This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Walsh, T. R. Articles by Nordmann, P. Search for Related Content PubMed PubMed Citation Articles by Walsh, T. R. Articles by Nordmann, P.

 Previous Article  |  Next Article 

Clinical Microbiology Reviews, April 2005, p. 306-325, Vol. 18, No. 2
0893-8512/05/$08.00+0     doi:10.1128/CMR.18.2.306-325.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.

Metallo-ß-Lactamases: the Quiet before the Storm?

Department of Pathology and Microbiology, University of Bristol, Bristol, United Kingdom,¹ Service de Bactériologie-Virologie, Hôpital de Bicêtre, Assistance Publique/Hôpitaux de Paris, Faculté de Médecine Paris-Sud, Le Kremlin-Bicêtre, France²

SUMMARY

INTRODUCTION

CLASSIFICATION OF MBLs

CHROMOSOMALLY ENCODED MBLs

GENETIC APPARATUS OF TRANSFERABLE MBLs

BIOCHEMISTRY OF MBLs

TRANSFERABLE MBLs

IMP-Type MBLs

VIM-Type MBLs

SPM-1

GIM-1

EXPERIMENTAL INHIBITORS OF MBLs

DETECTION OF MBLs

TREATMENT OF INFECTIONS WITH MBL-POSITIVE GRAM-NEGATIVE BACTERIA

EPIDEMIOLOGY OF MBLs: NOW AND IN THE FUTURE

CONCLUSION

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
The increase in antibiotic resistance among gram-negative bacteria is a notable example of how bacteria can procure, maintain, and express new genetic information that can confer resistance to one or several antibiotics. This genetic plasticity can occur both inter- and intragenerically. Gram-negative bacterial resistance possibly now equals or usurps that of gram-positive bacterial resistance and has prompted calls for similar infection control measures to curb their dissemination (134). Reports of resistance vary, but a general consensus appears to prevail that quinolone and broad-spectrum ß-lactam resistance is increasing in members of the family Enterobacteriaceae and Acinetobacter spp. and that treatment regimes for the eradication of Pseudomonas aeruginosa infections are becoming increasingly limited (90, 106). For example, a 5-year longitudinal study involving many centers from Latin America indicated that year after year, P. aeruginosa resistance has continually risen to the point where approximately 40% are resistant to "antipseudomonal" drugs, including carbapenems (3). While the advent of carbapenems in the 1980s heralded a new treatment option for serious bacterial infections, carbapenem resistance can now be observed in Enterobacteriaceae and Acinetobacter spp. and is becoming commonplace in P. aeruginosa.

Gram-negative bacteria have at their disposal a plethora of resistance mechanisms that they can sequester and/or evince, eluding the actions of carbapenems and other ß-lactams. The common form of resistance is either through lack of drug penetration (i.e., outer membrane protein [OMP] mutations and efflux pumps), hyperproduction of an AmpC-type ß-lactamase, and/or carbapenem-hydrolyzing ß-lactamases. Based on molecular studies, two types of carbapenem-hydrolyzing enzymes have been described: serine enzymes possessing a serine moiety at the active site, and metallo-ß-lactamases (MBLs), requiring divalent cations, usually zinc, as metal cofactors for enzyme activity (20, 23, 24, 45).

The serine carbapenemases are invariably derivatives of class A or class D enzymes and usually mediate carbapenem resistance in Enterobacteriaceae or Acinetobacter spp. The enzymes characterized from Enterobacteriaceae include NmcA, Sme1-3, IMI-1, KPC1-3, and GES-2 (107, 133, 139, 142, 197, 200). Despite the avidity of these enzymes for carbapenems, they do not always mediate high-level resistance and not all are inhibited by clavulanic acid (108). In contrast, the oxacillinases have been characterized from Acinetobacter baumannii only and include OXA 23 to 27 (1, 16), OXA-40 (61), and OXA-48 (128). These enzymes hydrolyze carbapenems poorly but are able to confer resistance and are only partially inhibited by clavulanic acid. The class A and class D carbapenemases are encoded by genes that have been procured by the bacterium and can be chromosomally encoded (sometime associated with integrons) or carried on plasmids (108).

MBLs, like all ß-lactamases, can be divided into those that are normally chromosomally mediated and those that are encoded by transferable genes. The early studies on chromosomally mediated MBLs mainly centered around Bacillus cereus (BC II) (83), and Stenotrophomonas maltophilia (L1) (178). However, primarily due to genomic sequencing, increasingly more chromosomally mediated genes are being discovered but are often found in obscure nonclinical bacteria (89, 103, 146, 149, 161).

Over the last decade there have been several articles summarizing the levels of MBLs in the bacterial community (20, 22, 23, 85, 86, 108, 117). However, in the past 3 to 4 years many new transferable types of MBLs have been studied and appear to have rapidly spread. In some countries, P. aeruginosa possessing MBLs constitute nearly 20% of all nosocomial isolates, whereas in other countries the number is still comparatively small (46, 80). In recent years MBL genes have spread from P. aeruginosa to Enterobacteriaceae, and a clinical scenario appears to be developing that could simulate the global spread of extended-spectrum ß-lactamases. Moreover, given that MBLs will hydrolyze virtually all classes of ß-lactams and that we are several years away from the implementation of a therapeutic inhibitor, their continued spread would be a clinical catastrophe. This review will focus on the biochemical and genetic characterization of MBLs and in particular transferable MBLs from Pseudomonas spp., Acinetobacter spp., and Enterobacteriaceae, including their epidemiology and methods for detecting them, and review those MBL experimental inhibitors studied thus far.

CLASSIFICATION OF MBLs

Top Previous Next References

 
MBLs were first formally categorized from serine ß-lactamases in 1980 in the classification scheme proposed by Ambler (2). At the time, very few MBLs had been sufficiently studied to merit inclusion, the most notable exceptions being L1 (from S. maltophilia) and BCII (from B. cereus). In 1989, Bush further classified MBLs into a separate group (group 3) according to their functional properties and remains the recommended referencing system for ß-lactamases generally (21). This scheme was primarily based on substrate profiles (in particular imipenem hydrolysis), their sensitivity to EDTA, and their lack of inhibition by serine ß-lactamase inhibitors. This scheme was updated in 1995 and further modified in 1997 to accommodate the growing number of group 3 enzymes continually being classified (24, 141). At the time, only two transferable types of MBLs had been studied, Bacteroides fragilis CcrA and IMP-1 from P. aeruginosa.

