Logo of aacPermissionsJournals.ASM.orgJournalAAC ArticleJournal InfoAuthorsReviewers
Antimicrob Agents Chemother. 2012 Jul; 56(7): 3481–3491.
PMCID: PMC3393435

Systematic Analysis of Metallo-β-Lactamases Using an Automated Database


Metallo-β-lactamases (MBLs) are enzymes that hydrolyze β-lactam antibiotics, resulting in bacterial resistance to these drugs. These proteins have caused concerns due to their facile transference, broad substrate spectra, and the absence of clinically useful inhibitors. To facilitate the classification, nomenclature, and analysis of MBLs, an automated database system was developed, the Metallo-β-Lactamase Engineering Database (MBLED) (http://www.mbled.uni-stuttgart.de). It contains information on MBLs retrieved from the NCBI peptide database while strictly following the nomenclature by Jacoby and Bush (http://www.lahey.org/Studies/) and the generally accepted class B β-lactamase (BBL) standard numbering scheme for MBLs. The database comprises 597 MBL protein sequences and enables systematic analyses of these sequences. A systematic analysis employing the database resulted in the generation of mutation profiles of assigned IMP- and VIM-type MBLs, the identification of five MBL protein entries from the NCBI peptide database that were inconsistent with the Jacoby and Bush nomenclature, and the identification of 15 new IMP candidates and 9 new VIM candidates. Furthermore, the database was used to identify residues with high mutation frequencies and variability (mutation hot spots) that were unexpectedly distant from the active site located in the ββ sandwich: positions 208 and 266 in the IMP family and positions 215 and 258 in the VIM family. We expect that the MBLED will be a valuable tool for systematically cataloguing and analyzing the increasing number of MBLs being reported.


β-Lactamases (EC catalyze the hydrolysis of the β-lactam ring of β-lactam antibiotics and are the main cause of resistance against these drugs, especially in Gram-negative bacteria (26). In contrast to serine β-lactamases (SBLs; enzymes of molecular β-lactamase classes A, C, and D according to Ambler [1] or functional classes 1 and 2 according to Bush et al. [6]), which employ a catalytic serine for β-lactam ring hydrolysis, metallo-β-lactamases (MBLs; molecular class B [1] or functional class 3 [6] β-lactamases) employ one or two Zn(II) ions in the catalytic mechanism (10, 14). These proteins have caused major concerns due to their efficient inactivation of most β-lactams (except monobactams and similar compounds, such as sulfactams [35]) and the absence of clinically useful inhibitors (12, 30). In addition, they are often encoded on mobile genetic elements, which facilitate their horizontal transfer (45). A better understanding of the sequence-structure-function relationship of these enzymes and how mutations may be acquired under the selective pressure of antibiotics and MBL inhibitors (once they become available) is critical for the design of novel antibacterial drugs with long-term efficacy (29). One important step toward better understanding these processes is to systematically analyze the current knowledge on MBL gene nucleotide sequences, MBL amino acid sequences, MBL structures, and MBL activities found in databases and in the scientific literature. To that end, we are developing the Metallo-β-Lactamase Engineering Database (MBLED). In addition to serving as a data repository for MBLs, the MBLED has been designed as an interactive tool for the investigation of variations and the potential evolutionary history of these enzymes. A deeper understanding of these facts is necessary to design new and efficacious β-lactam antibiotics that will withstand the action of β-lactamases. The active investigation of mutational hot spots and the prediction of possible variants that result in new resistances may allow for preemptive therapeutic approaches that lead to an improvement of public health. Here, we present the first version of the MBLED, which includes tools to retrieve, analyze, and compare amino acid sequence and structure data. It is publicly accessible at http://www.mbled.uni-stuttgart.de and updated as information on new variants becomes available.

Information on MBL amino acid sequences, their source organisms, and their designations can be found in the NCBI peptide database (2). This database is publicly accessible for submission. In many cases, protein entries show inconsistencies, such as the use of a name that has already been assigned to another protein, the use of different names for the same protein, or the disregard of mutations in the sequence that would warrant a new name. A website focusing on β-lactamases has been established (5) and is maintained by Jacoby and Bush (http://www.lahey.org/Studies/). This website allows the standardization of β-lactamase nomenclature and provides tables with detailed mutation profiles of the TEM, SHV, and OXA β-lactamase families.

IMP and VIM (30, 45), as well as the recent NDM enzymes (9, 22), are probably the clinically most significant MBLs. These enzymes are often encoded on plasmids and have been isolated from several opportunistic pathogenic bacteria, such as Serratia marcescens (34), Pseudomonas aeruginosa (23), Klebsiella pneumoniae (21), and Escherichia coli (22). In addition, many variants have been found in clinical isolates globally. As of January 2012, 33 IMP, 33 VIM, and 6 NDM variants are known and have been named or assigned (http://www.lahey.org/Studies/other.asp#table1). Not all of these variants have yet been identified through NCBI entries.

