Correspondence: Victoria Korolik, Institute for Glycomics, Griffith University, PO Box 50, Gold Coast Mail Centre, Gold Coast, 9726, Australia. Tel.: +61 7 55528321; fax: +61 7 55528908; e-mail: firstname.lastname@example.org
DNA fragments encoding two putative zinc-dependent hydrolases, designated GLX2-1 and GLX2-2, from a clinical isolate of Campylobacter jejuni, strain 012, were cloned and sequenced. GLX2-1 was encoded by a sequence of 798 bp and GLX2-2 by a sequence of 597 bp. The amino acid sequences deduced from C. jejuni DNA showed 99% and 100% identity, respectively, to putative zinc hydrolases reported from C. jejuni ATCC strain 11168, and also shared identity (28–43%) with several hypothetical conserved proteins and known zinc-dependent hydrolases and metallo-β-lactamase superfamily proteins. A strictly conserved motif, -H-X-H-X-D-, characteristic of the metallo-β-lactamase superfamily of proteins, including class B metallo-β-lactamases, was identified in both proteins. Other conserved metal-binding ligands, characteristic of the metallo-β-lactamase superfamily of proteins, were also identified. Functional β-lactamase could not be expressed in either Escherichia coli or Campylobacter coli transformed with C. jejuni hydrolase-containing plasmids, suggesting that they do not function as metallo-β-lactamases, although structurally they are consistent with the zinc metallo-hydrolase family of the β-lactamase fold.
The carbapenems imipenem and meropenem have the broadest spectrum of all β-lactams (Livermore & Woodford, 2000); however, reports of carbapenemases belonging to Ambler molecular class A (penicillinases), class B (metallo-hydrolases) and class D (oxacillinases) have increased over recent years (Nordmann & Poirel, 2002). Class B enzymes are the most prevalent of carbapenems, and are unique amongst β-lactamases in having a zinc ion (or ions) at their active site (Livermore & Woodford, 2000). β-Lactamases of the other molecular classes (A, C and D) have a serine active site, and generally lack significant carbapenemase activity (Bush et al., 1995).
With the accumulation of sequence and structural data, many relatives of class B β-lactamases have been identified. They constitute a diverse protein family, whose highly diverged members are involved in different biological functions (Daiyasu et al., 2001). Four protein families that share a common structural domain, the αβ/βα fold of class B β-lactamases, have been characterized: zinc β-lactamases, glyoxalases II, A-type flavoproteins and zinc phosphodiesterase from Escherichia coli (Schilling et al., 2003; Wenzel et al., 2004). Depending on its biological function, the metallo-β-lactamase fold is capable of binding several different metals (Gomes et al., 2002; Schilling et al., 2003).
In this study, two genes encoding putative zinc hydrolases of the metallo-β-lactamase superfamily were identified in a clinical strain of C. jejuni. The genes were cloned and expressed in both E. coli and C. coli using a shuttle cloning vector pGU0202 (Alfredson & Korolik, 2003), which has been successfully employed previously to express the β-lactamase OXA-61 from C. jejuni in Campylobacter and E. coli (Alfredson & Korolik, 2005). The expression of the putative β-lactamase from the zinc hydrolase genes in C. coli and E. coli was tested against the penicillin group and the carbapenems imipenem and meropenem. The deduced amino acid sequences and structural motifs of the C. jejuni hydrolases were compared with those of other members of the zinc metallo-hydrolase family of the β-lactamase fold.
Materials and methods
Bacterial strains, plasmids and media
Escherichia coli was grown on Luria–Bertani (LB) broth-based medium (Oxoid, Unipath Ltd, Basingstoke, Hampshire, UK) or on LB agar medium (Oxoid) at 37°C. When appropriate, the medium contained kanamycin (50 μg mL−1) (Sigma Chemical Co., St Louis, MO). Campylobacter jejuni strain 012 and C. coli strain 427 were obtained from frozen stock of human clinical isolates in our culture collection, and were grown on tryptic soy agar (Oxoid) supplemented with 5% defibrinated horse blood (bioMerieux, Brisbane, Australia) and 0.5% yeast extract (Oxoid) in a microaerobic atmosphere (5% O2, 10% CO2 in N2). The cultures were incubated initially at 42°C for 36–48 h, and for 24 h for subsequent cultivation.
