Association of the Emerging Carbapenemase NDM-1 with a Bleomycin Resistance Protein in Enterobacteriaceae and Acinetobacter baumannii

Abstract

The carbapenemase NDM-1 has been identified recently in Enterobacteriaceae and Acinetobacter baumannii as a source of multidrug resistance, including resistance to carbapenems. By analyzing the immediate genetic environment of the bla_NDM-1 carbapenemase gene among a series of NDM-1-producing enterobacterial isolates, a novel gene (ble_MBL, for ble gene associated with the metallo-β-lactamase NDM-1) was identified. The ble_MBL gene encodes a novel bleomycin resistance protein (BRP), named BRP_MBL, that shares weak similarities with known BRPs (less than 60% amino acid identity). The expression of BRP_MBL conferred resistance to bleomycin and to bleomycin-like molecules in Enterobacteriaceae and A. baumannii. The bla_NDM-1 and ble_MBL genes were coexpressed under the control of the same promoter, located upstream of the bla_NDM-1 gene and at the extremity of the insertion sequence ISAba125. Most of the NDM producers possessed the ble_MBL gene. Although BRP_MBL did not modify the growth or death rates of Escherichia coli under experimental conditions, it suppressed the mutation rate of hypermutable E. coli and therefore may stabilize the plasmid-borne bla_NDM-1 gene. This study suggests that the emerging carbapenemase NDM-1 is selected by bleomycin-like molecules, and that BRP_MBL producers (and consequently NDM producers) are better suited to various environments.

INTRODUCTION

Like other metallo-β-lactamases, NDM-1 inactivates all β-lactams (including carbapenems) except monobactams. It is the most recently discovered carbapenemase that is spreading rapidly worldwide (13, 15, 17, 31). NDM-1 producers have been identified mainly in the United Kingdom, India, and Pakistan (9), but numerous studies within the last year reported NDM-1 producers from many countries in Europe (18, 22, 25), Asia, Africa, Australia, and North America (4, 6, 16, 23). Most of the patients infected by NDM-1 producers are from India, Pakistan, or Bangladesh or have traveled in the Indian subcontinent. This indicates that the Indian subcontinent is a reservoir for the gene encoding NDM-1 (32). In addition, several isolates of NDM-1-producing Enterobacteriaceae have been reported from the Balkan states (10) and the Middle East (Oman and Iraq) (19, 21), suggesting that those areas are other reservoirs of NDM-1 producers (10). The bla_NDM-1 gene is associated with neither a single strain nor a single plasmid backbone, since it has been identified in unrelated Gram-negative bacterial isolates and species either on the chromosome (Acinetobacter baumannii) or on different plasmid types (6, 7, 20). NDM-1 has been identified in Escherichia coli, an agent of community-acquired infections that is widely disseminated in the environment and water (32). In addition, it has been identified recently in other environmental bacterial species, such as Vibrio cholerae (32). The NDM resistance trait is usually associated with multiresistance and pandrug resistance.

While the basic structure of imipenem derives from thienamycin, which is naturally produced by Streptomyces cattleya, the bleomycin (Bm) molecule is a glycopeptide antibiotic that is naturally produced by Streptomyces verticillus (30). It is a member of a family of molecules that are mostly used as anticancer agents in clinical settings to induce DNA strand breaks. To counteract the DNA cleavage induced by Bm-like molecules (28), Bm-producing Streptomyces strains express resistance genes (ble) that are often clustered in biosynthetic operons (27). However, acquired bleomycin resistance also occurred in clinically relevant bacterial species that likely do not produce Bm, including Gram-negative rods through the acquisition of transposon Tn5 (5) or Gram-positive bacteria (mainly Staphylococcus aureus) through the acquisition of plasmid pUB110 (26). In the case of Tn5, the bleomycin resistance gene is associated with a neomycin-kanamycin resistance gene (12). A recent metagenomic study revealed that the nonclinical environment might constitute a large reservoir of Bm resistance genes (14). Despite a high divergence between each other at the sequence level, those genes encode proteins conferring a similar resistance mechanism (14). It is based on the production of an acidic protein (named BRP, for Bm resistance protein) that is able to complex Bm and thus prevent the Bm-induced DNA cleavage.

