Lipodepsipeptide Empedopeptin Inhibits Cell Wall Biosynthesis through Ca2+-dependent Complex Formation with Peptidoglycan Precursors*

Background: The mechanism of action of the potent antibiotic empedopeptin was not known. Results: Empedopeptin forms calcium-dependent complexes with peptidoglycan precursors, particularly lipid II. Conclusion: Bacterial cell wall synthesis is blocked by sequestration of the bactoprenol carrier. Significance: This mechanism is proposed for a wider class of structurally related antibiotics including empedopeptin, tripropeptins, and plusbacins. Empedopeptin is a natural lipodepsipeptide antibiotic with potent antibacterial activity against multiresistant Gram-positive bacteria including methicillin-resistant Staphylococcus aureus and penicillin-resistant Streptococcus pneumoniae in vitro and in animal models of bacterial infection. Here, we describe its so far elusive mechanism of antibacterial action. Empedopeptin selectively interferes with late stages of cell wall biosynthesis in intact bacterial cells as demonstrated by inhibition of N-acetylglucosamine incorporation into polymeric cell wall and the accumulation of the ultimate soluble peptidoglycan precursor UDP-N-acetylmuramic acid-pentapeptide in the cytoplasm. Using membrane preparations and the complete cascade of purified, recombinant late stage peptidoglycan biosynthetic enzymes and their respective purified substrates, we show that empedopeptin forms complexes with undecaprenyl pyrophosphate containing peptidoglycan precursors. The primary physiological target of empedopeptin is undecaprenyl pyrophosphate-N-acetylmuramic acid(pentapeptide)-N-acetylglucosamine (lipid II), which is readily accessible at the outside of the cell and which forms a complex with the antibiotic in a 1:2 molar stoichiometry. Lipid II is bound in a region that involves at least the pyrophosphate group, the first sugar, and the proximal parts of stem peptide and undecaprenyl chain. Undecaprenyl pyrophosphate and also teichoic acid precursors are bound with lower affinity and constitute additional targets. Calcium ions are crucial for the antibacterial activity of empedopeptin as they promote stronger interaction with its targets and with negatively charged phospholipids in the membrane. Based on the high structural similarity of empedopeptin to the tripropeptins and plusbacins, we propose this mechanism of action for the whole compound class.


Empedopeptin is a natural lipodepsipeptide antibiotic with potent antibacterial activity against multiresistant Gram-positive bacteria including methicillin-resistant Staphylococcus aureus and penicillin-resistant Streptococcus pneumoniae in vitro and in animal models of bacterial infection. Here, we describe its so far elusive mechanism of antibacterial action. Empedopeptin selectively interferes with late stages of cell wall biosynthesis in intact bacterial cells as demonstrated by inhibition of N-acetylglucosamine incorporation into polymeric cell
wall and the accumulation of the ultimate soluble peptidoglycan precursor UDP-N-acetylmuramic acid-pentapeptide in the cytoplasm. Using membrane preparations and the complete cascade of purified, recombinant late stage peptidoglycan biosynthetic enzymes and their respective purified substrates, we show that empedopeptin forms complexes with undecaprenyl pyrophosphate containing peptidoglycan precursors. The primary physiological target of empedopeptin is undecaprenyl pyrophosphate-N-acetylmuramic acid(pentapeptide)-N-acetylglucosamine (lipid II), which is readily accessible at the outside of the cell and which forms a complex with the antibiotic in a 1:2 molar stoichiometry. Lipid II is bound in a region that involves at least the pyrophosphate group, the first sugar, and the proximal parts of stem peptide and undecaprenyl chain. Undecaprenyl pyrophosphate and also teichoic acid precursors are bound with lower affinity and constitute additional targets. Calcium ions are crucial for the antibacterial activity of empedopeptin as they promote stronger interaction with its targets and with negatively charged phospholipids in the membrane. Based on the high structural similarity of empedopeptin to the tripropeptins and plusbacins, we propose this mechanism of action for the whole compound class.
Empedopeptin is an amphoteric, cyclic lipodepsipeptide produced by the Gram-negative soil bacterium Empedobacter haloabium ATCC 31962. Two accompanying publications in 1984 described its isolation, structure determination, and antibacterial activity in vitro and in vivo (1,2). Empedopeptin showed potent antibacterial activity against a broad range of aerobic and anaerobic Gram-positive bacteria including the important pathogens Staphylococcus aureus, Streptococcus pneumoniae, and Clostridium difficile. Minimal inhibitory concentrations (MICs) 2 in the low g/ml range for susceptible as well as antibiotic-resistant isolates, substantial therapeutic efficacy in lethal bloodstream infections in mice, good pharmacokinetic parameters, and low acute toxicity form a promising profile for the compound (1) especially in light of rising vancomycin resistance among staphylococci and enterococci. Therefore, we revisited empedopeptin and elucidated its molecular mechanism of action.