Plasmid DNA preparation
Plasmids from E. coli were prepared using the Nucleospin® Plasmid Kit (Macherey-Nagel, Duren, Germany), according to the manufacturer's instructions.
The β-lactam antibiotics used in this study were ampicillin (Sigma Chemical Co.), piperacillin, carbenicillin, potassium clavulanate, cephalothin, cefotaxime (ICN Biomedicals Inc., Irvine, CA), meropenem (Oxoid) and imipenem (AB Biodisk, Solna, Sweden).
Minimal inhibitory concentration (MIC) testing
MICs of the β-lactam antibiotics ampicillin (Sigma Chemical Co.), piperacillin, carbenicillin, potassium clavulanate, cephalothin, cefotaxime (ICN Biomedicals Inc.) and meropenem (Oxoid) were determined by agar dilution according to the National Committee for Clinical Laboratory Standards (NCCLS) method (National Committee for Clinical Laboratory Standards, 2003). A final inoculum of c. 104 CFU was used, and was inoculated onto the antibiotic-containing medium. The growth conditions described above for E. coli and Campylobacter spp. were used. The concentrations of the antibiotics tested were 0.5–512 mg mL−1 as described previously (Alfredson et al., 2003). Imipenem MIC testing was performed using E test strips according to the manufacturer's instructions. Potassium clavulanate was incorporated into the culture medium at a concentration of 2 μg mL−1. The control strain used was E. coli ATCC 25922.
PCR and recombinant DNA methodology
The primers MetBla-F and MetBla2-F (Table 1) were designed to target the nucleotide sequences of the conserved metallo-β-lactamase motif -H-X-H-X-D-, and the primers MetBla-R and MetBla2-R (Table 1) were designed to target the nucleotide sequences of the zinc hydrolase domain -G-H-T-X-G-. The primer pairs used were MetBla-F/MetBla-R and MetBla2-F/MetBla2-R. The primer sequences were designed using the nucleotide sequences of the proteins Cj1589 and Cj0809c, respectively, encoding putative zinc hydrolases of the metallo-β-lactamase superfamily reported in the C. jejuni ATCC 11168 genome sequence (GenBank accession no. NC_002163). PCR was performed using Invitrogen Platinum®Pfx proofreading DNA polymerase in a master mix containing 200 μM dNTPs, 1.5 mM MgSO4 and 20 pmol of primers. The cycling parameters were 94°C for 3 min, followed by 32 cycles of 94°C for 30 s, annealing at 53°C for 30 s and extension at 68°C for 20 s, and ending with incubation at 68°C for 1 min. For both reactions, PCR products were amplified from a crude lysate prepared from C. jejuni strain 012, and ligated separately into BamHI-digested pGU0202 (Alfredson & Korolik, 2003), followed by CaCl2-mediated transformation into E. coli HB101. Recombinant E. coli clones harbouring 237-bp and 225-bp inserts were directly sequenced in both directions using primers pDA2-F and pDA2-R (Table 1), and the sequences were analysed.
All oligonucleotide primers were purchased from Invitrogen Life Technologies.
† Italic nucleotides indicate BamHI restriction site (GGATCC), SpeI restriction site (ACTAGT) and SacI restriction site (GAGCTC).
To determine the precise sequences of the putative zinc hydrolases, including their respective promoter regions, inverse PCR was performed. Inverse primers for both putative zinc hydrolase genes were designed from the known sequences of the respective inserts. Primer pairs InvMet-F/InvMet-R and IM2-F/IM2-R (Table 1) were used. Inverse PCR was performed using ClaI-digested, self-ligated, C. jejuni 012 genomic DNA as a template. Pfu turbo® proofreading DNA polymerase (Stratagene, La Jolla, CA) was used in a master mix containing 250 μM dNTPs and 20 pmol of primers. The cycling parameters were 94°C for 3 min, followed by 32 cycles of 94°C for 30 s, annealing at 50°C for 30 s and extension at 72°C for 3.5 min, and ending with incubation for 5 min at 72°C. Inverse PCR products were directly sequenced in both directions, and sequences were analysed in silico to construct the entire, respective, putative C. jejuni zinc hydrolase DNA sequences.
