Activation of the l,d-transpeptidation peptidoglycan cross-linking pathway by a metallo-d,d-carboxypeptidase in Enterococcus faecium

Authors

  • Emmanuelle Sacco,

    1. Centre de Recherche des Cordeliers, LRMA, Equipe 12, Université Pierre et Marie Curie – Paris6, UMR S 872, Paris, F-75006 France.
    2. Université Paris Descartes, UMR S 872, Paris, F-75006 France.
    3. INSERM, U872, Paris, F-75006 France.
    Search for more papers by this author
    • These authors contributed equally to the work.

  • Jean-Emmanuel Hugonnet,

    1. Centre de Recherche des Cordeliers, LRMA, Equipe 12, Université Pierre et Marie Curie – Paris6, UMR S 872, Paris, F-75006 France.
    2. Université Paris Descartes, UMR S 872, Paris, F-75006 France.
    3. INSERM, U872, Paris, F-75006 France.
    Search for more papers by this author
    • These authors contributed equally to the work.

  • Nathalie Josseaume,

    1. Centre de Recherche des Cordeliers, LRMA, Equipe 12, Université Pierre et Marie Curie – Paris6, UMR S 872, Paris, F-75006 France.
    2. Université Paris Descartes, UMR S 872, Paris, F-75006 France.
    3. INSERM, U872, Paris, F-75006 France.
    Search for more papers by this author
  • Julie Cremniter,

    1. Centre de Recherche des Cordeliers, LRMA, Equipe 12, Université Pierre et Marie Curie – Paris6, UMR S 872, Paris, F-75006 France.
    2. Université Paris Descartes, UMR S 872, Paris, F-75006 France.
    3. INSERM, U872, Paris, F-75006 France.
    Search for more papers by this author
  • Lionel Dubost,

    1. Muséum National d'Histoire Naturelle, Plateforme de Spectrométrie de Masse et de Protéomique du Muséum, Département Régulation Développement et Diversité Moléculaire, Paris, F-75005 France.
    2. Molécules de Communication et Adaptation des Micro-Organismes, FRE 3206 CNRS, Paris, F-75005 France.
    Search for more papers by this author
  • Arul Marie,

    1. Muséum National d'Histoire Naturelle, Plateforme de Spectrométrie de Masse et de Protéomique du Muséum, Département Régulation Développement et Diversité Moléculaire, Paris, F-75005 France.
    2. Molécules de Communication et Adaptation des Micro-Organismes, FRE 3206 CNRS, Paris, F-75005 France.
    Search for more papers by this author
  • Delphine Patin,

    1. Laboratoire des Enveloppes Bactériennes et Antibiotiques, Institut de Biochimie et Biophysique Moléculaire et Cellulaire, UMR 8619 CNRS, Univ Paris-Sud, 91405 Orsay, France.
    Search for more papers by this author
  • Didier Blanot,

    1. Laboratoire des Enveloppes Bactériennes et Antibiotiques, Institut de Biochimie et Biophysique Moléculaire et Cellulaire, UMR 8619 CNRS, Univ Paris-Sud, 91405 Orsay, France.
    Search for more papers by this author
  • Louis B. Rice,

    1. Medical and Research Services, Louis Stokes Cleveland VA Medical Center, Cleveland, OH 44106, USA.
    Search for more papers by this author
  • Jean-Luc Mainardi,

    1. Centre de Recherche des Cordeliers, LRMA, Equipe 12, Université Pierre et Marie Curie – Paris6, UMR S 872, Paris, F-75006 France.
    2. Université Paris Descartes, UMR S 872, Paris, F-75006 France.
    3. INSERM, U872, Paris, F-75006 France.
    4. AP-HP, Hôpital Européen Georges Pompidou, Paris, F-75015 France.
    Search for more papers by this author
  • Michel Arthur

    Corresponding author
    1. Centre de Recherche des Cordeliers, LRMA, Equipe 12, Université Pierre et Marie Curie – Paris6, UMR S 872, Paris, F-75006 France.
    2. Université Paris Descartes, UMR S 872, Paris, F-75006 France.
    3. INSERM, U872, Paris, F-75006 France.
      *E-mail michel.arthur@crc.jussieu.fr; Tel. (+33) 1 43 25 00; Fax (+33) 1 43 25 68 12.
    Search for more papers by this author

*E-mail michel.arthur@crc.jussieu.fr; Tel. (+33) 1 43 25 00; Fax (+33) 1 43 25 68 12.

Summary

Bypass of the penicillin-binding proteins by an l,d-transpeptidase (Ldtfm) confers cross-resistance to β-lactam and glycopeptide antibiotics in mutants of Enterococcus faecium selected in vitro. Ldtfm is produced by the parental strain D344S although it insignificantly contributes to peptidoglycan cross-linking as pentapeptide stems cannot be used as acyl donors by this enzyme. Here we show that production of the tetrapeptide substrate of Ldtfm is controlled by a two-component regulatory system (DdcRS) and a metallo-d,d-carboxypeptidase (DdcY). The locus was silent in D344S and its activation was due to amino acid substitutions in DdcS or DdcR that led to production of DdcY and hydrolysis of the C-terminal d-Ala residue of the cytoplasmic peptidoglycan precursor UDP-MurNAc-pentapeptide. The T161A and T161M substitutions affected a position of DdcS known to be essential for the phosphatase activity of related sensor kinases. Complete elimination of UDP-MurNAc-pentapeptide, which was required specifically for resistance to glycopeptides, involved substitutions in DdcY that increased the catalytic efficiency of the enzyme (E127K) and affected its interaction with the cell envelope (I14N). The ddc locus displays striking similarities with portions of the van vancomycin resistance gene clusters, suggesting possible routes of emergence of cross-resistance to glycopeptides and β-lactams in natural conditions.