The overall negatively charged, water-soluble empedopeptin consists of an octapeptide core cyclized by an ester bond and a C 14 -myristic acid tail (2) (Fig. 1A). The macrolactone core comprises alternating stretches of D-and L-configured amino acids (arginine, proline, and serine) as well as the non-proteinogenic amino acids hydroxyaspartic acid and hydroxyproline. Empedopeptin has considerable structural similarity to the lipodepsipeptides of the tripropeptin and plusbacin groups (3)(4)(5)(6)(7). The lower portion of the octapeptide core, which contains the three hydroxyamino acids plus arginine, is actually identical in empedopeptin, the tripropeptins (A-E and Z), and plusbacins (A1-B4) (4), whereas moderate variations occur in the amino acid composition of the upper ring segment as well as the fatty acid side chain (Fig. 1B). The lipid tails are branched or unbranched and are variable in length, the latter directly influencing antimicrobial activity. Although produced by different organisms, these lipodepsipeptides can basically be regarded as variants of the same structural scaffold.
It has been demonstrated previously that plusbacin A3 interferes with bacterial peptidoglycan biosynthesis (8), although the specific enzymatic reaction or target structure had remained elusive. While work on this publication was in progress, it was reported that tripropeptin C can bind to undecaprenyl-pyrophosphate (C 55 -PP), thereby inhibiting the activity of undecaprenyl-pyrophosphate phosphatase (UppP) (9). Here, we show that empedopeptin exerts its antibacterial action by inhibition of peptidoglycan synthesis. However, using a comprehensive set of purified late stage peptidoglycan synthetic enzymes and the corresponding purified substrates, we demon-strate that this class of lipopeptide antibiotics sequesters further peptidoglycan precursors besides C 55 -PP and that the primary target of empedopeptin is lipid II, which is bound in a Ca 2ϩ -dependent fashion.

EXPERIMENTAL PROCEDURES
Production and Purification of Empedopeptin-Production of empedopeptin by E. haloabium ATCC 31962 on a 3-liter scale and extraction of the broth with butanol were performed as described (1  Laboratory Standards Institute guidelines (10) in polypropylene microtiter plates (Nunc brand) using cation-adjusted Mueller-Hinton broth (Oxoid). Experiments were conducted in the presence and absence of 1.25 mM Ca 2ϩ , which equates to the concentration of ionized calcium in human serum (11).
Precursor Incorporation Studies-To study the effect of empedopeptin on the synthesis of the major bacterial biopolymers, Bacillus subtilis was grown in Belitzky defined minimal medium (12), and the incorporation of tritium-labeled precursors (uridine, thymidine, phenylalanine, and N-acetylglucosamine (GlcNAc); 0.02 MBq/ml) into the acid-precipitable cell fraction was measured as described previously (13). Cultures were incubated with empedopeptin at a concentration of 8 g/ml and with vancomycin at 0.25 g/ml, both corresponding to 8ϫ MIC in this medium. Belitzky medium contains 2 mM Ca 2ϩ .
Determination of Membrane Potential-The membrane potential was determined by a previously described method (14) using the lipophilic cation tetraphenylphosphonium (TPP ϩ ), which diffuses across the bacterial membrane in response to a trans-negative membrane potential (⌬⌿). 1 Ci/ml [ 3 H]TPP ϩ (26 Ci/mmol) was added to a culture of S. aureus SG511 (A 600 ϭ 0.7) in half-concentrated Mueller-Hinton broth at 37°C supplemented with 1.25 mM Ca 2ϩ , and cellbound radioactivity was determined after filtration. For calculation of the membrane potential, cell-associated versus free TPP ϩ concentrations were applied to the Nernst equation. Culture aliquots were treated with empedopeptin (10 g/ml, corresponding to 10-fold MIC) or the uncoupler carbonyl cyanide m-chlorophenylhydrazone (500 M) as a control.