All MBLs hydrolyze imipenem, but their ability to achieve this varies considerably and the rate of hydrolysis may or may not correlate with the bacterium's level of resistance to carbapenems (92). Accordingly, this classification further segregated these enzymes into subgroups primarily on the basis of imipenem and other ß-lactam hydrolysis (141). Essentially, group 3a enzymes possess a broad spectrum of activity; group 3b enzymes possess a preferential avidity for carbapenems; and group 3c enzymes hydrolyze carbapenems poorly compared to other ß-lactam substrates. However, these enzymes possess the characteristic hallmark of being universally inhibited by EDTA as well as other chelating agents of divalent cations, a quintessential feature of MBLs that correlates with their mechanistic function (141).

At a molecular level, the MBLs are a disparate group of proteins that make classifying and standardizing their structures virtually impossible (Fig. 1). Attempts have been made to subdivide class B enzymes based on sequence identity and other structural features (141). The phylogenetic tree (Fig. 1) indicates the relatedness of one enzyme to another as judged by nucleotide sequences. The rationale of class B1 is that the enzymes possess the key zinc coordinating residues of three histidines and one cysteine and accommodates the transferable MBLs IMP, VIM, GIM, and SPM-1. Class B2 include those that possess an asparagine instead of the histidine at the first position of the principal zinc-binding motif, NXHXD, and derive from Aeromonas spp. and the Serratia fonticola enzyme SFH-1. MBL L1 is the sole occupant of the class B3 enzymes, as it is singularly unique among all ß-lactamases in being functionally represented as a tetramer (140).

View larger version (11K): [in this window] [in a new window]   FIG. 1. Phylogeny of chromosomally encoded MBLs. Represented sequences of various MBLs were obtained from GenBank. IMP-1 and IMP-12 (most divergent from IMP-1) and VIM-1 and VIM-7 (most divergent from VIM-1) were also added for comparison. Signal peptides were removed prior to alignment. Sequences were aligned and phylogeny trees were constructed with Clustal W (PAM250 matrix; DNA Star) using the neighbor joining method.

CHROMOSOMALLY ENCODED MBLs

Top Previous Next References

 
Some bacteria, usually from environmental habitats, ubiquitously carry MBLs, although there is much debate as to why this is the case. One argument is that over a considerable period of time, the bacteria have been exposed to ß-lactams or ß-lactam-type compounds and the bacteria have been conscripted to acquire and maintain these genes and their products. Another argument is that these enzymes perform a normal cellular function that is yet to be fully elucidated. Regardless of the viewpoint, a number of these MBL genes are inducible, and the majority of the bacteria carrying them are or can become highly resistant to ß-lactams. Fortunately, these organisms are opportunistic pathogens, and with the arguable exception of S. maltophilia and Bacillus anthracis, seldom cause serious infections. Chromosomally encoding MBL bacteria include B. cereus (BC II) (83), Bacillus anthracis (30), S. maltophilia (L1) (178), Aeromonas hydrophilia (CphA) (92), Chryseobacterium meningosepticum (BlaB or GOB-1) (147), Chryseobacterium indologenes (IND-1) (13), Legionella gormannii (FEZ-1) (15), Caulobacter crescentus (Mbl1B) (161), Myroides spp. (TUS-1, MUS-1) (89), and Janthinobacterium lividium (THIN-B) (146) Flavobacterium johnsoniae (JOHN-1) (103) and S. fonticola (SFH-1) (149) are listed in Table 1.

One group of MBL genes that are often described as chromosomal but are in fact transferable are those associated with Bacteroides spp. Compared to other anaerobic bacteria, B. fragilis is relatively resistant to ß-lactams, not least due to the potential production of its MBL, designated CfiA or sometimes CcrA (143, 164). The cfiA MBL gene was first genetically characterized in 1990 and is one of the most intensely studied with respect to its mechanism of catalysis and structure-function properties, often providing a paradigm for similar enzymes (34, 37, 183, 185). cfiA is often quiescent and requires a surrogate sequence to provide an adequate promoter, thereby potentiating expression of the structural gene. Insertion elements, such as IS942, IS1186, and IS4351, have been shown to embed immediately upstream of the ribosome-binding site, providing enhanced transcriptional capabilities for the cfiA gene (125, 126, 143). Studies have shown that in most countries, the silent cfiA gene is present in B. fragilis at approximately 2 to 4% of strains (125, 191).

GENETIC APPARATUS OF TRANSFERABLE MBLs

Top Previous Next References

 
The dissemination of MBL genes is thought to be driven by the regional consumption of extended-spectrum cephalosporins or carbapenems (80, 87). Most, if not all, genes encoding IMP- and VIM-type as well as GIM-1 are found as gene cassettes in class 1 integrons (29, 79, 129, 131, 156, 192), although IMP MBL genes are also found on class 3 integrons (5, 32). Integrons are capable of procuring gene cassettes via a site-specific recombination event between two DNA sites, one in the integron and one in the gene cassette. Integrons consist of three regions: the 5' conserved region, the 3'conserved region, and a variable region. The 5' region consists of the integrase gene (intl), its adjacent recombination site (attI), and a promoter, which facilitates expression of the procured gene cassettes in the variable region. The 3' conserved region often consists of a partially deleted qac gene (qacE1) fused to a sul gene and, correspondingly, confers resistance to antiseptics and sulfonamide, respectively.

Gene cassettes are small pieces of circular DNA, approximately 1 kb in size, comprising a single gene together with a recombination site termed a 59-base element (14). bla_VIM genes from some European counties have been found with a truncated 59-base element and the gene cassettes are likely to be "fused" (114, 181). In most instances, this involves the MBL gene and an aacA4 gene that encodes kanamycin, neomycin, amikacin, and streptomycin resistance. Therefore, both aminoglycosides and ß-lactams will select clinical bacteria harboring this fused gene cassette, further compromising these antibiotic regimens (181).