In this work, the MBLED is used to analyze the relationships between different IMP and VIM variants. This includes the creation of mutation profiles and the identification of mutation hot spots.


Development and construction of the MBLED.

The MBLED was set up using the generic data warehouse system DWARF (13) and 25 representative characterized protein sequences as seeds (Table 1). Details can be found in the supplemental material. Protein sequences with high similarity were assigned to a single homologous family, which was named according to its characteristic member. Homologous families were grouped into different subclasses based on the Ambler classification: B1, B2, and B3 (16). For BLAST hits representing protein structures, monomers were extracted from the Protein Data Bank (PDB) database (40) and deposited as structure entries in the MBLED.

Table 1
Seed sequences for database building

Identification of IMP and VIM mutation profiles.

Data from nucleotide sequence entries of assigned IMPs and VIMs provided by Jacoby and Bush (http://www.lahey.org/Studies/other.asp#table1) were used to deduce the respective protein sequence. The protein sequence was then aligned with IMP-1 or VIM-2, respectively, using ClustalW (43) to identify amino acid substitutions. In order to consistently number each residue, the established class B β-lactamase (BBL) numbering scheme (15) was utilized. The reference amino acid sequences for the homologous IMP and VIM families are those of IMP-1 and VIM-2, respectively, the members that were isolated first (34, 36) and for which crystal structures were available first (8, 46). For these reference sequences, the complete numbering scheme for each residue was adapted and integrated into an automated process for the pairwise comparison and analysis of every protein in the respective protein family. All amino acid substitutions of each IMP variant relative to IMP-1 and each VIM variant relative to VIM-2 were considered the mutation profile of that protein (Tables 2 and and3).3). Phylogenetic trees illustrating the relationship between the IMP and VIM variants were generated and are shown as Fig. 1 and and22.

Table 2
IMP mutation profiles
Table 3
VIM mutation profiles
Fig 1
Phlylogenetic tree of the IMP family. Known and named variants are labeled in the format “IMP-n” (where n represents the specific number designation). Unnamed variants are labeled by their GenInfo Identifier (GI) numbers.
Fig 2
Phylogenetic tree of the VIM family. Known and named variants are labeled in the format “VIM-n” (where n represents the specific number designation). Unnamed variants are labeled with their GenInfo Identifier (GI) numbers.

Identification and naming of IMP and VIM sequences.

If a mutation profile matched an already existing mutation profile, it was named accordingly. Sequences with known mutation profiles that were shorter than the reference sequence by more than 5 amino acid residues were considered fragments. Differences in length can arise either from insertions or deletions inside the amino acid sequence or from additional or missing N- or C-terminal residues. Insertions and deletions inside the amino acid sequence were annotated as follows: “G150−” denotes the deletion of glycine at position 150, while “-91R” denotes the insertion of arginine after position 91.

For each protein entry in the IMP and VIM families, the protein name was retrieved using the “Definition” field description available in every protein's GenBank entry. If a protein had already been reported as a certain IMP or VIM variant, its mutation profile was compared to that of the reported sequence. If the mutation profile did not match the mutation profile of that variant, this sequence was declared as being labeled incorrectly and either needs to be reassigned to an already existing variant or is a new, yet unnamed variant. If a protein had not yet been reported as an IMP or VIM variant, its mutation profile was compared to those of existing variants in order to determine whether the unnamed protein was identical or should be designated a new variant.

Analysis of mutation sites.

The location, frequency, and diversity of mutations in the IMP and VIM families were analyzed systematically. The number of sequences with amino acid substitutions at a specific position was determined as a measure of mutation frequency, and the number of different amino acids observed at each position was determined as a measure of a specific residue's variability. The latter equals the effective number of amino acid types observed at a specific position, k*, as described by Materon et al. (27), that could theoretically range from 1 (only wild-type amino acid) to 20 (any of the 20 proteinogenic amino acids). The difference is that here we report naturally observed substitutions, whereas Materon et al. allowed for all possible substitutions via codon randomization. This analysis is visualized in primary and secondary structure diagrams (Fig. 3A and and4A;4A; secondary structure elements assigned by the DSSP program [18, 20]) and three-dimensional crystal structures of the reference sequences (Fig. 3B and and4B4B).

Fig 3
Primary, secondary, and tertiary structure representations of IMP-1. (A) Primary and secondary structure of IMP-1. The amino acid sequence is displayed with the numbering according to the BBL standard numbering scheme and amino acid identity. The “mutation ...
Fig 4
Primary, secondary, and tertiary structure representations of VIM-2. (A) Primary and secondary structure of VIM-2. The amino acid sequence is displayed with the numbering according to the BBL standard numbering scheme and amino acid identity. The “mutation ...


Data content of the MBLED.

