ABSTRACT
The mechanism of high-level resistance to vancomycin in enterococci consists of the synthesis of peptidoglycan terminating ind-alanyl-d-lactate instead of the usuald-alanyl-d-alanine. This alternate cell wall biosynthesis pathway is ensured by the collective actions of three enzymes: VanH, VanA, and VanX. The origin of this resistance mechanism is unknown. We have cloned three genes encoding homologs of VanH, VanA, and VanX from two organisms which produce glycopeptide antibiotics: the A47934 producer Streptomyces toyocaensis NRRL 15009 and the vancomycin producer Amycolatopsis orientalis C329.2. The predicted amino acid sequences are highly similar to those found in VRE: 54 to 61% identity for VanH, 59 to 63% identity for VanA, and 61 to 64% identity for VanX. Furthermore, the orientations of the genes,vanH, vanA, and vanX, are identical to the orientations found in vancomycin-resistant enterococci. Southern analysis of total DNA from other glycopeptide-producing organisms,A. orientalis 18098 (chloro-eremomycin producer), A. orientalis subsp. lurida (ristocetin producer), andAmycolatopsis coloradensis subsp. labeda(teicoplanin and avoparcin producer), with a probe derived from thevanH, vanA, and vanX cluster fromA. orientalis C329.2 revealed cross-hybridizing DNA in all strains. In addition, the vanH, vanA,vanX cluster was amplified from all glycopeptide-producing organisms by PCR with degenerate primers complementary to conserved regions in VanH and VanX. Thus, this gene sequence is common to all glycopeptide producers tested. These results suggest that glycopeptide-producing organisms may have been the source of resistance genes in vancomycin-resistant enterococci.
The glycopeptide antibiotic vancomycin and the structurally related antibiotic teicoplanin are considered to be the last lines of defense against a variety of serious infections caused by gram-positive organisms, such as enterococci, methicillin-resistant Staphylococcus aureus, andClostridium difficile (11). The recent rapid emergence and spread of vancomycin-resistant enterococci (VRE) has therefore become a grave concern of both medical practitioners and their patients (22, 27).
The molecular target of glycopeptide antibiotics is thed-alanyl-d-alanine (d-Ala-d-Ala) terminus of growing peptidoglycan, the rigid polymer which protects bacterial cells from osmotic lysis. By binding to this terminal dipeptide, glycopeptide antibiotics interfere with proper cell wall formation, which results in eventual cell death (6). Glycopeptide-resistant organisms avoid such a fate by modifying the drug’s peptide target, specifically, by modifying it to the depsipeptided-alanyl-d-lactate (d-Ala-d-Lac) (40). The biosynthetic machinery required to effect this transformation in vancomycin-resistant Enterococcus faecium is found on a transposable element, Tn1546, which incorporates five genes necessary and sufficient to confer high-level inducible glycopeptide resistance (2). Two of these gene products, VanR and VanS, are required for the vancomycin-induced resistance response and are members of a two-component regulatory system directing the transcription of vanH, vanA, and vanX. These genes encode the three proteins which are necessary ford-Ala-d-Lac synthesis and are thus essential for glycopeptide resistance (1). VanA is the pivotal enzyme, producing the ester d-Ala-d-Lac instead of the usual d-Ala-d-Ala dipeptide for incorporation into peptidoglycan precursors (9). VanH is an α-keto acid reductase (d-lactate dehydrogenase [d-LDH]) that supplies VanA with substrate by converting pyruvate intod-lactate (9), while VanX is a highly specificd,d-dipeptidase which hydrolyzes thed-Ala-d-Ala generated by the endogenous cell wall biosynthetic pathway but does not recognized-Ala-d-Lac (34, 43). This elegant system is the means by which VRE can overcome current antibiotic regimens. A second variant of this mechanism and gene structure is encountered and distributed on plasmids (42), conjugative chromosomal elements (32), or transposons (e.g., Tn1547 [31]) with thed-Ala-d-Lac ligase designated VanB (18, 33).
