New Combinations of Mutations in VanD-Type Vancomycin-Resistant Enterococcus faecium, Enterococcus faecalis, and Enterococcus avium Strains

ABSTRACT We studied the clinical isolates Enterococcus faecium NEF1, resistant to high levels of vancomycin (MIC, 512 μg/ml) and teicoplanin (MIC, 64 μg/ml); Enterococcus faecium BM4653 and BM4656 and Enterococcus avium BM4655, resistant to moderate levels of vancomycin (MIC, 32 μg/ml) and to low levels of teicoplanin (MIC, 4 μg/ml); and Enterococcus faecalis BM4654, moderately resistant to vancomycin (MIC, 16 μg/ml) but susceptible to teicoplanin (MIC, 0.5 μg/ml). The strains were distinct, were constitutively resistant via the synthesis of peptidoglycan precursors ending in d-alanyl-d-lactate, and harbored a chromosomal vanD gene cluster that was not transferable. New mutations were found in conserved domains of VanSD: at T170I near the phosphorylation site in NEF1, at V67A at the membrane surface in BM4653, at G340S in the G2 ATP-binding domain in BM4655, in the F domain in BM4656 (a 6-bp insertion), and in the G1 and G2 domains of BM4654 (three mutations). The mutations resulted in constitutivity, presumably through the loss of the phosphatase activity of the sensor. The chromosomal Ddl d-Ala:d-Ala ligase had an IS19 copy in NEF1, a mutation in the serine (S185F) or near the arginine (T289P) involved in d-Ala1 binding in BM4653 or BM4655, respectively, and a mutation next to the lysine (P180S) involved in d-Ala2 binding in BM4654, leading to the production of an impaired enzyme. In BM4653 vanYD, a new insertion sequence, ISEfa9, belonging to the IS3 family, resulted in the absence of d,d-carboxypeptidase activity. Strain BM4656 had a functional d-Ala:d-Ala ligase, associated with high levels of both VanXD and VanYD activities, and is the first example of a VanD-type strain with a functional Ddl enzyme. Study of these five clinical isolates, displaying various assortments of mutations, confirms that all VanD-type strains isolated so far have undergone mutations in the vanSD or vanRD gene, leading to constitutive resistance, but that the Ddl host ligase is not always impaired. Based on sequence differences, the vanD gene clusters could be assigned to two subtypes: vanD-1 and vanD-4.

Vancomycin-resistant enterococci have spread with unanticipated rapidity and are now encountered in hospitals in most countries. Three types, VanA, VanB, and VanD, with acquired resistance to the glycopeptides vancomycin and teicoplanin by means of production of peptidoglycan precursors ending in the depsipeptide D-alanyl-D-lactate (D-Ala-D-Lac) instead of the dipeptide D-Ala-D-alanine (D-Ala-D-Ala), have been characterized. VanA and VanB are the most prevalent phenotypes and have been found primarily in Enterococcus faecium and Enterococcus faecalis. VanD has been found only in a few strains of E. faecium (6,7,10,13,16,31), two E. faecalis strains (13), one Enterococcus gallinarum strain (8), and one Enterococcus raffinosus strain (37). Recently, the vanD operon was detected in a Ruminococcus sp., an anaerobic gram-positive bacterium from human bowel flora, suggesting a potential reservoir for vancomycin resistance genes in commensals of the digestive tract (17).
VanD-type strains share various characteristics that distinguish them from VanA-and VanB-type enterococci. In particular, resistance to moderate levels of the two glycopeptides is constitutively expressed and is not transferable by conjugation to other enterococci (10,13,16,31). The organization of the vanD gene cluster, located exclusively in the chromosome, is similar to those of vanA and vanB (13,16). The vanD operon contains the vanH D DX D resistance genes, encoding, respectively, a dehydrogenase that converts pyruvate to D-lactate, a ligase required for the synthesis of D-Ala-D-Lac, and a dipeptidase; these enzymes have high levels of sequence identity (59 to 70%) with the corresponding deduced proteins of the vanA and vanB operons (15). The VanY D D,D-carboxypeptidase belongs to the penicillin-binding protein (PBP) family of catalytic serine enzymes, which are susceptible to benzylpenicillin (10,13,16,31), and is distinct from VanY and VanY B , which are penicillin-insensitive Zn 2ϩ -dependent D,D-carboxypeptidases. The vanR D S D regulatory genes, encoding a two-component regulatory system composed of a membrane-bound histidine kinase (VanS D ) and a cytoplasmic response regulator (VanR D ) that acts as a transcriptional activator, are present upstream of the resistance genes and are only distantly related to VanRS and VanR B S B (19 to 58% identity). The regulatory and resistance genes are transcribed from distinct promoters (P R, P RB , and P RD for the regulatory genes and P H , P YB , and P YD for the resistance genes) that are coordinately regulated (15). No genes homologous to vanZ or vanW from the vanA and vanB operons, respectively, are present.