While gene cassettes carrying aminoglycoside and ß-lactam resistance genes can freely move from one integron to another, they cannot by themselves move from one organism to another and require the assistance of other genetic elements such as plasmids and transposons (14). The majority of MBL genes are found on plasmids usually between 120 and 180 kb; however some, such as bla_VIM-7 from the United States, are carried on a conjugative plasmid of 24 kb (166). The in vivo transfer of large native plasmids that carry MBL genes probably involves the promiscuity of the accommodating bacteria as much as the size of the plasmid or what other genes, besides the MBL gene, are carried on it. At present, there is very little information as to what other genes are carried on these plasmids or whether the functions they encode will aid or hinder their bacterial host. A recent study evaluated the possibility for an increase in virulence afforded by the bla_IMP gene using cell monolayers (4). While there was no increase in virulence, the study examined the effect of the MBL gene only and not the native plasmids normally accommodating these genes.

The genetic context of MBLs not only represents a measure of its plasticity but also enables reasonable speculation as to how transferable MBL genes spread internationally. In 2003, the first account of an MBL gene (bla_IMP-13) and its integron were reported to be embedded in a Tn5051-type transposon from an Italian P. aeruginosa isolate (165). The insertion of the elements containing bla_IMP-13 and bla_VIM-2 from a P. aeruginosa from Poland are in an identical site. Moreover, the tnpR genes of the transposon from both sites are identical, suggesting that the transposon is responsible for dissemination of the class 1 integron, which then procured the different MBL genes (165). These data are corroborated by the fact that the P. aeruginosa isolates possess different pulsed-field gel electrophoresis patterns and that there was no evidence of plasmid carriage. Some of the Japanese integron mediated bla_VIM-2 have also been speculated to be embedded in transposons (199).

Not all MBL genes are associated with integrons or transposons. The genetic context of bla_SPM-1 was shown to be unique, being adjacent to genes closely related to Salmonella enterica serovar Typhimurium and not associated with an integron or transposon (167). Interestingly, bla_SPM-1 and its surrounding genes are part of a mobile genomic pathogenicity island and found on a plasmid of approximately 180 kb. Salmonella pathogenicity islands have been associated with mobile common regions that have also been associated with other mobile elements called SXT regions. These regions can be mobilized under bacterial stress, as has been shown recently when resistant elements linked to SXT increased 300-fold when the bacterium was exposed to fluoroquinolones (9, 67). Further analysis of the Brazilian P. aeruginosa isolates demonstrated that the upstream DNA contained common regions or CR elements, in this case CR4 (130). Compared to integrons and transposons, very little is known about CR elements, particularly how they facilitate the mobilization of genes despite the fact that they have been associated with antibiotic resistance genes (115). SPM-1 genes in P. aeruginosa from Sao Paolo also contain common regions, although these are very different from the isolates from Recife, suggesting a different genetic origin (M. A. Toleman, unpublished data). At present, the genetic region immediately surrounding bla_SPM-1 is not associated with coresistance to other antibiotics.

BIOCHEMISTRY OF MBLs

Top Previous Next References

 
MBLs and serine ß-lactamase both mediate resistance to ß-lactams by cleaving the amide bond of the ß-lactam ring; however, the way in which the two groups of enzymes achieve this differs considerably (45). MBLs possess a distinct set of amino acids that define the finite architecture of the active site which coordinates the zinc ions. The zinc ions in turn usually coordinate two water molecules necessary for hydrolysis (186). The principal zinc-binding motif is histidine-X- histidine-X-aspartic acid (HXHXD), which is common to most MBLs apart from the class B2 enzymes (141). Without exception, the preferred metal is zinc, and while most MBLs accommodate two zinc ions in their active site, the class B2 enzymes possess just a single zinc ion (141). The proposed mechanism of hydrolysis suggests that the active site orients and polarizes the ß-lactam bond to facilitate nucleophilic attack by zinc-bound water/hydroxides (94, 162, 184).

The MBL mechanism of hydrolysis is complex and varies from one MBL to another (162). Elucidation of the crystal structure of MBLs has offered invaluable insights into their catalytic mechanisms. Despite the fact that MBLs may share less than 25% amino acid identity with one another, they all share the unique ßß fold and their active site architecture is virtually superimposable. While there are many crystal structures of MBLs to facilitate our understanding of their hydrolytic mechanism, there are only two structures of MBLs complexed to ß-lactams (J. Spencer, personal communication) (49), leaving much room for speculation about their catalytic steps (27, 28, 33, 34, 51, 52, 175). It appears that most MBLs have a loop that is flexible and this is thought to facilitate binding and hydrolysis of the ß-lactam substrates.

Unlike serine ß-lactamases, MBLs possess a wide plastic active-site groove and accordingly can accommodate most ß-lactam substrates, facilitating their very broad spectrum of activity. They are also impervious to the impeding effects of serine inhibitors such as clavulanic acid and sulbactum that are often treated as poor substrates (111, 112). Interestingly, none of the MBLs hydrolyze aztreonam particularly well, and it has been speculated that it could be considered a therapeutic MBL inhibitor (see section on inhibitors). However, in studies of animals with pneumonia caused by P. aeruginosa producing VIM-2, infection could not be eradicated with aztreonam even when the animals were given high drug doses (11).

The affinity of an enzyme for a substrate is reported as the K_m, the enzyme's ability to turn over the substrate is the k_cat, and k_cat/K_m is a measure of the enzyme's overall catalytic efficiency. Table 2 shows the K_m, k_cat and k_cat/K_m values of the MBLs GIM-1, IMP-1, VIM-1, VIM-2, and SPM-1 derived under similar experimental conditions.

Table 2 indicates that SPM-1 hydrolyzes most therapeutic ß-lactams well and is generally a more efficient enzyme (higher k_cat/K_m values) than IMP-1 (78) and GIM-1 (29), the exceptions being ampicillin, imipenem, and moxalactam for IMP-1. SPM-1 also binds cephalosporins, particularly cefoxitin, more tightly (lower K_m values) than penicillins and carbapenems. GIM-1 primarily functions as a penicillinase with moderate activity against narrow-spectrum cephalosporins and carbapenems, although it does bind most ß-lactams tightly with the notable exceptions of imipenem (K_m 287 µM), cefepime (K_m 431 µM), and moxalactam (K_m 1,035 µM).

