Introduction

Because of their broad spectrum and minimal side effects, β-lactam antibiotics, including penicillins, cephalosporins, monobactams, and carbapenems, currently account for more than 50% of the antibiotics prescribed worldwide (Livermore 1996). Penicillin was one of the earliest antibiotics to come into common clinical use, and over the past 60 years a plethora of β-lactam-resistant microorganisms has emerged (Medeiros 1997). The most common means of β-lactam resistance is the synthesis of β-lactamases, enzymes that inactivate β-lactam antibiotics by hydrolyzing the β-lactam ring (Nikaido and Normark 1987). The most frequently encountered β-lactamases are the serine-β-lactamases, which use a catalytic serine as part of the β-lactam ring hydrolysis mechanism.

A completely unrelated and less frequently encountered group of enzymes, the metallo-β-lactamases, also act by cleaving the β-lactam ring of β-lactam antibiotics but do so by an entirely different mechanism that involves a catalytic metal Zn2+ ion (Bush 1998). Metallo-β-lactamases were first identified nearly 40 years ago (Sabath and Abraham 1966) and are now known to be present in at least 20 species of bacteria (Hall et al. 2003). Metallo-β-lactamases are considered to be a particular threat because (1) they are not inactivated by clinically useful β-lactamase inhibitors and (2) they exhibit activity toward carbapenems, a class of β-lactam antibiotics that is generally not hydrolyzed by the serine β-lactamases. Metallo-β-lactamases have classically been categorized as Ambler Class B (Ambler 1980) and subdivided into three subclasses, B1, B2, and B3 (Rasmussen and Bush 1997). Although there is structural homology between subclass B1+B2 and subclass B3 metallo-β-lactamases (Galleni et al. 2001; Hall et al. 2003), there is no detectable sequence homology between members of subclass B1+B2 and members of subclass B3 (Hall et al. 2003).

We have previously presented phylogenies of the experimentally determined subclass B1+B2 and subclass B3 metallo-β-lactamases (Barlow and Hall 2003). Here we extend that study to include homologs of the experimentally determined metallo-β-lactamases, and we show that the β-lactam-hydrolyzing function has evolved twice, arising independently within the B1+B2 and the B3 subgroups.

Materials and Methods

Identification of Metallo-β-Lactamase Homologs

Metallo-β-lactamase homologs were identified by a tblastn search (Altschul et al. 1990, 1997) of the NCBI Microbial Genomes database using the following experimentally determined Subclass B1+B2 (Imp-I [gi15866617], BlaBl [gi9587056], CphA [gi38824], and VIM2 [gi7381449]) and Subclass B3 (CAU-I [gi21425614], FEZ1 [8980430], L1 [gi525299], and GOB1 [gi6164597]) protein sequences as queries. The query sequences were chosen to represent the major clades within their subgroups. Table 1 lists the accession numbers and organisms associated with those sequences. Sequences that aligned with a query over at least 69% of either sequence length and had an E-score ≤10−4 were considered to be candidates for homologous proteins. Most sequences aligned over >69% or <40% of the length of the query. A pairwise blast (Tatusova and Madden 1999) alignment was done between each candidate and the query sequence that had identified it. A candidate was confirmed as a homolog only if met the same criteria in the pairwise blast as it had in the original tblastn search. Table 1 lists all of the experimentally determined metallo-β-lactamases and homologs that were used in this study.

Table 1 Sequences, organisms, and accession numbers
Full size table

Phylogenetic Reconstructions

The Subclass B1+B2 and Subclass B3 protein sequences derived from the genes in Table 1 were aligned separately with ClustalX 1.8 (Thompson et al. 1997) using the Gonet 250 similarity matrix with a gap opening penalty of 35 and a gap extension penalty of 0.75 for the pairwise alignment stage and a gap opening penalty of 15 and a gap extension penalty of 0.3 for the multiple alignment stage.

The corresponding DNA coding sequences (see Table 1 for accession numbers) were aligned by introducing triplet gaps between codons corresponding to gaps in the aligned protein sequences by using the program CodonAlign (Hall 2001). CodonAlign for Macintosh and for PC (Windows) computers, and source code that can be compiled for other platforms, is available at no charge at http://www.rochester.edu/College/BIO/labs/HallLab/index.html. Both the protein and the DNA alignments, in Nexus format, are available from B.G.H. on request to drbh@mail.Rochester.edu.

