Biochemical Characterization of VIM-39, a VIM-1-Like Metallo-β-Lactamase Variant from a Multidrug-Resistant Klebsiella pneumoniae Isolate from Greece

Abstract

VIM-39, a VIM-1-like metallo-β-lactamase variant (VIM-1 Thr33Ala His224Leu) was identified in a clinical isolate of Klebsiella pneumoniae belonging to sequence type 147. VIM-39 hydrolyzed ampicillin, cephalothin, and imipenem more efficiently than did VIM-1 and VIM-26 (a VIM-1 variant with the His224Leu substitution) because of higher turnover rates.

TEXT

Acquired metallo-β-lactamases (MBLs) are important resistance determinants emerging in clinically relevant Gram-negative pathogens (1–3). MBLs are zinc-dependent enzymes commonly characterized by the ability to hydrolyze all β-lactams (with the exception of monobactams), including carbapenems. Their activity is not affected by the currently available β-lactamase inhibitors (i.e., clavulanic acid, tazobactam, sulbactam, or avibactam). The most prevalent acquired MBLs belong to the IMP, VIM, and NDM types, all of subclass B1 (4). Among them, the family of VIM-type enzymes currently includes >40 variants (www.lahey.org/studies) divided into five subgroups (VIM-1 [5], VIM-2 [6], VIM-7 [7], VIM-12 [8], and VIM-13 [9]) based on their structural relatedness. Interestingly, these variants may exhibit significant functional differences and represent an important model to better understand the structure-function relationships of these clinically relevant MBLs (6).

In this work, we performed a detailed biochemical characterization of VIM-39, a new VIM-1-related MBL variant found in a sequence type 147 (ST147) Klebsiella pneumoniae clinical isolate from Greece (10) that differs from VIM-1 by two substitutions, His224Leu (already found in VIM-26 [11]) and Thr33Ala.

The original isolate, ST147 K. pneumoniae Kpn-7994, a VIM-39-producing isolate that exhibits resistance to multiple drugs, including all β-lactams (except for aztreonam; Table 1), amikacin, chloramphenicol, trimethoprim-sulfamethoxazole, ciprofloxacin, and colistin, was from a patient treated at Attikon General Hospital (Athens, Greece) (10). Isoelectric focusing of sonicated cell extracts developed with 0.25 mM nitrocefin (in 50 mM HEPES buffer [pH 7.5] supplemented with 0.5 mM ZnCl₂) showed that Kpn-7994 produced two β-lactam-hydrolyzing protein bands with apparent pI values of 5.1 (consistent with that of a VIM MBL) and 7.6 (most likely consistent with the endogenous SHV-type β-lactamase). Additionally, a matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) meropenem hydrolysis assay confirmed the production of a carbapenemase by Kpn-7994 (12). The resistance phenotype of the Escherichia coli transconjugants, containing bla_VIM-39, was consistent with the production of an MBL conferring resistance or reduced susceptibility to penicillins, cephalosporins, and carbapenems but susceptibility to aztreonam (Table 1). They also acquired resistance to trimethoprim-sulfamethoxazole. E. coli transconjugants carried FIIk plasmids of approximately 200 kb that harbored the bla_VIM-39 gene as the first cassette of an In-e541-like integron.

TABLE 1.

MICs of β-lactam antibiotics for a VIM-39-producing K. pneumoniae clinical isolate and an E. coli A15 transconjugant and E. coli DH5α clones producing VIM MBLs under isogenic conditions

Antibiotic	MIC (μg/ml)
	K. pneumoniae Kpn-7994	E. coli DH5α/pBvim39	E. coli DH5α/pBvim26	E. coli DH5α/pBvim1	E. coli DH5α/pBvim1-T33A	E. coli DH5α/pBC-SK	E. coli A15/pKP-M7994	E. coli A15
Ampicillin	>1024	512	128	512	512	2	512	4
Piperacillin	256	64	64	64	64	2	32	≤0.5
Ticarcillin	>1024	1,024	512	1,024	1,024	2	1,024	4
Temocillin	64	16	16	64	64	8	16	8
Cephalothin	>256	256	64	256	256	4	256	4
Ceftriaxone	64	4	4	2	2	≤0.12	8	≤0.12
Ceftaroline	>64	64	64	32	64	≤0.12	>64	0.12
Cefoxitin	128	4	4	16	16	1	8	1
Cefotaxime	>256	16	16	8	32	≤0.12	8	≤0.6
Ceftazidime	256	32	16	32	128	0.25	16	≤0.25
Cefepime	32	1	1	4	4	≤0.12	1	≤0.12
Aztreonam	0.5	≤0.12	≤0.12	≤0.12	≤0.12	≤0.12	≤0.12	≤0.12
Imipenem	32	4	2	1	4	0.12	4	0.25
Meropenem	16	0.12	0.12	0.06	0.12	0.032	0.12	0.032
Ertapenem	8	0.032	0.032	0.032	0.032	≤0.016	0.032	≤0.016
Doripenem	32	0.12	0.12	0.06	0.12	0.032	0.25	0.032