The kinetic values demonstrated by the MBLs raise many questions about why there should be such variation in binding and hydrolysis when these enzymes are very similar. Transient kinetic studies have tried to probe the catalytic mechanism of MBLs in binding and hydrolyzing ß-lactam substrates. However, most of these studies have utilized the chromogenic substrate nitrocefin, which was shown to be atypical for some enzymes (162). Furthermore, many studies have centered on the MBLs BCII and CcrA rather than the more clinically relevant enzymes, those encoded by highly transferable genes (18, 95, 163).

TRANSFERABLE MBLs

Top Previous Next References

 

During a survey of clinical isolates of Shigella flexneri, S. marcescens, P. aeruginosa, and Alcaligenes spp. for the presence of MBLs, three minor variants of IMP-1 have been identified in Japan, IMP-3 (68), IMP-6 (198), and IMP-10 (69) (Fig. 2). IMP-3 has two amino acid changes from IMP-1 and was identified in an S. flexneri isolate. Genetic and kinetic studies determined that the substitution of glycine for serine at position 196 caused a reduction in the activity against penicillin (68). This same amino acid change is observed in IMP-6, which displays not only reduced activity against penicillin G and piperacillin but also a higher level of meropenem hydrolysis compared to imipenem, the opposite to IMP-1 (198) (Fig. 2).

View larger version (10K): [in this window] [in a new window]   FIG. 2. Phylogeny of IMP-type MBLs. Signal peptides were removed prior to alignment. Sequences were aligned and phylogeny trees were constructed with Clustal W (PAM250 matrix; DNA Star) using the neighbor-joining method.

The belief that mobile MBLs genes were solely a distant Japanese problem was negated with the advent of bla_IMP-2 in 1997 and bla_IMP-5 in 1998 from Italy and Portugal, respectively (36, 38). Subsequent reports described in detail the bla_IMP genes, their chromosomal location, and the details of their genetic environments (40, 144). IMP-2 had 36 amino acid changes relative to IMP-1, and IMP-5 had 17 amino acid changes relative to IMP-1, and both were different in terms of their genetic context, bla_IMP-2 being found next to two aminoglycoside resistance-conferring alleles (aacA4 followed by aadA1) and bla_IMP-5 being the sole gene cassette.

Subsequent to these reports, two other IMP variants have been described from Italy. IMP-12 was produced by a P. putida clinical isolate from Varesse in 2000 and its allele, bla_IMP-12, was located on a 50-kb nontransferable plasmid (43). IMP-12 is highly divergent from IMP-1, possessing 36 different amino acids and displaying poor activity against penicillin. bla_IMP-13 was cloned from a P. aeruginosa clinical specimen isolated in Rome and was 19 amino acids different from IMP-1 (165) and chromosomally encoded. The bla_IMP-13 allele has subsequently been found on a plasmid (M. A. Toleman, unpublished results) and has been associated with a minor epidemic in a hospital in the south of Italy (98).

The differences between the European IMPs and those from Southeast Asia cannot be reconciled with global dissemination of IMP alleles from Japan. It is more likely that these alleles represent local emergence. However, IMP-1 has very recently been found in England (172, 174) in isolates of Acinetobacter junii and A. baumannii. The A. junii isolate carrying bla_IMP-1 is identical in amino acid sequence, although the nucleotide sequence contains several silent changes. The A. baumanii strain was isolated from a female who had recently been on holiday in Spain and hospitalized there for 4 weeks before being transferred to Britain, and the origin of the resistant isolate and its genetic details have yet to be fully described.

Retrospective studies on resistant isolates collected as early as 1994 in Hong Kong and 1995 in Canada determined that the carbapenem resistance was due to IMP-7 (P. aeruginosa) in Canada (54) and IMP-4 (Acinetobacter spp.) in Hong Kong (31). The MBL gene bla_IMP-4 was detected in 66% of imipenem-resistant strains collected between 1994 and 1998 at the Prince of Wales hospital in Hong Kong (31). IMP-4 was 10 amino acids different from IMP-1 and 37 amino acids different from IMP-2. bla_IMP-4 was harbored on a plasmid and integron encoded along with three other resistance genes (qacG2, aacA4, and catB3) (64). Acinetobacter strains harboring IMP-4 were found in 1997 and 1998 at a prevalence of 14% of all Acinetobacter strains and disappeared in 1999, but the reasons for this is not clear (64). bla_IMP-4 was subsequently (1998 and 2000) found in Citrobacter youngae and P. aeruginosa isolates from Guangzhou in mainland China. Guangzhou is close to Hong Kong, which suggests local dissemination of this IMP allele (60). IMP-4 has now been found in Australia in Escherichia coli, Klebsiella pneumoniae, and P. aeruginosa, possibly "imported" from Southeast Asia (123, 132).

IMP-7 was responsible for a clonal P. aeruginosa outbreak in Canada at two hospitals. The integron was cloned, and bla_IMP-7 was found in the third gene cassette position with other gene cassettes encoding aminoglycoside resistance (orf1, aacC4, and aacC1) (54). Subsequently, an identical bla_IMP-7 gene was found in 1999 in a carbapenem-resistant P. aeruginosa isolate in Malaysia (63).

Further examination for MBL-positive isolates in Japan revealed that there is very little evidence of spread of these resistance alleles from Japan. In Taiwan, the IMP variant IMP-8 has been found in a single institution the National Cheung Kung Hospital. Here, the bla_IMP-8 gene cassette was found in a class 1 integron next to aacA4 in a K. pneumoniae specimen isolated in 1998 (194). IMP-8 is more similar to IMP-2 (two amino acids different) than IMP-1, suggesting no clear link with the Japanese IMP-1-producing isolates. Furthermore, hybridization studies with a bla_IMP-8 probe found bla_IMP-8 genes in 28.5% of 140 ceftazidime-resistant K. pneumoniae strains isolated from 1999 to 2000, although only 12.5% of these were carbapenem resistant (194). The bla_IMP-8 gene was also used to probe 9,082 enteric isolates, and 36 of 1,261 Enterobacter cloacae isolates were positive (193). Recently, IMP-1 and IMP-1-like MBLs have been reported from Singapore and Korea, respectively. IMP-1-producing K. pneumoniae was reported in a large tertiary-care hospital in Singapore in 2001 (74). Since then, one other bla_IMP-1-harboring isolate was detected in a Pseudomonas fluorescens isolate from 2001 (75). However, genetic analysis found that this IMP-1-producing allele was, like the bla_IMP-1 allele found in A. junii from Britain, more similar in nucleotide sequence to bla_IMP-3.