The interaction of glycopeptides with their target, the late peptidoglycan precursors ending in D-Ala-D-Ala, is prevented by the removal of precursors terminating in D-Ala (33). Two enzymes are involved in this process: a cytoplasmic D,D-dipeptidase (VanX, VanX B , or VanX D ) that hydrolyzes the dipeptide D-Ala-D-Ala synthesized by the host Ddl and a membranebound D,D-carboxypeptidase (VanY, VanY B , or VanY D ) that removes the C-terminal D-Ala residue of late peptidoglycan precursors when elimination of D-Ala-D-Ala by VanX is incomplete (2). However, the VanD-type strains investigated previously have negligible D,D-dipeptidase activity, encoded by vanX D alleles, despite the presence in VanX D of the critical residues implicated in the binding of Zn 2ϩ and in catalysis (27). The lack of such an activity should result in a glycopeptide-susceptible phenotype, since these bacteria are unable to eliminate peptidoglycan precursors ending in D-Ala-D-Ala. However, in VanD-type strains, the susceptible pathway does not function due to an inactive D-Ala:D-Ala ligase resulting from various mutations in the chromosomal ddl gene (6,7,13,16,31). Consequently, the strains should grow only in the presence of vancomycin, since they rely on the inducible resistance pathway for peptidoglycan synthesis. However, this is not the case, since the vanD clusters are expressed constitutively due to mutations either in the VanS D sensor (13,16) or in the VanR D regulator (13) of the two-component regulatory system.
We report the organization of the vanD gene clusters, the regulation of the expression of the resistance genes, and the sequences of the ddl genes in the clinical isolates E. faecium NEF1 (23), BM4653, and BM4656 and E. faecalis BM4654, and in the first VanD-type E. avium strain, BM4655.

MATERIALS AND METHODS
Strains, plasmids, and growth conditions. The origins and characteristics of the bacterial strains and plasmids used in this study are given in Table 1. E. faecium NEF1 (23), BM4653, and BM4656 and E. avium BM4655 were isolated from hospitalized patients in France (Nîmes, Chambéry, Toulouse, and Clamart, respectively), and E. faecalis BM4654 was isolated from a patient in Adelaide, South Australia. The susceptible E. avium strain ATCC 14025 was used as a control, and Escherichia coli Top10 (Invitrogen, Groningen, The Netherlands) was used as a host for recombinant plasmids. Kanamycin (50 g/ml) was the selective agent for cloning PCR products into the pCR-Blunt vector (Invitrogen). Spectinomycin (60 g/ml) or gentamicin (32 g/ml) was added to the culture medium to prevent the loss of plasmids derived from pAT79 (4) or pAT392 (3), respectively. Strains were grown in brain heart infusion (BHI) broth or agar (Difco Laboratories, Detroit, MI) at 37°C.  (10) and 4655-1 (ATACGAG CTCGCTAAAATGTTAGGGAT) each contain a SacI restriction site (italicized). Primers 4147-2 (10) and 4655-2 (GCCATCTAGATTATTCAAAACGC GCCT) each contain an XbaI restriction site (italicized). The SacI and XbaI restriction sites allowed directional cloning of the ddl gene under the control of the constitutive P 2 promoter and upstream of the cat reporter gene of the pAT79 shuttle vector (4). The 1,135-bp insert of the resulting pAT825 [P 2 ddl(S185F) cat] and pAT827 [P 2 ddl(P180S) cat] plasmids and the 1,105-bp insert of pAT829 [P 2 ddl(T289P) cat] contained the mutated ddl gene with the resulting single S 185 F, P 180 S, or T 289 P amino acid substitution, respectively, and their cognate RBS, whereas plasmids pAT826 (P 2 ddl BM4656 cat) from E. faecium BM4656 and pAT828 (P 2 ddl ATCC 14025 cat) from E. avium ATCC 14025 had no mutations in their ddl genes and were used as controls.
TAIL-PCR amplification. Thermal asymmetric interlaced PCR (TAIL-PCR) (26)  For the third PCR, the solution was the same as for the second PCR except that 3 l of a 1/10 dilution of the second or third PCR product was used as a template. Three TAIL-PCRs were performed with the AD1, AD2, AD3, and AD4 random primers designed by Liu and Whittier (26). The reaction products of the first, second, and third PCR steps were separated by agarose gel electrophoresis. The bands corresponding to the third PCRs, which showed expected decreases in length consistent with the positions of the specific primers along the genome, were purified and sequenced. In order to rule out mismatches during TAIL-PCR, the nucleotide sequences of the amplification products were confirmed by conventional PCR with new primers.
DNA sequencing. Plasmid DNA was extracted with the commercial Wizard Plus Minipreps DNA purification system (Promega, Madison, WI), and the PCR fragments were purified with a Qiagen PCR purification kit. Plasmid DNA or PCR products were labeled with a dye-labeled ddNTP Terminator cycle sequencing kit (Beckman Coulter UK Ltd.), and the samples were sequenced and analyzed with a CEQ 2000 automated sequencer (Beckman).