In total, 517 protein entries from the NCBI nonredundant protein database were collected, analyzed, and parsed into the MBLED. The actual version of the MBLED consists of 517 protein entries with 597 sequences classified into the three subclasses B1, B2, and B3. Within subclass B1, protein entries were grouped into 17 protein families, as suggested previously (24): IMP, CcrA, BlaB, IND, VIM, JOHN, EBR, CGB, MUS, TUS, SPM, GIM, SIM, BcII, NDM, KHM, and DIM (24). Although the sequence identity between IND-2 and CGB-1 (82.5%) is higher than that between IND-2 and IND-1 (78.6%), the properties of the CGB variants are unique, justifying separate families (17). Subclass B2 contains the two homologous families CphA, which includes ImiS, and Sfh. Subclass B3 contains nine homologous families: L1, GOB, FEZ, THIN-B, LRA, AIM, SMB, CAU, and BJP. Ninety MBL structures are also included in the MBLED for 13 out of 28 different homologous families (subclass B1: IMP, CcrA, BlaB, IND, VIM, SPM, BcII, and NDM; subclass B2: CphA and Sfh; subclass B3: L1, FEZ, and BJP). For the 90 structure entries, 154 individual chains were identified and stored separately in the database.

Web accessibility.

The MBLED is accessible at http://www.LacED.uni-stuttgart.de/classB or http://www.mbled.uni-stuttgart.de through a JavaScript-enabled World Wide Web browser. Protein tables provide information on the protein name, description, and the source organism. The GenInfo Identifier numbers (GIs) corresponding to each MBLED protein entry are also shown and linked to the NCBI nonredundant protein database. Multisequence alignments of each subclass as well as each homologous family were generated using ClustalW (43). For protein structures, all sequence entries are displayed with aligned secondary structure information. Annotation of individual residues is visualized by color coding in the alignment and by moving the cursor over the colored residues. For each family, the profile hidden Markov model (http://hmmer.janelia.org/) and the phylogenetic tree are provided.

IMP and VIM mutation profiles.

To date, 28 IMPs and 26 VIMs have been assigned and presented with the NCBI nucleotide (GenBank) accession numbers (http://www.lahey.org/Studies/other.asp#table1). The sequence alignment of each variant with the corresponding reference protein (IMP-1 or VIM-2) was analyzed to identify the variant's amino acid substitutions relative to the reference protein—i.e., its mutation profile (Tables 2 and and3).3). The number of substituted amino acid residues relative to IMP-1 varied considerably among the IMPs (Table 2), ranging from single residues (IMP-6, IMP-10, and IMP-30) to 50 residues (IMP-27). The VIM family also exhibited a broad range of variants, ranging from single amino acid residue substitutions relative to VIM-2 (VIM-8, VIM-9, VIM-10, VIM-11, VIM-15, VIM-16, VIM-17, VIM-23, VIM-24, and VIM-30) to 68 substitutions (VIM-7). This diversity was previously illustrated in the form of phylogenetic trees (9, 30). Updated phylogenetic trees, including novel IMP and VIM candidates (Fig. 1 and and2)2) demonstrate that the IMP variants are more evenly distributed over sequence space than the VIM variants. The latter are mostly found in the VIM-1 and VIM-2 clusters (30), with the exception of the isolated variants VIM-7 (44), VIM-12, which is located between the two clusters, being a VIM-1/VIM-2 hybrid (37), VIM-13 (19), and VIM-25. In contrast, in addition to the IMP-1 and IMP-2 clusters (30), several new clusters are forming around IMP-4, IMP-9, and IMP-13.

Naming of unidentified or misidentified MBLs.

Some sequence entries were found in which the protein was unidentified or misidentified but could easily be designated an already existing wild-type protein. Among the IMP-type enzymes, GI 4760643 was found to be identical to IMP-3 and GI 473726 was found to be identical to IMP-1 (see Table S1 in the supplemental material). Among the VIM-type enzymes, GI 49035769 was misidentified (see Table S2 in the supplemental material and see below). Other sequence entries were found to lack an identifier in the “Definition” section of their GenBank entries while being identified in other sections (see Tables S3 and S4 in the supplemental material). GI 8546574, which is defined as a “molecular class B beta-lactamase” is identified in the “Features” section by its coding sequence (CDS) blaIMP-2. Other entries like GI 134034959, which is simply defined as a “metallo-beta-lactamase,” can be identified by the title of the corresponding publication in the “Reference” section. While the information on the exact nature of the proteins is available, the lack of a distinct description in the “Definition” section makes it harder to access. Considering that widely used programs, such as BLAST, parse and display this information to the user, the lack of distinct descriptions requires additional research by the user.

Identification of new IMP and VIM variants.