Phylogenies were constructed by the Bayesian method (Mau and Newton 1997; Mau et al. 1999; Rannala and Yang 1996) as implemented by the program MrBayes (Huelsenbeck and Ronquist 2001). MrBayes is available at no charge from http://morphbank.ebc.uu.se/mrbayes3/. The evolutionary model was the general time reversible model (Tavaré 1986). Among-site variation in evolutionary rate was estimated separately for first, second, and third positions of sites within codons. Four chains, with a “temperature” of 0.2 for the heated chains, were run for 3,000,100 generations, sampling trees every 100 generations. The ln likelihood of the trees had converged on a constant value by generation 100,000, i.e., after saving 1000 trees. The consensus tree, with branch lengths, was calculated from the final 29,001 trees visited, well after convergence had occurred.

One of the advantages of Bayesian inference of phylogeny is that the results are easy to interpret. For example, the sum of the posterior probabilities of all trees is 1. Moreover, the posterior probability of any single clade is simply the sum of the posterior probabilities of all trees that contain that clade. The consensus trees calculated by MrBayes do not include the posterior probabilities of the clades, thus each entire set of trees was imported into PAUP* (Swofford 2000) and the same trees used by MrBayes to calculate a consensus were used to calculate a 50% majority rule consensus in PAUP* (Swofford 2000). The resulting tree shows the posterior probabilities of the clades, i.e., the percentage of time that those taxa are included in the clade.

The consensus trees calculated by MrBayes were imported into PAUP* for the purposes of displaying and printing the tree.

Cloning Metallo-β-Lactamase Homolog Genes

All genes were amplified using the Failsafe PCR System (Epicentre Technologies). Tm0681 was amplified from Thermotoga maritima genomic DNA (American Type Culture Collection, ATCC number 43589D) using primers F1 and R1 with Premix Buffer F. In order to increase the efficiency of protein synthesis, Primer Fl (5′-GGGGGGGTACCAGTATCCCCATAGAAAGGCCGATGC 3′) incorporated a mismatched base to change the wild-type F-methionine start codon (TTG) to (ATG). Primer Fl also included a KpnI site upstream of the start codon, and Primer Rl (5′-GGGGGGAGCTCATATCTATATTCGAACGATCACG-3′) included a SacI site downstream of the stop codon. SSO2519 was amplified from Sulfolobus solfataricus genomic DNA (American Type Culture Collection, ATCC number 35092D) using primers F2 and R2 with Premix Buffer C. Primer F2 (5′-GGGGGCCATGGAAGTTCAATATAAGTTCGAAA-3′) included a NcoI site that contains the start codon, and Primer R2 (5′-GGGGGGAGCTCGTCGCAACAACGTTAGAAACAATC-3′) included a SacI site downstream of the stop codon. NA1 was amplified from Novosphingobium aromaticivorans genomic DNA (American Type Culture Collection; ATCC number 700278D) using primers F3 and R3 with Premix Buffer G. Primer F3 (5′-GGGGGGGTACCTGCGAGTCCTATTAGTCGATAAGACC-3′) included a KpnI site upstream of the start codon. A TAA stop codon was included to terminate translation that initiates within the multiple cloning site of the plasmid. Primer R3 (5′- GGGGGGAGCTCCGAACACGGTAACTTTCCAGCTCA-3′) included a SacI site downstream of the stop codon. STM3737 was amplified from genomic DNA of Salmonella enterica serovar typhimurium strain LT2 using primers F4 and R4 with Premix Buffer H. Primer F4 (5′-GGGGGGGTACCGGCGCTTCAATCAGGATAACATT-3′) included a KpnI site upstream of the start codon. A TAA stop codon was included to terminate translation that initiates within the multiple cloning site of the plasmid. Primer R4 (5′-GGGGGGAGCTCGCCGCCGGTGAGTTATTTCAT-3′) included a SacI site downstream of the stop codon.

Amplicons were purified with QIAquick PCR Purification Kit (Qiagen). The SSO2519 amplicon was digested with restriction endonucleases NcoI and SacI (New England Biolabs), whereas the NA1, Tm0681, and STM3737 amplicons were digested with KpnI and SacI (New England Biolabs). Each was ligated into similarly digested plasmid pACSE2 (Barlow and Hall 2002a) and transformed into strain DH5α-E (F−φ80d1acZΔM15Δ(lacZYA–argF) U169 endA1 recA1 hsdR17(r−m+) deoR thi-1 phoA supE44 λ− gyrA96 relA1 gal-) (GIBCO). The products of the ligations were designated pSSO2519, pNAl, pTm0681, and pSTM3737, respectively. The authenticity of the sequence of each cloned gene was confirmed by sequencing using the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit (PE Applied Biosystems).