Quantification of Intracellular UDP-N-acetylmuramic Acidpentapeptide (UDP-MurNAc-pp)-We followed the protocol of Kohlrausch and Höltje (15) to analyze the cytoplasmic nucleotide precursor pool. S. aureus ATCC 29213 was grown in 20 ml of half-concentrated Mueller-Hinton broth with 1.25 mM Ca 2ϩ to an A 600 of 0.6 and incubated with 130 g/ml chloramphenicol for 15 min. Chloramphenicol is necessary to prevent (under the impact of the antibiotic under investigation) induction of autolytic processes and de novo synthesis of enzymes hydrolyzing the nucleotide-activated sugars, thereby interfering with determination of the soluble precursor (16). Empedopeptin was added at 40 g/ml and vancomycin was added at 5 g/ml (both 10ϫ MIC) and incubated for another 45 min. Extraction of UDP-linked peptidoglycan precursors with boiling water and their analysis by HPLC was performed as described previously (17). Mass spectrometric confirmation of UDP-MurNAc-pp was conducted using a MALDI-TOF mass spectrometer (Bruker Biflex, Bruker Daltronics, Bremen, Germany) working in a linear negative mode. An aliquot of the HPLC fraction containing UDP-MurNAc-pp was mixed with 6-aza-2-thiothymine in 50% (v/v) ethanol, 20 mM ammonium citrate as matrix material and spotted onto a ground steel MALDI target plate. Mass spectra were recorded in the range of 1000 -1500 Da and analyzed by Flexanalysis (Bruker Daltronics).
In Vitro Lipid II Synthesis with Membrane Preparations of Micrococcus luteus-In vitro lipid II synthesis was performed using membranes of M. luteus as described (18,19). In short, for analytical assays, membrane preparations (200 g of protein) were incubated in the presence of purified substrates (5 nmol undecaprenyl phosphate (C 55 -P), 50 nmol UDP-MurNAc-pp, and 50 nmol UDP-GlcNAc) in 60 mM Tris-HCl, 5 mM MgCl 2 , pH 7.5, 1.25 mM Ca 2ϩ , 0.5% (w/v) Triton X-100 in a total volume of 50 l for 1 h at 30°C. Incorporation of [ 14 C] UDP-GlcNAc was used to quantify lipid II synthesized. Bactoprenol-containing products were extracted with an equal volume (50 l) of butanol/pyridine acetate, pH 4.2 (2:1, v/v) and analyzed by thin layer chromatography (TLC) using chloroform/methanol/water/ammonia (88:48:10:1, v/v/v/v) as the solvent (20) and phosphomolybdic acid staining (21). Quantification was carried out by phosphorimaging in a Storm TM imaging system (GE Healthcare). Empedopeptin (molecular mass, 1126.2 Da) was added in molar ratios of 0.5-2 with respect to C 55 -P. Synthesis and purification of lipid II on a preparative scale were performed as described (21,22). In Vitro Peptidoglycan Synthesis Reactions Using Purified Proteins and Substrates-The MurG (UDP-N-acetylglucosamine:N-acetylmuramic acid(pentapeptide) undecaprenyl-pyrophosphate N-acetylglucosamine transferase) assay was performed in a final volume of 30 l containing 2.5 nmol of purified lipid I (25) and 25 nmol of [ 14 C]UDP-GlcNAc in 200 mM Tris-HCl, 5.7 mM MgCl 2 , pH 7.5, 0.8% Triton X-100 in the presence of 0.45 g of purified MurG-His 6 enzyme. The reaction mixture was incubated for 60 min at 30°C. The assay for synthesis of lipid II-Gly 1 catalyzed by FemX was performed as described previously (21). In vitro transglycosylation was determined by incubating 2 nmol of [ 14 C]lipid II in 100 mM MES, 10 mM MgCl 2 , pH 5.5 in a total volume of 50 l. The reaction was initiated by the addition of 7.5 g of PBP2-His 6 and incubated for 1.5 h at 30°C (24). In all in vitro assays, empedopeptin was added in molar ratios ranging from 0.5 to 2 with respect to the amount of C 55 -P, lipid I, and lipid II, respectively. 1.25 mM Ca 2ϩ was added as indicated. Synthesized lipid intermediates were extracted from the reaction mixtures with n-butanol/pyridine acetate, pH 4.2 (2:1, v/v), analyzed by TLC as described above, and quantified by phosphorimaging. In the PBP2 assay, reaction mixtures were applied directly onto the thin layer plates without extraction and were developed in solvent B (butanol/ acetic acid/water/pyridine (15:3:12:10, v/v/v/v)). Experimental details related to assaying MraY, TagO, and UppP are provided in supplemental Figs. S1-S3.
Complex Formation of Empedopeptin with Lipid I/II and C 55 -PP-Binding of empedopeptin to C 55 -PP, lipid I, or lipid II was analyzed by incubating 2 nmol of each purified cell wall precursor with 1-8 nmol of empedopeptin in 50 mM Tris-HCl, pH 7.5 supplemented with 1.25 mM Ca 2ϩ for 30 min. Analysis of complex formation between empedopeptin and the respective peptidoglycan precursor was carried out by applying reaction mixtures directly onto TLC plates and developing the plates in solvent B. Lipid-containing peptidoglycan precursors were detected by phosphomolybdic acid staining.