The origin of the vanH, vanA/B, andvanX (vanHAX) resistance cassette has remained obscure. Both the intrinsically vancomycin-resistant lactic acid bacteria, such as organisms from the generaLeuconostoc, Pediococcus, andLactobacillus, and the glycopeptide-producing organisms are potential reservoirs. Alternatively, the genes may have arisen through mutations in homologous genes within enterococci or other organisms. The vancomycin-resistant lactic acid bacteria such as Leuconostoc mesenteroides have both the requisited-Ala-d-Lac ligase (15, 29) andd-LDH (12) enzymes, although thed-Ala-d-Lac ligases differ in their primary sequence from VanA/B, notably in the ω-loop region which is critical to catalysis (16). We have recently cloned a gene encoding a VanA/B homolog, designated DdlM, from the glycopeptide antibiotic-producing organism Streptomyces toyocaensisNRRL 15009 (23). The striking homology between VanA/B and DdlM (62% identity, >77% similarity) suggested that these enzymes evolved from a common ancestor. Similarly, we reported a segment of the ligase gene (ddlN) cloned from the vancomycin-producing organism Amycolatopsis orientalisC329.2 which also encoded a clear VanA/B homolog (23). Importantly, the ω-loop segment is conserved between VanA, VanB, DdlM, and DdlN. Thus, these results suggested that the VanA/B enzymes originated in glycopeptide-producing organisms but that the origins of the VanH and VanX accessory proteins were unknown. We now report on the cloning and sequencing of genes encoding VanH and VanX homologs from the same glycopeptide producers as well as the presence ofvanHAX gene clusters in several other glycopeptide-producing organisms. The strong homology of the gene products and the arrangement of the cloned genes indicate a common origin for the vancomycin resistance genes in VRE and antibiotic-producing organisms.
MATERIALS AND METHODS
Cloning of vanX from S. toyocaensis NRRL 15009.Cloning of a vanX gene from S. toyocaensis NRRL 15009 (vanXst) was achieved by the isolation of a specific gene probe by PCR amplification and degenerate deoxyoligonucleotide primers p1 and p2 designed to amplify an internal vanX fragment of approximately 360 bp (Table 1). PCR amplification reactions contained S. toyocaensis NRRL 15009 total DNA as template (500 ng), 1 U of Taq polymerase, each deoxynucleoside triphosphate (dNTP) at a concentration of 0.4 mM, 1 μM primers, and 2 mM MgCl2. The 360-bp product was cloned into the pGEM-T plasmid vector (Promega), and both strands were sequenced. To isolate a clone encompassing the entire vanXst gene, the probe fragment was randomly labeled with 32P by using the Klenow fragment (4). Labeling reactions contained 200 ng of template, 1 μl of each PCR primer, 1× Klenow buffer (10 mM Tris-Cl [pH 7.5], 5 mM MgCl2, 7.5 mM dithiothreitol), and 10 U of the Klenow fragment in a final volume of 55 μl. This probe was used to screen Southern hybridization blots of total S. toyocaensis NRRL 15009 DNA which had been digested to completion with several restriction enzymes in separate experiments. Hybridizing fragments (5.2 kb) of DNA from digestion with PstI were cloned into pBluescript II KS+ (Stratagene), transformed intoEscherichia coli DH5α, and confirmed to contain thevanXst gene by colony hybridization. A 5.2-kb fragment was found to contain the vanXst gene and was sequenced in its entirety.
Sequences of PCR primers used in this study
The vanH gene was located immediately upstream of theddlM gene, whose sequence we have reported previously (23).