The metallo-β-lactamase homolog genes carried on the plasmid constructs are expressed from the strong pTAC promoter under the control of the plasmid-encoded lacI q gene by induction with IPTG (isopropyl-β-D-thiogalactopyranoside). Gene expression in all plasmids except pSSO2519 was achieved using IPTG at a concentration of 1 mM. Induction of plasmid pSSO2519 with 1 mM IPTG was found to prevent growth of the strain carrying it, therefore gene expression in that plasmid was induced with 125 μM IPTG.

Antibiotics and Determination of Antibiotic Resistance

Ampicillin (Sigma), cefotaxime (Sigma), cefepime (Bristol–Myers Squibb), cefoxitin (Merck), cefuroxime (Sigma), ceftazidime (Glaxo Wellcome), aztreonam (Bristol-Myers Squibb), imipenem (Merck), meropenem (Zeneca), and piperacillin (Sigma) were used in this project. Stock solutions of antibiotics were prepared in 0.1 M NaPO4 buffer, pH 7.0, filter sterilized, and stored at −80°C in single-use aliquots. All minimum inhibitory concentrations (MICs) were determined in Mueller Hinton (Difco) broth containing an appropriate concentration of IPTG (1 mM or 125 μM) according to Barlow and Hall (2002a). Disk Diffusion tests employed disks obtained from BBL that contained aztreonam (30 μg), cefotaxime (30 μg), ceftazadime (30 μg), cefuroxime (30 μg), cefepime (30 μg), imipenem (10 μg), meropenem (10 μg) or pipericillin (100 μg). Muller–Hinton agar containing 1 mM or 125 μM IPTG was spread with 2.7 × 106 cells of the strain to be tested, a single disk was placed in the center of the plate, and the plate was incubated overnight at 37°C. For each test five replicate plates were spread with DH5α-E carrying plasmid pACSE2 (control) and five plates were spread with DH5α-E carrying the gene of interest cloned into the same plasmid vector (experimental). The following day the diameters of the zones of growth inhibition were measured and the mean and standard error of zone diameters were calculated. The smaller the zone, the more resistant is the strain to the antibiotic.

Results and Discussion

Homologs of the Subclass B1+B2 and Subclass B3 metallo-β-lactamases were identified as described in Materials and Methods. No homolog was identified by both a Subclass B1+B2 and a Subclass B3 query sequence. However, the sequence SSO2519 from the archaebacterial organism Sulfolobus solfataricus is homologous to the Subclass B1+B2 metallo-β-lactamase BlaBl (E = 3 x 10−6) and is also homologous to two of the Subclass B3 homologs, AF1748 (E = 2 × 10−8) and PH1213 (E = 2 × 10−6).

It is important to distinguish between the families of metallo-β-lactamase proteins that are identified on the basis of sequence homology and the subsets of those proteins that have biologically detectable β-lactam-hydrolyzing activity. We refer to those proteins for which β-lactam hydrolysis has been experimentally demonstrated as “true” metallo-β-lactamases, and to the remainder as “homologs.”

Subclass B1+B2 Phylogeny

DNA sequences of the true Subclass B1+B2 metallo-β-lactamases and their homologs were aligned by Clustal 1.8 and CodonAlign as described in Materials and Methods.

Figure 1 shows the Bayesian phylogeny of the Subclass B1+B2 metallo-β-lactamases and their homologs. With the single exception of the SSO2519 sequence of the archaebacterium Sulfolobus solfataricus, all of the Subclass B1+B2 sequences are found in the Eubacteria. Sequence SSO2519 was therefore used as an outgroup to root the Subclass B1+B2 tree. The true metallo-β-lactamases, shown in boldface in Figure 1, form a single clade arising from the node designated “A” in Fig. 1.

Figure 1
figure 1

Bayesian phylogeny of Subclass B1+B2 metallo-β-lactamases and their homologs. Boldface indicates experimentally determined metallo-β-lactamases. Asterisks indicate plasmid-borne alleles. Lines are proportional to branch lengths. Posterior probabilities of clades were ≥80% except where indicated by a percentage value within a circle at a node. Except for the Sulfobolus solfataricus sequence that is enclosed within a dashed box, all sequences are from the Eubacteria. The true metallo-β-lactamases are descended from Node A. The branches of Clade A have been slightly thickened for clarity. The significance of Nodes labeled B and C is discussed in the text.