Introduction

Peptidoglycan synthesis (Fig. 1A) is the target of the two main classes of drugs available to treat severe infections due to Gram-positive bacteria, the β-lactams and the glycopeptides (vancomycin and teicoplanin). The success of β-lactams, as the first and still the most frequently used class of antibacterial agents, has been attributed to their site of action, the periplasm, that prevents acquisition of resistance by decreased transport of the antibiotic through the cytoplasmic membrane (Zapun et al., 2008). In addition, the β-lactams often have multiple essential targets, the d,d-transpeptidase catalytic domains of penicillin-binding proteins (PBPs) (Sauvage et al., 2008). This limits the emergence of resistance because alteration of multiple genes may be required as found for the development of β-lactam resistance in clinical isolates and in vitro-selected mutants of Streptococcus pneumoniae (Zapun et al., 2008). Resistance mediated by a single low-affinity PBP is currently limited to a few bacterial species, methicillin resistance in Staphylococcus aureus (Pbp2a) and ampicillin resistance in enterococci (Pbp5) (Zapun et al., 2008). The success of β-lactams is also due to the fact that the essential cross-linking reaction of peptidoglycan polymerization was found to be catalysed by members of a unique class of transpeptidases, the active-site serine PBPs, in all bacteria in which the reaction was investigated (Sauvage et al., 2008). In a prospective study investigating potential PBP-independent mechanisms of β-lactam resistance, we started from Enterococcus faecium strain D344S, which had spontaneously lost pbp5, and obtained in four selection steps mutant M512, which is highly resistant to ampicillin (MIC > 2000 µg ml−1) (Fig. 1B) (Mainardi et al., 2000). The selection procedure was performed in a Δpbp5 background to avoid selection of mutations affecting low-affinity Pbp5 that can also result in ampicillin resistance (Mainardi et al., 2000). Analysis of peptidoglycan structure of mutant M512 indicated that the d,d-transpeptidase activity of the PBPs (Fig. 1A) had been bypassed by an l,d-transpeptidase that catalysed the formation of 3→3 cross-links (Fig. 1C) instead of the classical 4→3 cross-links (Mainardi et al., 2000). This observation has shown for the first time that the d,d-transpeptidase activity of PBPs can be dispensable to the cross-linking of a functional stress-bearing peptidoglycan layer (Mainardi et al., 2000).

Figure 1.

Activation of the l,d-transpeptidation pathway for peptidoglycan cross-linking in E. faecium mutants co-resistant to β-lactams and glycopeptides.
A. Peptidoglycan polymerization in the parental strain of Enterococcus faecium. The membrane steps lead to the synthesis of lipid II that consists of the complete disaccharide-peptide unit linked to undecaprenyl lipid carrier (black box) by a pyrophosphoryl bond. The side-chain of l-Lys at the third position of the stem pentapeptide is substituted by an aspartyl residue (β→ε peptide bond) and its α-carboxyl is partially amidated (d-iAsx). d-iGlu at the second position of the stem pentapeptide is connected to l-Lys3 by a γ→α peptide bond and its α-carboxyl is amidated (d-iGln). Following translocation of lipid II across the cytoplasmic membrane, the disaccharide-peptide subunit is polymerized by glycosyltransferases that form nascent glycan chains. The glycan chains are cross-linked by d,d-transpeptidases that cleave the d-Ala4-d-Ala5 peptide bond of an acyl donor and form a peptide bond between the carbonyl of d-Ala4 and the side-chain amine at the third position of an acyl acceptor (4→3 cross-links). Glycopeptide antibiotics bind to the peptidyl-d-Ala4-d-Ala5 extremity of peptidoglycan stem at the cell surface and block by steric hindrance both the transglycosylation and transpeptidation reactions. β-lactam antibiotics are structure analogues of the d-Ala4-d-Ala5 extremity of peptidoglycan precursors and act as suicide substrates of the d,d-transpeptidases leading to irreversible inactivation of the enzyme by acylation of their catalytic serine. The PBP family also includes unessential d,d-carboxypeptidases that hydrolyse d-Ala5 but do not form peptide bonds.
B. Representation of the eight sequential selection steps used to obtain mutant M9. The mutants were selected with ampicillin (Ap), vancomycin (Vm) and teicoplanin (Te) at the concentration (µg ml−1) indicated in the squares (Mainardi et al., 2002; Cremniter et al., 2006). The fourth, fifth and eighth selection steps led to the acquisition of mutations in the ddc locus, which comprises three genes, ddcR, ddcS and ddcY, flanked by two housekeeping genes (murA and fda). The locus is cryptic in wild-type E. faecium but can be activated by mutational alterations of ddcR and ddcS in response to selection by ampicillin. The resulting amino acid substitutions in DdcS or DdcY and the inferred functional alterations are indicated within the body of arrows. M4, M4a, M4b and M4c are three independent mutants selected from M3. The sequence of the ddc locus has been deposited in GenBank under accession number GQ217519.
C. Schematic representation of the l,d-transpeptidation pathway. The DdcY d,d-carboxypeptidase hydrolyses the C-terminal residue (d-Ala5) of UDP-MurNAc-pentapeptide in the cytoplasm. Tetrapeptide stems lacking d-Ala5 do not bind glycopeptides, thereby accounting for acquisition of high-level resistance to the antibiotics by mutant M9. Cross-linking of the tetrapeptide is performed by an l,d-transpeptidase (Ldtfm) that cleaves the l-Lys3-d-Ala4 peptide bond of an acyl donor and forms a peptide bond between the carbonyl of l-Lys3 and the side-chain amine at the third position of an acyl acceptor (3→3 cross-links). The DdcY d,d-carboxypeptidase insensitive to β-lactams produces the essential tetrapeptide donor of Ldtfm that bypasses the PBPs and confers high-level resistance to ampicillin as the active-site cysteine l,d-transpeptidase is not acylated by this antibiotic in contrast to the classical active-site serine d,d-transpeptidases. The amino acid sequence of DdcR and DdcS suggests that these two proteins form a classical two-component regulatory system catalysing phosphotransfer reactions between ATP, a histidyl residue of the sensor protein (DdcS-P), an aspartyl residue of the response regulator (DdcR-P), and, ultimately, water (inorganic phosphate).