Cloning of van genes from A. orientalisC329.2.Cloning of the A. orientalis C329.2vanHAX cluster was achieved essentially as described for thevanX gene of S. toyocaensis NRRL 15009. Avan gene probe was generated from A. orientalisC329.2 genomic DNA by degenerate PCR with primers p2 and p3 complementary to conserved sequences in vanA andvanX (Table 1). Each amplification reaction mixture consisted of 500 ng of total DNA from A. orientalis C329.2, 1× Taq buffer, 5% dimethyl sulfoxide, 2 mM MgCl2, each dNTP at a concentration of 0.4 mM, 1 U ofTaq polymerase, and 1 μM primers. The 1.3-kb product was cloned into pGEM-T (Promega), and this clone was used as a template to generate a 32P-radiolabeled oligonucleotide probe with the Klenow fragment. This probe was used to screen Southern blots of restriction endonuclease-digested A. orientalis C329.2 genomic DNA to identify a 3.5-kb BamHI product containingvan genes. This fragment was cloned into pGEM-7Z and was isolated by standard methods.
DNA sequencing.The nucleotide sequence of the S. toyocaensis NRRL 15009 gene cluster was determined with a series of nested deletions constructed by ExoIII-S1 nuclease digestion and through the synthesis of specific oligonucleotide primers. The gene cluster for A. orientalis C329.2 was sequenced by primer walking. DNA cycle sequencing was performed by B. Allore on an Applied Biosystems automated DNA sequencer at the MOBIX central facility at McMaster University.
Analysis of gene clusters in glycopeptide producing organisms.Total DNA was isolated from glycopeptide-producing organisms, and 10 to 20 μg was separated on a 1% TAE (Tris-acetate-EDTA)–agarose gel after digestion to completion withBamHI. The DNA was blotted onto nitrocellulose by standard methods. A vanHAX cluster oligonucleotide probe radiolabeled with 32P was generated as described above by using the cloned 3.5-kb BamHI fragment from A. orientalisC329.2 as template and primers p3 and p4 (Table 1). This probe was hybridized to genomic DNA restriction products, and the blot was washed under stringent conditions prior to exposure.
PCR amplification of vanHAX from glycopeptide-producing organisms.The presence of thevanHAX cluster in glycopeptide producers was also assessed by amplification of the cluster by PCR. Degenerate oligonucleotide primers complementary to conserved regions in VanH (primer p5) and VanX (primer p2) were designed to amplify nearly the entire cluster (Table1). The PCR mixtures (100 μl) contained 500 ng of total DNA, 1 U ofTaq polymerase, 1× Taq buffer, 5% dimethyl sulfoxide, 4 mM MgCl2, each dNTP at a concentration of 0.4 mM, and 1 μM primers. Forty cycles of 94, 55, and 72°C (1, 1, and 2 min, respectively) were required to produce a visible band of approximately 2.6 kb. This product was isolated, applied to a 1% TAE agarose gel, and blotted onto nitrocellulose. A 32P-labeledvanHAX probe was generated from A. orientalisC329.2 total DNA and primers p3 and p4 (Table 1).
Analysis of sequence data.Contiguous sequences were constructed by using the MegAlign software from DNA Star (Madison, Wis.). Open reading frames (ORFs) were screened in all six frames by using the program FRAME (8) to identify probable streptomycete ORFs. Protein sequence alignments were performed by using CLUSTAL W software (39), version 1.7, with the Blosum matrix with a gap penalty of 10 and a gap extension of 0.05.
RESULTS AND DISCUSSION
Cloning of van genes from S. toyocaensisNRRL 15009 and A. orientalis C329.2. S. toyocaensis NRRL 15009 and A. orientalis C329.2 are gram-positive microbes which produce the “aglyco” glycopeptide antibiotic A47934, a structural homolog of teicoplanin (44), and vancomycin, respectively. We cloned and sequenced an expanse ofS. toyocaensis NRRL 15009 DNA spanning 2,650 bp and a 3,470-bp fragment of DNA from A. orientalis C329.2. Each of these contains three ORFs whose products bear striking homology to VanH, VanA (23), and VanX of VRE (Fig.1). These genes have been designatedvanHst, ddlM (23), andvanXst, respectively, in S. toyocaensis and vanHaov, ddlN, and vanHaov, respectively, in A. orientalis C329.2. (This rather ponderous terminology is necessitated by the fact that various A. orientalisstrains produce different glycopeptide antibiotics; thus, the subscriptao designates A. orientalis and the subscript v indicates that it is vancomycin producer.)
