AAGAAACGATTCTGAGCC-3′) and crrB Reverse 1 (5′-CTACCGACAACCCTAAAC). The resulting amplicons were subjected to nucleotide sequencing using primers crrB Forward 1 and crrB Reverse 1. When necessary, the beginning and end of the crrB gene were amplified using primer pairs designed on crrA gene and crrB for the beginning of crrB gene (crrAB_F1crrA: GTGAGGTTTCCACGCCAGTA; crrAB_R1crrB: TGCAACGTCATTCAGTGGGT) and on crrB and glycosyltransferase genes for the end (crrB-GTase_F2crrB: GTTTGCGATCCCTTACGCAT; crrB-GTase_R2GTase: TATTGCCGCCCAGGACTTAG). The crrB sequences from the CST-resistant strains were compared to those from K. pneumoniae ATCC BAA-2146 or K. pneumoniae NCTC 13442 strains to identify CrrB amino acid substitutions.

Determination of mRNA expression levels by qRT-PCR.

Bacterial culture in cation-adjusted Mueller-Hinton broth (CAMHB) was grown to an optical density at 600 nm (OD₆₀₀) of between 0.5 and 1. Total RNA extraction was achieved with the RNeasy Protect Bacteria 50 preps kit (Qiagen no. 74524) according to the manufacturer’s instructions and was performed on 3 independent biological replicates.

RNA integrity number (RIN) was determined, and reverse transcription was performed using SuperScript II reverse transcriptase (Invitrogen ref. 18064-022) and random hexamers from Applied Biosystems (ref. N8080127). qRT-PCR was carried out using a LightCycler 480 (Roche) with Sensi-Fast SYBR no rox commercialized by Bioline (BIO-98050) and was performed in triplicate on each cDNA sample.

As a control, a blank sample (distilled water) and a no-reverse-transcriptase control were included to exclude DNA contamination. The recA gene was used as the reference housekeeping gene (Table 5).

TABLE 5.

Primers used in this study

Primer name	Primer sequence (5′–3′)	Purpose	Reference
crrB Forward 1	AAGAAACGATTCTGAGCC	PCR and sequencing	This study
crrB Reverse 1	CTACCGACAACCCTAAAC	PCR and sequencing	This study
crrAB_F1crrA	GTGAGGTTTCCACGCCAGTA	PCR and sequencing	This study
crrAB_R1crrB	TGCAACGTCATTCAGTGGGT	PCR and sequencing	This study
crrB-GTase_F2crrB	GTTTGCGATCCCTTACGCAT	PCR and sequencing	This study
crrB-GTase_R2GTase	TATTGCCGCCCAGGACTTAG	PCR and sequencing	This study
H239_3064 left1	TTATTGTTGACGATGCTATTGTTGT	qRT-PCR	This study
H239_3064 right1	GCTGAACTGTCGATAAATAATACCC	qRT-PCR	This study
recA left1	CTGATCTTTATCAACCAGATCCGTA	qRT-PCR	This study
recA right1	CAATCTTGTTTTTCACCACTTTGAC	qRT-PCR	This study

The data for each sample are expressed relative to the level of recA, using REST software 2009 (36) and the following equation (Pfaffl):

$$\text{ratio}=\frac{{\left(E_{\text{target}}\right)}^{\mathrm{\Delta }{\text{CP}}_{\text{target}}(\text{control}-\text{sample})}}{{\left(E_{\text{ref}}\right)}^{\mathrm{\Delta }{\text{CP}}_{\text{ref}}(\text{control}-\text{sample})}}$$ (1)

This method quantified the expression of a target gene relative to that of a reference gene, for comparisons of parental strains K. pneumoniae ATCC BAA-2146 or K. pneumoniae NCTC 13442 with the NOSO-502-resistant mutant.

Genetic manipulations for gene deletion and complementation.

Inactivation of genes crrC, H239_3063, and H239_3064 was performed following the method described by Kaniga et al. with some modifications (37). Briefly, 5′ and 3′ flanking regions of each gene (900 bp) were cloned into the suicide plasmid pKNG101 by assembly cloning using the NEBuilder HiFi DNA assembly kit following the manufacturer’s recommendations (New England BioLabs) and using chemically competent CC118λpir cells as the host for plasmid replication. Next, recombinant plasmids were transferred into K. pneumoniae strains by tripartite conjugation using Escherichia coli HB101 (pRK2013) as a helper strain. For this purpose, an overnight culture of each strain was used to spot 50 μl of each strain onto a sterile LB plate and the mix was incubated at 37°C for 6 h. Transconjugants of K. pneumoniae were selected on Klebsiella ChromoSelect selective agar base (Sigma-Aldrich) supplemented with 100 μg/ml carbenicillin and 50 μg/ml streptomycin. The excision of undesired pKNG101 sequences was obtained by plating transformants on M9 plates containing 10% (wt/vol) sucrose. Negative selection on streptomycin was carried out to confirm the loss of the plasmid in transconjugants. The allelic exchanges were verified by PCR and sequencing.

Gene trans-complementation was performed by cloning the full-length operon crrC-H239_3063-H239_3064-H239_3065 or a shorter version containing genes H239_3063-H239_3064-H239_3065 into the inducible plasmid pJN105 using assembly cloning as described (38). Recombinant plasmids were transferred into K. pneumoniae strains by electroporation using one pulse of 3 kV in 0.2 cm² cuvettes. Recombinant strains were selected on LB medium supplemented with 10 μg/ml gentamicin. Finally, gene expression was induced in LB medium supplemented with 1% arabinose.