The bla_VIM-39 gene was amplified by PCR with primers VIM-1-EXP/f (5′-GGAATTCCATATGTTAAAAGTTATTAGTAGTTTAT-3′), which added EcoRI (underlined) and NdeI (bold) restriction sites at the 5′ end of the gene, and VIM-1-EXP/r (5′-CCGGATCCTGCTACTCGGCGACTGAGCG-3′), which added a BamHI (underlined) restriction site after the bla_VIM-39 stop codon. Total DNA from Kpn-7994 was used as the template. The resulting 821-bp amplicon was digested with NdeI and BamHI and cloned into the pLB-II vector (13), resulting in plasmid pBvim39, where overexpression of the bla_VIM-39 gene was under the transcriptional control of the P_lac promoter. Similar plasmids encoding VIM-1 (pBvim1) and VIM-26 (pBvim26) were constructed by following the same procedure and using total DNA from K. pneumoniae isolates Kpn-1192 (expressing VIM-1) and Kpn-1151 (expressing VIM-26) as the templates (10). The identities of the inserts were verified by sequencing, and recombinant plasmids were introduced into E. coli DH5α.

E. coli DH5α strains harboring plasmids pBvim39 and pBvim26 were used as sources of the VIM-39 and VIM-26 enzymes, respectively. The two MBL variants were then purified by chromatography from the supernatants of 500-ml cultures grown in ZYP-5052 medium (14) supplemented with 50 μg/ml chloramphenicol. The cultures were grown aerobically at 37°C for 24 h. Under these conditions, most of the enzyme activity was found in the cultures' supernatants. Cells were removed by centrifugation (10,000 × g for 30 min at 4°C), and the supernatant was concentrated by ultrafiltration with a cellulose membrane (Millipore Corporation, Billerica, MA, USA) with a nominal molecular weight limit of 10,000. The concentrated samples were loaded onto an XK26/20 column packed with 75 ml of Q Sepharose HP ion-exchange medium (GE Healthcare, Uppsala, Sweden) previously equilibrated with 20 mM triethanolamine (pH 7.2; buffer A). Elution of the bounded proteins was performed with the same buffer by using a linear NaCl gradient (0 to 0.25 M in 800 ml). The fractions containing β-lactamase activity were pooled, and their buffer was changed to 20 mM diethanolamine (pH 8.5) supplemented with 50 μM ZnSO₄ (buffer B) with a HiPrep 26/10 desalting column (GE Healthcare). The resulting sample was loaded onto a Source 15Q column (bed volume, 1 ml; GE Healthcare) previously equilibrated with buffer B, and proteins were eluted with a linear NaCl gradient (0 to 0.25 M in 50 ml) in the same buffer. Fractions containing β-lactamase activity were pooled and stored at −20°C. During the purification procedure, the presence of β-lactamase activity was monitored by measuring the hydrolysis of 150 μM imipenem in 50 mM HEPES buffer supplemented with 50 μM ZnSO₄ (pH 7.5) (buffer H). This protocol yielded approximately 2.1 mg/liter purified VIM-39 and 0.9 mg/liter VIM-26 (specific activities, 1,941 ± 59 and 1,666 ± 34 U, respectively [1 U is the amount of enzyme hydrolyzing 1 μmol of imipenem/min/mg of protein]). The purity of the final preparations was >95%, as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (15). The authenticity of the enzyme preparations was confirmed by MALDI-TOF MS with a Microflex LT mass spectrometer (Bruker Daltonics). The measured molecular masses of the two enzymes were in agreement with the expected theoretical values (25,266.74 versus 25,268.01 Da for VIM-39 and 25,300.15 versus 25,298.04 Da for VIM-26) calculated for the mature proteins (signal peptide, residues 1 to 26).

β-Lactam hydrolysis by the purified VIM-26 and VIM-39 enzymes was monitored by measuring the absorbance variation with a Cary 100 UV-Vis spectrophotometer (Varian Instruments, Walnut Greek, CA, USA) at 30°C in buffer H in a final reaction volume of 500 μl. The wavelengths and changes in extinction coefficients used in spectrophotometric assays were as described previously (5). The purified VIM enzymes were diluted in buffer H supplemented with 20 μg/ml of bovine serum albumin to prevent denaturation. The steady-state kinetic parameters (K_m and k_cat) were determined under the initial reaction rates by using the Hanes-Woolf plots (6). K_m values of <10 μM were measured as inhibition constants by using a competitive inhibition model (6) with 100 μM nitrocefin as the reporter substrate. The kinetic parameters for the VIM-26 and VIM-39 MBLs were compared with those already reported for VIM-1 (5).