In Korea, two studies have revealed that Korea has a considerable problem with carbapenem-resistant isolates due to MBLs. One study found IMP-like MBL genes in 35% of 130 carbapenem- or ceftazidime-resistant isolates (including two P. aeruginosa) (109). Another study examined isolates from 28 hospitals located in six cities and provinces across Korea in 2000 to 2001 (80). MBLs were present in 60% of all Korean hospitals tested, and IMP-like MBLs in 41.7% of all hospitals, accounting for 28.9% of all Acinetobacter isolates. Imipenem resistance has risen in Korea from 6% of all isolates in 1996 to 19% in 2001. Unfortunately, the lack of genetic data on these strains does not give us the necessary information to say whether these IMP-like genes are actually evidence of dissemination from Japan.

The only information in the scientific literature on IMP-like MBLs in the Americas has principally come from Brazil, where there is a serious problem with isolates of multidrug-resistant Acinetobacter spp. Sequencing of a PCR product amplified with IMP-specific primers gave 100% identity with bla_IMP-6. However, the PCR product represented only partial gene sequence (47). Further recent studies of isolates collected through the SENTRY worldwide antimicrobial surveillance program have identified five further isolates from Brazil harboring IMP-1 and a new divergent IMP allele, bla_IMP-16 (Fig. 2) (97). The most recent IMP-like MBL (IMP-18) has been found in a P. aeruginosa isolate from Las Cruces, New Mexico (59).

Additionally, VIM-1 has been detected in three clonally related P. putida isolates in Italy as a source of nosocomial infections, underlining that environmental isolates are either the source or at least vectors of MBLs (87). These isolates were recovered in the same hospital in Varese, Italy. The MICs of imipenem and meropenem were >32 µg/ml, whereas that of aztreonam was 32 µg/ml. These isolates harbored a ca. 52-kb plasmid that encoded the VIM-1 determinant (87).

VIM-1 has been also detected in E. coli (154) and in several K. pneumoniae isolates in Greece (53). A similar VIM-1-positive K. pneumoniae strain has been detected very recently in France (P. Nordmann, unpublished data) that was associated with the extended-spectrum ß-lactamase SHV-5. Interestingly, the carbapenem resistance level among the enterobacterial isolates was variable. In a study performed by Scoulica et al., the MICs of imipenem and meropenem for the E. coli clinical isolates were below the proposed breakpoint definition for resistance (154).

bla_VIM-2 was first identified in southern France from a P. aeruginosa isolate in a blood culture from a neutropenic patient in 1996 (127). This isolate was resistant to most ß-lactams, including ceftazidime, cefepime, and imipenem, but remained susceptible to aztreonam. VIM-2 is closely related to VIM-1 (90% amino acid identity) and was encoded by a gene cassette, the only resistance gene identified in the bla_VIM-2-positive class 1 integron in that isolate (Fig. 3). The bla_VIM-2 gene was located on a nonconjugative ca. 45-kb plasmid. However, this plasmid was transferable by electroporation to P. aeruginosa. ß-Lactamases VIM-1 and VIM-2 have identical amino acid residues that may be involved near or in the active site of these enzymes (41). Sequence heterogeneity is mostly observed in the NH₂- and carboxy-terminal regions of VIM-1 and VIM-2.

View larger version (7K): [in this window] [in a new window]   FIG. 3. Phylogeny of VIM-type MBLs. Signal peptides were removed prior to alignment. Sequences were aligned and phylogeny trees were constructed with Clustal W (PAM250 matrix; DNA Star) using the neighbor-joining method.

In addition, a retrospective epidemiological study in the hospital in Marseilles, France, where the first VIM-2-producing P. aeruginosa strain was isolated, revealed that from 1995 to 1999, 10 other VIM-2-positive P. aeruginosa isolates were identified in patients hospitalized in different units. These isolates had indistinguishable genotypic patterns, but the class 1 integrons containing the bla_VIM-2 gene may vary in size and structure (P. Nordmann, upublished data).

Similarly, VIM-2-producing P. aeruginosa isolates were found to be the source of outbreaks in two university hospitals in Italy and Greece during the same period (77, 135). VIM-2-producing P. aeruginosa strains have also been reported from other countries, such as Japan, South Korea, Portugal, Spain, Poland, Croatia, Chile, Venezuela, Argentina, Belgium, and most recently in the United States (26, 96, 137, 152, 181, 199, 203). The U.S. "outbreak" involved four patients in an intensive care unit and typically the P. aeruginosa harboring VIM-2 was sensitive to aztreonam only (150). The VIM-2-producing P. aeruginosa isolates from the other cases were often involved in serious infections, such as septicemia and pneumonia in different patients, and they exhibited a high level of resistance to imipenem. In addition, VIM-2 has been detected in Citrobacter freundii in Taiwan, in S. marcescens in South Korea, and in Enterbacter cloacae in South Korea (70, 193, 203). In the last strain, the MICs of imipenem and meropenem were 4 µg/ml.

Recently, VIM-2 and a novel variant of the VIM series, VIM-3, have been identified in P. aeruginosa isolates in Taiwan (192). The amino acid sequence of VIM-3 differs from that of VIM-2 by two amino acid substitutions. The precise genetic environment of the bla_VIM-3 gene remains unknown, even if a chromosomal location was noticed (192).