Identification, biochemical and structural analyses of the transpeptidase for the synthesis of 3→3 cross-links in E. faecium M512 showed that the enzyme is unrelated to the PBPs (Mainardi et al., 2005; Biarrotte-Sorin et al., 2006). The l,d-transpeptidase (Ldtfm) has been characterized as the first member of a new class of active-site cysteine peptidases (Mainardi et al., 2005; Biarrotte-Sorin et al., 2006). Recently, the protein family was found to include enzymes for attachment of lipoproteins to peptidoglycan (Magnet et al., 2007b) in addition to peptidoglycan cross-linking enzymes (Magnet et al., 2007a; 2008) that are sporadically distributed in several Gram-positive species (Mainardi et al., 2005), including Mycobacterium tuberculosis (Lavollay et al., 2008) and Corynebacterium jeikeium (Lavollay et al., 2009), which were lately investigated. In stationary phase culture of M. tuberculosis, 80% of the cross-links are of the 3→3 type, revealing for the first time that PBPs may have a minor role in transpeptidation in a wild-type bacterial strain (Lavollay et al., 2008). The ldtMt1 gene, which encodes one of the five Ldtfm homologues of M. tuberculosis, was shown to be upregulated in response to starvation (Betts et al., 2002; Lavollay et al., 2008). This observation suggests that the l,d-transpeptidase participates in the adaptative response that generates ‘dormant’ (non-replicative) forms of the bacilli, which are particularly resistant to antibiotics (Betts et al., 2002; Lavollay et al., 2008). As Ldtfm from E. faecium, LdtMt1 is functional in an in vitro model of peptidoglycan cross-linking, which is based on the use of peptidoglycan cell-wall fragments as substrates (Lavollay et al., 2008). Both enzymes are inactivated by a single class of β-lactams, the carbapenems, that act as suicide substrates in a nucleophilic attack of their β-lactam ring by the active-site cysteine (Mainardi et al., 2007; Lavollay et al., 2008). Together, these results led to the proposal that LdtMt1 and carbapenems could respectively represent a target and a drug for the eradication of the ‘dormant’ forms of M. tuberculosis (Lavollay et al., 2008). In agreement, a carbapenem, meropenem, in combination with a β-lactamase inhibitor, clavulanic acid, has been shown to be highly active against non-replicative forms of M. tuberculosis (Hugonnet and Blanchard, 2007; Hugonnet et al., 2009).

The molecular switches that control the relative contributions of PBPs and l,d-transpeptidases are unknown but appear to differ in M. tuberculosis and E. faecium. As described above, the ldtMt1 gene is induced in stationary phase cultures of M. tuberculosis (Betts et al., 2002; Lavollay et al., 2008) whereas ldtfm is constitutively expressed in E. faecium (Mainardi et al., 2005). In the latter bacteria, the contribution of Ldtfm to peptidoglycan cross-linking appears to be controlled by the availability of the substrate (Mainardi et al., 2002; 2005). Ldtfm cleaves the l-Lys3-d-Ala4 peptide bond of tetrapeptide stems that act as acyl donors in the cross-linking reaction but is devoid of any activity for stem pentapeptides (Fig. 1C) (Mainardi et al., 2005). The specificity of Ldtfm provides a rational basis for substrate-level control of its activity because the peptidoglycan subunits, as assembled in the cytoplasm, contain a stem pentapeptide produced by ligation of the dipeptide d-Ala-d-Ala onto a tripeptide stem (Barreteau et al., 2008). This hypothesis has originally been proposed to account for the early observation of increased d,d-carboxypeptidase activity and UDP-MurNAc-tetrapeptide content in crude extracts of the ampicillin-resistant mutant M512 (Mainardi et al., 2002). To gain insight into this molecular switch, we have identified the enzyme that generates the substrate of Ldtfm, characterized its activity and analysed the modifications responsible for emergence of high-level β-lactam and glycopeptide resistance.

Results

Identification of the d,d-carboxypeptidase for production of the substrate of Ldtfm

Two families of structurally unrelated d,d-carboxypeptidases participate in peptidoglycan metabolism (Sauvage et al., 2008). The first family includes low-molecular weight PBPs that are inhibited by β-lactams (Sauvage et al., 2008). The second includes metallo-enzymes of the VanY superfamily that are not inhibited by these drugs (Leyh-Bouille et al., 1970; Dideberg et al., 1982; Arthur et al., 1998; Reynolds, 1998; Reynolds et al., 2001; Depardieu et al., 2007). Extracts from parental strain D344S and mutant M3 contained similar low levels of d,d-carboxypeptidase activity that was fully inhibited by ampicillin and should therefore correspond to the activity of PBPs (Table 1). The d,d-carboxypeptidase activity was increased c. fivefold in M512 and no inhibition was detected, indicating the production of a different enzyme. Because the d,d-carboxypeptidase activity was not inhibited by ampicillin, we searched for homologues of the VanY metallo-d,d-carboxypeptidase (Arthur et al., 1998) in the partially sequenced genome of E. faecium and identified one, which was designated DdcY (30% identity over 189 residues). A truncated version of the protein, which lacked the putative anchor, was produced in Escherichia coli and purified by affinity, anion exchange and size exclusion chromatographies. The purified protein, designated DdcY6H, was incubated with the cytoplasmic peptidoglycan precursor UDP-MurNAc-pentapeptide and the products of the reaction were analysed by mass spectrometry (Fig. S1). The observed monoisotopic mass of the substrate (1149.38) matched the calculated mass of UDP-MurNAc-pentapeptide (monoisotopic mass of 1149.38). Incubation of UDP-MurNAc-pentapeptide with the purified protein led to new peaks (deduced mass of 1078.37) that matched the monoisotopic mass of UDP-MurNAc-tetrapeptide (1078.36) generated by hydrolysis of the d-Ala4-d-Ala5 peptide bond of UDP-MurNAc-pentapeptide. Tandem mass spectrometry confirmed the structure of UDP-MurNAc-pentapeptide and UDP-MurNAc-tetrapeptide (data not shown). The spectrophotometric assay indicated that the enzyme released stoichiometric amounts of d-Ala from UDP-MurNAc-pentapeptide (data not shown). The d,d-carboxypeptidase activity of DdcY was not inhibited by ampicillin (Table S1). These data indicated that DdcY6H displayed all the characteristics required for production of the substrate of the Ldtfml,d-transpeptidase.