The analysis of the immediate genetic environment located downstream of the bla_NDM-1 gene in E. coli 271 (23) revealed a short open reading frame (ORF), referred to as ble_MBL (for ble gene associated with the metallo-β-lactamase NDM-1), encoding a protein sharing structural homologies with BRPs. Considering that Bm-like molecules could also be a natural selective agent for NDM-1 producers in the environment, the objectives of the study were to characterize this BRP and to evaluate its role in conjunction with NDM-1 and its prevalence among bla_NDM-1-positive isolates.

MATERIALS AND METHODS

Bacterial strains.

The screening panel included 22 bla_NDM-positive clinical isolates of worldwide origin (see Table 4). E. coli 271 was used as a reference bla_NDM-1-positive strain for cloning experiments (23). E. coli TOP10 was used as the host for recombinant constructs in cloning experiments. E. coli XL1-Red (Stratagene, Amsterdam, The Netherlands), which contains the mutS, mutT, and mutD mutator alleles, was used as a hypermutable strain for the evaluation of mutation rates in analyses with isogenic and wild-type E. coli XL1-Blue.

Table 4.

Detection of the ble_MBL gene and corresponding susceptibility patterns among a collection of bla_NDM-positive isolates^a

Strain	Name	Country of isolation	bleMBL status	Phenotypic susceptibility to zeocinb	Reference or source
E. coli	5649	Switzerland	+	R	25
E. coli	RIC	France	−	S	22
E. coli	GUE	France	+	R	20
E. coli	271	Australia	+	R	23
E. coli	RK1	Germany	+	R	18
E. coli	IR5 TW	India	+	R	This study
K. pneumoniae	UK	United Kingdom	+	R	This study
K. pneumoniae	6642	Switzerland	+	R	25
K. pneumoniae	6759	Switzerland	−	S	25
K. pneumoniae	601	Sultanate of Oman	+	R	19
K. pneumoniae	419	Sultanate of Oman	+	R	19
K. pneumoniae	Kp7	Kenya	+	R	24
K. pneumoniae	IND	India	+	R	This study
K. pneumoniae	IBN	France	+	R	21
K. pneumoniae	DIN	France	+	R	This study
K. pneumoniae	MAR	Morocco	+	R	This study
E. cloacae	IR38	India	+	R	This study
P. stuartii	PS1	India	+	R	20
P. mirabilis	7892	Switzerland	−	S	25
C. freundii	STE	France	+	R	24
A. baumannii	GEN	Switzerland	+	R	This study
A. baumannii	NRZ	Egypt	+	R	6

^aR, resistant; S, susceptible; −, negative by PCR screening; +, positive by PCR screening.

^bMICs of >256 μg/ml were considered resistant.

Molecular techniques.

To express the bla_NDM-1 and ble_MBL genes, amplifications of four different coding regions were performed in isogenic backgrounds as indicated in Fig. 3A, using total DNA of E. coli 271 as the template and primers described in Table 1. The corresponding amplicons were cloned into pCR2.1 TOPO vector (Invitrogen, Cergy-Pontoise, France) and transferred into E. coli TOP10 according to the manufacturer's instructions. The selection of recombinant clones was performed on kanamycin (30 μg/ml)-containing agar plates. The PCR-based screening of the ble_MBL gene was performed by using the specific primers ble-for and ble-rev (Table 1; also see Fig. 3A). For mutation frequency experiments, the same recombinant plasmids were electroporated in E. coli XL1-Red.

Fig 3.

Genetic constructions used for evaluating the expression analysis of the bla_NDM-1 and ble_MBL genes. (A) Scheme of primers used for cloning and their hybridization location. (B) Genetic structures of the recombinant E. coli strains pCR2.1-(P_NDM-1-bla_NDM-1-ble_MBL), pCR2.1-(P_NDM-1-bla_NDM-1), pCR2.1-(ble_MBL), and pCR2.1-(bla_NDM-1-ble_MBL).

Table 1.