VIM-4 was reported from a P. aeruginosa isolate from Larissa, Greece (136). This strain was recovered in 2001 from a patient who had received imipenem. This strain was resistant to all ß-lactams but kept some antibacterial activity for aztreonam (MIC of 16 µg/ml) (82). VIM-4 differs from VIM-1 by a single amino acid change (Ser175Arg) that also differs between VIM-2 and VIM-3. Interestingly, a carbapenem-resistant VIM-4-producing P. aeruginosa isolate was also identified in Sweden but from a patient that was transferred from Greece (55). Nothing is known about its genetic support. Very recently, Luzzaro et al. identified the same MBL gene in K. pneumoniae and E. cloacae isolates of a single patient hospitalized in May 2002 in Varese, Italy (88). This patient had received a carbapenem-containing therapy that likely contributed to selection of the VIM producers. The MICs of imipenem and meropenem for the K. pneumoniae isolate were 2 and 0.5 µg/ml, respectively, whereas those of the E. cloacae clinical isolate were 0.25 and 0.12 µg/ml, respectively, the latter having abnormally low carbapenem MICs. Thus, since it was demonstrated that bla_VIM-4 was encoded by the same plasmid in both isolates, it is interesting that carbapenem MIC levels may vary significantly among enterobacterial isolates despite a common resistance mechanism (88).

A survey performed on carbapenem-resistant P. aeruginosa isolates recovered from children in Warsaw, Poland, evidenced that several clonally distinguishable VIM-4-producing P. aeruginosa strains were present in that country, one of the clones now being considered an endemic strain (116). All the isolates possessed an identical class 1 integron with the aminoglycoside resistance gene cassette aacA4 at the first position and the bla_VIM-4 gene cassette at the second and last position. This allele has now been found in P. putida from the same institute (T. R. Walsh, unpublished data).

VIM-5 differs from VIM-1 by five amino acid changes (Fig. 3) (8). It has been identified in K. pneumoniae (P. Nordmann, unpublished data) and in P. aeruginosa isolates from Ankara, Turkey (8). The P. aeruginosa isolate in which bla_VIM-5 has been identified was resistant to all ß-lactams, including aztreonam. ß-Lactamase VIM-6 was identified from two P. putida isolates from Singapore (75). These isolates were highly resistant to ß-lactams, with MICs of >32 µg/ml for imipenem and meropenem, >256 µg/ml for ceftazidime, and 128 µg/ml for aztreonam. VIM-6 differs from VIM-2 by two amino acid changes (glutamine/arginine at position 59 and asparagine/serine at position 165) and from VIM-3 by only one amino acid (75).

The latest VIM-type ß-lactamase to be fully characterized is VIM-7, which has been characterized from a carbapenem-resistant P. aeruginosa isolate from Houston, Texas (166). It shares only 77% identity with VIM-1 and 74% with VIM-2 and therefore constitutes a third subgroup among the VIM-type ß-lactamases (Fig. 3). The bla_VIM-7 gene was located on a ca. 24-kb plasmid and likely to be integron borne. It was identified from a clinical isolate that was resistant to all ß-lactams, including aztreonam, and to all other available antibiotics except polymyxin B.

Whereas reports indicate that VIM-type ß-lactamases may be identified in distantly related geographical areas, several studies have been performed to evaluate the spread of such enzymes in certain areas. Although VIM-1 and -2 have been identified in several enterobacterial species, P. aeruginosa remains the most important known reservoir of these enzymes (Table 4). Thus, Lagatolla et al. evaluated the occurrence of MBL-encoding genes in P. aeruginosa isolates at Trieste University Hospital in Italy and found that 20% of P. aeruginosa isolates and 70% of the carbapenem-resistant P. aeruginosa isolates produced the VIM-1 and -2 enzymes. They demonstrated a clonal diversity among VIM-positive isolates and showed heterogeneity of the coresistance determinants. In addition, they identified VIM-positive P. aeruginosa isolates from outpatients (77).

Similarly, a retrospective survey has been performed in a tertiary-care hospital in Korea since 1995. Imipenem resistance reached 16% of all P. aeruginosa isolates, and 9% of the resistant isolates were producing VIM-2 ß-lactamase (81). These isolates were mostly clonally related, and their carbapenem resistance determinant was transferable by conjugation. The authors noted that the MIC range of ß-lactams for these VIM-2-producing isolates was quite large. For example, the MICs of aztreonam, imipenem, and meropenem all varied from 8 to 128 µg/ml and that of ceftazidime from 32 to 128 µg/ml. Thus, identification of VIM producers on the sole basis of the ß-lactam susceptibility profile remains uncertain.

Another study has been performed in Greece in which all carbapenem-resistant P. aeruginosa isolates recovered from separate patients during a 1-year period at the University Hospital of Thessaly, Larissa, were studied for MBL production. bla_VIM-like genes were detected in 47 of the 53 (88.7%) carbapenem-resistant P. aeruginosa isolates that corresponded to seven genotypes. Four genotypes possessed bla_VIM-2 and three possessed bla_VIM-4. They were carried as single gene cassettes or along with an aminoglycoside resistance gene (aacA29a) in class 1 integrons. In addition, the spread of VIM-producing-P. aeruginosa isolates in Greece was confirmed (135). In France, VIM-producing strains have recently been reported in a series of P. aeruginosa that are scattered throughout the territory (Poirel and Nordmann, unpublished results). Thus, it seems that repeated reports of VIM-producing strains in southern Europe and in Southeast Asia correspond to a true spread of these isolates rather than to a special interest of research teams located in theses areas. An extended North American survey was performed from 1999 to 2002, studying 1,111 P. aeruginosa and 236 A. baumannii strains from 23 medical centers (72) and found only a single VIM-positive isolate. This was denoted VIM-7, although it sequence is very different from that of the other VIMs and it probably arose from a different ancestral source (166).

Several sequences of new bla_VIM genes have been submitted to the EMBL database but have not been formally published (Table 4). The first of these is bla_VIM-8 which was isolated from P. aeruginosa in Colombia, adding to the growing number of bla_VIM genes discovered in South America (96). Two others, bla_VIM-9 and bal_VIM-10 (accession numbers AY534988 and AY534989, respectively), are from the United Kingdom and differ by just two amino acids. At present, there exist two bla_VIM-11 EMBL submissions isolated from P. aeruginosa in Argentina and Italy (AY605049 and AY635904, respectively). This is the second incidence of an MBL from Argentina (113).

