Functional conservation of the lipid II biosynthesis pathway in the cell wall-less bacteria Chlamydia and Wolbachia: why is lipid II needed?

Authors


*E-mail pfarr@microbiology-bonn.de; Tel. (+49) 228 28711207; Fax (+49) 228 28714330.

Summary

Cell division and cell wall biosynthesis in prokaryotes are driven by partially overlapping multiprotein machineries whose activities are tightly controlled and co-ordinated. So far, a number of protein components have been identified and acknowledged as essential for both fundamental cellular processes. Genes for enzymes of both machineries have been found in the genomes of the cell wall-less genera Chlamydia and Wolbachia, raising questions as to the functionality of the lipid II biosynthesis pathway and reasons for its conservation. We provide evidence on three levels that the lipid II biosynthesis pathway is indeed functional and essential in both genera: (i) fosfomycin, an inhibitor of MurA, catalysing the initial reaction in lipid II biosynthesis, has a detrimental effect on growth of Wolbachia cells; (ii) isolated cytoplasmic membranes from Wolbachia synthesize lipid II ex vivo; and (iii) recombinant MraY and MurG from Chlamydia and Wolbachia exhibit in vitro activity, synthesizing lipid I and lipid II respectively. We discuss the hypothesis that the necessity for maintaining lipid II biosynthesis in cell wall-lacking bacteria reflects an essential role of the precursor in prokaryotic cell division. Our results also indicate that the lipid II pathway may be exploited as an antibacterial target for chlamydial and filarial infections.

Introduction

Cell wall biosynthesis and cell division in prokaryotes are both fundamental cell biological processes that are known to be driven by multiprotein machineries whose activity needs to be tightly controlled and co-ordinated to maintain cell integrity. The cell wall is the shape-maintaining element in prokaryotes that protects the protoplast from osmotic lysis (Scheffers and Pinho, 2005). It consists of peptidoglycan, a polymer of long glycan chains with alternating units of N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc), which are cross-linked via flexible peptide bridges. Peptidoglycan is found in all eubacteria except for some obligate intracellular species. The biosynthesis of peptidoglycan takes place in three compartments of prokaryotic cells and requires a considerable number of biosynthetic enzymes. In the cytoplasm, six enzymes (MurA to MurF) catalyse the formation of the UDP-MurNAc-pentapeptide precursor. The synthesis of the first membrane bound intermediate is catalysed by MraY, which transfers UDP-MurNAc-pentapeptide to the lipid carrier undecaprenyl phosphate (C55-P) yielding lipid I. With the addition of UDP-GlcNAc by MurG, lipid II, the final peptidoglycan precursor, is synthesized. Lipid II is then translocated from the cytoplasm to the outside of the cell and incorporated into the peptidoglycan network.

Notably, there is increasing evidence that peptidoglycan synthesis is organized in a membrane spanning, multienzyme machinery that is closely interlinked with the cell division apparatus, e.g. by sharing components such as FtsI. This enzyme is a transpeptidase that incorporates precursors via peptide bridges into the cell wall and has been identified as part of the molecular machine that drives cell division in prokaryotes and that assembles in a defined, concerted order at the site of cell division (Errington et al., 2003). Despite enormous progress in the last decade towards the identification of proteins involved in cell division and their mode of assembly in the model organism Escherichia coli, the biochemical functions of several division proteins are still poorly understood (Errington et al., 2003; Vicente and Rico, 2006). Currently, the following stages of procaryotic cell division are recognized: at the site of division a cytoplasmic apparatus linked to the membrane assembles; then additional elements form an extracellular connector; and finally, proteins (e.g. FtsI) involved in the incorporation of septal peptidoglycan assemble. The process ends with constriction of the septum.

Intracellular bacteria such as Chlamydia and Wolbachia do not need cell walls for osmotic stabilization, and peptidoglycan has never been reliably detected in both. Whereas Chlamydia pneumoniae is a human pathogen, wolbachiae are obligate endobacteria of many arthropods and filarial nematodes. A striking feature of both bacterial genera is the reduced size of their genomes, ∼1.2 Mb and ∼1.3 Mb, respectively, compared with E. coli (∼4.6 Mb), reflecting the evolutionary adaptation to the environment in the host cell. These bacteria are thought to harbour a minimal genome encoding only genes essential for their parasitic or mutualistic lifestyle (Foster et al., 2005; Pfarr and Hoerauf, 2005). Surprisingly, searching the genome of Chlamydia and Wolbachia revealed that these cell wall-lacking bacteria harbour the genes for the production of the cell wall precursor lipid II (Stephens et al., 1998; Foster et al., 2005). Until now, nothing was known about the functionality of the cell wall biosynthesis pathway in Wolbachia. In contrast, the functionality of some chlamydial enzymes (MurA, MurC/Ddl, CT390, DapF), catalysing early cytoplasmic steps of peptidoglycan biosynthesis, has been characterized (Hesse et al., 2003; McCoy et al., 2003; McCoy and Maurelli, 2005). However, examination of the published genomes reveals that both genera lack transglycosylases that would catalyse the formation of linear glycan chains of peptidoglycan (Fig. 1) (Stephens et al., 1998; Ghuysen and Goffin, 1999; Wu et al., 2004; Foster et al., 2005; McCoy et al., 2006).