Sequences of primers used in the study

Amplicon	Primer name	Sequence (5′ to 3′)
PNDM-1-blaNDM-1-bleMBL	Pre-NDM A	CACCTCATGTTTGAATTCGCC
	ble-for	CCGCAAATCTTGATTTTC
PNDM-1-blaNDM-1	Pre-NDM A	CACCTCATGTTTGAATTCGCC
	Pre-NDM B	CTCTGTCACATCGAAATCGC
blaNDM-1- bleMBL	ble-NDM-rev	GGGGTTTTTAATGCTGAATA
	ble-for	CCGCAAATCTTGATTTTC
bleMBL	ble-rev	CGCCGCAATCACTCATAC
	ble-for	CCGCAAATCTTGATTTTC

Culture media, susceptibility testing, and reagents.

MICs of bleomycin and its analogue zeocin were determined by broth microdilution. Briefly, 100 μl of Mueller-Hinton (MH) medium containing serial dilutions of bleomycin (BleomycineBellon; Sanofi Aventis, Paris, France) or zeocin (Invitrogen, Cergy-Pontoise, France) were inoculated with ∼10⁴ CFU of the studied strain and incubated at 37°C for 16 h. MICs corresponded to the first dilution that inhibits bacterial growth. The MICs of imipenem were determined by Etest (AB bioMérieux, Solna, Sweden) on MH agar plates after an overnight incubation at 37°C. To further assess any decreased susceptibility to bleomycin-like molecules, disk diffusion tests were additionally performed on MH agar plates using a sterile disk supplemented with 200 μg of zeocin. Wild-type reference strains were used as controls in each experiment for each species studied.

Survival analysis.

Survival analysis designed to evaluate the death rate of bacteria expressing BRP was performed as previously described (2). Briefly, 40 ml of Luria broth (LB) medium containing 30 μg/ml of kanamycin was inoculated with 1 ml of overnight culture of each clone and gently shaken at 37°C for 10 days. Aliquots were diluted and plated on LB agar containing 30 μg/ml of kanamycin, and numbers of CFU per ml were determined. Statistical analysis was performed on the results of three independent experiments using Student's t test. P < 0.05 was considered statistically different.

Evaluation of mutation rates.

To compare the effect of the protein BRP_MBL on the bacterial mutagenesis rate, E. coli XL1-Red recombinant strains expressing NDM-1, BRP_MBL, or both proteins were grown overnight in a Trypticase soy broth medium and then plated on rifampin (40 μg/ml)-containing Mueller-Hinton agar plates. Colony counts were performed using serial dilutions and compared to those obtained for isogenic E. coli XL1-Blue.

RESULTS AND DISCUSSION

The ble_MBL gene encodes a functional bleomycin resistance protein.

The analysis of the direct genetic environment of the bla_NDM-1 gene in E. coli 271 revealed the presence of a 366-bp-long ORF downstream of the bla_NDM-1 gene, with both genes being separated by 3 bp (Fig. 1). BlastX analysis identified this ORF in frame with the bla_NDM-1 gene as sharing similarities with genes encoding BRPs. This ble_MBL gene encoded a 121-amino-acid-long protein (BRP_MBL) with weak identity to BRPs from Gram-positive bacteria, being 19.2 and 22% identical to those of Streptomyces verticillus and Streptoalloteichus hindustanus, respectively (Fig. 2). The best matches were obtained with the BRP encoded by Tn5 (54% amino acid identity) and putative BRPs from Burkholderia species (ca. 60% amino acid identity) (Fig. 2). The analysis of the amino acid sequence of BRP_MBL revealed two proline residues at the N-terminal extremity that are known to be critical for BRP dimer formation (8). The computerized evaluation of the pI indicated a predicted acidic value of 4.8, which is in accordance with the acidic features of BRPs.

Fig 1.