When the sequence of SPM-1 was compared to that of other MBLs, the highest identities were seen with to IMP-1 (35.5%), ImiS (32.2%), CphA (32.1%), BCII (30.0%), and CcrA (27.0%) (167) (Fig. 1). The sequence of SPM-1 differs significantly from that of both IMP and VIM, not least due to the presence of an "insertion" of 24 amino acids just after the active site, HFHLD. This insertion has been shown to be very flexible and acts as a "loop," probably augmenting the binding and hydrolysis of ß-lactams (T. R. Walsh, unpublished data). The genetic context of bla_SPM-1 is unique in that it is immediately associated with common region elements and not with transposons or integrons (130). Interestingly, these common elements differ significantly in P. aeruginosa strains collected from different areas of Brazil even though the bla_SPM genes are identical (M. A. Toleman, unpublished results).

The ability of SPM-1 to hydrolyze various ß-lactams is summarized in Table 2. As judged by k_cat values, the preferred substrates of SPM-1 are penillins: penicillin (108), ampicillin (117), piperacillin (117), carbencillin (74), azlocillin (53), and cephalothin (43). Generally, SPM-1 binds cephalosporins more tightly than penicillins, which give relatively large K_m values (38 to 814 µM). Like IMP-1 and VIM-1, SPM-1 does not hydrolyze clavulanic acid or aztreonam particularly efficiently, which can act as competitive inhibitors (K_m of >0.1 and <0.3, respectively; Table 2) (102).

The amino acid sequence of GIM-1 displayed most identity with IMP variants IMP-6, IMP-1, and IMP-4 (43.5, 43.1, and 43.1%, respectively), with identity to VIM variants ranging from a high of 31.2% compared to VIM-7, 28.8% compared with VIM-1, VIM-4, and VIM-5, and only 28.0% similarity with SPM-1 (Fig. 1). GIM-1 possesses the major consensus features of the MBL class B1 family, such as the principal zinc-binding motif (HXHXD), and has been shown to contain two zincs at its active site (141). GIM-1 demonstrates a hydrolytic profile similar to that of IMP-1 but is arguably a weaker enzyme, as denoted by its lower k_cat and higher K_m values (Table 2).

Similar to the majority of MBL genes, bla_GIM-1 was found on a class 1 integron that is carried on relatively small plasmid of 45 kb. This integron also harbors three other resistance genes, two aminoglycoside resistance genes, aacA4 and aadA1, and a ß-lactamase gene, bla_OXA-2. Unusually, the bla_GIM-1 and aacA4 genes appeared to be accommodated in a single gene cassette that has probably been generated from individual cassettes by deletion of most of the intervening 59-base element. Gene fusions have also been seen with bla_VIM-type and aacA4 genes as well. This convoluted genetic arrangement implies that either aminoglycoside or ß-lactam therapy will select for bla_GIM-1.

EXPERIMENTAL INHIBITORS OF MBLs

Top Previous Next References

 
The introduction of amoxicillin-clavulanate in the 1980s set a paradigm for therapeutic potentiation between a ß-lactam (ampicillin) and a ß-lactamase inhibitor (clavulanic acid). Imbued by the success of amoxicillin-clavulanate, other combinations have been studied with inhibitors directed against all classes of ß-lactamases, including MBLs (100, 168). Theoretically, the panoply of ß-lactamases could be inhibited with a similar approach; however, in the case of MBLs, there are additional obstacles to circumvent (121). First, MBLs possess subtle but significant variations in their active site architecture, so that designing a single inhibitor efficacious against even the transferable MBLs will be problematic (38). Moreover, many inhibitor-screening studies have not included the more clinically important enzymes, i.e., IMP, VIM, and SPM, but used the older, better-characterized enzymes as models, which may or may not be appropriate. Second, unlike clavulanic acid, which interacts directly with class A enzymes and forms a stable covalent intermediate, MBLs do not form highly populated metastable reaction intermediates. Therefore, given their very broad spectrum of activity, attempting to inhibit MBLs with ß-lactam-like derivatives may not meet with the same success.

Third, while many studies have used an array of compounds to inhibit these enzymes at a kinetic level, few have examined the potentiation of these compounds with potent ß-lactams on P. aeruginosa containing MBL genes. Demonstrating adequate affinity of the inhibitor for the enzyme does not necessarily correlate to lower MICs in the presence of an antipseudomonal ß-lactam. Fourth, part of the success of clavulanic acid was due to the fact that there was no homologous mammalian target, i.e., relatively low toxicity. Unfortunately, MBLs have active site motifs similar to those for mammalian enzymes that are highly likely to be quintessential for cellular functions. For instance, human glyoxalase II has a similar protein fold and shares most of the key zinc binding residues of MBLs and accordingly possesses a similar active site architecture (38). Glyoxalase II is a thioesterase and is crucial for the catabolism of toxic 2-oxoaldehydes (153). Therefore, it is likely to be extremely taxing to design compounds that inhibit IMP, VIM, and SPM but do not interact with, for instance, human glyoxalase II. Comparative studies have occasionally used carboxypeptidase A (58); however, these enzymes possess only a single zinc ion and their overall fold is substantially different (189). Studies on MBL inhibitors have hitherto not included human glyoxalase II or other mammalian binuclear enzymes for toxicity screening.

A variety of structurally disparate compounds have been examined as MBL inhibitors, including thioester derivatives (44, 58, 118, 119, 121), trifluoromethyl alcohols and ketones (182), thiols (6, 18, 44, 56, 57, 71, 76, 101, 155, 159), sulfonyl hydrazones (160), tricyclic natural products (122), succinic acid derivatives (171), biphenyl tetrazoles (169, 170), cysteinyl peptides (17), mercaptocarboxylates (57, 121), 1-ß-methylcarbapenem (104, 105) cefotetan (140), thioxocephalosporins (173), and penicillin derivatives (25). The activities of these compounds against selected MBLs are summarized in Table 5.