Figure 1.

Proposed lipid II pathway in Chlamydia and Wolbachia. Enzymes that are encoded in the genomes of both genera are depicted. Genes not found in the genomes are highlighted by red crosses. Question marks and dashed arrows indicate crucial catalytic steps of the proposed pathway that remain to be elucidated. In the cytoplasm, the UDP-linked monosaccharide pentapeptide of lipid II is synthesized by the Mur enzymes. The exact composition of the stem peptide-chain of the precursor is unknown, since racemases necessary to produce D-Ala and D-Glu are not present in both genera. The only amino acid which can be synthesized de novo in Wolbachia is m-DAP (Foster et al., 2005) and chlamydiae produce m-DAP using a pathway that is unique among bacteria (McCoy et al., 2006), suggesting that m-DAP is in position three of the stem peptide of both genera. The soluble building block is then linked to the sugar carrier bactoprenol phosphate (C55-P) by MraY and lipid II synthesis is completed by addition of N-acetylglucosamine through MurG. Lipid II, by a yet uncharacterized process, is then translocated to the outside where in free-living bacteria, through the activity of the penicillin-binding proteins (PBPs), the building block is polymerized into the peptidoglycan network. In Chlamydia and Wolbachia the only PBPs found are monofunctional transpeptidases (PBP2 in both genera and FtsI in Chlamydia and the Wolbachia endosymbiont of Drosophila melanogaster (wMel)) and D-alanyl-D-alanine carboxypeptidase PBP6a (encoded in both genera) (Chopra et al., 1998; Ghuysen and Goffin, 1999; Wu et al., 2004; Foster et al., 2005).

Here we demonstrate that lipid II is indeed synthesized by isolated Wolbachia membranes ex vivo and that MraY and MurG from both species are active in vitro. The conservation of lipid II biosynthesis in cell wall-lacking bacteria indicates that the bactoprenol-bound peptidoglycan building block, in addition to the essential proteins identified so far, may play a vital role in cell division.

Results

Lipid II biosynthesis in Chlamydia and Wolbachia

The genes encoding enzymes that are involved in the biosynthesis of the peptidoglycan precursor lipid II were annotated in the genomes of Chlamydia and Wolbachia (Stephens et al., 1998; Foster et al., 2005). We confirmed the original annotation results using blast alignments of each protein with the respective E. coli counterpart. For each lipid II biosynthesis enzyme an orthologue was present in both genera with the exception of the racemases MurI and Alr for producing the D-isomers of alanine and glutamate (Table 1). Therefore, both genera have the genetic information to synthesize a lipid II molecule whereas the exact composition and structure of the peptide-sidechain remain to be determined.

Table 1.  Enzymes involved in the biosynthesis of lipid II.
E. coliCtCpwBmwMelwPawRi
  • a. 

    In chlamydiae the murC and ddl genes are fused.

  • Locus tags of genes coding for the different lipid II biosynthesis enzymes are shown for two exemplary chlamydial and all completely sequenced Wolbachia strains. We confirmed (bold texted genes) the original annotation results using blast alignments of each protein with the respective counterpart of E. coli (W3110; AC_000091). The Expected (E) values of all alignments between the sequences of the annotated enzymes and the orthologues of E. coli ranged from 4 × 10−71 to 3 × 10–12 for Chlamydia and 2 × 10−91 to 5 ×10−11 for Wolbachia. The functionality of the chlamydial enzymes MurA, MurC and Ddl has been demonstrated previously (Hesse et al., 2003; McCoy et al., 2003; McCoy and Maurelli, 2005). Genes encoding the racemases MurI and Alr are missing from the genomes of both genera; thus, structural differences in the wolbachial and chlamydial lipid II as compared with the E. coli molecule, e.g. due to incorporation of alanine and glutamate l-isomers in the peptide-side chain, may occur.

  • Ct, Chlamydia trachomatis UW-3/Cx (NC_000117); Cp, Chlamydia pneumoniae CWL029 (NC_000922); wBm, Wolbachia endosymbiont of B. malayi (NC_006833); wMel, Wolbachia endosymbiont of D. melanogaster (NC_002978); wPa, Wolbachia endosymbiont of C. quinquefasciatus (NC_010981); wRi, Wolbachia endosymbiont D. simulans strain Riverside (NC_012416).