Table 1.  Characteristics of E. faecium D344S and derivatives selected on ampicillin.
Strain (substitutions in DdcS)MICsad,d-carboxypeptidase activityb
ApVmTeWithout ApWith Ap
  • a. 

    MICs (µg ml−1) of ampicillin (Ap), vancomycin (Vm) and teicoplanin (Te).

  • b. 

    The d,d-carboxypeptidase activity (nmol min−1 mg−1) in membrane fractions was defined as the number of nmol of d-Ala released from pentapeptide l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala (7.5 mM) per min and per mg of protein in the extracts. The activity was determined in the absence (without Ap) or in the presence (with Ap) of ampicillin (20 µg ml−1). d-Ala was assayed using d-aminoacid oxidase coupled to peroxidase.

  • ND, not determined.

D344S(None)0.0640.2513 ± 2< 5
M1(None)0.540.25NDND
M2(None)140.25NDND
M3(None)240.517 ± 6< 5
M512(T161A)> 200080.570 ± 1279 ± 11

To establish that production of DdcY accounted for the d,d-carboxypeptidase activity specifically detected in mutant M512, an antiserum was raised against the protein. Western blot analysis of crude extracts revealed that the protein was produced by mutant M512 (Fig. 2A, lane 3) but not by parental strain D344S (lane 1) or by mutants M1 (data not shown), M2 (data not shown) and M3 (lane 2). Together, these data indicated that the ddcY gene was cryptic in D344S and that its expression was activated at the fourth selection step that produced M512 from M3 (Table 1 and Fig. 1B).

Figure 2.

Western blot analysis of crude extracts from D344S and derivatives resistant to ampicillin and glycopeptides. DdcY (A), DdcS (B) and Ldtfm (C) were detected with rabbit antisera in crude lysates (1.5 µg of protein) of E. faecium D344S and of mutants M3 to M9. The masses deduced from the sequence of DdcY, DdcS and Ldtfm were 25 023, 43 015 and 55 042 Da.

Substrate specificity of DdcY

Dipeptide d-Ala-d-Ala was not a substrate of DdcY6H whereas the mono-acetylated tripeptide and the pentapeptide were hydrolysed with similar catalytic efficiencies (Table 2). Similar turnover numbers were obtained for the pentapeptide and UDP-MurNAc-pentapeptide. Thus, substrate recognition appeared to mainly involve the C-terminal l-Lys-d-Ala-d-Ala extremity of peptidoglycan precursors. Comparison of the mono- and di-acetylated tripeptides revealed preferential hydrolysis of substrates containing a free amine on the ε carbon of l-Lys. This observation suggests that the activity of DdcY is optimized for hydrolysis of cytoplasmic peptidoglycan precursors prior to the addition of d-Asp onto l-Lys of pentapeptide stems (Fig. 1C). In agreement, UDP-MurNAc-tetrapeptide was detected in the cytoplasm of M512 (see below), thereby suggesting a cytoplasmic location of DdcY which may allow hydrolysis of the nucleotide precursor.

Table 2.  Substrate specificity of DdcY6H and impact of the E127K amino acid substitution.
 Turnover (min−1)aRelative
  • a. 

    The catalytic efficiencies of DdcY6H and DdcY6HE127K were compared by determining d-Ala released from various substrates at 37°C using d-aminoacid oxidase coupled to peroxidase as an indicator reaction.

  • b. 

    Because the turnover numbers of DdcY were proportional to the substrate concentration up to the highest achievable concentrations, Km values could not be determined. Ac-l-Lys-d-Ala-d-Ala, Nα-acetyl-l-Lys-d-Ala-d-Ala; Ac2-l-Lys-d-Ala-d-Ala, Nα,Nε-diacetyl-l-Lys-d-Ala-d-Ala; Pentapeptide, l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala; UDP-MurNAc-pentapeptide, UDP-N-acetylmuramoyl-l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala.

  • NA, not applicable.

Substrate (concentration)bDdcY6HDdcY6HE127Kturnover
Ac-l-Lys-d-Ala-d-Ala (18 mM)440 ± 501280 ± 1702.9
Ac2-l-Lys-d-Ala-d-Ala (18 mM)90 ± 14330 ± 303.6
Pentapeptide (18 mM)310 ± 30950 ± 403.0
UDP-MurNAc-pentapeptide (9 mM)150 ± 20390 ± 302.5
d-Ala-d-Ala (18 mM)< 5< 5NA

Activation of the ddcY gene in E. faecium M512

Analysis of the environment of ddcY revealed two genes encoding a putative two-component regulatory system (Fig. 1B). The putative response regulator, DdcR, consisted of an N-terminal effector domain and a C-terminal DNA-binding domain belonging to the OmpR-PhoB subclass (Itou and Tanaka, 2001). The putative sensor, DdcS, contained a C-terminal kinase domain. The N-terminal region of DdcS harboured two clusters of hydrophobic amino acids that could correspond to trans-membrane segments delineating a periplasmic region of 26 residues. As usually found in sensor kinases that recognize different signals, this region of the protein was unrelated to known proteins from databases. The ddcY gene was followed by murA encoding UDP-N-acetylglucosamine 1-carboxyvinyltransferase, an enzyme dedicated to peptidoglycan synthesis, which catalyses the first reaction for the formation of UDP-MurNAc from UDP-GlcNAc. The ddcR gene was preceded by fda encoding the glycolytic enzyme fructose 1,6-biphosphate aldolase (class II). Thus, the sequence comprising ddcR, ddcS and ddcY, referred to as the ddc locus, was flanked by two housekeeping genes.