Fifteen IMP-type protein sequences originating from microbial sources with new mutation profiles were found through the database analysis (see Table S5 in the supplemental material). For all but one entry (GI 90101507), names of existing variants had been assigned; however, they matched neither the mutation profile after which they were named nor any of the other mutation profiles of named IMPs. Closer examination of the entries revealed that seven of them were artificially generated mutants labeled as such (31, 32) and that one (GI 50897036) was only missing the N-terminal methionine, possibly a result of PCR amplification or a sequencing artifact. Additional details can be found in the supplemental material.

Six VIM entries were found by the database analysis to have names that did not match the amino acid sequence assigned to that name plus one enzyme named “VIM-2-like” (see Table S6 in the supplemental material). Details are provided in the supplemental material.

All of these enzymes that deviate from already named IMPs and VIMs are candidates for new variants, given that their clinical significance can be established and that sequencing artifacts can be excluded.

Data inconsistencies and reconciliation.

Two IMPs were annotated by names that were inconsistent with their amino acid sequences and matched existing profiles. For example, the protein entry with GI 83583501 was reported as IMP-4, but its mutation profile is an exact match with IMP-26 (see Table S1 in the supplemental material). We also noticed that the sequence of VIM-14, which was published in 2011 (28), was deposited in 2004 as VIM-11 (GenBank accession no. AY635904 and GI 49035768) (see Table S2 in the supplemental material).

The discovery of identical proteins in GenBank carrying different names and entries with missing information in the intuitively accessed “Definition” field demonstrates the need for a standardized and consistent method for submitting and publishing protein entries. The current status of some GenBank entries is bound to lead to confusion among the scientific community.

Analysis of mutation frequency and variability in IMP- and VIM-type MBLs.

The MBLED was employed to study amino acid substitutions in MBLs in more detail. Conserved residues are generally assumed to be critical for enzyme stability and/or function, while variable residues are assumed to be less critical. In order to further investigate how frequently and to what degree variable residues are mutated, we determined their mutation frequency and variability. Figure 3 illustrates these properties in the IMP family. High mutation frequencies and variability are observed in the leader sequence as well as the N and C termini of the mature protein, which is expected. In addition, mutations are spread out over the entire sequence. Mutated residues can be categorized as follows.

(i) Low mutation frequency and variability.

Residues with low mutation frequency (values of 1 or 2) and variability (value of 2) are assumed to be random events that do not significantly affect enzyme stability or function.

(ii) High mutation frequency and low variability.

Residues with high mutation frequency (double-digit values) and low variability (value of 2) are assumed to provide equal or improved fitness; otherwise, they would not be observed frequently. However, there seem to be restrictions as to the nature of amino acids tolerated, otherwise there would be higher variability.

(iii) Variability higher than 2 and almost as high as, equal to, or higher than the mutation frequency.

Residues with variability higher than 2 and almost as high as, equal to, or higher than the mutation frequency are assumed to be inessential due to their promiscuity and are located in structurally and functionally flexible positions, such as on the protein surface.

(iv) High mutation frequency (double-digit values) and high variability (values higher than 2).

Residues with high mutation frequency (double-digit values) and high variability (values higher than 2) are also assumed to be inessential like those in category iii, but there may be some structural or functional significance, or they may have been carried on in the course of evolution; otherwise they would not be observed frequently.