Genetic environment of the bla_NDM-1 gene in E. coli 271. In this isolate, the bla_NDM-1 gene was bracketed by two insertion sequences (IS), ISEc33 on the upstream end and ISSen4 on the downstream end, from which right (IRR) and left (IRL) inverted repeats are shown (20). Horizontal arrows indicate the different open reading frames, including the insertion sequence transposase genes, bla_NDM-1, and the 3′ end of the truncated phosphoribosyl anthronilate isomerase (ΔPh ribosyl anthronilate isomerase). P corresponds to the location of the bla_NDM-1 promoter, which is partly constituted from the 3′ end of ISAba125 (20).

Fig 2.

Unrooted tree of putative bleomycin resistance proteins and the corresponding bacterial species. Amino acid sequences were obtained from the in silico analysis of the GenBank database. pUB110 corresponds to the plasmid-carried bleomycin resistance determinant from Staphylococcus aureus (26), and Tn5 indicates the determinant encoded by transposon Tn5 (5, 12). The scale bar represents 0.2 substitutions per nucleotide site.

To determine whether the ble_MBL gene encodes a functional BRP, MICs of bleomycin and zeocin (a bleomycin-like analogue) were determined for E. coli 271, for four additional NDM-1-producing clinical isolates, and for five NDM-1-positive and carbapenem-resistant E. coli transconjugants. MICs of bleomycin and zeocin were >256 μg/ml in all cases, with MICs of the E. coli J53 recipient strain used for conjugation being 8 and 4 μg/ml, respectively.

The bla_NDM-1-ble_MBL locus then was cloned and expressed in E. coli TOP10. MICs of bleomycin and zeocin for the recombinant strain were >256 μg/ml, respectively. Those results indicate that the ble_MBL gene encodes a functional BRP conferring resistance to bleomycin and zeocin.

The bla_NDM-1 and bleMBL genes form an operon with a common promoter.

Since the 3′ end of the bla_NDM-1 gene and the 5′ end of the ble_MBL gene were separated by 3 bp, both resistance genes might be coexpressed under the control of a common promoter. To test this hypothesis, different constructs were obtained (Fig. 3) and evaluated for their ability to confer resistance to bleomycin/zeocin and imipenem (Table 2). The E. coli TOP10 recombinant strain harboring the bla_NDM-1 promoter (P_NDM-1), together with the bla_NDM-1 and ble_MBL genes, exhibited a MIC of imipenem at >32 μg/ml, and MICs of bleomycin and zeocin were >256 μg/ml. As expected, the E. coli TOP10 recombinant strain with bla_NDM-1 but lacking the ble_MBL gene was susceptible to bleomycin/zeocin and resistant to carbapenems. Compared to the MICs of the E. coli TOP10 recipient strain (8 and 2 μg/ml, respectively), MICs of bleomycin (16 μg/ml) and zeocin (4 μg/ml) were slightly increased for E. coli recombinant strain possessing the ble_MBL gene preceded by 49 bp corresponding to the 3′ end of the bla_NDM-1 gene. This slight decrease of bleomycin susceptibility could be explained by the low-level expression of the ble_MBL gene from the constitutive promoter of the pCR2.1 vector. The cloning of the bla_NDM-1-ble_MBL locus lacking the P_NDM-1 promoter led to recombinant strains exhibiting only a slightly decreased susceptibility to bleomycin/zeocin and imipenem. These results showed that the bla_NDM-1 and ble_MBL genes were coexpressed under the control of a common and strong promoter, P_NDM-1. The determination of the +1 transcription initiation sites performed by the 5′ rapid amplification of cDNA ends technique (Invitrogen) as described previously (11) confirmed that they were the same for the bla_NDM-1 and ble_MBL genes, thus confirming that both genes were cotranscribed.

Table 2.

MICs of bleomycin, zeocin, and imipenem for the different E. coli recombinant strains

Strain	MIC (μg/ml)
	Bleomycin	Zeocin	Imipenem
E. coli TOP10	8	2	0.12
E. coli TOP10 pCR2.1	8	2	0.12
E. coli TOP10 pCR2.1-(PNDM-1-blaNDM-1)	8	2	>32
E. coli TOP10 pCR2.1-(PNDM-1-blaNDM-1-bleMBL)	>256	>256	>32
E. coli TOP10 pCR2.1-(blaNDM-1-bleMBL)	16	4	0.5
E. coli TOP10 pCR2.1-(bleMBL)	16	4	0.12

The BRP_MBL protein influences E. coli mutability.