Comparison of glycopeptide resistance gene clusters from VRE and glycopeptide antibiotic-producing organisms.
No genes encoding the homologs of the two-component regulatory elements (VanR and VanS) of VRE, thed,d-carboxypeptidase (VanY), or the vancomycin-resistant enterococcus-associated proteins of unknown function (VanW or VanZ) were found in the DNA flanking thevanHAX cluster in either organism, although we did not explicitly screen for these and they may be located elsewhere in the chromosome. The 5′ region immediately upstream ofvanHst contains an ORF which encodes a putatived-Ala-d-Ala-adding enzyme (MurF) homolog, which we have designated MurX (23), and part of a gene homologous to femA which is required for the synthesis of the penta-Gly bridge that joins the peptidoglycan peptide strands and is associated with high-level methicillin resistance in staphylococci (13).
A 1.6-kb ORF, designated ORF1, is located 195 bp downstream fromvanXst. This ORF has associated promoter −30 and −10 sequences and thus does not appear to be part of avanHst-ddM-vanXst polycistronic message. The ORF1 amino acid sequence shows a high degree of homology [P(N)10−170 to 10−42 by BLAST search] to other bacterial proteins of unknown function. Analysis against the PROSITE database did not reveal any distinguishing or revealing features. Further sequencing of 2 kb of DNA downstream of ORF1 did not reveal any genes encoding proteins which could obviously be ascribed to peptidoglycan or glycopeptide antibiotic biosynthesis.
No complete ORFs were detected in the regions flanking thevanHAX cluster from A. orientalis, although only approximately 800 bp of flanking DNA was sequenced.
Putative streptomycete −20 (CGGGC) and −10 (CACATA) promoter sequences (36) are located upstream ofvanHst. No additional promoter sequences were found upstream of ddlM or vanXst; thus, these three genes appear to be driven by a single promoter, as is the case with the sequences of VRE. Putative −20 (CGGGG) and −10 (CCCATA) promoter sequences were also identified in the A. orientalis C329.2 cluster. Additionally, a possible terminator sequence located between bases 3228 and 3273 was identified.
The vanHst gene terminates 10 bases within the predicted 5′ end of the ddlM gene, and similarly,vanHaov also terminates 10 bases withinddlN (Fig. 2). These observations directly parallel the arrangement ofvanHA and vanA in Tn1546, where the vanHA termination codon begins 5 bases into the vanA gene (3) and in the VanB gene cluster where the vanHB termination codon similarly begins 5 bases into the vanB gene (Fig. 2) (17).
Overlap of vanH and vanA homolog genes.
Sequence analysis of vanH and vanX from glycopeptide producing organisms.The predicted amino acid sequence of VanHst and VanHaov clearly identify them as d-LDHs. Amino acid residues which are thought to participate in catalysis based on site-directed mutagenesis studies (21, 37, 38) of d-LDHs, as well as the three-dimensional structure of the Lactobacillus pentosusd-LDH (35), are conserved in VanHstand VanHaov. For example, a conserved Arg (Arg237 in VanHst and Arg255 in VanHaov) is predicted to bind to the carboxylate and polarize the ketone carbonyl for reduction, and an invariant His (His298 in VanHst and His316 in VanHaov) is predicted to act as an active-site general acid donating a proton during pyruvate reduction. Protein sequence alignment of VanHst and VanHaov with the VanH enzymes from VRE yielded overall amino acid identities of 51 to 55%, while alignment with other d-LDHs resulted in identities ranging from 26 to 31%. Phylogenetic analysis clearly places all the VanH enzymes in a distinct branch of the d-LDH family (Fig.3), an observation which parallels the clustering of DdlM and DdlN with VanA and VanB within thed-Ala-d-Ala ligase family (Fig.4) (23).
