Genes homologous to glycopeptide resistance vanA are widespread in soil microbial communities


  • Editor: Stefan Schwarz

Correspondence: Luca Guardabassi, Department of Veterinary Pathobiology, The Royal Veterinary and Agricultural University, Stigbøjlen 4, 1870 Frederiksberg C., Denmark. Tel.: +45 35 28 27 45; fax: +45 35 28 27 51; e-mail:


The occurrence of d-Ala : d-Lac ligase genes homologous to glycopeptide resistance vanA was studied in samples of agricultural (n=9) and garden (n=3) soil by culture-independent methods. Cloning and sequencing of nested degenerate PCR products obtained from soil DNA revealed the occurrence of d-Ala : d-Ala ligase genes unrelated to vanA. In order to enhance detection of vanA-homologous genes, a third PCR step was added using primers targeting vanA in soil Paenibacillus. Sequencing of 25 clones obtained by this method allowed recovery of 23 novel sequences having 86–100% identity with vanA in enterococci. Such sequences were recovered from all agricultural samples as well as from two garden samples with no history of organic fertilization. The results indicated that soil is a rich and assorted reservoir of genes closely related to those conferring glycopeptide resistance in clinical bacteria.


The occurrence and diversity of antibiotic resistance genes has been increasingly investigated in environmental habitats such as soil, groundwater or surface water (Chee-Sanford et al., 2001; Stokes et al., 2001; Heuer et al., 2002; Tolba et al., 2002; Schwartz et al., 2003; Agerso et al., 2004; Riesenfeld et al., 2004). This type of research generates useful data on the ecology and evolution of antibiotic resistance, and provides an early warning system for future clinically relevant antibiotic resistance mechanisms. Recently, D'Costa et al. (2006) screened a library of 480 spore-forming bacteria isolated from soil and without any exception, every strain in the library was found to be resistant to at least seven of 21 antimicrobials tested, including synthetic compounds and novel drugs recently introduced in human medicine. The level of genetic and phenotypic diversity of antimicrobial resistance in soil cannot be entirely appreciated by culture methods since the vast majority of soil bacteria are nonculturable (Ward et al., 1990). Various culture-independent approaches can be used to overcome this limitation such as PCR on total DNA extracts from soil (Stokes et al., 2001; Heuer et al., 2002; Schwartz et al., 2003; Agerso et al., 2004), denaturing gradient gel electrophoresis (Chee-Sanford et al., 2001) or metagenomics (Riesenfeld et al., 2004). Such methods have revealed the presence in soil of both known and unknown genes coding for resistance to tetracyclines or aminoglycosides (Stokes et al., 2001; Heuer et al., 2002; Agerso et al., 2004; Riesenfeld et al., 2004). However, limited information is available on the occurrence and diversity of genes encoding resistance to other antimicrobial classes in soil microbial communities.

Glycopeptide antibiotics (i.e. vancomycin and teicoplanin) are drugs of primary importance in the treatment of nosocomial infections caused by staphylococci and enterococci. In contrast with tetracycline and aminoglycoside resistance, which can be mediated by various mechanisms (i.e. drug efflux, drug inactivation, target protection or modification) and by multitude of genetic determinants (Shaw et al., 1993; Roberts, 2005), glycopeptide resistance is mediated by a single mechanism (i.e. replacement of the drug target) and one genetic determinant (vanA) is the main responsible for resistance in clinical isolates (Cetinkaya et al., 2000). Based on these characteristics, glycopeptide resistance can be more easily tracked compared with resistance to other antibiotics and thus is particularly useful to study evolution of antibiotic resistance. Glycopeptide resistance vanA operons are designated according to the name of the gene encoding d-Ala : d-Lac ligase, an enzyme that catalyzes synthesis of peptidoglycan precursors with low affinity for glycopeptides (Arthur et al., 1996). Operons sharing up to 94% sequence identity with vanA were recently described in Paenibacillus thiaminolyticus and Paenibacillus apiarius isolated from Danish soil (Guardabassi et al., 2004, 2005). In this study, we developed a nested degenerate PCR for detection d-Ala : d-Lac ligase genes previously reported in Paenibacillus and in other Gram-positive genera. The method was used to study the occurrence and diversity of such genes in agricultural and garden soil collected from different geographical locations in Denmark.

Materials and methods

Soil DNA extraction

Soil was collected 5–10 cm below surface at six agricultural fields periodically amended with manure and at three old gardens with no history of organic fertilization (Table 1). Total DNA was extracted using a bead beater (Biospec Products Inc., Bartlesville, OK) and UltraClean Soil DNA Isolation kit (MoBio Laboratories Inc. Carlsbad, CA). Twelve composite DNA samples were prepared by pooling 10 μL aliquots of five extracts representative of the same sampling site and time, and used as a template for PCR.