DETECTION OF MBLs

Top Previous Next References

 
Rather like the accepted ethos for the early detection of extended-spectrum ß-lactamases, it is judicious to detect MBLs for precisely the same reasons. Unfortunately, there are no standardized phenotypic methods available and the testing criteria are likely to depend on whether the gene is carried by P. aeruginosa or a member of the Enterobacteriaceae, i.e., the evincible level of resistance. For example, most Enterobacteriaceae and some Acinetobacter spp. carrying MBL genes will appear sensitive, with imipenem MICs of between 1 and 2 µg/ml (154, 194). Therefore, the implementation of a screening plate to detect MBLs, as has been advocated for extended-spectrum ß-lactamases, must take account of the genus of the bacterium, i.e., pseudomonads intrinsically have higher carbapenem MICs than Enterobacteriaceae. It is plausible that for screening Enterobacteriaceae for the presence of MBLs, a plate could contain ceftazidime with and without EDTA, but this would only be effective if the bacterium did not also produce an extended-spectrum ß-lactamase, which cannot be assumed.

The identification of some ß-lactamases has been aided by isoelectric focusing with the aid of counterstaining the gel with the chromogenic substrate nitrocefin to determine the enzyme's isoelectric point. This technique is based on the surface charge properties of these enzymes, which are neutralized at a certain pH. For closely related enzymes e.g., TEM and SHV, the isoelectric point represents a valuable tool in the identification process. However, MBLs, even the transferable types, differ considerably from one another, and thus, isoelectric focusing is not recommended as a tool to identify them, although it can provide useful information as to the isoelectric point of unknown MBLs by using EDTA inhibition (preincubated with the enzyme prior to electrophoresis or soaking the gel with EDTA after electrophoresis) as part of the isoelectric focusing process (120).

Given the fact that all MBLs are affected by the removal of zinc from the active site, in principle, their detection should be straightforward, and studies have seized upon this principle and used a variety of inhibitor-ß-lactam combinations to detect strains possessing these clinically important enzymes (Table 6) (195). However, MBLs vary in their level of inhibition with certain compounds and also vary in their ability to confer resistance to ceftazidime or imipenem, two substrates commonly used in screening MBLs. As previously mentioned, Enterobacteriaceae carrying MBLs are often carbapenem intermediate or susceptible and can be missed when using imipenem or meropenem in the detection method. Consequently, there is no perfect inhibitor-ß-lactam combination to detect all transferable MBLs.

The problem of using EDTA in combination with imipenem is further complicated by the fact that a minority of MBL-negative P. aeruginosa produced reduced imipenem MICs in the presence of EDTA. This is due, in part, to the effect of zinc on OprD and the newly described CzcR-CzcS system (35, 124). With the advent of new and increasing numbers of MBLs, the phenotypic methods listed in Table 6 must be continually evaluated for sensitivity and specificity, particularly for detection in Enterobacteriaceae.

For clinical laboratories concerned about implementing a reasonable screening system, we suggest the following. First, target key isolates based on ceftazidime and carbapenem MIC data. For example, P. aeruginosa isolates with an imipenem MIC of 16 µg/ml may be considered appropriate candidates. For Acinetobacter spp. isolates, an imipenem MIC of 8 µg/ml, whereas for Enterobacteriaceae, an MIC of 2 µg/ml may be appropriate. For ease of application for most microbiology laboratories, the Etest MBL strip is recommended (177), where one half of the strip is impregnated with an imipenem gradient across seven dilutions and the other half with another imipenem gradient overlaid with a constant concentration of EDTA (177) (Fig. 4). However, the current strip will not detect all MBL-positive Enterobacteriaceae due to the low level of "resistance" and will need to be supplemented by the disk approximation test for some Enterobacteriaceae isolates. However, to increase the sensitivity of this technique, several substrates (imipenem, ceftazidime, and meropenem) should be used, preferably with more than one inhibitor (EDTA and mercaptopropionic acid) (6, 202). Positive isolates should be forwarded to a state or national reference laboratory where molecular techniques can verify the phenotypic observations. The excellence of a screening program is dictated by the sensitivity and specificity of the methods it employs; the screening program is also not perfect, for instance, some P. aeruginosa strains are likely to give false-positive results due to altered OprD levels and not the presence of an MBL (35).

View larger version (81K): [in this window] [in a new window]   FIG. 4. Etest MBL strip (reprinted with permission from AB BIODISK, Solna, Sweden) and an Acinetobacter sp. expressing a VIM-2 MBL. The intersection of the ellipses at the strip is read from two halves, i.e., at the section with imipenem alone (IP) and imipenem plus EDTA (IPI). A reduction in the MIC of imipenem of 3 dilutions in the presence of EDTA is interpreted as a positive test.

TREATMENT OF INFECTIONS WITH MBL-POSITIVE GRAM-NEGATIVE BACTERIA

Top Previous Next References

 
The MBL producers that are most clinically significant are primarily those where the gene encoding the enzyme is transferable and include P. aeruginosa and Acinetobacter spp. and to a lesser extent enterobacterial species. While S. maltophilia is also clinically important, its MBL carriage can be predicted, as can its ß-lactam resistance profile. Thus, these broad-spectrum ß-lactamases are mostly identified from bacterial species that already have a high degree of natural resistance to many antibiotic classes. Concerning ß-lactam resistance, these species express a cephalosporinase, have efficient efflux pumps, and have low intrinsic outer membrane permeability to many hydrophilic molecules. Thus, multidrug resistance may be easily observed in those species as a result of combined mechanisms of resistance. The unique problem with MBLs is their unrivalled broad-spectrum resistance profile. In addition, in many cases the MBL genes may be located on plasmids with genes encoding other antibiotic resistance determinants, i.e., aminoglycoside resistance genes. These MBL-positive strains are usually resistant to ß-lactams, aminoglycosides, and fluoroquinolones. However, they usually remain susceptible to polymyxins.

No extended survey with a series of human infections with MBL-positive isolates has been performed to determine the optimal treatment. Thus, suitable therapy for treating those infections remains unknown. Using an animal model of pneumonia infection with a VIM-2-positive P. aeruginosa isolate, it was shown that aztreonam at a high dose reduced the bacterial load and may be a useful