MurACT455CPn0572Wbm0740WD1197WPa_0760WRi_011710
MurBCT831CPn0988Wbm0778WD0541WPa_0473WRi_003550
MurCCT762aCPn0905aWbm0118WD0495WPa_0207WRi_002500
MurDCTA758CPn0901Wbm0508WD0849WPa_0533WRi_008130
MurECT269CPn0418Wbm0492WD0924WPa_0137WRi_008770
MurFCT756CPn0899Wbm0238WD1128WPa_0848WRi_013010
MurGCT761CPn0904Wbm0557WD0323WPa_0898WRi_004740
MraYCT757CPn0900Wbm0643WD1102WPa_0869WRi_012770
DdlCT762aCPn0905aWbm0570WD0095WPa_0403WRi_001550
Alr
MurI

Fosfomycin inhibits the growth of Wolbachia

To investigate if lipid II biosynthesis is essential in endobacteria, we examined the influence of an antibiotic that targets this pathway on the growth of Wolbachia. For this, we used fosfomycin to inhibit the cytoplasmic enzyme MurA, catalysing the first step in the synthesis of the soluble building block UDP-MurNAc-pentapeptide (Fig. 1). Over a period of 14 days, Wolbachia infected C6/36 cells were treated with different concentrations of fosfomycin, tetracycline as positive control or no antibiotic. The lower concentrations of fosfomycin did not affect Wolbachia growth as compared with the control without antibiotic, whereas at higher concentrations (32 mg l−1 and 128 mg l−1) fosfomycin depleted Wolbachia from the C6/36 cells (Fig. 2). Although the depletion of Wolbachia with the highest fosfomycin concentration did not reach the level of the tetracycline treated cells, it is obvious that fosfomycin is active against Wolbachia indicating that the lipid II pathway is essential for these endobacteria.

Figure 2.

Susceptibility of Wolbachia to fosfomycin and tetracycline after 14 days treatment. To investigate the susceptibility of Wolbachia to fosfomycin, infected C6/36 cells were treated with different concentrations of fosfomycin or with 10 mg l−1 tetracycline for 14 days. After extraction of genomic DNA, the copy number of the 16srRNA and actin genes was detemined by real-time PCR. White bar: no antibiotic (negative control); black bar: tetracycline treated cells (positive control); grey bars: fosfomycin treated cells.

This has also been shown for Chlamydia, although, due to an amino acid exchange in the active site of MurA, this genus appears to be intrinsically resistant to fosfomycin (McCoy et al., 2003). Therefore, it was necessary to use d-cycloserine, a specific inhibitor of the D-Ala-D-Ala-ligase (Fig. 1), to show the requirement for the lipid II pathway (Barker, 1968; Gordon and Quan, 1972; McCoy and Maurelli, 2005) The reciprocal experiment of incubating Wolbachia infected C6/36 cells with d-cycloserine did not inhibit the growth of the endobacteria in our experiments (data not shown).

Wolbachia cytoplasmic membranes synthesize lipid II in ex vivo assays

Next we searched for evidence that lipid II is indeed synthesized. Therefore, we isolated Wolbachia membranes from Aedes albopictus cells which were then incubated with the substrates C55-P, UDP-MurNAc-pentapeptide and UDP-[14C]-GlcNAc. Under the same optimized conditions used for Staphylococcus simulans (Schneider et al., 2004), the isolated membranes were able to synthesize lipid II (Fig. 3). To preclude interference of contaminating glycosyltransferases from the Aedes host system, membranes from uninfected insect cells were isolated and used in the same assay. Clearly, such membranes were not able to produce a product that could migrate with the same mobility as the lipid II spot on TLC. In addition to the expected product (lipid II) we detected another [14C]-labelled product with the same mobility as C55-P-P-GlcNAc (T. Schneider et al., in preparation). A candidate enzyme to conduct such a transfer of UDP-GcNAc onto C55-P is a predicted glycosyltransferase (locus_tag:WPa_0619).

Figure 3.

Synthesis of lipid II by membranes from Wolbachia. To investigate whether Wolbachia membranes contain functional MraY and MurG, an ex vivo activity assay was established. For this assay, isolated membranes (280 μg on protein basis) were incubated with 5 nmol C55-P, 50 nmol UDP-MurNAc-pentapeptide and 100 nmol UDP-[14C]-GlcNAc. The reaction products were extracted and separated by TLC. Subsequently, the radiolabelled spots were visualized using storage phosphor screen technology. Lane 1: membranes from Wolbachia isolated from C6/36 cells; lane 2: membranes from uninfected C6/36 cells to control for interfering eukaryotic glycosyltransferases; lane 3: assay without addition of membranes; lane 4: purified [14C]-lipid II (0.4 nmol). The additional band in lane 1 corresponds to C55-P-P-GlcNAc (T. Schneider et al., in preparation).