Table 1.   Description of sampling sites
Sample codeLocalitySoil typeSampling timeTreatment with manure
AJenslevAgriculturalMarch 1999Once a year, last treatment 1 year before sampling
BNy BjergbyAgriculturalMarch 2000Once a year, last treatment 1 year before sampling
CKirke HyllingeAgriculturalMarch 2000Once a year, last treatment 1 year before sampling
DKirke HyllingeAgriculturalApril 2000Once a year, last treatment 1 year before sampling
EKirke HyllingeAgriculturalMay 2000Once a year, last treatment few days before sampling
FFersslevAgriculturalApril 2000Once a year, last treatment 1 year before sampling
GVenslevAgriculturalApril 2000Once a year, last treatment 1 year before sampling
HTåstrupAgriculturalApril 2000Once in 13 years, last treatment 1 year before sampling
ITåstrupAgriculturalMarch 1999Once in 13 years, last treatment 2 years before sampling
LNødeboGardenJanuary 2001None (35 years old garden)
MBrønshøjGardenJanuary 2001None (50 years old garden)
NNødeboGardenApril 2003None (35 years old garden)

Nested PCR

d-Ala-d-Lac ligase genes homologous to vanA were amplified by nested PCR using degenerate primers DP-fwd1 (5′-GGI GAR GAY GGI KCI ATR CAR GG-3′), DP-fwd2 (5′-TAI CCI GTI TTY GTK AAR CCB GC-3′), and DP-rev (5′-GTI ARI CCS GGI ARI GTR TTG AC-3′). Such primers were designed using the sequences of enterococcal vanA (accession no. M97297) vanB (accession no. U00457), and related d-Ala : d-Lac ligase genes in Bacillus circulans (accession no. Y15704), Paenibacillus popilliae (vanF, accession no. AF155139), Streptococcus bovis (accession no. Z70527), Oerskovia turbata (accession no. X79049), Amycolatopsis orientalis (accession no. AF060799), and Streptomyces toyocaensis (accession no. U82965). PCR reactions were prepared using Ready-To-Go PCR beads (Pharmacia Amersham Biotech) and amplified under the following conditions: 5 min at 95°C, 35 cycles at 95°C for 1 min, 50°C (first PCR step) or 56°C (second and third PCR step) for 1 min and 72°C for 2 min, followed by a final extension at 72°C for 10 min. In order to increase the specificity of the nested PCR, a third step was added using specific primers PSP-fwd (5′-ATT GGA CGC CGC AAT TGA AT-3′) and PSP-rev (5′-ACT GCG TTT TCA GAG CCT TT-3′) targeting a highly conserved region of the vanA genes in soil Paenibacillus (Guardabassi et al., 2004, 2005).

Cloning and sequencing

PCR products were purified using the Qiagen PCR Purification kit (Qiagen S.A., Courtaboeuf, France) and cloned in plasmid pCR®2.1 into Escherichia coli TOP10F′ using the TA Cloning® Kit (Invitrogen Corporation, Carlsbad, CA). Randomly selected clones obtained were screened for the presence of an insert by plasmid isolation (Birnboim & Doly, 1979) and restriction analysis with EcoRI (Invitrogen Corporation). Inserts were sequenced on an ABI 3700 automated DNA sequencer (Perkin-Elmer, Branchberg) using universal M13 primers.

Sequence analysis

The 26 sequences obtained in this study were compared with the corresponding sequences of vanA, vanB, vanF and homologous gene in the teicoplanin producing actinomycetes retrieved from GenBank using the Blast 2.0. A phylogenetic tree was constructed by Clustal X version 1.81 using the neighbour joining algorithm. The sequence from the teicoplanin producer Streptomyces toyocaensis was used for rooting the tree.


The second step of the nested degenerate PCR yielded products of the expected size (378 bp) from all samples except one (N). We sequenced 14 clones obtained from such PCR products and 13 of them contained sequences homologous (44–58% amino acid identity) to d-Ala : d-Ala ligase genes in Synechococcus elongatus (accession no. YP 172440), Symbiobacterium thermophilum (accession no. YP 074341.1), Thermoanaerobacter tengcongensis (accession no. NP 624066.1), Clostridium acetobutylicum (accession no. Q97F58), Solibacter usitatus (accession no. ZP 00520902), Mycobacterium leprae (access no. NP 302152.1), and Desulfibacterium afniense (accession. no. ZP 00097930.2). The only one putative d-Ala : d-Lac ligase gene (D90) detected was distantly related to enterococcal vanA (67% aa identity) and more closely related to the homologous gene in the teicoplanin producer Streptomyces toyocaensis (80% aa identity) (Fig. 1). Based on these results, we decided to add a third PCR step using primers specific for the vanA genes in soil Paenibacillus in order to enhance detection of d-Ala-d-Lac ligase genes homologous to vanA. PCR products of the expected size (218 bp) were obtained from all the 11 samples. Sequencing of the inserts in 25 clones obtained from such PCR products led to detection of 24 distinct sequences having 86–100% identity to enterococcal vanA (Table 2), 23 of which were novel sequences. Phylogenetic analysis by neighbor-joining (Clustal X version 1.81) evidenced the high diversity of the obtained sequences and their close evolutionary relationship with vanA in Enterococcus and in Paenibacillus (Fig. 1).

Figure 1.