Phylogenetic analysis of the d-Ala-d-Ala ligases indicates four major groups (Fig. 4): the A and B groups, which are bona fide d-Ala-d-Ala ligases but which also contain the d-Ala-d-Lac ligases from lactic acid bacteria; the VanC group, which ared-Ala-d-Ser ligases and which confer constitutive low-level vancomycin resistance in several organisms such as Enterococcus gallinarum and Enterococcus casseliflavus (14, 28); and the VanA group, which consists of the d-Ala-d-Lac ligases from VRE and glycopeptide-producing organisms.
Amino acid sequence alignment of VanXst and VanXaov with the homologs from VRE confirms the presence of the Zn2+ binding ligands His116, Asp123, and His184 as well as the putative active site base Glu181 (25) (Fig.5). The VanXst and VanXaov proteins show 61 to 63% identity (>75% similarity) to the homologous proteins in VRE.
Overlap of the Zn2+ binding regions of VanX proteins. The alignment was as calculated by the CLUSTAL W method (39). The Zn2+ ligands are boxed, and the putative active site base is indicated with an arrow.
Identification of vanHAX gene clusters in other glycopeptide antibiotic-producing organisms.Several other glycopeptide-producing organisms were screened for the presence of thevanHAX genes by two methods. First, a part of the gene cluster from A. orientalis C329.2 was amplified and used to probe total DNA that had been isolated from several glycopeptide producers and digested to completion withBamHI. A single hybridizing band was observed in producers of chloro-eremomycin, ristocetin, vancomycin, and teicoplanin-avoparcin (Fig. 6A), indicating that these organisms harbor similar genes. Second, a PCR screen was developed with degenerate primers complementary to conserved sequences in VanH and VanX. The presence of a vanHAX gene cluster should give rise to an amplified product of approximately 2.7 kb. By using total DNA from several glycopeptide-producing organisms, a band of the expected size was observed in all organisms tested, and the sequence represented by this band was confirmed to encode thevanHAX cluster by Southern hybridization with theA. orientalis C329.2 van cluster probe (Fig.6B). Thus, this gene cluster appears to be ubiquitously present in glycopeptide-producing bacteria.
The presence of the vanHAX gene cluster in glycopeptide-producing organisms. (A) Southern blot of completeBamHI digests of A. orientalis 18098 (chloro-eremomycin producer) (lane 1), A. orientalissubsp. lurida (ristocetin producer) (lane 2), A. coloradensis subsp. labeda (teicoplanin and avoparcin producer) (lane 3), and A. orientalis C329.2 (vancomycin producer) (lane 4). The blot was probed with a portion of the ddlN gene amplified from A. orientalisC329.2. (B) Results of PCR amplification of the vanHAXcluster from glycopeptide-producing organism. The lane assignments and the blot probe are the same as described for panel A. Numbers to the left of the blots are in kilobases.
Implications of the van gene cluster in glycopeptide-producing organisms.In addition to the considerable similarity at the amino acid level noted above, the orientation and alignment of the vanHst, ddlM, andvanXst genes and thevanHaov, ddlN, andvanXaov genes in the glycopeptide producers is identical to the orientation of the homologous genes in VRE (Fig. 1). There are several implications of these findings. First, the fact that glycopeptide producers possess this gene cluster implies that they use a glycopeptide resistance mechanism which is similar to that found in VRE. Second, the presence of the vanXd,d-dipeptidase gene suggests that at some point during growth, the organisms switch from producing the conventional d-Ala-d-Ala peptidoglycan terminus to producing d-Ala-d-Lac. This is supported by our recent isolation of a dedicated d-Ala-d-Ala ligase which is incapable of d-Ala-d-Lac synthesis from S. toyocaensis NRRL 15009 cultures that are not producing the A47934 glycopeptide antibiotic (24).