Nucleotide sequencing of the entire ddc locus revealed that M512 differs from D344S, M1, M2 and M3 by a single mutation in ddcS (Fig. 1B). The point mutation led to the substitution of Thr at position 161 by Ala (T161A). As two-component regulatory systems often control their own expression (Arthur and Quintiliani, 2001), we purified a soluble fragment of DdcS in order to raise an antiserum against this protein. Western blot analysis (Fig. 2B) revealed that DdcS and DdcY were co-regulated as the proteins were only detected in M512 and in the following mutants (M6 to M9). The ddc locus was cryptic in the parental strain D344S and its activation occurred at the fourth selection step that led to the acquisition of the T161A substitution in the kinase domain of DdcS (Fig. 1B). Introduction of plasmid pJC25 expressing the ddcS gene of M512 into M3 resulted in ampicillin resistance and production of the DdcY d,d-carboxypeptidase (Fig. S2). Thus, the T161A substitution in DdcS was responsible for activation of the ddc locus.

Additional mutants derived from M3 were selected on ampicillin to determine whether the selection procedure reproducibly resulted in activation of the ddc locus following alteration of the DdcRS two-component regulatory system (Fig. 1B). Characterization of four derivatives of M3, in addition to M512, revealed that the ddc locus was similarly activated based on determination of the d,d-carboxypeptidase activity insensitive to inhibition by ampicillin in membrane fractions and direct detection of the DdcY and DdcS proteins by Western blot analysis (Fig. 3). Sequencing of the ddc locus revealed in each case a single mutation that had led to substitutions in DdcS (L100I, T161M and L163S) or DdcR (M55I).

Figure 3.

Properties of derivatives of M3 selected on ampicillin. Western blot analysis was performed on crude lysates (1.5 µg of protein) with rabbit antisera raised against DdcY (A), DdcS (B) and Ldtfm (C). D. d,d-carboxypeptidase activity in membrane fractions. The d,d-carboxypeptidase activity (nmol min−1 mg−1) was defined as the number of nmol of d-Ala released from pentapeptide l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala (7.5 mM) per min and per mg of protein in the membrane fractions. The activity was determined in the absence (white bars) or in the presence (solid bars) of ampicillin (20 µg ml−1). d-Ala was assayed using d-aminoacid oxidase coupled to peroxidase.

Selection of amino acid substitutions in DdcY by glycopeptides

Enterococcus faecium M512 is resistant to ampicillin but remains susceptible to glycopeptides. Four additional selection steps using glycopeptides yielded mutants M6, M7, M8 and M9 that were increasingly resistant to vancomycin and teicoplanin (Fig. 1B) (Cremniter et al., 2006). Sequencing of the ddc locus of M9 revealed two mutations in the ddcY gene that were not present in M512. The first mutation, which introduced a Glu to Lys substitution at position 127 of DdcY, was present in M6, M7, M8 and M9 but not in M512. The second mutation caused a substitution (I14N) in the putative membrane anchor of DdcY and was only present in M9. We concluded that the E127K and I14N substitutions in DdcY had been acquired at the fifth and eight steps that led to selection of M6 from M512 and of M9 from M8 respectively (Fig. 1B).

A soluble fragment of DdcY containing the E127K substitution, designated DdcY6HE127K, was produced in E. coli and purified to evaluate the impact of the E127K substitution on the in vitro enzyme activity for UDP-MurNAc-pentapeptide and several different peptidoglycan precursor analogues (Table 2). The DdcY6HE127K protein was identical to DdcY6H described above except for the amino acid substitution at position 127. As the E127K substitution led to similar (2.5–3.6-fold) increases in the turnover numbers of the enzyme for the different substrates, the substitution moderately increased the catalytic efficiency of the enzyme but did not alter its specificity. As expected, DdcY6HE127K was not inhibited by β-lactam antibiotics (Table S1). DdcYE127K contributed to β-lactam and glycopeptide resistance as introduction of plasmid pES1 encoding this protein into M3 led to increases in the MIC of ampicillin (500-fold), vancomycin (1.5-fold) and teicoplanin (threefold). The recombinant plasmid did not confer ampicillin or glycopeptide resistance to parental strain D344S indicating that unknown mutations selected at the first and second steps (mutants M1 and M2) were essential for expression of resistance to these antibiotics.

Cytoplasmic precursors were extracted from exponentially growing cultures, identified by mass spectrometry, and sequenced by tandem mass spectrometry (Table 3). UDP-MurNAc-pentapeptide and UDP-MurNAc-tetrapeptide were detected in various proportions in the mutants revealing increased d,d-carboxypeptidase activity at different selection steps. As expected, the T161A substitution in DdcS, which led to activation of the ddc locus at the fourth selection step, was associated with accumulation of UDP-MurNAc-tetrapeptide (49.3 ± 7.5% for M512) that was only present in minor amounts in M3 (2.8 ± 0.8%). Acquisition of the E127K substitution in DdcY at the following step led to a further increase in the proportion of UDP-MurNAc-tetrapeptide from 49.3 ± 7.5% (M512) to 66.1 ± 2.1% (M6). Although no additional mutation was detected in the ddc locus of mutants M7 and M8, the proportion of UDP-MurNAc-tetrapeptide increased stepwise at the sixth and seventh selection steps to reach 98.3 ± 2.1% (M7) and > 99% (M8) respectively. Thus, increased production of tetrapeptide at the corresponding selection steps did not involve any modification of DdcY. These results indicated that serial selection for ampicillin and glycopeptides resistance led to the gradual replacement of the UDP-MurNAc-pentapeptide pool by UDP-MurNAc-tetrapeptide, a switch that provided the essential substrate of the l,d-transpeptidase and eliminated precursors containing the target of glycopeptides.

Table 3.  Characteristics of E. faecium M512 and derivatives selected on glycopeptides.
Strain (substitution in DdcY)MICsad,d-carboxypeptidase activitybPrecursorc
ApVmTeWithout ApWith ApPentaTetra
  • a. 

    MICs (µg ml−1) of ampicillin (Ap), vancomycin (Vm) and teicoplanin (Te).