Computer analysis of sequence data. The degrees of identity and similarity with known proteins were determined using BLASTN, BLASTX, and BLASTP (1) and FASTA (30) from the Genetics Computer Group (GCG) suite of programs.
Contour-clamped homogeneous electric field gel electrophoresis. Genomic DNA embedded in agarose plugs was digested overnight at 27°C with 25 U of SmaI or for 3 h at 37°C with 0.01 U of I-CeuI, an intron-encoded endonuclease specific for rRNA genes (25). The products were separated on a 0.8% agarose gel using a CHEF-DRIII system (Bio-Rad Laboratories) under conditions described previously (16). Fragments were hybridized successively, as described previously (16), to an [␣-32 P]-labeled 16S rRNA (rrs) probe obtained by amplification of an internal portion of the rrs gene with primers RWO1 and DG74 (21) and to a vanD probe obtained by PCR with primers VanD1 (5Ј-TAAGGCGCTTGCAT ATACCG) and VanD2 (5Ј-TGCAGCCAAGTATCCGGTAA), internal to the gene, with BM4653 total DNA as a template (Fig. 1).
Analysis of peptidoglycan precursors. Peptidoglycan precursors were extracted and analyzed by high-performance liquid chromatography as described previously (3). Enterococci were grown in BHI medium without or with vancomycin (3 g/ml) to the mid-exponential phase (A 600 , 1). Ramoplanin (3 g/ml) was added to inhibit peptidoglycan synthesis, and incubation was continued for 15 min to cause accumulation of peptidoglycan precursors.  A superscript "r" indicates resistance, and a superscript "s" indicates sensitivity. For strains, base pair changes are indicated in the genotypes; for plasmids, the amino acid substitutions in the deduced proteins encoded by the inserts are given. b 5bp37, a 5-bp insertion at position 37 in the ddl gene.
were assayed as described previously (3). Strains were grown to an optical density at 600 nm of 0.7 in the absence or presence of vancomycin (8 g/ml) and with gentamicin (32 g/ml) to counterselect the loss of derivatives of pAT392. Following cell breakage, the supernatant (S100) and the resuspended pellet (C100) were collected and assayed for D,D-peptidase (VanX or VanY) activities by measurement of the D-Ala released from substrate hydrolysis (6.56 mM D-Ala-D-Ala or 5 mM L-Ala-D-Glu-L-Lys-D-Ala-D-Ala) as described previously (3). Nucleotide sequence accession numbers. The 1,384-bp fragments containing the ddl genes of strains ATCC 14025 and BM4655 and the 1,043-bp fragment containing the int D gene downstream of vanX D in BM4655 were submitted to GenBank and assigned accession no. EU999032, EU999033, and EU999034, respectively. The 1,366-bp ISEfa9 sequence inserted in the vanY D gene of BM4653 has been deposited in the GenBank database under accession no. EU999035. The entire vanD cluster of E. faecium A902 (13), sometimes designated VanD2, was submitted to GenBank and assigned accession no. EU999036.

RESULTS AND DISCUSSION
Characterization of VanD-type E. faecium NEF1, BM4653, and BM4656, E. faecalis BM4654, and E. avium BM4655. E. faecium NEF1 (23) was resistant to high levels of vancomycin (MIC, 512 g/ml) and teicoplanin (MIC, 64 g/ml), whereas E. faecium BM4653 and BM4656 and E. avium BM4655 were resistant to a moderate level of vancomycin (MIC, 32 g/ml) and to a low level of teicoplanin (MIC, 4 g/ml) (Fig. 2). E. faecalis BM4654 was moderately resistant to vancomycin (MIC, 16 g/ml) but remained susceptible to teicoplanin (MIC, 0.5 g/ml) (Fig. 2). The glycopeptide resistance genotype of these strains was determined by multiplex PCR (14) and found to be vanD. The patterns of the five strains obtained by pulsed-field gel electrophoresis after digestion of total DNA by SmaI differed, and a vanD probe hybridized with a DNA fragment of a different size in each strain, indicating that the clinical isolates were distinct (data not shown). Repeated attempts to transfer vancomycin resistance from the five strains to E. faecalis JH2-2 (22) or E. faecium 64/3 (38) by filter mating were unsuccessful (frequency Ͻ 10 Ϫ9 per recipient).
Chromosomal location of the vanD gene cluster. The vanD gene cluster was assigned to a chromosomal fragment of ca. 380 kb in E. faecium NEF1, 365 kb in E. faecium BM4653, 350 kb in E. faecium BM4656, 450 kb in E. faecalis BM4654, and 565 kb in E. avium BM4655 by pulsed-field gel electrophoresis of total DNA digested with I-CeuI, followed by successive hybridizations with a 16S rRNA (rrs) probe and a probe internal to vanD (data not shown). This analysis confirmed the data obtained with SmaI, indicating that the VanD-type isolates were distinct and differed from those reported previously (13,16).
