Phenotypic and molecular detection of MBL.

Hodge and imipenem-EDTA double-disc synergy tests as well as Etest were used to screen for MBL production. In addition, the carbapenemase activities of cell sonicates from overnight broth cultures (Mueller-Hinton broth; BD) were determined by spectrophotometric assays. These were undertaken by using 150 μM imipenem as the substrate and 299 nm for the measurement of hydrolysis. The assays were performed with or without EDTA (25 mM) to examine the inhibition of carbapenemase activity. The strains were screened for the presence of known mobile MBL genes (bla_VIM, bla_IMP, bla_SPM-1, bla_GIM-1, bla_SIM-1, bla_AIM-1) by PCR with the primers reported previously (22). The strains were also screened for the presence of other β-lactamase genes (bla_CTX, bla_CMY, etc.) as well as class 1, 2, and 3 integron structures (24). Strains known to harbor MBL genes were used as positive controls for the PCR analysis.

Conjugation.

E. coli J53 was used as the recipient in the conjugation studies, and mating was carried out on blood agar medium without selection (25). After 18 h, the mixed cultures were taken from the plates, suspended in saline, and plated onto MacConkey medium containing sodium azide (100 μg/ml) and ceftazidime (8 μg/ml). Confirmation that conjugation had taken place in E. coli J53 was carried out by indole testing and testing for the presence of the MBL gene by PCR analysis.

DNA cloning.

Cloning experiments were performed by using the cloning vector pK18, as described previously (35). Restriction endonucleases BamHI, EcoRI, and HindIII (Invitrogen, Carlsbad, CA) were used to cut both plasmid and K. pneumoniae 05-506 chromosomal DNA. Transformation was carried out by using electroporation and E. coli TOP10 cells (Invitrogen Ltd, Paisley, United Kingdom). Transformants were selected on LB agar plates containing 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (30 μg/ml), ceftazidime (6 μg/ml), and kanamycin (25 μg/ml). Recombinant plasmids were recovered with a QIAprep spin miniprep kit (Qiagen, West Sussex, United Kingdom). Restriction endonuclease analysis of the plasmid was carried out with a variety of restriction enzymes to determine the size of the ligated nucleotide segment.

The PCR-amplified open reading frame of NDM-1 was digested with the appropriate enzymes and was subsequently ligated into the pVFT3S (six-His—trx fusion vector) bacterial expression vector. The construct was transformed into E. coli BL21 groEL/groES (pGro7/BL21; Takara Bio Inc., Japan).

DNA sequencing and computer analysis.

Both strands of the clones from the libraries that were produced were sequenced by using M13 forward and M13 reverse primers. The cloned insets containing bla_NDM-1, the complex class 1 integron, and bla_CMY-4 were fully sequenced. The nucleotide sequences, deduced amino acid sequences, and phylogenetic relationships were analyzed by using the Lasergene software package (DNAStar, Madison, WI). The sequences obtained were compared to the sequences available on the Internet (http://www.ebi.ac.uk/fasta33/).

Pulsed-field gel electrophoresis (PFGE) and genomic DNA digestion with endonucleases.

Genomic DNA was prepared in agarose blocks and digested with the restriction enzymes XbaI and SpeI (Roche Diagnostics, Mannheim, Germany), I-CeuI (New England Biolabs, Beverly, MA), and S1 nuclease (Invitrogen, Abingdon, United Kingdom). The DNA fragments were separated by use of a CHEF-DR III apparatus (Bio-Rad, Hercules, CA) for 20 h at 6 V/cm and 14°C with initial and final pulse times of 0.5 and 30 s, respectively.

Southern blot analysis.

Southern blotting was performed on agarose gels by in-gel hybridization with the bla_NDM-1 probe labeled with ³²P (Strategene) and by a recently described random primer method (22).

MLST.

Multilocus sequence typing (MLST) of K. pneumoniae was performed as described by Diancourt et al. (7). Experimentally determined DNA sequences were uploaded into the MLST database (http://www.pasteur.fr/recherche/genopole/PF8/mlst/Kpneumoniae.html), and allelic numbers as well as sequence types (STs) were obtained.

ß-Lactamase purification.