Like other subclass B1 MBLs, VIM-39 showed a broad substrate profile and efficiently hydrolyzed all of the compounds tested (k_cat/K_m, >10⁵ M⁻¹ · s⁻¹) except aztreonam (Table 2) and exhibited a strong preference for ampicillin, cephalothin, and the carbapenems (except, to some extent, ertapenem), with k_cat/K_m values of ≥5.5 × 10⁶ M⁻¹ · s⁻¹. Overall, rather high turnover rates were measured, although substrates with an α-methoxy group, such as temocillin and cefoxitin, were associated with the lowest turnover rates (k_cat, ≤45 s⁻¹). Piperacillin was hydrolyzed very efficiently, despite a significantly high K_m value, thanks to a very high turnover rate. Interestingly, VIM-39 is very similar to VIM-1 in this regard and slightly divergent from VIM-26. VIM-39 was more active against imipenem, meropenem, and doripenem than against ertapenem (32.8-, 7.9-, and 10.4-fold, respectively). In particular, its low ertapenem hydrolytic efficiency was due to a lower turnover rate (k_cat, 70 s⁻¹). Compared to VIM-1, the most dramatic differences could be summarized as an important increase in the turnover rates, especially for carbapenems, and a significant decrease in the K_m values of ampicillin, ticarcillin, and ceftazidime.

TABLE 2.

Kinetic parameters of the purified VIM-39 MBL compared with those of VIM-1 and VIM-26

Substrate	kcat (s−1)			Km (μM)			kcat/KM (μM−1 · s−1)
	VIM-39	VIM-26	VIM-1a	VIM-39	VIM-26	VIM-1a	VIM-39	VIM-26	VIM-1a
Penicillins
Ampicillin	350	75	37	37	22	917	9.4	3.5	0.040
Piperacillin	1,400	270	1,860	800	145	3,500	1.7	1.9	0.53
Ticarcillin	220	41	452	75	22	1,117	3.0	1.9	0.41
Temocillin	45	6.9	0.5	17	14	22	2.7	0.47	0.023
Cephamycin cefoxitin	17	4	26	135	46	131	0.12	0.087	0.2
Cephalosporins
Cephalothin	370	36	280	59	20	53	6.2	1.8	5.3
Ceftazidime	110	17	60	140	42	794	0.76	0.42	0.076
Monobactam aztreonam	<0.01	<0.01	<0.01	>1,000	>1,000	>1,000	<1 × 10−5	<1 × 10−5	<1 × 10−5
Carbapenems
Imipenem	640	71	2	28	6.8	1.5	23	10	1.3
Meropenem	670	840	13	120	135	48	5.5	6.1	0.27
Ertapenem	70	74	NDb	99	41	ND	0.7	1.8	ND
Doripenem	480	160	ND	65	28	ND	7.3	5.7	ND

^aData for VIM-1 are from reference 5. Values are the means of three independent measurements. Standard deviations were always <10%.

^bND, not determined.

Comparison of the kinetic parameters of VIM-39 with those of VIM-26, determined under the same conditions, showed that the former enzyme was a slightly more efficient β-lactamase, exhibiting higher or similar (piperacillin, meropenem) k_cat/K_m values for most of the substrates tested. The most marked increases in efficiency were against ampicillin, temocillin, cephalothin, and imipenem (2.7-, 5.7-, 3.4-, and 2.3-fold, respectively), thanks to higher k_cat values. Additionally, kinetic constants showed that VIM-26 hydrolyzes penicillins more efficiently than cephalosporins, with a marked preference for ampicillin. For the cephalosporins, VIM-26 was more active against cephalothin and ceftazidime than against cefoxitin, mainly because of a very low turnover rate (k_cat, 4 s⁻¹). VIM-26 exhibited similar hydrolytic efficiencies for imipenem, meropenem, and doripenem but a lower one for ertapenem. These findings are in overall accordance with the data published in a recent study describing the structural and biochemical characterization of VIM-26 (16), although differences in the kinetic parameters were found. The latter differences could be explained by the different experimental conditions used.