Organization of the vanD operons. The products obtained following PCR mapping with primers complementary to the vanD operon ( Fig. 1) from E. faecium BM4339 (10) were of the expected size for all the strains except E. faecium BM4653, for which PCR with primer pair VDB-YD1 COOH gave rise to a product approximately 1,400 bp larger than expected due to an insertion sequence in the vanY D gene. All the genes of the vanD operon were present in the five strains, and their organization was identical to that in BM4339, i.e., two adjacent clusters, one containing the regulatory genes vanR D S D and the other containing the vanY D H D DX D resistance genes (Fig. 1). No large insertions or deletions in the noncoding regions were detected.
The deduced sequences from the vanR D and vanS D PCR products (Fig. 1) were almost identical to the VanR D response   Tables 2 and 3; also data not shown). The gene for the VanD D-Ala:D-Lac ligase from E. faecium NEF1 has been characterized previously, and the enzyme had the highest degree of identity with that of BM4339 (Table 2) (23). The deduced sequences of the VanH D , VanD, and VanX D proteins of the five strains displayed 83 to 100% identity with the corresponding proteins from the previously described VanD-type strains (Tables 2 and 3; also data not shown). Analysis of the translation products of the vanY D genes of E. faecium BM4656, E. faecalis BM4654, and E. avium BM4655 indicated that they contained the three motifs SXXK, SG(C/ N), and KTG, characteristic of the penicillin binding domains of PBPs (16). However, the VanY D D,D-carboxypeptidase of E. faecium NEF1 was terminated prematurely after amino acid 183 due to a substitution (TGG to TGA) leading to a stop codon, resulting in a polypeptide with a conserved membranespanning portion at the N terminus and the first conserved active-site sequence (SXXK) but lacking the other two sequences that form the PBP active site. In E. faecium BM4653, the vanY D gene was interrupted by insertion sequence ISEfa9, belonging to the IS3 family, which generated a 3-bp (GAA) duplication of target DNA corresponding to nucleotides 583 to 585 of vanY D . The ISEfa9 copy was flanked by 40-bp imperfect inverted repeats and contained two open reading frames (ORFs), ORFA and ORFB, in the same orientation, which exhibited 55 and 54% identity with the structural genes for  a For comparative purposes, the putative full-length nonmutated VanS D and VanY D proteins were "reconstructed" as follows. For VanS D , an A nucleotide was substituted at position 668 for the 1-bp deletion in BM4416 (31) (also designated N97-330 ͓6͔), and a T nucleotide was substituted at position 657 for the 1-bp deletion in A902 (13) (accession no. 999036). The ISEfa4 insertion sequence in 10/96A (16), the 7-bp insertion in BM4539 (13), and the 6-bp insertion in BM4656 were ignored. For VanY D , the substitution mutation TGA was changed to TGG at position 549 in the NEF1 gene to eliminate the stop codon. The 1-bp insertions in 10/96A (16) and N03-0072 (7) and the insertion sequence in BM4653 were ignored. a For comparative purposes, the putative full-length nonmutated VanS D and VanY D proteins were "reconstructed" as follows. For VanS D , an A nucleotide was substituted at position 668 for the 1-bp deletion in BM4416 (31) (also designated N97-330 ͓6͔), and a T nucleotide was substituted at position 657 for the 1-bp deletion in A902 (13) (accession number 999036). The ISEfa4 insertion sequence in 10/96A (16), the 7-bp insertion in BM4539 (13), and the 6-bp insertion in BM4656 were ignored. For VanY D , the substitution mutation TGA was changed to TGG at position 549 in the NEF1 gene to eliminate the stop codon. The 1-bp insertions in 10/96A (16) and N03-0072 (7) and the insertion sequence in BM4653 were ignored.  (Table 3 and data not shown).
In both E. faecium NEF1 and E. avium BM4655, downstream of vanX D and associated with the vanD gene cluster, an intD gene for a putative integrase-like protein, displaying 100% and 94% amino acid sequence identity, respectively, with that of E. faecium BM4339, was present (Fig. 1). The vanD cluster of BM4655 was almost identical to that of 10/96A (Table 3), and an intD gene 100% identical to that of BM4655 was found downstream of vanX D in 10/96A.