Cell lysates were centrifuged at 12,000 rpm for 30 min at 4°C. The supernatant was obtained and loaded onto Ni-nitrilotriacetic acid agarose. The NDM-1 protein was eluted by the use of elution buffer with 300 mM imidazole. The protein was separated from the His-trx tag by tobacco etch virus protease and was finally purified by gel filtration on a GS320 column.

Kinetic measurements.

Purified ß-lactamase was used to determine the kinetic parameters k_cat and K_m. Reactions were performed at 30°C in 1 ml of assay buffer (10 mM HEPES, pH 7.5). The assays were performed in a UV-1601PC spectrophotometer (Shimadzu Corp., Tokyo, Japan) by observing the changes in absorption resulting from the opening of the ß-lactam ring at the specific wavelengths for each of the each antimicrobial agents evaluated, as described previously (18).

Typing and phenotypic characterization of K. pneumoniae 05-560, E. coli NF-NDM-1, and transformants.

ST14 is a single-locus variant of ST15, which has been encountered with KPC- and CTX-M-producing K. pneumoniae strains (19, 30). We have also found ST14 among U.S. KPC producers (19).

The susceptibility patterns of K. pneumoniae 05-560, E. coli NF-NDM-1, E. coli TOP10, E. coli TOP10(pNDM-1), E. coli J53, and E. coli J53(pNDM-1) are listed in Table ​Table1.1. K. pneumoniae 05-560 was resistant to all β-lactams and was sensitive only to colistin (Table ​(Table1).1). The susceptibility profile of E. coli NF-NDM-1 largely reflects that of 05-560, apart from susceptibility to aztreonam, cefepime, and ciprofloxacin, to which the level of resistance of 05-560 was not quite as high. When the cloned MBL gene and its adjacent DNA were expressed in an E. coli background, the strain was highly resistant to all cephalosporins, apart from cefepime (MIC, 8 μg/ml), and was sensitive to aztreonam. However, the aztreonam MICs differed significantly for TOP10 bla_NDM-1 (MIC, 0.25 μg/ml) and J53 bla_NDM-1 (MIC, 24 μg/ml), even though the aztreonam MICs for TOP10 and J53 were very similar. Interestingly, the MICs of imipenem and meropenem were relatively high for both E. coli TOP10 and J53 compared with those for strains with the other cloned MBL genes (31, 41).

Phenotypic and molecular screening for MBLs.

The results of the MBL screening tests were positive by the imipenem-EDTA double-disc synergy test, the MBL Etest (256 μg/ml with imipenem and 3 μg/ml with imipenem plus inhibitor), and the determination of carbapenemase activities by spectrophotometry (data not shown). However, PCR analysis failed to detect previously known MBL genes when the control genes bla_VIM, bla_IMP, bla_SPM-1, bla_GIM-1, bla_SIM-1, bla_AIM-1 were used. PCR did detect a class 1 integron (intI and qacEΔ1/sul) of 4.8 kb.

Nucleotide and deduced amino acid sequences of three resistance-conferring regions.

Total DNA from K. pneumoniae 05-560 was used to construct a genomic library to isolate all genes conferring β-lactam resistance. After the plasmids of resistant colonies were analyzed by restriction profiling, it was evident that there were two different types of β-lactam resistance. One group carried a putative AmpC enzyme, on the basis of the fact that it was resistant to the effects of EDTA but sensitive to inhibition with cloxacillin. The other group possessed a putative MBL on the basis of the fact that it was resistant to the effects of cloxacillin and clavulanic acid but sensitive to EDTA. Restriction of the recombinant plasmids showed that the AmpC-type genes are carried on a 2.1-kb fragment and that the MBL gene is carried on a 4.3-kb fragment. Additionally, the 4.8-kb fragment containing the class 1 integron was also sequenced.

The class 1 integron contains a unique set of genes, some of which have been previously reported from Asia. Immediately next to the intI gene is an arr-2 cassette that mediates resistance to rifampin (rifampicin) and that has been found in isolates of Pseudomonas aeruginosa and Acinetobacter baumannii from China and Hong Kong, respectively (GenBank accession numbers EU886979 and AY038837, respectively) (1, 14, 27, 38). It has also been identified on the chromosome of Streptomyces coelicolor (GenBank accession number NC003888). Next to aar-2 is a novel erythromycin esterase gene (ereC) which has 92% nucleotide sequence identity with ereA from E. coli (GenBank accession number AY183453) (Fig. ​(Fig.1).1). The predicted protein sequence differs from the EreA sequence at an additional 10 amino acids at the N terminus and also has 31 amino acid substitutions throughout the sequence. At present, only two erythromycin esterases, ereA and ereB, have been found (GenBank accession numbers DQ157752 and X03988, respectively), and these share only 25% identity in their amino acid sequences (Fig. ​(Fig.1).1). Adjacent to the esterase gene is aadA1, which mediates resistance to gentamicin, and the cmlA7 gene cassette, which mediates resistance to chloramphenicol, is present between aadA1 and qac/sul (Fig. ​(Fig.22).