  • b. 

    The d,d-carboxypeptidase activity (nmol min−1 mg−1) in membrane fractions was defined as the number of nmol of d-Ala released from pentapeptide l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala (7.5 mM) per min and per mg of protein in the extracts. The activity was determined in the absence (without Ap) or in the presence (with Ap) of ampicillin (20 µg ml−1). d-Ala was assayed using d-aminoacid oxidase coupled to peroxidase. The d,d-carboxypeptidase activity was c. 8- to 11-fold lower in the cytoplasmic fractions than in the membrane fractions and this ratio was similar in all the mutants (data not shown).

  • c. 

    Cytoplasmic peptidoglycan precursors were extracted from exponentially growing bacteria with formic acid, purified by rp-HPLC, identified by mass spectrometry and sequenced by tandem mass spectrometry. The relative abundance (%) of UDP-MurNAc-pentapeptide (Penta) and UDP-MurNAc-tetrapeptide (Tetra) was calculated by integration of absorbance at 262 nm (mean ± standard deviation from a minimum of three independent cultures).

M3(None)240.517 ± 6< 597.2 ± 0.82.8 ± 0.8
M512(None)> 200080.570 ± 1279 ± 1151.7 ± 7.549.3 ± 7.5
M6(E127K)> 200016159 ± 556 ± 633.9 ± 2.166.1 ± 2.1
M7(E127K)> 200032272 ± 1066 ± 101.7 ± 2.198.3 ± 2.1
M8(E127K)> 2000250457 ± 662 ± 13< 1> 99
M9(E127K and I14N)> 20001000100076 ± 882 ± 4< 1> 99

Discussion

Enterococcus faecium has the dual capacity to synthesize 4→3 and 3→3 peptidoglycan cross-links depending on the relative contribution of the d,d-transpeptidase activity of the classical PBPs and of the l,d-transpeptidase activity of Ldtfm respectively (Mainardi et al., 2002). The balance between the two modes of transpeptidation was found to be altered under the selective pressure of ampicillin (Mainardi et al., 2002) because this antibiotic inhibits the PBPs but not Ldtfm (Mainardi et al., 2005; 2007). Strikingly, the mutations responsible for the emergence of ampicillin resistance did not affect the structure or the level of synthesis of Ldtfm but the production of its substrate (Figs 1 and 2) (Mainardi et al., 2005). Ldtfm specifically uses acyl donors containing a tetrapeptide stem (Mainardi et al., 2005) whereas peptidoglycan cross-linking by PBPs requires a pentapeptide donor (Sauvage et al., 2008) (Fig. 1). Thus, a switch between productions of peptidoglycan precursors containing the specific acyl donor stems of the PBPs (pentapeptide) and of Ldtfm (tetrapeptide) controls the relative contribution of the two types of cross-linking activities to peptidoglycan synthesis. In this report, we have identified the enzyme, a d,d-carboxypeptidase designated DdcY, that controls this switch and shown that activation of the l,d-transpeptidation pathway involves mutations that turn on production of the enzyme, affect its activity, and probably modify its interaction with membrane components.

Selection of mutations that increased the level of ampicillin resistance in independent mutants derived from M3 led to amino acid substitutions in the regulatory proteins DdcR or DdcS (Figs 1B and 3). Among these substitutions, T161A and T161M were located four residues downstream from His157, the putative autophosphorylation site of DdcS. Substitutions affecting the homologous Thr residues of the related sensor kinases VanSB (T237K and T237M), EnvZ (T247K) and PhoR (T220N) selectively inactivate the phosphatase but not the kinase activity of these bi-functional enzymes (Aiba et al., 1989; Tokishita et al., 1992; van Bambeke et al., 1999; Dutta et al., 2000; Arthur and Quintiliani, 2001). In the corresponding mutants, the cognate regulators are activated by phosphorylation irrespective to the signalling status leading to constitutive expression of the target genes. Activation of the ddcY gene may therefore result from the release of a negative control mediated by the phosphatase activity of DdcS.

Mutant M512 was resistant to ampicillin only and four additional selection steps were required for acquisition of high-level glycopeptide resistance (Fig. 1B). Here, we show that each of the four selection steps affected either or both the sequence of the DdcY d,d-carboxypeptidase and the extent of elimination of UDP-MurNAc-pentapeptide (Table 3). As previously discussed (Arthur et al., 1996; Cremniter et al., 2006), complete elimination of pentapeptide stems containing the d-Ala4-d-Ala5 target of glycopeptides was required for resistance to these drugs, whereas a supply of tetrapeptide stems in sufficient amount for peptidoglycan cross-linking by Ldtfm is sufficient for resistance to ampicillin. The E127K substitution detected in mutant M6 increased the catalytic activity of the enzyme in the absence of any modification of its substrate specificity (Table 2). The modest increase in the catalytic efficacy resulted in a modest increase in the pool of UDP-MurNAc-tetrapeptide (Table 3). The following two selection steps led also to increases in the pool of UDP-MurNAc-tetrapeptide (Table 3) although the corresponding mutations did not map in the ddc locus. Strikingly, the mutations selected at these steps did not increase the production of DdcY (Fig. 2) or the d,d-carboxypeptidase activity present in membrane fractions (Table 3). The missense mutation selected in the last step introduced a polar amino acid in the putative membrane anchor of DdcY (substitution I14N) and could therefore affect the localization of the protein. However, comparison of membrane and cytoplasmic fractions by Western blot analysis and determination of d,d-carboxypeptidase activity did not reveal any difference between mutants M8 and M9 (data not shown). Thus, alterations of the catalytic activity of DdcY and of its localization did not appear to account for the development of glycopeptide resistance in the last three selection steps. The corresponding mutations may therefore affect interactions between DdcY with partner proteins in the membrane that would optimize presentation of the stem pentapeptide substrate to the DdcY active site.