As already mentioned, the vanR D and vanS D genes are highly conserved among all VanD-type strains, whereas the vanH D , vanD, and vanX D resistance genes are more distant. However, among the proteins encoded by the vanD clusters, VanY D presents the highest variability, with percentages of identity ranging from 78 to 100% (Tables 2 and 3; also data not shown). On the basis of the sequences of VanY D and VanD, VanD-type strains can be assigned to two subtypes: vanD-1 and vanD-4. The vanD clusters of the VanD-1 strains are closer to that of BM4339 (10), whereas those of VanD-4 strains are more closely related to that of 10/96A (16). In E. faecium NEF1, BM4653, BM4656, BM4416, and A902 and in E. faecalis BM4538 and BM4539, the vanD operon was similar to that of BM4339 (Table 2 and data not shown) except for vanY D in E. faecium A902, which had the highest identity with vanY D in E. faecium 10/96A (16). The vanD operon of E. avium BM4655 was identical to that of E. faecium 10/96A (Table 3). These two subtypes also hold true for other Enterococcus species and other bacterial genera. In E. gallinarum N04-0414 (8) and a Ruminococcus sp. (17), the vanD clusters are, respectively, 94% and 98.2% identical with that of BM4339, whereas in E. raffinosus GV5 (37), the vanD cluster is highly similar (99.5% identity) to that in 10/96A, except for the presence of a frameshift mutation in vanY D in the latter strain. However, E. faecalis BM4654 and E. faecium N03-0072 (7), which are identical, harbor hybrid operons: the predicted VanS D , VanY D , and VanD proteins are more similar to their counterparts in 10/96A than in BM4339 and its derivatives, whereas the opposite is true for the remaining proteins (VanR D , VanH D , and VanX D ) ( Table 3).

Study of the D-Ala:D-Ala ligases.
Every VanD-type strain reported so far has undergone a mutation in the ddl gene that prevents the synthesis of peptidoglycan by the susceptible chromosomal pathway. The ddl genes for the D-Ala:D-Ala ligases of the VanD-type strains were amplified, and PCR products with the expected length of 1,154 bp for E. faecium BM4653 and BM4656 or 1,063 bp for E. faecalis BM4654 were sequenced. For E. faecium NEF1, amplification of the entire ddl gene gave rise to a 2,194-bp fragment, which was also sequenced. The sequences upstream and downstream of the 600-bp known portions of the ddl genes of E. avium ATCC 14025 (susceptible) (19) and BM4655 were obtained by TAIL-PCR, and the corresponding proteins were compared (Fig. 3).
In E. faecium NEF1, an insertion sequence of 1,040 bp, similar to IS19 and ISEfm1, which belong to the IS982 family, was present at the beginning of the gene (Fig. 3) instead of in the middle as in E. faecium strain BM4416 (31), also designated N97-330 (6). Insertion of the element generated a 9-bp (ATTTATTAT) duplication of the target DNA corresponding to nucleotides 82 to 90 of the ddl gene. The IS19-like copy was terminated by 22-bp perfect inverted repeats (ACCCGTGGT GCTAGTTAATTTT) and contained an ORF that exhibited 100% identity with the structural gene for the putative transposase of IS4 from E. faecium DO and 99% identity with the genes for the putative transposases of IS19 and ISEfm1. In contrast to that of ISEfm1, the transposase of IS19-like, like that of IS19, was not truncated following a frameshift.
Comparative ddl analysis revealed a point mutation in codon 185 of E. faecium BM4653 and codon 180 of E. faecalis BM4654 relative to the ddl gene of E. faecium BM4147, resulting in a Ser-to-Phe or Pro-to-Ser substitution at a position involved in the binding of D-Ala1 or next to the lysine involved in the binding of D-Ala2, respectively; both of these point mutations presumably lead to a nonfunctional protein (Fig. 3). For E. avium BM4655, analysis relative to the ddl gene of the susceptible E. avium strain ATCC 14025 revealed a point mutation in codon 289 resulting in a Thr-to-Pro substitution located near the arginine involved in the binding of D-Ala1 (Fig.  3). The Ddl ligases of E. avium ATCC 14025 and BM4655 had the highest degree of identity (98%) with that of E. raffinosus (37) and only 79% and 72% identities with those of E. faecium and E. faecalis, respectively. Surprisingly, no mutations were present in the ddl gene of E. faecium BM4656. Consequently, the deduced protein contained all the amino acids implicated in the binding of D-Ala1, D-Ala2, and ATP that are conserved in E. faecium, E. faecalis, and E. avium (Fig. 3).
Strain BM4339 has an impaired Ddl enzyme, and the introduction of an intact ddl gene under the control of a constitutive promoter restores its susceptibility to glycopeptides (10). This was also observed with plasmids pAT826 (P 2 ddl BM4656 cat) from E. faecium BM4656 and pAT828 (P 2 ddl ATCC 14025 cat) from the susceptible E. avium strain ATCC 14025 ( Table 1). The decrease in glycopeptide resistance is due to the production of the heterologous Ddl enzyme, since BM4339 possesses only weak VanX D,D-dipeptidase activity (10). To test if the S 185 F, P 180 S, and T 289 P mutations were responsible for the impairment of the Ddl enzyme, plasmids pAT825 [P 2 ddl(S185F) cat], pAT827 [P 2 ddl(P180S) cat], and pAT829 [P 2 ddl(T289P) cat], containing the ddl mutant genes and their RBS sequences from BM4653, BM4654, and BM4655, respectively, were cloned under the control of the constitutive P 2 promoter and electrotransformed into BM4339 ( Table 1). The resulting transformants, BM4663, BM4665, and BM4667 (Table 1), remained vancomycin resistant, confirming that the D-Ala:D-Ala ligases from the three strains were probably inactive.