More significant, however, are the evolutionary implications of these results. Phylogenetic analysis reveals that the gene products fromS. toyocaensis NRRL 15009, A. orientalis C329.2, and VRE form distinct subfamilies ofd-Ala-d-Ala ligases and d-LDHs. Furthermore, the spatial arrangements of the genes are maintained in both the antibiotic-producing and clinically resistant organisms, consistent with a common ancestry. Additionally, the signature overlap of the 5′ end of the vanA and homologous genes and the 3′ end of the vanH genes is conserved. Thus, the cumulative evidence indicates that the origin of clinically relevant vancomycin resistance lies within the glycopeptide-producing organisms and not the lactic acid bacteria, which also possess d-LDH andd-Ala-d-Lact ligase enzymes but whose enzymes are of a distinct lineage.
The G+C contents of the S. toyocaensis NRRL 15009 andA. orientalis C329.2 glycopeptide resistance genes are 65.3 and 63.6%, respectively, while the genes of VRE are less purine rich, with 44 and 49% purines for the VanA and VanB clusters, respectively. This weighs against recent direct transfer of the genes from either S. toyocaensis NRRL 15009 or A. orientalis C329.2 to enterococci, although other glycopeptide-producing organisms, with different G+C contents, may have served as donor. We note that the G+C contents of the VREvanH, vanA/B, and vanX elements are 5 to 10% higher than those of the adjacent vanR,vanS, vanY, and vanZ (orvanW) genes, consistent with mobilization of thevanH-vanA/B-vanX genes as a unit from another source into a context with appropriate control elements. It is also interesting that a difference in G+C content is also found in the flanking DNA of the gene clusters cloned from the glycopeptide-producing organisms, but here the flanking sequences show higher G+C contents: 69.9% forS. toyocaensis NRRL 15009 and 68.4% for A. orientalis C329.2. Thus, the resistance cluster may have been acquired by these organisms as well. On the basis of G+C content analysis and in the absence of more sequence information from other glycopeptide antibiotic-producing organisms, the source of the vanHAX cluster in VRE is not likely to be an actinomycete.
Alternatively, the resistance genes may have been acquired by non-glycopeptide-producing soil organisms as a method of defense, probably originally from a glycopeptide-producing organism, and then mobilized and passed on to other organisms, thus eventually diluting the original G+C content bias. It has been observed invanA-harboring Oerskovia (G+C content of 70 to 75 mol%) that two of the three nucleotide mutations present in its version of vanA serve to increase the G+C content (30). Whatever the mechanism of mobilization of thevanHAX cluster, if it originated in antibiotic-producing organisms, the transfer was not a recent event. However, the use of vast quantities of the glycopeptide antibiotic avoparcin over the past two decades in European agriculture may have provided the selective pressure for this mobilization or the enrichment of resistant organisms in the environment. The fact that enterococci harboring thevanA gene cluster on transposons such as Tn1546can be isolated from farm environments in Europe (5, 7) but not the United States (10, 26) is telling. Furthermore, the observation that VRE first emerged in Europe may therefore not be a coincidence, although another reservoir for VRE in the United States cannot be ruled out (26). For example, there has been a dramatic rise in the use of vancomycin in clinical settings since the early 1980s, especially in the United States (from 2,000 kg/year in 1984 to 11,200 kg/year in 1996 [20]), and this increase in antibiotic use must be considered when discussing the rise in the prevalence of VRE over the past decade. As an additional cautionary note, it has recently been demonstrated that many antibiotic formulations are contaminated with DNA from producing organisms (41); thus, a possible source for the dissemination of resistance genes exists, possibly as a contaminant in agricultural or clinical glycopeptide antibiotic formulations. The results described here should emphasize that caution must be employed in our use of antibiotics that, while not themselves clinically useful, have the potential to select for donors of resistance genes or preexisting resistant populations.
ACKNOWLEDGMENTS
This work was supported by an operating grant from the Natural Sciences and Engineering Research Council of Canada (NSERC), by a Medical Research Council of Canada Graduate (MRC) Studentship to C.G.M., by an NSERC Postdoctoral Fellowship to I.A.D.L., and by an MRC Scholar Award to G.D.W.
FOOTNOTES
- Received 12 December 1997.
- Returned for modification 10 February 1998.
- Accepted 15 June 1998.
- Copyright © 1998 American Society for Microbiology