The first category is distributed relatively evenly throughout the protein, including β strands and α helices, and amino acid residue substitutions seem to be quite random in nature (e.g., charged to neutral, positively charged to negatively charged, polar to nonpolar, and nonpolar to nonpolar). The second category is mostly found outside or at the termini of secondary structure elements, and substitutions are often conservative (e.g., D49E, E79D, or I223V), in agreement with the assumption above. In the third category are cases in which almost every observed mutant explores a different amino acid substitution, suggesting that the amino acid identity plays a minor role. None of these are found in the center of a secondary structure element. For instance, position 67 can be a valine (IMP-1), isoleucine, phenylalanine, and alanine. While these are all nonpolar amino acids, they are different in size. Position 67 is at the base of the β-hairpin loop covering the active site with the side chain pointing toward the substrate binding site. Therefore, in the free enzyme, it is exposed to solvent and not limited by packing requirements; however, it may play an important role in substrate recognition when substrate is being bound and turned over. Interestingly, Materon and coworkers also selected valine, isoleucine, and phenylalanine at this position with cefotaxime following codon randomization (27). In addition, they found glutamine, tyrosine, and serine with this antibiotic, plus threonine, cysteine, glycine, and arginine with other antibiotics. The variety of selected variants depended on the antibiotic used for selection and was highest with cefotaxime (V, Q, Y, S, I, and F), intermediate with imipenem (Y, C, S, and T) and cephaloridine (G, R, F, and I), and lowest with ampicillin (V and I) (27). These data support that residue 67 is highly variable and that the selection of its identity depends on the antibiotic, suggesting a role in antibiotic recognition. These observations also suggest that our assumptions made for category iii above do not give a complete picture. In the case of residue 67, valine is not essential for protein function, but mutations to other residues may actually improve substrate recognition of specific antibiotics and thus play an important role in evolution. The natural variability of positions 50, 62, 77, 108, 183, 215, 235, 246, 254b, and 297 can be explained by their location on the protein surface. Position 235 is located at the edge of the substrate binding site, and mutation of glycine to serine has an effect on substrate recognition (25; unpublished results). In the fourth category, the variability is also high, suggesting that similar principles apply as for the third group, with the difference that the same mutations are realized in more enzymes, possibly because they occurred earlier in evolution. P68 is also at the base of the β-hairpin loop covering the active site, next to V67, and it can be replaced with serine or threonine. Materon et al. also found great diversity at this position when selecting after codon randomization with different antibiotics, including the serine, but not the threonine found in nature (27). Positions 78, 97, 208, 227, 252, and 266 are all exposed to solvent. According to our analysis and excluding the N and C termini, positions 208 and 266 can be considered the mutation “hot spots” in the IMP family, with mutation frequencies of 20 and 19, respectively, among 28 investigated IMP variants and amino acid variability of 5 out of the 20 proteinogenic amino acids, or 25%. Both of these positions are in the C-terminal β sheet, located on the surface at the interface of the ββ sandwich (7) (Fig. 5). Positions 218 and 262 are also in the ββ sandwich, close to the zinc ligands C221 and H263, respectively. They have been found to be important for a hydrogen bonding network that favors catalytic efficiency (31, 33). R208 is found in enzymes of the IMP-1 cluster, IMP-4, and IMP-26 (Fig. 1). It can be mutated to other positively charged residues: lysine in enzymes of the IMP-2 cluster, IMP-14, IMP-16, IMP-18, IMP-22, and IMP-27, or histidine in IMP-7. Alternatively, residue 208 can be neutral and polar: asparagine in IMP-5, IMP-9, IMP-11, IMP-12, IMP-15, IMP-21, and IMP-29 or serine in IMP-13 and IMP-33. Different types of substitutions at this position coincide with specific branches of the phylogenetic tree (Fig. 1): that is, no mutation to the original or other amino acids in those branches is observed, suggesting that this residue plays a pivotal role in evolution. V266 can be mutated to other hydrophobic residues (isoleucine and alanine), the positively charged lysine (39), or the neutral polar threonine. Again, variants with the same type of substitution are often found in the same region of the phylogenetic tree (Fig. 1).

Fig 5
Overlaid structures of IMP-1 and VIM-2 (same as in Fig. 3B and and4B4B but in a different orientation) with mutation hot spots highlighted. The backbones of IMP-1 and VIM-2 are shown as blue and gray ribbons, respectively. The zinc ions in IMP-1 ...

In the VIM family (Fig. 4), the highest mutation frequency and variability are found in the leader sequence and in the termini of the mature protein, but to a significantly lesser degree throughout the mature protein, also in comparison to the IMP family. Comparison of Fig. 4B and and3B3B indicates that residues with high mutation frequencies (shown in red) are found mostly around the substrate binding site and are missing in the N-terminal αβ domain in the lower left part of the VIM enzymes shown in Fig. 4B (apart from residue 36 close to the N terminus and residue 148 close to the C-terminal βα domain), while they are found throughout the protein in IMP enzymes (Fig. 3B). This observation requires further study but seems to imply that VIM enzymes are less variable than IMP enzymes, especially with respect to the N-terminal αβ domain. Nevertheless, all four categories of mutations as described for the IMP family are also found in the VIM family. Again, mutations in the first category are quite random in nature and position. Mutations in the second category (high mutation frequency and low variability) are absent in the N-terminal αβ domain, except for the very C-terminal residue 28, but abundant in the C-terminal βα domain. Some positions in the third category are 59 (glutamine in VIM-2 and lysine, arginine, and histidine in variants), 60 (serine in VIM-2, lysine in VIM-7, and threonine in VIM-13), and 65 (alanine in VIM-2, threonine in VIM-7, and valine in VIM-13). The combination of mutations at positions 60 and 65 in VIM-7 and VIM-13 may not be coincidental, and altered catalytic efficiencies have been observed in both enzymes (19, 38) and attributed to these residues in VIM-7 (4). These residues are in the β-hairpin loop covering the active site and may therefore be important for substrate recognition, just like residues 67 and 68 in the IMP family. However, here they are shielded from the substrate binding site by the conserved W87 (residues 59 and 60) and Y67 (residue 65). W87, which is a nonconserved F87 in the IMP family, has been shown to be important for stability and folding in VIM-2 (3). Other residues in the third category are T142, E149, R228, and T246. R228 is in a surface loop reaching into the substrate binding site and is mutated to serine in some enzymes, possibly to accommodate a bigger substrate, or leucine in VIM-24, possibly to accommodate a hydrophobic substrate. The role of this residue has been discussed previously (11). There are no mutations of the fourth category in the N-terminal αβ domain, except in residues 26, 27, and 36 close to the N terminus and residue 148 at the beginning of a loop to the C-terminal βα domain. Position 215 can be seen as a mutation hot spot (almost the highest mutation frequency and the highest variability among the VIM enzymes), analogous to position 208 in the IMP family; when overlaying the IMP-1 and VIM-2 structures, they are at almost the exact same position (Fig. 5). Residue 215 in the VIM family can be serine (enzymes of the VIM-2 cluster), asparagine (most enzymes of the VIM-1 cluster and VIM-12), arginine (VIM-7), and lysine (VIM-19, part of the VIM-1 cluster, but the closest relative to VIM-7) (Fig. 2). Again, the correspondence between types of mutations at this position with branches of the phylogenetic tree suggests a pivotal role of this residue. Other residues of the fourth category are Y224, located in a loop inside the protein, which can be mutated to histidine or leucine (also discussed in reference 11), and F258 in a β strand before the zinc ligand His263 and at the ββ sandwich interface, which can be mutated to valine (all enzymes of the VIM-1 cluster plus the isolated VIM-7 and VIM-13) or tyrosine (VIM-10). In space, this residue is not far from the mutation hot spot residue 266 in IMP-1 (7-Å distance between Cα atoms in overlaid structures) (Fig. 5). Once again, the correspondence between the substitution type at VIM position 258 and the region in the phylogenetic tree (Fig. 2) is obvious.