Characterization of the peptidoglycan precursors. The cytoplasmic peptidoglycan precursors from cells grown in the absence or presence of vancomycin were analyzed (Fig. 2). E. faecium NEF1 synthesized essentially pentadepsipeptide precursors (96%), even in the absence of induction, indicating constitutive peptidoglycan synthesis using only the resistance pathway (Fig. 2). Therefore, the elevated glycopeptide resistance of strain NEF1 was associated with increased synthesis of UDP-MurNAc-pentadepsipeptide to the detriment of UDP-MurNAc-pentapeptide (Ͻ1%), as a consequence of the alter-  3. Alignment of the deduced amino acid sequences of the Ddl ligases of VanD-type strains. Numbers on the left and right of each line refer to the and last amino acids in the corresponding sequence. Identical amino acids are indicated below the alignment by dashed lines. The numbers in boldface above the alignment indicate the amino acids implicated in the binding of D-Ala1 (E13, H98, S185, and R293) (1), D-Ala2 (K179, K253, and E308) (2), and ATP (Y254, D295, and S319) (3) (20,35), which are conserved in E. faecium, E. faecalis, and E. avium. In E. faecium NEF1, the position of IS19 is indicated by a vertical arrow between the two flanking deduced amino acids. In E. faecium BM4653, E. faecalis BM4654, and E. avium BM4655, the single-base differences relative to the ddl genes of E. faecium BM4147, E. faecalis V583, and the susceptible E. avium strain ATCC 14025, leading, respectively, to a Ser-to-Phe substitution at position 185, a Pro-to-Ser substitution at position 180, and a Thr-to-Pro substitution at position 289, are indicated by boldface italics. The Ddl sequence of E. faecium BM4656 is identical to that of BM4147, and no mutations were observed. The Ddl sequence of ATCC 14025, determined in this study, is shown for comparison with that of E. avium BM4655.
ation of the regulation of the resistance genes and the absence of the host D-Ala:D-Ala ligase. It has been shown that in VanAtype strains, increased transcription of the vanHAX operon is associated with increased incorporation of D-Ala-D-Lac into peptidoglycan precursors to the detriment of D-Ala-D-Ala and with gradual increases in vancomycin resistance levels (3). However, more complete elimination of D-Ala-D-Ala-containing precursors is required for teicoplanin resistance (3). This situation was also observed in NEF1, which has a higher percentage of UDP-Mur-NAc-pentadepsipeptide (96%) than the other VanD-type strains of this study and the highest resistance to both vancomycin (512 g/ml) and teicoplanin (64 g/ml) (Fig. 2). For the remaining strains, the results also showed that, whether in the absence or in the presence of vancomycin, UDP-MurNAc-pentadepsipeptide was the main precursor. However, in addition to the pentadepsipeptide, UDP-MurNAcpentapeptide was present at 6 to 19% in E. faecium BM4653 and BM4656, E. faecalis BM4654, and E. avium BM4655, which could account for the lower levels of resistance of these four strains to vancomycin and teicoplanin in comparison with those of E. faecium NEF1. UDP-MurNAc-tetrapeptide (9 to 33%) was also present in E. faecium BM4656, E. faecalis BM4654, and E. avium BM4655, whereas in E. faecium NEF1 and BM4653 its levels were very low (4%) (Fig. 2) due to the synthesis of a truncated, and presumably nonfunctional, VanY D D,D-carboxypeptidase. The small amount of pentapeptide present could have been synthesized by the VanD ligase, as has already been shown for the VanA ligase (9). Since the D-Ala:D-Ala ligases of these strains probably have no activity (Fig. 3), little, if any, UDP-MurNAc-pentapeptide should be present; therefore, the tetrapeptide is likely to result from the removal of D-lactate from UDP-MurNAc-pentadepsipeptide. PBPs that function as D,D-carboxypeptidases hydrolyze esters and peptide bonds (32). It is thus presumed that the VanY D enzymes, which are PBPs and whose activities were inhibited by low concentrations of benzylpenicillin (data not shown), would hydrolyze UDP-MurNAc-pentadepsipeptide, with the production of tetrapeptide.