Full size image

Although the homologs have not been reported to have metallo-β-lactamase activity, that does not necessarily mean that the activity is absent. We cloned the closest relative of the true Subclass B1+B2 metallo-β-lactamases, sequence TM0681 from Thermotoga maritima, and the most distant relative, sequence SSO2519 from Sulfolobus solfataricus, into the expression vector pACSE2. To determine whether either protein had biologically significant metallo-β-lactamase activity, we expressed each in E. coli Kl2 strain DH5-αE and compared the minimum inhibitory concentrations of each of 10 β-lactam antibiotics (see Materials and Methods) in strains that expressed the homolog enzymes with a control strain that carried just the vector (Table 2). For neither enzyme did expression of the homolog increase the MIC. MIC determinations depend upon twofold increases in drug concentration, limiting their resolution to twofold differences. We previously found that some alleles that conferred identical MICs could be resolved by the disk diffusion test for antibiotic resistance (Barlow and Hall 2003). Disk diffusion tests measure a zone of growth inhibition when a disk containing a standard amount of drug is placed on a plate that has been spread with about 106 cells of the strain to be tested. The smaller the zone of inhibition, the more resistant is the strain. Table 3 shows that neither of the Subclass B1+B2 homologs significantly decreased the diameter of the zone of inhibition, indicating that neither homolog possesses biologically detectable metallo-β-lactamase activity. It is therefore reasonable to conclude that metallo-β-lactamase activity arose once at Node A.

Table 2 MICs for DHSα-E carrying cloned homolog genes
Full size table
Table 3 Disk diffusion tests for antibiotic sensitivity
Full size table

The lowest-branching eubacterial lineage of the Subclass B1+B2 phylogeny is actIORF5 from the Gram-positive organism Streptomyces coelicolor. The immediate descendants of the node labeled “B” are members of the Proteobacteria bacteria. The polytomy at the node labeled “C” includes branches that lead to (1) a clade of the Gram-positive Bacillus/Clostridium group, (2) a clade of the CFB group Flavobacteria, (3) a clade that includes both Proteobacteria of the γ subdivision and CFB group Bacteriodetes, and (4) another clade of Proteobacteria of the γ subdivision. While there are disagreements about the detailed branching order of the Universal Tree (Brown et al. 2001; Gupta 2001; Woese 1987), it is generally agreed that the Proteobacteria are the most recent group of the Eubacteria and that the CFB and Bacillus/Clostridium group arose before the Proteobacteria. Within the Proteobacteria, the γ subdivision is believed to be the most recent. Thermotoga is believed to be one of the most ancient of the Eubacteria. The most reasonable explanation for the topology of the Subclass B1+B2 tree is that those sequences that are closely related to the true Subclass B1+B2 metallo-β-lactamases arose in the Proteobacteria (node B), and that the true metallo-β-lactamases arose shortly after the γ subdivision diverged from the β subdivision of the Proteobacteria (node A). Shortly after that time it appears that there was a horizontal transfer into the Bacillus/Clostridium group of the Firmicutes and another transfer to the CFB Flavobactaeria group. Slightly after that, there appears to have been another horizontal transfer to the CFB Bacteroidetes group. In addition, there appears to have been a horizontal transfer into Themotoga maritima just before the true metallo-β-lactamases evolved.

The γ subgroup of the Proteobacteria diverged from the β subgroup about 995 million years ago (Barlow and Hall 2002b). If, as the phylogeny suggests, the true Subclass B1+B2 metallo-β-lactamases arose around the time that the γ diverged from the β subgroup of the Proteobacteria, it means that true metallo-β-lactamases are nearly 1 billion years old.

The majority of the Subclass B1+B2 genes are located on the chromosomes of their hosts, but there are several plasmid-borne alleles, indicated by asterisks in Fig. 1. True Subclass B1+B2 metallo-β-lactamases have been mobilized to plasmids twice, once for the IMP group and once for the VIM group; and a homolog has been mobilized once in Ralstonia solanacearum.

Subclass B3 Phylogeny

DNA sequences of the true Subclass B3 metallo-β-lactamases and their homologs were aligned by Clustal 1.8 and CodonAlign as described in Materials and Methods.