Open in a separate window

FIG. 1.

Alignment of novel erythromycin esterase EreC (this study) with EreA and EreA2. Amino acid changes between EreC and EreA are indicated with asterisks. The single amino acid difference between EreA and EreA2 is highlighted in gray.

Open in a separate window

FIG. 2.

Three characterized antibiotic resistance-conferring regions from K. pneumoniae 05-506. (A) The 4.3-kb region is linked to the 4.8-kb complex class 1 integron region. The genes encoding the efflux pump and lactate dehydrogenase (gray diagonal lines) are of Klebsiella origin. bla_NDM-1 (dark gray) is flanked between the pathogenicity island (vertical black lines) and IS26/Tn3 (black small squares). This region lies downstream of the 4.8-kb complex class 1 integron containing Int (checkered area), arr-2, ere2A, aadA1, and cmlA7 as gene cassettes and qacΔ1 (white boxes). Downstream is an intact copy of ISCR1 (black and white diagonal lines). Arrows, direction of transcription; black ellipses, 59-base elements; Δ, genes that are truncated. (B) bla_CMY-4 (gray) is located between ISEcP1 (black) and blc (white), as reported previously (16).

The AmpC fragment is shown in Fig. ​Fig.22 and carries bla_CMY-4, which is responsible for the AmpC phenotype. Upstream of bla_CMY-4 is an intact copy of ISEcP1, which is probably responsible for the transposition and mobility of bla_CMY-4. Downstream is blc, which is a common feature in many bla_CMY regions and which has been reported by Kang et al. (16).

The 4.3-kb fragment contained a novel MBL gene, designated bla_NDM-1 (Fig. ​(Fig.2).2). This section of DNA was a chimera of truncated and intact genes. The bla_NDM-1 gene was found between a section of K. pneumoniae chromosomal DNA and truncated sections of the IS26 and Tn3 transposase genes. The Klebsiella DNA consisted of a putative efflux protein gene adjacent to another gene encoding l-lactate hydrogenase, and both were 99.5% identical to a section of chromosomal DNA from K. pneumoniae MGH78578 (GenBank accession number CP000647) and found in the same orientation in which they are present on the Klebsiella chromosome. The lactate dehydrogenase gene was truncated and missing approximately 750 bp of the 5′ end of the gene. Immediately upstream of the lactate dehydrogenase gene was a short internal part of the ampC gene, bla_DHA-1 (base pair coordinates 1268 to 1478), consisting of the DNA encoding amino acids 196 to 265. This gene was truncated at both ends both by the lactate dehydrogenase gene and by a section of DNA encoding the 5′ end of a phosphoribosyl anthalinate isomerase gene. This section of DNA (coordinates 1478 to 2100) has identities ranging from 55% to 68% to the same gene from various members of the alphaproteobacteria, including Rhodobacter spp., Rhodopseudomonas spp., Caulobacter spp., and Novosphingobium spp. The highest degree of identity was found with Novosphingobium aromaticivor, with 68% nucleotide sequence identity over 575 nucleotides. Again, this gene was truncated and missing the DNA encoding the last 19 amino acids of the C terminus of the protein. The bla_NDM-1 gene is encoded within a 1,350-bp section of DNA with a lower G+C content (57%) than that of the surrounding DNA, which has a G+C content of 62 to 65% and which consists of a gene of 807 nucleotides and flanking DNA on either side. Upstream of the bla_NDM-1 gene is 250 nucleotides of DNA that contains several sequences that show similarity to conventional promoter sequences, and it is likely that the promoter driving the expression of bla_NDM-1 has been acquired along with the gene (Fig. ​(Fig.2).2). Farther upstream is a truncated left-hand end of an IS26 insertion sequence, including 270 bp of the left-hand end of the element with the left-hand inverted repeat and DNA encoding the first 74 amino acids of the transposase. This element lies adjacent to a tnpA gene of a Tn3-like transposon (coordinates 3748 to 4318), which is itself truncated and missing the 5′ end of the gene encoding the first 128 amino acids of the transposase (Fig. ​(Fig.22).