Taken together, these data indicate that the strains were constitutively resistant to vancomycin by means of the production of precursors ending in D-Ala-D-Lac and that they therefore used the resistance pathway exclusively. Thus, in four of the five strains, the production of an impaired Ddl enzyme (Fig. 3) accounts for the lack of peptidoglycan precursors terminating in D-Ala-D-Ala (Fig. 2). In contrast, the functionality of the Ddl enzyme from BM4656 explains the presence of the higher percentage of pentapeptide (14%) in comparison with those (Ͻ1% to 9%) in the other VanD-type E. faecium or E. faecalis strains (Fig. 2). D,D-Peptidase activities. The regulation of the expression of the resistance genes was studied by analysis of VanX D D,Ddipeptidase and VanY D D,D-carboxypeptidase activities (Fig.  4). D,D-Dipeptidase activity was measured in the 100,000 ϫ g supernatant of lysed bacteria that had been grown in the absence or presence of 4 or 8 g/ml of vancomycin as an inducer. Weak D,D-dipeptidase activity (VanX D ) was found in the cytoplasmic extracts from induced or uninduced E. faecalis BM4654 and E. avium BM4655 (Fig. 4). Since these strains do not produce D-Ala-D-Ala-containing peptidoglycan precursors due to mutations in the chromosomal ddl gene, no D,D-dipeptidase activity is required for glycopeptide resistance in this genetic background.
In E. faecium NEF1, BM4653, and BM4656, VanX D was constitutively synthesized at moderate levels. In the case of BM4656, the D,D-dipeptidase was required, since the D-Ala:D-Ala ligase was functional. Constitutive production of VanX D was more surprising in E. faecium NEF1 and BM4653 (Fig. 4), since their Ddl enzymes were probably nonfunctional (Fig. 3). However, in these strains, VanY D , which is prematurely truncated after the first of the three motifs involved in the scaffolding of the active site (27) by a stop codon in NEF1 or by ISEfa9 in BM4653, is also likely to be nonfunctional, as confirmed by the very low D,D-carboxypeptidase activities (Fig. 4) and by the lack of tetrapeptide (Fig. 2); consequently, VanX D D,D-dipeptidase activity is required to eliminate the D-Ala-D-Ala that may be synthesized by the VanD ligase.
In E. faecium BM4656, E. faecalis BM4654, and E. avium BM4655, the D,D-carboxypeptidase specific activities in membrane extracts were high and were expressed constitutively (Fig. 4). This accounts for the presence of UDP-Mur-NActetrapeptide in these strains. VanY D activity compensated for the very weak VanX D activity in BM4654 and BM4655. In BM4654 and BM4655, all the critical residues implicated in the binding of Zn 2ϩ and in catalysis were present in VanX D . However, in BM4654 the DFM motif implicated in the binding of VANCOMYCIN RESISTANCE IN ENTEROCOCCI D-Ala-D-Ala was altered to DYM, and in BM4655 a mutation was observed next to the same motif. Both these mutations could explain the very weak VanX D activity observed in these strains (Fig. 4). In BM4656, since the D-Ala:D-Ala ligase was functional, VanX D hydrolyzed the D-Ala-D-Ala synthesized by the host Ddl, and VanY D could contribute to resistance by removing the C-terminal D-Ala residue of late peptidoglycan precursors when elimination of D-Ala-D-Ala by VanX D was incomplete. However, when VanY D is inactive, as in NEF1 and BM4653, VanX D is more active (Fig. 4). As in the VanD-type strains previously studied (13,16,31), D,D-carboxypeptidase activities were inhibited by benzylpenicillin, consistent with the fact that VanY D belongs to the PBP family of catalytic serine enzymes, which are susceptible to ␤-lactam antibiotics.

Mutations in the VanS D sensors.
Mutants with an impaired D-Ala:D-Ala ligase require vancomycin for growth, since D-Ala-D-Lac is produced only under inducing conditions. However, this is not the case for the VanD-type strains studied previously, which express the vanD cluster constitutively due to mutations in either the VanS D sensor (6,13,16) or the VanR D regulator (13) of the two-component regulatory system.
In wild-type VanS D , five blocks of amino acids (H, N, G1, F, and G2) in the kinase domain are highly conserved (Fig. 5).
The histidine residue of VanS D at position 166 in the H block is the putative site of autophosphorylation of the sensor. This block is responsible for both autophosphorylation and kinase/ phosphatase activities, while the G1 and G2 blocks are involved in ATP binding. Sequence alignment of the sensors revealed a point mutation (T 170 I) in a critical region near the putative autophosphorylation site in NEF1; a point mutation (V 67 A) at the outer membrane surface in BM4653; a point mutation (G 340 S) in the G2 ATP-binding domain in BM4655; and three mutations (K 308 Q and I 309 V in the G1 ATP-binding domain and R 335 Q near the G2 ATP-binding domain) in BM4654 (Fig. 3). In BM4656, a duplication resulting in two tandemly arranged hexanucleotide (TTTACC) direct repeats occurred in the F domain, whose function remains unknown but which is always conserved (Fig. 5). In E. faecium BM4653, the V 67 A mutation, located at the outer surface of the membrane (Fig. 5), was presumably responsible for constitutive expression, as observed for other two-component regulatory systems, such as OmpR-EnvZ, which is related to the VanRS systems (34), BvgA-BvgS (28), NarL-NarX (11), and VirG-VirA (18). Although the five blocks of conserved amino acids characteristic of transmitter modules in histidine protein kinases were present in the C-terminal portion of VanS D in the five strains studied, the production of VanS D with mutations in these conserved domains may have resulted in a high steadystate level of phosphorylated VanR D and thus could account for the constitutive expression of the vanD operon in these strains.