To amplify the putative zinc hydrolase genes from C. jejuni strain 012, four primer pairs (Met1-F/Met1-R and Met1.in-F/Met1.in-R, and M2.3-F/M2-R and M2.1-F/M2-R; Table 1) were used. PCR was performed using Eppendorf Triple Master proofreading DNA polymerase in a master mix containing 200 μM dNTPs and 20 pmol of primers. The cycling parameters were 94°C for 3 min, followed by 32 cycles of 94°C for 30 s, annealing at 51°C for 30 s and extension at 72°C for 1 min, and ending with incubation for 3 min at 72°C. PCR products were amplified from a crude lysate prepared from C. jejuni strain 012 and ligated separately into BamHI-/SacI-digested pGU0202, or SpeI-/SacI-digested pGU0202 (Alfredson & Korolik, 2003), followed by CaCl2-mediated transformation into E. coli HB101.
DNA sequencing and sequence analysis
Sequencing of products was performed using the ABI 377 sequencer (Applied Biosystems, Foster City, CA) after dye terminator cycle sequencing (ABI Prism BigDye Terminator Sequencing Kit, Perkin-Elmer), according to the manufacturer's instructions. Both DNA strands were sequenced for each gene. The sequences were analysed using the software program chromas, version 1.56 (Technelysium Pty Ltd, Helensvale, Qld, Australia). The analysis of nucleotide and protein sequences was performed using the software program macvector, version 7.0 (Oxford Molecular, Oxford, Oxfordshire, UK). Database searches were carried out through the National Centre for Biotechnology Information (NCBI) with the blast and ORF Finder search programs (Altschul et al., 1997).
The complete nucleotide sequences of the putative zinc hydrolases have been deposited in GenBank under accession numbers AY701787 and AY674053.
Cloning of the C. jejuni 012 zinc hydrolase genes in E. coli
The putative zinc hydrolase genes, designated glx2-1 and glx2-2, from C. jejuni strain 012 were amplified using the primer pairs Met1-F/Met1-R and M2.3-F/M2-R (Table 1), respectively. The c. 1-kb amplicon encoding GLX2-1 was ligated into BamHI-/SacI-digested pGU0202, resulting in the plasmid construct pGU0404; the c. 0.8-kb amplicon encoding GLX2-2 was ligated into SpeI-/SacI-digested pGU0202, resulting in the plasmid construct pGU0405. Sequence analysis of both constructs, using primers pDA2-F and pDA2-R (Table 1), determined that the putative zinc hydrolase genes, including the promoter regions, were downstream of, and out of frame with, the proximal kanamycin resistance gene, aphA(3′)-III, encoded upstream of the multicloning site in pGU0202 (Alfredson & Korolik, 2003).
Further constructs were generated to place the putative zinc hydrolase genes in frame with the proximal aphA(3′)-III to determine whether possible translation of a fusion protein between the products of aphA(3′)-III and the putative zinc hydrolase genes was able to confer β-lactam resistance. The putative zinc hydrolase genes, glx2-1 and glx2-2, were amplified using the primer pairs Met1.in-F/Met1.in-R and M2.1-F/M2-R (Table 1) respectively, and the amplicons were ligated into SpeI-/SacI-digested pGU0202. This resulted in the plasmid constructs pGU0406, which harboured glx2-1, and pGU0407, which harboured glx2-2, respectively. Sequence analysis of both of these constructs determined that the putative zinc hydrolase genes, including the promoter regions, were located downstream of, and in the same reading frame as, the proximal kanamycin resistance gene, aphA(3′)-III. No transcription stops were detected between the ORFs.
Transformation of Campylobacter using cloned C. jejuni 012 zinc hydrolase genes
Recombinant plasmid pGU0404, pGU0405, pGU0406 and pGU0407 DNAs from E. coli were used to transform a β-lactamase-negative wild-type strain of C. coli, strain 427, which does not harbour homologues of the C. jejuni putative zinc hydrolases (data not shown) and has an ampicillin MIC of 4 μg mL−1. Campylobacter coli 427 has been shown previously to readily accept plasmid pGU0202 DNA from E. coli (Alfredson & Korolik, 2003). The analysis of plasmid DNA extracted from kanamycin-resistant transformants of C. coli 427 showed that they contained a single recombinant plasmid.