The discovery of mutation hot spots in the ββ sandwich distant from the active site that coincide to a certain degree in both IMP and VIM enzymes was unexpected. In addition, the fact that, once established, the different substitutions remain stable within their phylogenetic branch suggests that they have a favorable or at least no unfavorable impact on enzyme stability and activity. These substitutions require further investigation in order to fully understand their significance.

Concluding remarks.

The Metallo-β-Lactamase Engineering Database (MBLED) represents another module of the Lactamase Engineering Database (LacED) that already contains modules for the class A enzyme families TEM (41) and SHV (42) and can be found at http://www.laced.uni-stuttgart.de. The MBLED was established by gathering publicly available information on MBLs and is publicly accessible. It allows for a systematic analysis of sequence and annotation information and was used to generate mutation profiles of IMP and VIM variants to identify inconsistencies in the public databases and to find new IMP and VIM candidates. Based on these mutation profiles, mutation frequencies and variability were analyzed systematically. A major and unexpected finding was that the mutation hot spots with the highest mutation frequencies and variability were not found in or near the active site, but distant from the active site, on the protein surface, and at the interface of the two β sheets. Another interesting observation was that, although the IMP and VIM families have an equal number of variants, the IMP variants are more spread out over sequence space. Notably, the N-terminal αβ domain of VIM enzymes has undergone relatively few mutations.

We expect that the MBLED in its current version will be a valuable tool for investigating other MBL families as well as the growing number of variants in each family. In future versions, we plan to include nucleotide sequence and functional information to help us better understand the origin, evolution, epidemiology, and clinical significance of MBLs.

Supplementary Material

Supplemental material:


This research was supported by the Federal Ministry of Education and Research of Germany (VNB 04/B12 and FKZ 0315406 to J.P.) and an award from the Research Corporation for Science Advancement (ID 10493 to P.O.).

We acknowledge Florian Wagner for support with building the database and Quan Ke Thai for preliminary work on the database and the manuscript.


Published ahead of print 30 April 2012

Supplemental material for this article may be found at http://aac.asm.org/.