In contrast, in the VanD-type strains previously studied, constitutive vancomycin resistance was due to an insertion (ISEfa4 in E. faecium 10/96A [16] or 7 bp in E. faecalis BM4539 and BM4540 [13]) or a deletion (1 bp in E. faecium N97-330 [7], also designated BM4416 [31], and A902 [13] [accession no. EU999036]). These mutations resulted in a frameshift in the VanS D sensor and presumably led to truncated, nonfunctional proteins. In BM4539, BM4540, BM4416, and A902, only the H block, responsible for both autophosphorylation and kinase/ phosphatase activities, was present, whereas the other conserved blocks (N, G1, F, and G2) in the kinase domains of the sensors were absent, strongly suggesting a loss of activity. This is in agreement with reports indicating that the G2 block could play a role in the modulation of enzyme conformation (29) and that the G2 box of VanS B (12), as well as that of EnvZ and its surrounding residues (39), plays an important role in modulating phosphatase activity.
The absence of mutations in BM4656 Ddl and the fact that two VanD-type E. faecalis strains harbor identical vanD operons but different mutations in their D-Ala:D-Ala ligases (13) strongly suggest that the mutations in the sensor or in the regulator were acquired before those in the Ddl ligase; otherwise, the strain would have been transiently dependent on glycopeptide for growth.
Study of the functionality of the VanS D sensors. The introduction of an intact vanS D gene under the control of a constitutive promoter into VanD-type E. faecium BM4458 and BM4459 (Table 1), both of which possess a functional Ddl enzyme at different locations in the chromosome but an impaired VanS D protein with a mutated vanS D gene, restores inducible expression of the resistance genes (13).
To investigate whether the mutated vanS D genes were responsible for the constitutive expression of the resistance genes, plasmids pAT830 [P 2 vanS D (V67A) aac(6Ј)-aph(2Љ)], pAT831 [P 2 vanS D ::6bp330 aac(6Ј)-aph(2Љ)], pAT832 [P 2 vanS D (K308Q I309V R335Q) aac(6Ј)-aph(2Љ)], and pAT833 [P 2 vanS D (G340S) aac(6Ј)-aph(2Љ)], containing the mutated vanS D genes (Fig. 1), including an RBS, of BM4653, BM4656, BM4654, and BM4655, respectively, were electrotransformed into E. faecium BM4458 and BM4459 ( Table 1). The effects of the presence of the various mutated VanS D sensors in the two strains on peptidoglycan biosynthesis were analyzed by estimating the relative amounts of peptidoglycan precursors (Fig.  6A). The regulation of the expression of the resistance genes was also studied by determining VanY D D,D-carboxypeptidase activities in membrane extracts from bacteria grown in the absence or presence of vancomycin (Fig. 6B). The introduction   (Table 1), as in BM4459 carrying these plasmids, UDP-MurNAc-pentadepsipeptide, UDP-MurNAc-pentapeptide, UDP-MurNAc-tetrapeptide, and UDP-MurNAc-tripeptide were synthesized in similar quantities in the absence or presence of vancomycin (2 g/ml) (Fig. 6A and data not shown). The presence of the tripeptide could indicate that the VanD ligase was not sufficiently active to synthesize D-Ala-D-Lac as rapidly as the tripeptide was produced. It therefore appears that the production of VanS D sensors from E. faecium BM4653 and BM4656, E. faecalis BM4654, and E. avium BM4655 in strains BM4458 and BM4459 did not restore inducible expression of the resistance genes, as indicated by the constitutivity of D,D-carboxypeptidase activity ( Fig. 6B and data not shown), confirming that the VanS D sensors from the former strains had lost their phosphatase activity.
In summary, in all the strains, acquired VanD-type resistance is due to constitutive production of peptidoglycan precursors, with the majority ending in D-Ala-D-Lac. The vanD operon located exclusively in the chromosome can be divided into two subtypes, vanD-1 and vanD-4, on the basis of sequence differences in VanD and VanY D . In almost all VanD-type strains, the susceptible pathway does not function due to an inactive D-Ala:D-Ala ligase resulting from various mutations in the chromosomal ddl gene. These mutations have most probably been acquired following those in the VanS D sensor or the VanR D regulator and circumvent the weakness of VanX D activity, allowing higher levels of antibiotic resistance.