The threefold increase observed for the activity of DdcY6HE127K with respect to DdcY6H (Table 2) was not observed for the comparison of the d,d-carboxypeptidase activity in membrane fractions from M512 and M6 despite the fact that these strains produce wild-type DdcY and DdcYE127K respectively (Table 3). The basis for this discrepancy is unknown. One possible explanation is that the E127K substitution had opposite effects involving decreased stability of the protein in membrane fractions and increased activity. It is also possible that the impact of the E127K substitution on the hydrolysis of the C-terminal d-Ala residue of UDP-MurNAc-pentapeptide is less important for the complete enzyme than for the purified protein that contains a 6× His tag instead of the putative membrane anchor. The membrane anchor of DdcY may be relevant to the in vivo activity of the enzyme as the reaction leading to synthesis of lipid I from UDP-MurNAc-pentapeptide and undecaprenyl-phosphate by the transferase MraY is reversible (Bouhss et al., 2008). Thus, the formation of UDP-MurNAc-tetrapeptide from UDP-MurNAc-pentapeptide by DdcY may involve in part the hydrolysis of the C-terminal residue of lipid I.

In E. faecium D344S and its derivatives, the ddc locus was flanked by two housekeeping genes (fda and murA) (Fig. 1B). PCR analyses showed that the ddcR, ddcS and ddcY genes were present in only two representatives of a set of eight unrelated clinical isolates of E. faecium (Fig. 4). In the remaining six isolates, fda and murA were adjacent indicating that E. faecium isolates have diverged by the gain or loss of the ddc locus. Western blot analysis showed that Ldtfm was produced by all isolates. The DdcS and DdcY proteins were not detected as the ddc locus was cryptic in two out of eight isolates and absent from the remaining isolates. In these clinical isolates, Ldtfm may contribute to a minor extent to peptidoglycan cross-linking in the absence of ampicillin because small amounts of the tetrapeptide substrate can be formed extracellularly by the β-lactam sensitive d,d-carboxypeptidase activity of low molecular weight PBPs, as previously discussed (Mainardi et al., 2008).

Figure 4.

Detection of the ddc locus in clinical isolates of E. faecium.
A. Characteristics of the E. faecium strains.
B. Map of the fda-murA region in the chromosome of strains D344S and H80721.
C. Sequence of the fda-murA region in the chromosome of H80721. The point of divergence between the sequences from D344S and H80721 is located immediately downstream from the translation termination codon of fda (TAA). In H80721, this codon is followed by a 63-bp element which is absent from D344S. In D344S, the ddc locus is inserted 101 bp upstream from the murA translation initiation codon. These 101 bp and the sequenced portion of murA are identical in D344S and H80721.
D. Western blot analysis of protein extracts with polyclonal antibodies raised against DdcS, DdcY and Ldtfm.

Enterococci resistant to glycopeptide by production of d-lactate-ending precursors are at risk of developing resistance to ampicillin by l,d-transpeptidation as most van gene clusters encode a d,d-carboxypeptidase that can generate the essential tetrapeptide donor of Ldtfm. In addition, constitutive expression of van gene clusters has been detected in various clinical isolates. Thus, the striking similarities detected between the ddc locus and portions of the van clusters suggest possible routes of emergence of cross-resistance to glycopeptides and β-lactams in natural conditions.

Experimental procedures

Bacterial strains, growth conditions and selection of mutants

All cultures were performed at 37°C in brain heart infusion (BHI) agar or broth (Difco Laboratories). Serial selection with ampicillin was previously used to obtain mutants M1, M2, M3, M4 and M512 from E. faecium D344S, which lacks the pbp5 gene encoding low-affinity PBP5 (Mainardi et al., 2000; Mainardi et al., 2002). Mutants M6 to M9 derive from M512 following serial selection with glycopeptides (Cremniter et al., 2006). In this study, mutants M4a, M4b and M4c were obtained by plating mutant M3 on agar containing ampicillin (16 µg ml−1). Mutants appeared after 5 days of incubation with a frequency of about 10−9. No mutagenic treatment was used for these selection procedures. Minimal inhibitory concentrations of ampicillin (Bristol-Myers, Paris, France), vancomycin (Merck, Lyon, France) and teicoplanin (Aventis, Paris, France) were determined by the agar dilution method after 48 h of incubation (Mainardi et al., 2002).

Production and purification of DdcY6H

A soluble fragment of DdcY lacking the putative membrane anchor of the protein (residues 26–220) was produced as a fusion protein with an N-terminal 6 histidines tag in E. coli and purified by four chromatographic steps as described in Supporting information.

Substrate of the DdcY d,d -carboxypeptidase

Na,Nε-diacetyl-l-Lys-d-Ala-d-Ala (Ac2-l-Lys-d-Ala-d-Ala), Na-acetyl-l-Lys-d-Ala-d-Ala (Ac-l-Lys-d-Ala-d-Ala), l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala (pentapeptide) and d-Ala-d-Ala were purchased from Sigma. UDP-N-acetyl-muramoyl-l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala (UDP-MurNAc-pentapeptide) was obtained from UDP-MurNAc by enzymatic synthesis using the Mur ligases as previously described (Babic et al., 2007).

Determination of the d,d-carboxypeptidase activity of recombinant DdcY6H and DdcY6HE127K

The standard assay contained 50 mM Tris-HCl (pH 7.5) and 1 mM ZnCl2. The release of d-Ala from peptidoglycan precursors and analogues by the d,d-carboxypeptidases was determined spectrophotometrically using d-aminoacid oxidase coupled to peroxidase as previously described (Arthur et al., 1998). The assay (120 µl) was carried out in microcuvettes and the absorbance was recorded at 37°C at 460 nm using a Carry 100 Bio spectrophotometer (Varian).

To determine turnover numbers, enzyme dilutions were prepared in 50 mM Tris-HCl (pH 7.5) containing BSA (0.5 mg ml−1). Enzyme was incubated with Ac2-l-Lys-d-Ala-d-Ala (18 mM), Ac-l-Lys-d-Ala-d-Ala (18 mM), pentapeptide (18 mM), UDP-MurNAc-pentapeptide (9 mM) and the dipeptide d-Ala-d-Ala (18 mM) in 50 mM Tris-HCl (pH 7.5) containing ZnCl2 (1 mM) at 37°C. Aliquots were withdrawn at 0, 10, 18, 30, 45 and 60 min to determine d-Ala released from the different substrates using the d-aminoacid oxidase assay (Arthur et al., 1998). The turnover number was defined as the ratio of the initial velocity of the reaction to the enzyme concentration.