 Rooted phylogenetic tree based on partial nucleotide sequences of vanA-homologous genes detected in soil and of representative d-Ala : d-Lac ligase genes encoding glycopeptide resistance. Numbers indicate support for the tree nodes (>500) obtained from MP bootstrap analysis (1000 replicates). Novel sequences obtained from soil were designated with a letter indicating the sampling site (A–M) followed by a number indicating the specific sequence. The GenBank accession numbers of previously described d-Ala-d-Lac ligase genes are reported after the names of the host species.

Table 2.   Identity to vanA (%), GC contents (%), and accession numbers of the 25 sequences of putative d-Ala : d-Lac ligase genes detected in soil
Sequence*Identity with vanA in PaenibacillusIdentity with vanA in EnterococcusGC contentAccession no.
  • *

    Letters indicate the sampling sites from which the sequences were obtained (Table 1).

  • Reference Paenibacillus thiaminolyticus PT-2B1, accession no. DQ018710.

  • Reference Enterococcus faecium BM4147, accession no. M97297.

B3, I1959049DQ116920


Genes nearly identical to vanA were recovered from all agricultural samples as well as from two garden samples with no history of organic fertilization. The widespread occurrence and high diversity of vanA-homologous sequences reported in this study indicates that soil is a rich and assorted reservoir of genes closely related to the main glycopeptide resistance determinant in clinical bacteria. The 23 novel sequences detected in soil generally resembled those of vanA genes in P. thiaminolyticus and P. apiarius isolated from soil (Guardabassi et al., 2005) (Table 2, Fig. 1), suggesting members of this genus or related genera could contribute significantly to the natural reservoir of glycopeptide resistance genes. Various authors have suggested that glycopeptide resistance could have originated from glycopeptide-producing actinomycetes, which contain gene clusters phylogenetically related to those conferring glycopeptide resistance in enterococci (Marshall et al., 1998; Gholizadeh & Courvalin, 2000; Patel, 2000). However, our recent findings indicate that vanA operons have probably originated from Paenibacillus or other endospore-forming bacilli living in soil. In fact, the vanA gene clusters in soil Paenibacillus are nearly identical to those occurring in enterococci in relation to both gene organization and sequence (Guardabassi et al., 2004), and can be heterologously expressed in Enterococcus faecalis resulting in high-level resistance (MIC >32 mg L−1) (Hasman et al., 2006). On the contrary, the gene clusters in glycopeptide-producing actinomycetes are distantly related to vanA (59–63% predicted amino acid identity) (Fig. 1), lack the two-component regulatory system that is typical of vanA operons (Marshall et al., 1998), and do not confer glycopeptide resistance when cloned in E. faecalis (Hasman et al., 2006).

Nested degenerate PCR coupled by normal PCR and by production and screening of clone libraries resulted to be a useful approach for studying the occurrence and diversity of glycopeptide resistance genes in environmental samples as it combined the sensitivity of degenerate PCR with the specificity of normal PCR. Other genetic methods recently described such as soil metagenomics (Riesenfeld et al., 2004), i.e. cloning of soil DNA in Escherichia coli followed by selection of clones expressing antibiotic resistance, are not appropriate for studying glycopeptide resistance genes since Gram-negative bacteria such as E. coli are intrinsically resistant to these antibiotics. As for all culture-independent methods, the main disadvantage is that sequences are obtained without information on the bacterial hosts from which they originate. For example, it would have been interesting to know whether the sequence M1 (100% identity to vanA) was harbored in a strain belonging to a genus other than Enterococcus. Although the sequence was recovered from garden soil with no history of organic fertilization, glycopeptide-resistant enterococci harboring vanA may occur in soil not exposed to manure as well as in natural environments not heavily contaminated with feces (Guardabassi & Dalsgaard, 2004) and the gene has been previously detected in surface and drinking water biofilms (Schwartz et al., 2003).

The widespread occurrence of putative d-Ala-d-Lac ligase genes in soil raises questions about their function in indigenous bacteria. Similarly to clinical bacteria, d-Ala : d-Lac ligase could have a resistance function and protect the bacterial host from glycopeptides produced in soil by actinomycetes. However, the frequency of glycopeptide-producing actinomycetes in soil and the antibiotic concentrations produced in situ by these organisms are unknown. Alternatively, as in lactic acid bacteria (Handwerger et al., 1994), peptidoglycan terminating in d-Ala-d-Lac could be the normal cell wall precursors in uncultured bacteria living in soil. A third, less studied hypothesis (exaptation) is that such ligases are involved in other biological functions in soil bacteria and acquire a resistance function when expressed by heterologous hosts. The latter hypothesis is particularly fascinating and would require further investigation in view of the recent discovery in the genome of various glycopeptide-susceptible Bacillus species of genes homologous to the regulatory (vanR and vanS) and accessory genes (vanY) of vanA operons in enterococci (Integrated Genomics,


This work was supported by grants no. 23-01-0170 and no. 23-02-0169 from the Danish Agricultural and Veterinary Research Council and grant no. 274-05-0117 from the Danish Research Agency.