Recombinant MraY from C. pneumoniae and the Wolbachia (wBm) endosymbiont of the nematode Brugia malayi exhibit in vitro activity

We overproduced MraY, an integral membrane protein with 10 predicted transmembrane domains, from C. pneumoniae and wBm in E. coli and purified the recombinant His6-tagged proteins. The enzymes from both species exhibited in vitro activity, catalysing the transfer of the phospho-N-acetylmuramoyl-pentapeptide from UDP-MurNAc-pentapeptide onto C55-P, yielding lipid I (Fig. 4). For MraY in vitro activity assays we used UDP-MurNAc-pentapeptide purified from S. simulans as substrate. As in many Gram-positive bacteria, staphylococcal UDP-MurNAc-pentapeptide contains L-Lys in the pentapeptide chain whereas UDP-MurNAc-pentapeptide from the genera Chlamydia and Wolbachia is predicted to contain meso-diaminopimelic acid (m-DAP) instead of L-Lys; in addition, both might lack d-amino acids because of the absence of any racemase encoding genes in the genome (Stephens et al., 1998; Foster et al., 2005). Despite hypothetical structural differences in the pentapeptide chain, MraY from C. pneumoniae and wBm were able to utilize UDP-MurNAC-pentapeptide from S. simulans, indicating that the structure of the peptide chain is not crucial for MraY binding of UDP-MurNAc-pentapeptide.

Figure 4.

Synthesis of lipid I by recombinant MraY from C. pneumoniae (Cp) and wBm. In vitro activity assays for MraY were performed with 2 μg of enzyme incubated with 5 nmol C55-P and 50 nmol UDP-MurNAc-pentapeptide. Reaction products were separated by TLC and subsequently quantified by storage phosphor screen technology. Lane 1: C. pneumoniae MraY obtained from heterologous overproduction in E. coli purified as described in the material and methods section and in (Schneider et al., 2009); lane 2: wBm MraY obtained as described above; lane 3: E. coli host control after overproduction and purification of ORF1055 as described for MraY; lane 4: assay without addition of enzyme; lane 5: assay without enzyme; addition of purified lipid I (1 nmol).

Previous work in our laboratory (J. Esche and T. Schneider, in preparation) has shown that heterologous MraY-His6 produced in E. coli is contaminated with host MraY, which copurifies on Ni-NTA columns, possibly due to a stretch of three His residues in position 325–327. To control for this host MraY, we purified ORF1055 produced in the same strain as described for MraY-His6 (Schneider et al., 2009). ORF1055 is a staphylococcal protein of unknown function with physicochemical properties similar to MraY (localization and number of transmembrane domains). MraY background activity found in the ORF1055 preparation was due to E. coli host MraY and amounted to 0.53 ± 0.13 nmol lipid I. The purified preparation of recombinant chlamydial MraY yielded 1.31 ± 0.33 nmol lipid I (net production 0.78 ± 0.18 nmol lipid I) and the Wolbachia MraY 1.01 ± 0.31 nmol lipid I (net production 0.48 ± 0.17 nmol lipid I).

Recombinant MurG from C. pneumoniae and wBm exhibit in vitro activity

The membrane-associated MurG-His6 proteins from C. pneumoniae and wBm were also overproduced in E. coli and purified by Ni2+-NTA chromatography. MurG from both species exhibited in vitro activity and transferred GlcNAc from UDP-[14C]-GlcNAc onto lipid I, yielding [14C]-lipid II. A protein solution purified from the E. coli host strain harbouring the empty vector did not synthesize [14C]-lipid II, demonstrating that native MurG was not a contaminant in the rMurG (Fig. 5).

Figure 5.

Synthesis of lipid II by recombinant MurG from C. pneumoniae (Cp) and wBm. In vitro activity assays for MurG were performed with 4 μg of the enzyme incubated with 2.5 nmol lipid I and 25.0 nmol UDP-[14C]-GlcNAc. The reaction products were separated by TLC. Radiolabelled spots were visualized using storage phosphor screen technology. Lipid II synthesis by C. pneumoniae wild-type (lane 1), C. pneumoniae S104G mutant (lane 2) and wBm (lane 3) MurG. Lane 4: Reaction mix without addition of enzyme but with [14C]-lipid II (0.4 nmol) as positive control. Lane 5: Purified protein solution from the E. coli host harbouring the empty vector. Lane 6: Reaction mix without addition of enzyme.

Functional analysis of C. pneumoniae MurG wild-type and mutant proteins

In a previous report on the crystal structure of E. coli MurG (Ha et al., 2000) two highly conserved glycine rich loops (G loops) were predicted to be involved in binding to the diphosphate of lipid I. The G loops are located in the N-terminal domain and are reminiscent of the phosphate binding loops of nucleotide binding proteins, suggesting that MurG binds to the diphosphate and not to the pentapeptide chain of lipid I. Our results with recombinant MurG support this interpretation in that rMurG from Chlamydia and Wolbachia synthesized lipid II using UDP-MurNAc-pentapeptide from S. simulans as a substrate.