1. Ambler RP. 1980. The structure of β-lactamases. Philos. Trans. R. Soc. Lond. B Biol. Sci. 289:321–331 [PubMed]
2. Benson DA, Karsch-Mizrachi I, Lipman DJ, Ostell J, Sayers EW. 2011. GenBank. Nucleic Acids Res. 39:D32–D37 doi:10.1093/nar/gkq1079 [PMC free article] [PubMed]
3. Borgianni L, et al. 2010. Mutational analysis of VIM-2 reveals an essential determinant for metallo-β-lactamase stability and folding. Antimicrob. Agents Chemother. 54:3197–3204 [PMC free article] [PubMed]
4. Borra PS, et al. 2011. Structural and computational investigations of VIM-7: insights into the substrate specificity of VIM metallo-β-lactamases. J. Mol. Biol. 411:174–189 [PubMed]
5. Bush K, Jacoby G. 1997. Nomenclature of TEM β-lactamases. J. Antimicrob. Chemother. 39:1–3 [PubMed]
6. Bush K, Jacoby GA, Medeiros AA. 1995. A functional classification scheme for β-lactamases and its correlation with molecular structure. Antimicrob. Agents Chemother. 39:1211–1233 [PMC free article] [PubMed]
7. Carfi A, et al. 1995. The 3-D structure of a zinc metallo-β-lactamase from Bacillus cereus reveals a new type of protein fold. EMBO J. 14:4914–4921 [PMC free article] [PubMed]
8. Concha NO, et al. 2000. Crystal structure of the IMP-1 metallo β-lactamase from Pseudomonas aeruginosa and its complex with a mercaptocarboxylate inhibitor: binding determinants of a potent, broad-spectrum inhibitor. Biochemistry 39:4288–4298 [PubMed]
9. Cornaglia G, Giamarellou H, Rossolini GM. 2011. Metallo-β-lactamases: a last frontier for β-lactams? Lancet Infect. Dis. 11:381–393 [PubMed]
10. Crowder MW, Spencer J, Vila AJ. 2006. Metallo-β-lactamases: novel weaponry for antibiotic resistance in bacteria. Acc. Chem. Res. 39:721–728 [PubMed]
11. Docquier JD, et al. 2003. On functional and structural heterogeneity of VIM-type metallo-β-lactamases. J. Antimicrob. Chemother. 51:257–266 [PubMed]
12. Drawz SM, Bonomo RA. 2010. Three decades of β-lactamase inhibitors. Clin. Microbiol. Rev. 23:160–201 [PMC free article] [PubMed]
13. Fischer M, Thai QK, Grieb M, Pleiss J. 2006. DWARF—a data warehouse system for analyzing protein families. BMC Bioinformatics 7:495 doi:10.1186/1471-2105-7-495 [PMC free article] [PubMed]
14. Fisher JF, Meroueh SO, Mobashery S. 2005. Bacterial resistance to β-lactam antibiotics: compelling opportunism, compelling opportunity. Chem. Rev. 105:395–424 [PubMed]
15. Garau G, et al. 2004. Update of the standard numbering scheme for class B β-lactamases. Antimicrob. Agents Chemother. 48:2347–2349 [PMC free article] [PubMed]
16. Hall BG, Barlow M. 2005. Revised Ambler classification of β-lactamases. J. Antimicrob. Chemother. 55:1050–1051 [PubMed]
17. Hall BG, Salipante SJ, Barlow M. 2004. Independent origins of subgroup B1+B2 and subgroup B3 metallo-β-lactamases. J. Mol. Evol. 59:133–141 [PubMed]
18. Joosten RP, et al. 2011. A series of PDB related databases for everyday needs. Nucleic Acids Res. 39:D411–D419 doi:10.1093/nar/gkq1105 [PMC free article] [PubMed]
19. Juan C, et al. 2008. Characterization of the new metallo-β-lactamase VIM-13 and its integron-borne gene from a Pseudomonas aeruginosa clinical isolate in Spain. Antimicrob. Agents Chemother. 52:3589–3596 [PMC free article] [PubMed]
20. Kabsch W, Sander C. 1983. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22:2577–2637 [PubMed]
21. Koh TH, et al. 1999. Carbapenem-hydrolysing IMP-1 β-lactamase in Klebsiella pneumoniae from Singapore. Lancet 353:2162 doi:10.1016/S0140-6736(05)75604-X [PubMed]
22. Kumarasamy KK, et al. 2010. Emergence of a new antibiotic resistance mechanism in India, Pakistan, and the UK: a molecular, biological, and epidemiological study. Lancet Infect. Dis. 10:597–602 [PMC free article] [PubMed]
23. Laraki N, et al. 1999. Structure of In31, a blaIMP-containing Pseudomonas aeruginosa integron phyletically related to In5, which carries an unusual array of gene cassettes. Antimicrob. Agents Chemother. 43:890–901 [PMC free article] [PubMed]
24. Lee K, et al. 2005. Novel acquired metallo-β-lactamase gene, blaSIM-1, in a class 1 integron from Acinetobacter baumannii clinical isolates from Korea. Antimicrob. Agents Chemother. 49:4485–4491 [PMC free article] [PubMed]
25. Liu EM, Pegg KM, Oelschlaeger P. 2011. Biochemical characterization of the metallo-β-lactamases IMP-6 and IMP-25, a single and a double mutant of IMP-1 (C1-1222b). Abstr. 51st Intersc. Conf. Antimicrob. Agents Chemotheri. American Society for Microbiology, Chicago, IL http://www.icaac.org/
26. Llarrull LI, Testero SA, Fisher JF, Mobashery S. 2010. The future of the β-lactams. Curr. Opin. Microbiol. 13:551–557 [PubMed]