Since Bostock and colleagues (3) reported that a zeocin resistance protein may influence the hypermutable property of E. coli by suppressing the mutation rate, the impact of the production of BRP_MBL in a hypermutable E. coli background was evaluated. We observed that the expression of the ble_MBL gene reduced the generation of rifampin-resistant mutants by ca. 2.4-fold (Table 3), indicating that the BRP_MBL protein was counteracting the defective DNA repair system of E. coli XL1-Red. This observation suggests that BRP_MBL-expressing strains contribute to the stabilization of the NDM-related resistance trait, with the BRP_MBL producers therefore being more prone to resist mutational inactivation.

Table 3.

Mutation frequencies for rifampin resistance for E. coli recombinant strains

Strain	Mutation frequencya	Pb
XL1-Blue	(2.8 ± 1.2) × 10−8	3 × 10−6
XL1-Red	(1.3 ± 0.4) × 10−4
XL1-Red pCR2.1-(bleMBL)	(1.2 ± 0.2) × 10−4	NS
XL1-Red pCR2.1-(PNDM-1-blaNDM-1)	(1.3 ± 0.3) × 10−4	NS
XL1-Red pCR2.1-(blaNDM-1-bleMBL)	(1.2 ± 0.2) × 10−4	NS
XL1-Red pCR2.1-(PNDM-1-blaNDM-1-bleMBL)	(5.4 ± 0.2) × 10−5	0.03

^aValues given are means from three independent experiments ± standard deviations.

^bP value was calculated using Student's t test by comparison to the XL1-Red mutation frequency, which was used as the reference. NS, no significant difference.

The ble_MBL gene is distributed largely among NDM-1 producers.

The close genetic association between the ble_MBL and bla_NDM-1 genes observed in E. coli 271 prompted us to evaluate whether this colinearity was frequent among bla_NDM-1-positive isolates. PCR mapping was performed using a collection of NDM-1 (n = 21) or NDM-2 (n = 1) producers, including E. coli, Klebsiella pneumoniae, Enterobacter cloacae, Citrobacter freundii, Providencia stuartii, Proteus mirabilis, and A. baumannii isolates (Table 4). Nineteen isolates carried the ble_MBL gene, which was always identified downstream of the bla_NDM-1 gene. Disk diffusion antibiograms of zeocin and bleomycin confirmed the resistance to those molecules for all of these ble_MBL-positive isolates. The prevalence of the ble_MBL gene among NDM producers was very high, revealing that both genes probably had been mobilized together from the same progenitor. That hypothesis is strengthened by the fact that the genes exhibit a very similar G+C content (61.5% for bla_NDM-1 and 60.9% for ble_MBL). A bleomycin-based selective pressure therefore could be at the origin of the selection of ble_MBL-harboring plasmids and consequently coselect for NDM-1 producers, leading to carbapenem resistance.

Conclusion.

A novel bleomycin resistance protein has been characterized that was associated with the bla_NDM-1 gene as part of the same operon. We showed that this BRP was functional and caused decreased susceptibility to bleomycin and bleomycin-like molecules, such as zeocin. Since Walsh et al. demonstrated the large spread of the bla_NDM-1 gene in bacteria recovered from environmental samples in New Delhi (32), it can be hypothesized that bleomycin or bleomycin-like molecules exert a selective pressure in the environment, in accordance with a metagenomic study that showed that BRP-encoding genes could be retrieved from activated sludge (14). In the United Kingdom and Germany, studies showed that bleomycin was present in sewage effluents, rivers, and potable water (1, 29). It is possible that bleomycin-like molecules, secreted by Streptomyces spp. or resulting from contaminations due to the medical use of bleomycin as an anticancer agent, contribute to a selective pressure leading to the further spread of NDM producers in the environment.

We believe that results presented here may have important clinical implications, since the spread of NDM-1 producers may result not only from antibiotic selection pressure, such as ß-lactams, but also selective pressure by anticancer drugs and environmental molecules.