A previously published alignment of 73 MurG sequences revealed highly conserved glycine residues in the G loops of the lipid I binding site (Crouvoisier et al., 2007). The G loop 2 from C. pneumoniae MurG deviates from the conserved pattern seen in E. coli and Wolbachia (Table 2) with G104 (93% conservation among 73 MurG enzymes) being replaced by a serine residue. Based on the above model, such an alteration could result in decreased affinity to lipid I in comparison with the E. coli enzyme. To address this question, we mutated G loop 2 in the chlamydial enzyme by replacing Ser104 (wild-type) with glycine, as found in E. coli (Table 2), and compared kinetic constants of the S104G mutant and the wild-type proteins. Interestingly, the S104G mutant protein did not significantly alter the Km value of lipid I (15.0 ± 5.8 μM versus 12.4 ± 3.3 μM) but showed improvement in the Km value of UDP-GlcNAc (150.9 ± 35.5 μM versus 237.1 ± 13.2 μM) in comparison to the wild-type protein (Table 3). This suggests that the E. coli structure-activity model derived with a substrate analogue (MurNAc (Nε-dansylpentapeptide)-pyrophoshoryl) (Ha et al., 2000; Crouvoisier et al., 2007) may not fully apply to the chlamydial enzyme.

Table 2.  Alignment of G loop 2 sequences of MurG from E. coli, C. pneumoniae and wBm.
G loop 2
% of conservationb
G101a
70
XG103
100
G104
93
Y105
78
  • a. 

    Numbering of E. coli.

  • b. 

    Percentage of conservation among 73 MurG enzymes (Crouvoisier et al., 2007).

  • Bold highlights the amino acid change in Chlamydia versus the other species. This amino acid was mutated in later experiments.

E. coliGMGGY
C. pneumoniae wild-typeGFGSY
C. pneumoniae S104G mutantGFGGY
wBmGFGGY
Table 3.  Kinetic parameters of MurG wild-type and mutant proteins from C. pneumoniae.
MurG proteinKm lipid Ic (μM)Km UDP-GlcNAc (μM)Source
  • a. 

    Numbering as in E. coli.

  • b. 

    Assay conditions: 200 mM Tris-HCl (pH 7.5), 10 mM MgCl2, 35% (v/v) DMSO, 30 min, 37°C using the lipid I analogue MurNAc (Nε-dansylpentapeptide)-pyrophoshoryl.

  • c. 

    Apparent constant, determined with UDP-GlcNAc substrate at 833.3 and 25 μM for this study and for Crouvoisier et al. (2007), respectively.

C. pneumoniae wild-type12.4 ± 3.3237.1 ± 13.2This study
C. pneumoniae S104Ga mutant15.0 ± 5.8150.9 ± 35.5This study
E. coli wild-typeb2.8 ± 1.0b27 ± 11bCrouvoisier et al. (2007)

Discussion

Adaptation of pathogenic microorganisms to a specific host tends to result in reduction of genome size. Particularly, evolution within the stable environment of host cells has allowed endobacteria to tolerate losses of genes that are no longer needed. Many genes considered essential for free-living bacteria cannot be found in various endosymbionts (Feldhaar and Gross, 2009). Both Wolbachia and Chlamydia have lost essential biosynthetic pathways, e.g. wolbachiae are dependent on the host's production of amino acids, vitamins and cofactors (Wu et al., 2004; Foster et al., 2005) and chlamydiae are thought to be ‘energy parasites’ because they import ATP from the host cell (Moulder, 1991).

For intracellular bacteria, the highly energy consuming biosynthesis of a polymer, such as peptidoglycan, appears dispensable. However, our findings, together with previous results on chlamydial MurA, MurC/Ddl and meso-DAP pathway enzymes (Hesse et al., 2003; McCoy et al., 2003; 2006; McCoy and Maurelli, 2005), clearly demonstrate that the final cell wall building block, attached to a membrane carrier, the so-called lipid II, is indeed synthesized and essential in Wolbachia and Chlamydia, raising questions as to the role of lipid II in the cell biology of eubacteria besides its function in cell wall biosynthesis.

In prokaryotes, cell division and cell wall biosynthesis are tightly controlled and co-ordinated cellular processes. Therefore, we propose the hypothesis that the bactoprenol-bound peptidoglycan subunit lipid II provides a molecular link between cell wall biosynthesis and cell division and is, along with the various proteins identified so far for both processes, essential (Fig. 6). Several recent reports support this hypothesis: In Staphyloccocus aureus the penicillin-binding protein PBP2 delocalizes rapidly from the septation site when its substrate lipid II is depleted by addition of d-cycloserine, which blocks synthesis of UDP-MurNAc-pentapeptide (Fig. 1). This observation suggested that PBP2 is recruited to and kept at the division site by the availability of its substrate, which is localized to that particular site of the growing cell (Pinho and Errington, 2005). In turn, using fluorescence microscopy, Hasper et al. (2006) showed that lipid II sequestration by lantibiotics leads to the delocalization of lipid II from cell division sites and blocks cell wall biosynthesis. Yet another study revealed that the sequestration of lipid II by nisin results in aberrant, accelerated cell division and multiple division sites in the midcell region of rod-shaped bacterial cells (Hyde et al., 2006). These results strongly suggest that the cell wall biosynthesis machinery and hence the divisome machine, will disassemble when the interaction of lipid II processing enzymes with their substrate is blocked. Finally, our hypothesis is strongly supported by a report on the essential role of the lipid II synthesis pathway for the division of chloroplasts in mosses in which disruption of MurE, a cytoplasmatic enzyme essential for lipid II synthesis (Fig. 1), resulted in the formation of macrochloroplasts which were unable to divide (Machida et al., 2006).