27. Materon IC, Beharry Z, Huang W, Perez C, Palzkill T. 2004. Analysis of the context dependent sequence requirements of active site residues in the metallo-β-lactamase IMP-1. J. Mol. Biol. 344:653–663 [PubMed]
28. Mazzariol A, et al. 2011. A novel VIM-type metallo-β-lactamase (VIM-14) in a Pseudomonas aeruginosa clinical isolate from a neonatal intensive care unit. Clin. Microbiol. Infect. 17:722–724 [PubMed]
29. Oelschlaeger P. 2008. Outsmarting metallo-β-lactamases by mimicking their natural evolution. J. Inorg. Biochem. 102:2043–2051 [PubMed]
30. Oelschlaeger P, Ai N, Duprez KT, Welsh WJ, Toney JH. 2010. Evolving carbapenemases: can medicinal chemists advance one step ahead of the coming storm? J. Med. Chem. 53:3013–3027 [PMC free article] [PubMed]
31. Oelschlaeger P, Mayo SL. 2005. Hydroxyl groups in the ββ sandwich of metallo-β-lactamases favor enzyme activity: a computational protein design study. J. Mol. Biol. 350:395–401 [PubMed]
32. Oelschlaeger P, Mayo SL, Pleiss J. 2005. Impact of remote mutations on metallo-β-lactamase substrate specificity: implications for the evolution of antibiotic resistance. Protein Sci. 14:765–774 [PMC free article] [PubMed]
33. Oelschlaeger P, Pleiss J. 2007. Hydroxyl groups in the ββ sandwich of metallo-β-lactamases favor enzyme activity: Tyr218 and Ser262 pull down the lid. J. Mol. Biol. 366:316–329 [PubMed]
34. Osano E, et al. 1994. Molecular characterization of an enterobacterial metallo β-lactamase found in a clinical isolate of Serratia marcescens that shows imipenem resistance. Antimicrob. Agents Chemother. 38:71–78 [PMC free article] [PubMed]
35. Page MG, Dantier C, Desarbre E. 2010. In vitro properties of BAL30072, a novel siderophore sulfactam with activity against multiresistant Gram-negative bacilli. Antimicrob. Agents Chemother. 54:2291–2302 [PMC free article] [PubMed]
36. Poirel L, et al. 2000. Characterization of VIM-2, a carbapenem-hydrolyzing metallo-β-lactamase and its plasmid- and integron-borne gene from a Pseudomonas aeruginosa clinical isolate in France. Antimicrob. Agents Chemother. 44:891–897 [PMC free article] [PubMed]
37. Pournaras S, et al. 2005. VIM-12, a novel plasmid-mediated metallo-β-lactamase from Klebsiella pneumoniae that resembles a VIM-1/VIM-2 hybrid. Antimicrob. Agents Chemother. 49:5153–5156 [PMC free article] [PubMed]
38. Samuelsen O, Castanheira M, Walsh TR, Spencer J. 2008. Kinetic characterization of VIM-7, a divergent member of the VIM metallo-β-lactamase family. Antimicrob. Agents Chemother. 52:2905–2908 [PMC free article] [PubMed]
39. Santella G, Docquier JD, Gutkind G, Rossolini GM, Radice M. 2011. Purification and biochemical characterization of IMP-13 metallo-β-lactamase. Antimicrob. Agents Chemother. 55:399–401 [PMC free article] [PubMed]
40. Schwede T, Diemand A, Guex N, Peitsch MC. 2000. Protein structure computing in the genomic era. Res. Microbiol. 151:107–112 [PubMed]
41. Thai QK, Bos F, Pleiss J. 2009. The Lactamase Engineering Database: a critical survey of TEM sequences in public databases. BMC Genomics 10:390 doi:10.1186/1471-2164-10-390 [PMC free article] [PubMed]
42. Thai QK, Pleiss J. 2010. SHV Lactamase Engineering Database: a reconciliation tool for SHV β-lactamases in public databases. BMC Genomics 11:563 doi:10.1186/1471-2164-11-563 [PMC free article] [PubMed]
43. Thompson JD, Higgins DG, Gibson TJ. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673–4680 [PMC free article] [PubMed]
44. Toleman MA, Rolston K, Jones RN, Walsh TR. 2004. blaVIM-7, an evolutionarily distinct metallo-β-lactamase gene in a Pseudomonas aeruginosa isolate from the United States. Antimicrob. Agents Chemother. 48:329–332 [PMC free article] [PubMed]
45. Walsh TR, Toleman MA, Poirel L, Nordmann P. 2005. Metallo-β-lactamases: the quiet before the storm? Clin. Microbiol. Rev. 18:306–325 [PMC free article] [PubMed]
46. Yamaguchi Y, et al. 2007. Crystallographic investigation of the inhibition mode of a VIM-2 metallo-β-lactamase from Pseudomonas aeruginosa by a mercaptocarboxylate inhibitor. J. Med. Chem. 50:6647–6653 [PubMed]

Articles from Antimicrobial Agents and Chemotherapy are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • Nucleotide
    Primary database (GenBank) nucleotide records reported in the current articles as well as Reference Sequences (RefSeqs) that include the articles as references.
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem chemical substance records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records.

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...