Figure 2 shows the Bayesian phylogeny of the Subclass B3 metallo-β-lactamases and their homologs. The Subclass B3 phylogeny was rooted with the archeal sequences AF1748, PH1213, and MTH1267 based upon a conceptual rooting with the SSO2519 sequence that is homologous to two of the Subclass B3 homologs, AF1748 and PH1213. The conceptual rooting was done by constructing a phylogeny of the Subclass B3 metallo-β-lactamases and their homologs plus SSO2519 and using SSO2519 as the outgroup. SSO2519 is not homologous to all of the taxa and its inclusion distorts the alignment, thus that is not a legitimate tree. In general, however, the topology of the illegitimate tree was similar to that of the legitimate tree that lacked SSO2519. In particular, with SSO2519 as the outgroup, AF1748, PH1213, and MTH1267 were the most basal taxa. That, together with the biological rationale that Archeae should be an outgroup to the Eubacteria, justifies the choice of outgroup. Indeed, even if all of the archaeal sequences are used as an outgroup the topology of the tree is unchanged from that shown.

Figure 2
figure 2

Bayesian phylogeny of Subclass B3 metallo-β-lactamases and their homologs. Boldface indicates experimentally determined metallo-β-lactamases. Asterisks indicate plasmid-borne alleles (but see text). Lines are proportional to branch lengths. Posterior probabilities of clades were ≥80% except where indicated by a percentage value within a circle at a node. The true metallo-β-lactamases are descended from Node A. The branches of Clade A have been slightly thickened for clarity.

Full size image

The true Subclass B3 metallo-β-lactamases are all members of a clade that is descended from the node labeled A in Fig. 2. However, that clade also includes four sequences that are not known to be true metallo-β-lactamases (NA1, STM3737, MS1, and EC1). Two of those sequences, NA1 and STM3737, were cloned and expressed in E. coli to determine whether they encode proteins with biologically detectable metallo-β-lactamase activity. The STM3737 sequence encoded no significant metallo-β-lactamase activity as determined from the MICs of 10 β-lactam antibiotics, but the NA1 sequence encoded an activity that increased the MIC from 0.03125 to 0.125 μg/ml for cefotaxime and from 2 to 8μg/ml for cefuroxime (Table 2). The more sensitive disk diffusion tests (Table 3) show that STM3737 increases resistance to six drugs, and that NAl increases resistance to four drugs including cefoxitin. It appears likely that true metallo-β-lactamase activity arose once in the Subclass B3 lineage, at Node A, but that the function has been at least partially inactivated in several descendant lines.

Only one clade of Subclass B3 metallo-β-lactamases is found on plasmids, those found in Stenotrophomonas maltophilia. Those are indicated as being on plasmids because they have been reported to be located on a large, 200-kilobase, plasmid (Avison et al. 2001). Those authors, however, refer to the element as “plasmid-like” and point out that there is no evidence that the element can transfer from one cell to another. They suggest that it may be more accurate to consider the element as part of a fragmented chromosome.

While homologs of the Subclass B3 metallo-β-lactamases are present in both the Archaea and the Eubacteria, true metallo-β-lactamases evolved only in the Eubacteria. The presence of the true metallo-β-lactamase, GOBI, in Chryseobacterium meningosepticum, a member of the CFB group, suggests that the function may well have evolved before the divergence of the Gram-positive from the Gram-negative Eubacteria about 2.2 billion years ago (Feng et al. 1997).

The SSO2519 sequence of the archaebacterial organism Sulfolobus solfataricus is a legitimate homolog of the Subclass B1+B2 metallo-β-lactamases and is also a homolog of the Subclass B3 Archaeal sequences AF1748 and PH1213. That is entirely consistent with the conclusion, based on protein structural considerations, that the Subclass B1+B2 and Subclass B3 metallo-β-lactamases are derived from a common ancestor (Hall et al. 2003).

Conclusions

Taken together, the Subclass B1+B2 and Subclass B3 phylogenies suggest that the metallo-β-lactamase family probably originated in the Archaea. The true β-lactamase function has clearly evolved twice, independently, in the Eubacteria. That finding supports and reinforces our earlier conclusion (Hall et al. 2003) that the Subclass B1+B2 and the Subclass B3 metallo-β-lactamases should be considered separate classes of the metallo-β-lactamases, just as Classes A, C, and D are separate classes of the serine β-lactamases.