Figure 6.

Proposed model for a molecular link between lipid II and cell division: in endobacteria lipid II is synthesized via a fully conserved pathway and then processed by a rudimentary cell wall biosynthesis – cell division machinery which may consist of MraY, MurG, PBPs and division proteins (FtsK, FtsW and FtsY are encoded in the genomes of both genera, individual species harbour further orthologues of fts genes). Processing of lipid II, including translocation to the outside and recycling of the bactoprenol carrier, is required for co-ordinated function of the divisome machinery and therefore the lipid II biosynthesis pathway needs to be conserved in cell wall-lacking endobacteria.

The synthesis of lipid II in Chlamydia and Wolbachia and the obvious absence of a murein layer raise questions as to the presence of a pathway for lipid II processing and subsequent bactoprenol-P recycling (Fig. 1). Neither Chlamydia nor Wolbachia encode proteins which would be considered essential for a complete cycle, e.g. pyrophosphorylases that catalyse dephoshorylation of C55-PP as found in E. coli. (YeiU, YbjG, PgpB, UppP) (Tatar et al., 2007). Additionally, the fate of sugar and peptide units from the lipid II molecule is unclear when peptidoglycan is not found. The genomes of both genera lack genes that encode the transglycosylases needed to link the sugar moieties of lipid II to form glycan chains, the backbone of the cell wall. Furthermore, both genomes do not encode endopeptidase genes although they do harbour genes for monofunctional transpeptidases, which cross-link glycan chains via flexible peptide bridges in free-living bacteria (PBP2 in both genera as well as FtsI (PBP3) in Chlamydia and the Wolbachia endosymbiont of Drosophila melanogaster (wMel)). Moreover, D-alanyl-D-alanine carboxypeptidase (PBP6a, encoded in both genera) and N-acetylmuramoyl-L-alanine amidase (in Chlamydia, wMel and Wolbachia of Culex pipientis, but not in wBm) can be identified (Chopra et al., 1998; Ghuysen and Goffin, 1999; Wu et al., 2004; Foster et al., 2005). The physical presence of the transpeptidase proteins might be beneficial for the correct assembly of the divisome complex. Moreover, the transpeptidase (and amidase) activity could catalyse the release of the C55-P carrier and a rudimentary peptidoglycan could result as a by-product. In the context of the enigmatic chlamydial anomaly (Moulder, 1993), which describes their susceptibility to cell wall targeting antibiotics, a rudimentary glycan-less cell wall in which the peptide chains are cross-linked by peptide bonds has been proposed (Ghuysen and Goffin, 1999). Alternatively, one may suggest a structure with unconnected disaccharide units, just cross-linked by peptide bridges (Foster et al., 2005). Noteworthy, besides modulating the host immune response, McCoy and Maurelli (2006) discussed a role of peptidoglycan in chlamydial cell division and proposed, as chlamydial genomes do not encode an FtsZ orthologue, that peptidoglycan might be a functional equivalent of an FtsZ-ring in this genus. Here we propose that it is not a peptidoglycan macromolecule, but rather the monomeric precursor lipid II itself and its sequential processing that plays a crucial role in cell division.

Future perspectives

A unique feature of Chlamydiaceae is their developmental cycle that alternates between an extracellular, infectious form and an intracellular, non-infectious, replicating form called elementary body and reticular body respectively (Ward, 1983). This enables the investigation of lipid II production regarding its role in cell division disconnected from cell wall biosynthesis. Microarray-based transcriptional analysis of the whole C. trachomatis genome revealed a marked increase in the transcription of the lipid II synthesis machinery genes during reticular body replication (McCoy and Maurelli, 2006), supporting the essential role of lipid II in cell division. We are currently isolating lipid II from reticular bodies of Chlamydia and from Wolbachia cells. With the purified molecules we will be able to analyse its chemical structure since the composition of the peptide-chain of the precursor may not be the same as in free-living bacteria, which can be concluded from the absence of racemases necessary to produce D-Ala and D-Glu, from both genera (Stephens et al., 1998; Foster et al., 2005). The racemization of alanine and glutamate could be catalysed by other epimerases, or the pentapeptide-sidechain could contain l-isomers instead of the usually found d-isomers (Foster et al., 2005).

Practically, our results open the way to exploit the lipid II pathway for antichlamydial and antifilarial therapies. Filarial nematodes infect animals and humans and cause lymphatic filariasis or onchocerciasis in more than 120 million people (WHO, 2006; 2007). The Wolbachia of filarial worms are essential for embryogenesis, larval development and adult survival, making them prime targets for antifilarial chemotherapy. This has allowed the introduction of doxycycline as a method of antifilarial therapy (Bandi et al., 1999; Hoerauf et al., 1999; Hoerauf et al., 2000; Casiraghi et al., 2002; Hoerauf, 2008). However, for a broad application, new antibacterial compounds are needed that produce worm sterility and/or death with a shorter regimen.