Structural properties of NDM-1.

The bla_NDM-1 open reading frame encodes a putative protein of 269 amino acids with a molecular mass of approximately 27.5 kDa (Fig. ​(Fig.3).3). The NDM-1 sequence and the behavior of NDM-1 during gel filtration and mass spectrometry (data not shown) indicate that it is actively present as a monomer. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis showed an approximate molecular mass of 28 kDa (data not shown). NDM-1 possesses a leader peptide, with the probable cleavage site occurring at position 19 between two alanines which are in a very similar position to the cleavage site for the VIM MBLs (Fig. ​(Fig.3).3). NDM-1 also shares the additional loop of the VIM enzymes at positions 34 and 47 but also possesses a unique additional sequence at positions 162 to 166 (bla_NDM-1 and not BBL numbering) (Fig. ​(Fig.3)3) (4). NDM-1 shares very little identity with other MBLs and is the most closely related to VIM-1/VIM-2, with which it has only 32.4% identity (Fig. ​(Fig.4).4). Curiously, NDM-1 also has a unique HXHXD motif among the mobile MBLs, as it contains an alanine between the two histidines. NDM-1 also possesses a tyrosine at position 222 instead of the universally conserved tryptophan (Fig. ​(Fig.3).3). The mature peptide has a theoretical pI of 6.9.

Open in a separate window

FIG. 3.

Alignment of the amino acid sequences of NDM-1 with the amino acid sequences of IMP-1, IMP-2, IMP-8, VIM-1, VIM-2, GIM-1, SPM-1, SIM-1, and KHM-1. Conserved residues coordinating zinc ions are denoted with asterisks. Key amino acids are also numbered according to the BBL scheme (12). Additional amino acids unique to NDM-1 are enclosed in a light gray box. Amino acid substitutions near the active site are highlighted in plain boxes (13).

Open in a separate window

FIG. 4.

Phylogenetic tree obtained for MBL subgroup B1. The numbers for the substitutions should be multiplied by 100. The alignment used was prepared with the CLUSTALW program. The GenBank accession numbers for the sequences are as follows: IMP-1, S71932; IMP-2, AJ243491; IMP-8, AF322577; VIM-1, Y18050; VIM-2, AF191564; GIM-1, AJ620678; SPM-1, AJ492820; SIM-1, AY887066; KHM-1, AB364006.

Functional properties of NDM-1.

Analysis of the purified NDM-1 enzyme by sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis showed a single band corresponding to a size of 28 kDa (data not shown). Table ​Table22 lists the k_cat and K_m values and compares them with the kinetic values of IMP-1 and VIM-2. NDM-1 shows tight binding (low K_m values) to most cephalosporins and, in particular, to cefuroxime, cefotaxime, and cephalothin (cefalotin). It also shows relatively tight binding to the penicillins, which is unusual for an MBL, as both IMP-1 and VIM-2 possess higher K_m values (Table ​(Table2).2). Interestingly, NDM-1 does not bind to the carbapenems as tightly as IMP-1 or VIM-2. The rate of turnover (k_cat) of penicillins by NDM-1 is lower than that by IMP-1 and VIM-2 and is generally unusual for MBLs. The k_cat values for the cephalosporins are more similar to those of IMP-1 than to those of VIM-2, yet the k_cat values for the carbapenems (20 and 12 s⁻¹ for imipenem and meropenem, respectively) are more similar to those for VIM-2 (10 and 1 s⁻¹, respectively) than IMP-1 (46 and 50 s⁻¹, respectively) (Table ​(Table2).2). Due to the fact that most of the K_m values are low, the k_cat/K_m ratios are reasonably high and comparable to those of VIM-2. The exception is that of ceftazidime, which has a K_m of 181 μM and a k_cat of 5 s⁻¹ and which thus presents with a k_cat/K_m of 0.03.

This work was funded by EU grant LSHM-CT-2005-018705 and Wellcome Trust grant 084627/Z/08.

D. Yong thanks Seok H. Jeong and Yunsop Chong for their valuable advice.