Antimicrobial susceptibility testing
β-Lactam MICs for E. coli HB101 and C. coli 427 were compared with those for E. coli HB101 and C. coli 427 harbouring recombinant plasmids pGU0404, pGU0405, pGU0406 and pGU0407. No difference in MIC to the penicillin group, cephalosporins or carbapenems tested could be demonstrated between E. coli HB101 and C. coli 427 cells harbouring the recombinant plasmids and the respective host strains alone.
Sequence analysis of the C. jejuni zinc hydrolases, and their deduced amino acid sequences
The 1.0-kb and 0.8-kb DNA inserts, encoding glx2-1 and glx2-2, respectively, were sequenced. Analysis of the pGU0404 insert harbouring glx2-1 revealed the presence of an ORF of 798 bp, which encoded a putative 265-amino acid protein. Analysis of the pGU0405 insert harbouring glx2-2 revealed the presence of an ORF of 597 bp, which encoded a putative 198-amino acid protein. The -H116-X-H118-X-D120-H121- motif, which is conserved in class B, subclass B3, metallo-β-lactamases and glyoxalases II, was found within both proteins. Other conserved residues acting as metal-binding ligands, consistent with zinc hydrolases of the metallo-β-lactamase fold, were also found within the proteins: histidine 196 (His196), His263 and aspartic acid 221 (Asp221, Fig. 1). The putative start codon (ATG) was found at position 136 in the pGU0404 insert harbouring glx2-1, and at position 184 in the pGU0405 insert harbouring glx2-2. glx2-1 and glx2-2 start codons were preceded by the ribosome-binding sites AAGGA and AAAGA, respectively. Putative −10 and −35 promoter sequences were also identified according to previously reported C. jejuni consensus promoter sequences (Wosten et al., 1998). The overall GC contents of glx2-1 and glx2-2 were 27.1% and 30.2%, respectively, which is similar to the overall GC content (30.6%) of the C. jejuni ATCC 11168 chromosome (Parkhill et al., 2000).
Sequence homology with other zinc hydrolases of the metallo-β-lactamase fold
GenBank database blast analysis of C. jejuni 012 glx2-1 and glx2-2 showed 99% and 100% identity to sequences encoding a hypothetical protein (Cj1589) and a probable hydrolase (Cj0809c), respectively, reported from C. jejuni ATCC 11168 (GenBank accession no. NC_002163). The putative C. jejuni hydrolase GLX2-1 showed the highest identity of 34% to a metallo-hydrolase from Erwinia chrysanthemi (Genbank accession no. CAC83617), and shared 31% identity to a zinc-dependent hydrolase from Burkholderia cepacia (Genbank accession no. ZP_00223547). The putative C. jejuni hydrolase GLX2-2 showed the highest identity of 43% to a predicted hydrolase from Helicobacter hepaticus ATCC 51449 (Genbank accession no. AAP77097), and shared 38% identity to a putative zinc-dependent hydrolase of the metallo-β-lactamase superfamily reported from Corynebacterium glutamicum ATCC 13032 (Genbank accession no. CAF20032), and 34% identity to a glyoxalase II protein reported from Bacillus cereus ATCC 14579 (Genbank accession no. AAP11173). The homology of the C. jejuni putative hydrolases to other metallo-β-lactamase superfamily proteins varied between 28% and 35%. An alignment in the region of the metal-binding ligands was constructed to relate the C. jejuni zinc hydrolase to representatives of the class B β-lactamases and the glyoxalase II family of proteins (Fig. 1).