Experimental procedures

Bacterial strains and growth conditions

Chlamydia pneumoniae strain GiD were cultivated in HEp-2 cells (ATCC CCL-23) as previously described (Jantos et al., 1997). Wolbachia strain Aedes albopictus B were grown in the Aedes albopictus cell line C6/36 as previously described (Turner et al., 2006).

Cloning, expression and purification of cell wall biosynthesis enzymes from C. pneumoniae and the Wolbachia (wBm) endosymbiont of the nematode Brugia malayi

The mraY gene was amplified with primers containing a BamHI or an XhoI site (mraY_1_Cp: 5′-ATAGGATCCATCCCCTTAATTCCAATGTTT-3′ and mraY_2_Cp: 5′-ACACTCGAGTCTCCATAAGACAGCCGC-3′) for C. pneumoniae; and with primers containing a BamHI or a NotI site (wBm-MraY-Fw: 5′-AATAGGGGATCCAACTTACCTACAAAAATA-3′ and wBm-MraY-NotI-Rev: 5′-CATAGCGGCCGCCAACAAGAAAGCAACAG-3′) for wBm.

Polymerase chain reaction (PCR) products and a modified pET20 vector (kindly provided by B. Berger-Bächi, Zürich) were digested with BamHI and XhoI or BamHI and NotI and subsequently ligated with the Quick Ligation™ Kit (New England BioLabs, Frankfurt, Germany) as per the protocol from the manufacturer. In the obtained constructs, the mraY gene was expressed under control of a strong IPTG-inducible promoter such that the encoded MraY protein carried a C-terminal His6 extension. Overproduction and purification of the MraY-His6 fusion protein followed the protocol of Schneider et al. (2009).

The murG genes from C. pneumoniae and wBm were amplified and cloned into the pET21b vector (Novagen, VWR International, Darmstadt, Germany) using Chlamydia specific primers containing NheI and NotI sites (murG_1_Cpn 5′-AGGGCTAGCATGAAGAAAATTCGAAAAG-3′, murG_2_Cpn: 5′-TAAGCGGCCGCTAAGCATTCACAAATGAAT-3′), or Wolbachia specific primers containing NheI and XhoI sites (wBm-MurG-Fw: 5′-ATTCAGCTAGCGATATCGTTCTAGCAACAG-3′ wBm-MurG-Rev: 5′-CATATCTCGAGACCAAGTTTGTGAATTAC-3′). Both clones generated C-terminal His6 fusion proteins. Overproduction and purification of the MurG-His6 proteins was performed as previously described (Schneider et al., 2009).

Site-directed mutagenesis

S104 in MurG from C. pneumoniae was changed to glycine using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies, Waldbronn, Germany). Briefly, primers a304g 5′-GGTCATAGGATTTGGGGGCTACCACTCTCTTCC-3′ and a304g_antisense 5′-GGAAGAGAGTGGTAGCCCCCAAATCCTATGACC-3′ (a304g, base changes are underlined) were used to amplify unmethylated copies of pET21b containing the chlamydial murG. After digestion with DpnI, the remaining mutation-containing molecules were transformed into competent cells for nick repair. The correct base change was verified by sequencing.

MraY in vitro activity assay

In vitro activity assays for MraY were carried out in a final volume of 70 μl containing 2 μg MraY, 5 nmol C55-P, 50 nmol UDP-MurNAc-pentapeptide [from S. simulans (Schneider et al., 2004)] in 0.43% Triton X-100 (w/v), 10% DMSO, 71.4 mM Tris-HCl (pH 7.5), 5.4 mM MgCl2. The reaction mixture was incubated for 1 h at 30°C. The reaction products were extracted with 1 volume of n-butanol/pyridine acetate (2:1 v/v, pH 4.2) and separated by thin layer chromatography (TLC; silica plates, 60F254; Merck, Darmstadt, Germany) using chloroform-methanol-water-ammonia (88:48:10:1) as solvent (Rick et al., 1998). Spots were visualized by phosphomolybdic acid (PMA) staining and quantified using ImageQuant TL v2005 software (GE Healthcare Europe, Munich, Germany).

MurG in vitro activity assay

The activity of wild-type and mutant MurG was measured using standard in vitro activity assays in a final volume of 30 μl containing 4 μg MurG, 2.5 nmol lipid I [synthesized using UDP-MurNAc-pentapeptide from S. simulans (Schneider et al., 2004)], 25.0 nmol UDP-[14C]GlcNAc (GE Healthcare Europe, 5.6 kBq) in 100 mM Tris-HCl (pH 7.5), 5.7 mM MgCl2, 0.83% (w/v) Triton X-100. The reaction mixture was incubated for 30 min at 30°C. The n-butanol/pyridine-acetate-extracted products were separated by TLC as described above and the resultant radiolabelled spots visualized and quantified using storage phosphor screen technology (Storm 820, GE Healthcare Europe).