To determine inhibition of the d,d-carboxypeptidase activity by β-lactams, DdcY6H (258 nM) and DdcY6HE127K (89.5 nM) were incubated in 50 mM Tris-HCl (pH 7.5) containing ZnCl2 (1 mM) with Ac-l-Lys-d-Ala-d-Ala (18 mM) and various concentrations (0, 20, 60, 200, 600 and 2000 µg ml−1) of ampicillin (Bristol-Myers), ceftriaxone (Roche) and imipenem (Merck-Sharp and Dohme). The amount of d-Ala released from Ac-l-Lys-d-Ala-d-Ala was determined after 30 min of incubation at 37°C using the d-aminoacid oxidase assay.

Determination of d,d-carboxypeptidase activity in enterococcal extracts

Bacteria were cultured in 100 ml of BHI broth to an OD600 of 0.7 (200 ml at an OD600 of 0.35 for mutants M8 and M9), harvested and washed by centrifugation (5000 g for 10 min at 4°C). Bacteria were resuspended in 2.2 ml of 0.1 M phosphate buffer (pH 7.0) containing lysozyme (1 mg ml−1), incubated for 1 h at 37°C, and lysed by sonication. The lysate was centrifuged (5000 g for 10 min at 4°C) to remove cell debris and membranes were collected by centrifugation at 100 000 g for 45 min at 4°C. The supernatant was collected (cytoplasmic fraction) and the pellet was resuspended in 0.2 ml of 0.1 M phosphate buffer (pH 7.0) (membrane fraction). Protein concentration was estimated by the Bradford method using the Bio-Rad protein assay with BSA as a standard. d,d-carboxypeptidase activity was determined at 37°C in 50 mM Tris-HCl (pH 7.5) containing 1 mM ZnCl2 and 7.5 mM l-Ala-d-iGlu-l-Lys-d-Ala-d-Ala. d-Ala released by the d,d-carboxypeptidases was determined spectrophotometrically using the d-aminoacid oxidase assay. In order to compare the d,d-carboxypeptidase activity of membrane extracts, the activity was expressed as the number of nmol of d-Ala released from the pentapeptide per min and per mg of protein in the extracts.

Preparation and analysis of cytoplasmic peptidoglycan precursors

Bacteria were grown to an OD650 of 0.7 in 200 ml of BHI broth (mutants M512, M6 and M7) or to an OD650 of 0.35 in 400 ml of BHI broth (mutants M8 and M9). Bacteria were treated with bacitracin (200 µg ml−1) for 5 min to accumulate the precursors. Extraction of the precursors with ice-cold formic acid and rp-HPLC analysis were performed as previously described (Mainardi et al., 2002). The relative abundance of the UDP-MurNAc-peptides was estimated by the percentage of the integrated area of peaks detected by the absorbance at 262 nm. Products isolated by rp-HPLC were analysed by mass spectrometry (MS) using an electrospray time-of-flight mass spectrometer operating in the positive mode (Qstar Pulsar I, Applied Biosystem, Courtaboeuf, France). Tandem mass spectrometry (MS-MS) was performed with nitrogen as the collision gas (Bouhss et al., 2002).

Western blot analysis

Anti DdcY, DdcS and Ldtfm antibodies were obtained in New-Zealand rabbits and used for Western blot analyses as described in Supporting information.

Expression of the ddcS gene of mutant M512 in E. faecium

The ddcS gene of M512 was amplified by PCR using primers Sf (5′-GTGAGCTCAACTGTTTGGGGAGTGG-3′) and Sr (5′-GGACTAGTCCTACGAAAGCGGCAAATC-3′), which contained SacI and SpeI restriction sites respectively (underlined). The amplicon was digested with SacI plus SpeI and cloned between the SacI and XbaI sites of the shuttle expression vector pNJ3, a derivative of pNJ2 (Arbeloa et al., 2004) harbouring the aac(6′)-aph(2′) gene conferring resistance to kanamycin and gentamicin instead of the cat chloramphenicol resistance gene (our laboratory collection). The resulting plasmid, pJC25, was introduced by electroporation into JH2-2:Tn916 and subsequently transferred by conjugation to E. faecium M3 as previously described (Arbeloa et al., 2004).

Expression of the gene encoding DdcYE127K in D344S and M3

The ddcY gene of M6, encoding DdcYE127K, was amplified by PCR, cloned into the expression vector pJEH11, and the resulting plasmid (pES1) was introduced into D344S and M3 by electroporation as described in supplementary experimental procedures.

Detection of the ddc locus in clinical isolates of E. faecium by PCR

Genomic DNA of strains D344S (Mainardi et al., 2000), BM4107 (Leclercq et al., 1989), D3 (Rybkine et al., 1998), AR9 (Rybkine et al., 1998), 7090 (Rybkine et al., 1998), H80721 (Zorzi et al., 1996), 6885 (Rybkine et al., 1998) and 34 (Klare et al., 1992) was prepared as previously described (Cremniter et al., 2006). PCR amplification was performed with primers ddc31 (5′-TCAATGGGTATCTCAAAAG-3′) and ddc32 (5′-TGGAAAAATGGACAGTTGC-3′) that were complementary to the distal portion of the fda gene and the proximal portion murA.

Acknowledgements

This work was supported by the Fondation pour la Recherche Médicale (Equipe FRM 2006; DEQ200661107918) and the National Institute of Allergy and Infectious Diseases (Grant RO1 AI45626). Jean-Emmanuel Hugonnet was the recipient of fellowship from the FRM (FDT 20080914133). This manuscript is dedicated to the memory of Jean-Marc Reyrat.

Ancillary