MICs of β-lactam antibiotics against E. coli DH5α/pBvim39, DH5α/pBvim26, and DH5α/pBvim1 were determined by the broth microdilution method (17) and interpreted on the basis of the EUCAST criteria (http://www.eucast.org/). As expected, expression of the bla_VIM-39, bla_VIM-26, and bla_VIM-1 genes in E. coli DH5α conferred resistance or reduced susceptibility to all β-lactams, including penicillins, cephalosporins, and carbapenems (Table 1). Only aztreonam MICs were unaffected by the presence of the MBL gene. Interestingly, the MICs of imipenem were higher for the VIM-39-producing strain than for the VIM-26- and VIM-1-producing ones, which nicely correlates with the kinetic data and apparently underlines the potential involvement of residue Ala33 in this regard. The MICs of ampicillin and cephalothin for E. coli DH5α/pBvim39 were higher than those for E. coli DH5α/pBvim26 but similar to those for E. coli DH5α/pBvim1. In contrast, the MICs of temocillin, cefoxitin, and cefepime for E. coli DH5α/pBvim1 were higher than those for E. coli DH5α/pBvim39 and DH5α/pBvim26. The MICs were generally in line with the kinetic data obtained under the same conditions. However, discrepancies between MICs and kinetic data were observed for some substrates. These discrepancies might be explained by several factors, like differences in enzyme production (specific activities, 358.4 U for VIM-39 and 219.5 U for VIM-26) and/or stability, as previously demonstrated (13).

The VIM-39 β-lactamase differed from VIM-26 by a single amino acid substitution (Thr33Ala) and from VIM-1 by two substitutions (Thr33Ala and His224Leu, the latter being present in VIM-26). Therefore, we aimed to evaluate the role of the single Thr33Ala substitution in increased “imipenemase” activity. For this purpose, plasmid pBvim1 was used in mutagenesis experiments. Plasmid pBvim1-T33A, which encoded the β-lactamase VIM-1-T33A, was obtained with a QuikChange mutagenesis kit (Stratagene) and two mutagenic primers, m33V-F (5′-GCCGAGTGGTGAGTATCCGGCAGTCAACGAAATTCCGGT-3′) and m33V-R (5′-ACCGGAATTTCGTTGACTGCCGGATACTCACCACTCGGC-3′) (nucleotides that differ from those in the original sequence are underlined). The presence of the expected mutation in the bla_VIM-containing insert was verified by sequencing of at least two PCR products, and the recombinant plasmid was introduced into E. coli DH5α. Interestingly, the antibacterial susceptibility data showed that the MICs of imipenem for E. coli DH5α/pBvim1-T33A were similar to those for E. coli DH5α/pBvim39 (Table 1). This finding strengthened the role of the Ala33 residue in the higher activity of VIM-39 against imipenem.

In this study, we examined a novel VIM-type MBL variant, VIM-39, that is characterized by strong hydrolytic activity on carbapenems, in particular, imipenem. The Ala33 residue in VIM-39 is unique among VIM-type amino acid sequences and is located at the N terminus, precluding direct interaction with amino acids important for catalysis. However, we showed here that this substitution alone was apparently sufficient to confer an increase in the MICs of imipenem observed with VIM-39. This observation might be explained by an indirect impact in the positioning of the active-site residues by the replacement of the polar Thr residue with the aliphatic Ala residue. Previous studies have shown that amino acid substitutions located outside the active site may modulate the hydrolytic properties of the enzyme (18–20).

The second residue differing from VIM-1, Leu224, was previously observed in VIM-26. The importance of position 224, which represents variability among VIM-type enzymes, in the orientation and stabilization of the substrate in the active site has been hypothesized previously (6). This hypothesis is in agreement with the findings of a recent study that demonstrated the importance of the Leu224 residue in VIM-26, exhibiting higher hydrolytic activity against penicillins than VIM-1 and VIM-2 but lower activity against cephalosporins (16). Analysis of VIM-26 crystal structures showed that the L3 loop in VIM-26, which includes Leu224 and Ser228, makes the R2 drug binding site both open and neutral, explaining the reduced cephalosporinase of VIM-26 (16). On the basis of our kinetic data, VIM-26 is a less efficient β-lactamase than VIM-39. Assuming that VIM-39 is a direct descendant of VIM-26, which was produced by 26.6% (17/64) of the VIM-producing Enterobacteriaceae isolates recovered from patients in two Greek hospitals (10), a positive antibiotic selection driving its emergence can be supported. Furthermore, in addition to several VIM-type enzymes (VIM-1, VIM-4, VIM-12, VIM-19, VIM-26, and VIM-27) found in Enterobacteriaceae of Greek origin (11, 20–24), the emergence of VIM-39 documents the ongoing evolution of VIM-1-type enzymes in this country.

Nucleotide sequence accession number.

The nucleotide sequence of bla_VIM-39 has been submitted to the GenBank database and assigned accession no. KF131539.