Kinetic analysis of the MurG wild-type and mutant proteins

Km values of lipid I and UDP-GlcNAc were determined with a standard assay using various concentrations of one substrate while maintaining the concentration of the other substrate at a fixed value [lipid I 0.0625 up to onefold the concentration of UDP-GlcNAc (833.3 μM); UDP-GlcNAc 0.5 up to 10-fold the concentration of lipid I (83.3 μM)]. For the calculation of kinetic parameters the Michaelis-Menten model and non-linear regression were used (GraphPad Prism version 5.01). Because saturating concentrations were not achievable for UDP-GlcNAc, apparent Km values of lipid I were given.

Membrane preparation from Wolbachia isolated from C6/36 cells

Wolbachia infected Aedes albopictus C6/36 cells were grown in 50 cm2 culture flasks to ∼90% confluence. The isolation of Wolbachia was done according to the method described by Rasgon et al. (2006) with some modifications. Briefly, the pellet from 70 flasks of C6/36 cells was resuspended in 4 ml PBS and vortexed for 5 min with 100 sterile 3 mm borosilicate glass beads to lyse the cells. The supernatant was centrifuged at 2500 g, 4°C for 10 min to pellet cellular debris. Then the lysate was subsequently passed through 5.0 μm and 2.7 μm syringe filters (Whatman, Florham Park NJ) to remove residual cellular debris. To solubilize the Wolbachia, the suspension was sonicated 3 × 30 s (Sonopulse, cycle 0, 50% power, Bandelin Electronic, Berlin, Germany) on ice and centrifuged at 21 460 g, 4°C for 10 min. The pellet containing the Wolbachia membrane debris was then suspended in 100 μl of 50 mM Tris-HCl, 10 mM MgCl2, pH 7 and stored at −20°C or used directly for the ex vivo synthesis of lipid II. Uninfected C6/36 cells were used as negative control and treated as described above, without the centrifugation step after the lysis with glass beads.

Lipid II synthesis using membranes from Wolbachia

The functionality of MraY and MurG from Wolbachia membranes was determined using an ex vivo activity assay in a final volume of 120 μl containing isolated membranes (280 μg on protein basis), 5 nmol C55-P, 50 nmol UDP-MurNAc-pentapeptide [from S. Simulans (Schneider et al., 2004)], 100 nmol UDP-[14C]GlcNAc (GE Healthcare Europe, 5.6 kBq) in 72 mM Tris-HCl (pH 7.5), 5 mM MgCl2, 0.4% (w/v) Triton X-100. The reaction mixture was incubated for 1.5 h at 30°C. The reaction products were extracted and separated by TLC as described above and the resultant radiolabelled spots visualized using storage phosphor screen technology (Storm 820, GE Healthcare Europe).

Antibiotic susceptibility of Wolbachia

C6/36 cells infected with Wolbachia were cultured in 24-well plates by seeding each well with 2 × 105 insect cells. The cells were incubated for 14 days at 26°C in L15 Leibovitz medium (2 mM l-glutamine, 5% FCS, 1% non-essential amino acids, 2% tryptose phosphate broth and 1% penicillin/streptomycin) with and without the different antibiotic concentrations (10 mg l−1 tetracycline or 1, 8, 32 and 128 mg l−1 fosfomycin) which were replaced every day. The treated cells were harvested on day 14. Extraction of genomic DNA was performed with TRIzol Reagent (Invitrogen, Karlsruhe, Germany) according to the instructions of the manufacturer. Depletion of Wolbachia was monitored by real-time PCR using the primers 16s-rRNA-Fw (5′-TTGCTATTAGATGAGCCTATATTAG-3′) and 16s-rRNA-Rev (5′-GTGTGGCTGATCATCCTCT-3′) (Makepeace et al., 2006) which target the 16s-rRNA gene of Wolbachia and Ac-Fw (5′-ACGAACTGGGACGATATGGA-3′) and Ac-Rev (5′-GCCTCTGTCAGGAGAACTGG-3′) for the actin of the C6/36 cells. The copy number of each transcript was calculated using a plasmid containing the appropriate insert as standard curve.

Acknowledgements

We thank D. Alborn for providing us with purified ORF1055 protein from S. aureus for control experiments with MraY. We also thank M. Taylor for supplying us with Wolbachia infected and uninfected Aedes albopictus C6/36 cells. We thank I. Luhmer-Becker and M. Josten for excellent technical assistance and C. Poretschkin and C. Körner for help with protein purification.

Funding for this work was provided by the German Research Foundation (DFG) with grants to A. Hoerauf, K. Pfarr, I. Wiedemann and H.G. Sahl within the special research unit FOR 854 (PF673/3-1, WI1912/2-1, SA292/11-1, 13-1). K. Johnston was funded through the A-WOL consortium grant from the Bill and Melinda Gates Foundation awarded to the Liverpool School of Tropical Medicine. Financial support was also obtained from the intramural funding scheme of the Medical Faculty of Bonn (BONFOR). C. Poellinger received a fellowship from the NRW Graduate Research School Biotech Pharma.

Ancillary