Within the metallo-β-lactamase superfamily of proteins, members are characterized by the same folding pattern and conserved sequence motifs; however, they are involved in different biological functions (Daiyasu et al., 2001). Enzymes with β-lactamase activity exclusively bind zinc and require one or two zinc ions for enzymatic catalysis, whereas glyoxalase II enzymes are capable of binding zinc, iron and manganese (Schilling et al., 2003). The class B metallo-β-lactamases can be divided into three subclasses (B1–B3 Rasmussen & Bush, 1997), with the subclass members related only at a structural level (Galleni et al., 2001; Hall et al., 2003). Class B β-lactamases are generally chromosomally encoded and widespread (Livermore & Woodford, 2000), and are clinically important because they are capable of hydrolysing carbapenems, which are often the drugs of choice for multidrug-resistant organisms. To date, no carbapenem resistance has been reported in Campylobacter. Glyoxalase II enzymes have an N-terminal region of about 200 amino acids containing a tertiary structure similar to that of the β-lactamases, but, in the glyoxalase system, which represents the main pathway for the removal of cytotoxic methylglyoxal from cells (Melino et al., 1998), the catalytic activity involves the hydrolysis of the thioester of S-d-lactoglutathione to produce glutathione and d-lactic acid (Daiyasu et al., 2001). The metal-binding ligands of glyoxalase II most closely resemble those of the metallo-β-lactamase L1 from Stenotrophomonas maltophilia (Ullah et al., 1998), although the two enzymes exhibit different metal-binding preferences.
In this study, we cloned two putative zinc-dependent hydrolases from a clinical isolate of C. jejuni. The deduced amino acid sequences of the genes showed varying degrees of identity (28–43%) with members of the metallo-β-lactamase superfamily. The C. jejuni hydrolases contained the conserved sequence motif -H-X-H-X-D-H-, characteristic of class B, subclass B3, metallo-β-lactamases and glyoxalase II enzymes. Other metal-binding ligands consistent with zinc hydrolases were also found within the proteins. Putative β-lactamase, however, could not be expressed in either E. coli or C. coli harbouring the C. jejuni hydrolase-carrying plasmids pGU0404, pGU0405, pGU0406 and pGU0407. Putative promoter sequences and the ribosome-binding site were present upstream of the zinc hydrolase start codon in all plasmids (pGU0404, pGU0405, pGU0406 and pGU0407), allowing transcription of the gene in C. coli. An increase in MIC, however, could not be demonstrated against the β-lactams tested, suggesting that, in C. coli, the biological function of this protein is not a β-lactamase.
Structurally, the C. jejuni hydrolases appear to be single domain proteins whose predicted architecture is consistent with the zinc-dependent hydrolases of the metallo-β-lactamase superfamily, in particular glyoxalase II. All species of glyoxalase II, including human, yeast and Arabidopsis thaliana, contain a highly conserved metal-binding domain (-H-X-H-X-D-H-) that is also present in the family of metallo-β-lactamases, which are known to require zinc(II) (Zang et al., 2001). A sequence comparison of the C. jejuni putative hydrolases with the metal-binding regions of two glyoxalase II isozymes from A. thaliana (GLX2-1 and GLX2-2), human glyoxalase II and glyoxalase II from Saccharomyces cerevisiae, and representatives of the class B metallo-β-lactamases, revealed a significant homology between these enzymes (Fig. 1). All of the ligands to the first zinc(II) site in the class B metallo-β-lactamases (His116, His118, His196 and Asp120) were strictly conserved in both of the C. jejuni hydrolases and the glyoxalase II enzymes. The second zinc(II) site in the metallo-β-lactamases contained His263, and, in the subclass B1 and B2 metallo-β-lactamases, cysteine 221 (Cys221); in Stenotrophomonas maltophilia L1 metallo-β-lactamase, Cys221 was replaced by serine 221 (Ser221) (Gomes et al., 2002). The C. jejuni hydrolases and the glyoxalase II family contain the His263 ligand in the second zinc(II) site but, in the position equivalent to that in which Cys or Ser is found in metallo-β-lactamases, they both contain Asp (Asp221). Of the residues at this site (Cys, Ser and Asp), Asp is typical of iron-ligating proteins, but not of the zinc β-lactamases (Gomes et al., 2002). The relative spacing between the metal-binding ligands of the C. jejuni hydrolases is also equivalent to that of the metallo-β-lactamases and glyoxalase II enzymes (Fig. 1). These observations suggest that both of the C. jejuni hydrolases possibly have the ability to bind both iron and zinc, which is not seen in the metallo-β-lactamases, and is consistent with glyoxalase II enzymes. The exact biological function of these putative C. jejuni proteins remains to be determined.