Antagonization Assays-Antagonization of the antibiotic activity of empedopeptin by potential target molecules was performed by an MIC-type assay setup in microtiter plates. Empedopeptin (8 g/ml, corresponding to 8ϫ MIC) in Mueller-Hinton broth with 1.25 mM Ca 2ϩ was mixed with potential antagonists (UDP-GlcNAc, UDP-MurNAc-pp, C 55 -P, farnesyl-PP, C 55 -PP, lipid I, lipid II, undecaprenyl-PP-GlcNAc (lipid III), and lipid I variants with shortened stem peptides) in 0.156 -10fold molar excess with respect to the antibiotic. B. subtilis 168 (5 ϫ 10 5 cfu/ml) was added, and visible bacterial growth was recorded after overnight incubation. The concentration of purified peptidoglycan precursors was quantified on the basis of their phosphate content as described (26).
Quartz Crystal Microbalance-Quartz crystals were prepared and cleaned as described previously (27). Supported bilayers were completed by transferring a monolayer consisting of 90 mol % 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine and 10 mol % 1,2-dioleoyl-sn-glycero-3-phospho-(1Ј-racemic glycerol) onto a covalently fixed C 16 H 33 SH monolayer using the Langmuir-Blodgett technique. Fitting the quartz crystals into the flow chamber was performed under water to guarantee the integrity of the bilayer. Empedopeptin was investigated at 1 M in the presence and absence of 1.25 mM CaCl 2 . A LiquiLab21 quartz crystal microbalance (ifake.V., Barleben, Germany), which enables the simultaneous detection of frequency and damping changes in real time, was used. Measurements and calculation of the kinetic binding constants were performed as described (28,29).

RESULTS
Antimicrobial Activity of Empedopeptin-Empedopeptin exhibited potent antibacterial activity against multiresistant staphylococci and streptococci including methicillin-resistant S. aureus and penicillin-resistant S. pneumoniae ( Table 1). The activity against enterococci was moderate in general but notably included vancomycin-resistant isolates. Escherichia coli was only marginally affected in accordance with the lack of activity reported earlier for other Gram-negative species (1), which probably reflects insufficient penetration of the rather large (1126.2 Da), net negatively charged compound through the outer membrane. Minimal inhibitory concentrations of empedopeptin were strongly influenced by the concentration of calcium ions in the culture medium. Addition of 1.25 mM Ca 2ϩ , equating to the Ca 2ϩ concentration in human serum (11), to standard Mueller Hinton broth reduced the MIC values 2-16-fold depending on the test strain. It is important to note in this context that the calcium levels recommended by the Clinical Laboratory Standards Institute for cation-adjusted Mueller-Hinton broth (20 -25 mg/liter Ca 2ϩ ) (10) were not sufficient to achieve full empedopeptin activity.
Metabolic Pathway Targeted by Empedopeptin-To study the impact of the antibiotic on the biosynthetic capacity of intact bacterial cells, we followed the incorporation of radiolabeled metabolic precursors into bacterial macromolecules. Empedopeptin at 8-fold the MIC rapidly and strongly interfered with the incorporation of [ 3 H]GlcNAc into the cell wall of B. subtilis, whereas DNA and protein biosynthesis remained unaffected (Fig. 2). Incorporation of [ 3 H]uridine into RNA was also diminished after prolonged exposure of the cells to the antibiotic (30 min). This can be explained by the fact that empedopeptin triggered a substantial accumulation of a soluble UDP-activated peptidoglycan precursor (UDP-N-acetylmuramic acid-pentapeptide) in the bacterial cytoplasm as outlined below, thereby depleting the uridine pool for RNA synthesis. Vancomycin, an established and specific inhibitor of peptidoglycan synthesis, induced a comparable effect at 8ϫ MIC (Fig. 2).
Peptidoglycan synthesis takes place in two different cellular compartments. At first, several cytoplasmic enzymes assemble the soluble, activated UDP-MurNAc-pp. The sugar-peptide moiety is then transferred to the membrane anchor C 55 -P to yield a membrane-standing monomer, which is further decorated, flipped across the membrane to the exterior, and incorporated into the three-dimensional peptidoglycan network. To distinguish whether empedopeptin interferes with peptidoglycan synthesis in the cytoplasmic versus the membrane-associated phase, we determined the cytoplasmic levels of UDP-Mur-NAc-pp. Antibiotics that interfere with the late stages of peptidoglycan synthesis, such as vancomycin, are known to trigger an accumulation of this ultimate soluble peptidoglycan precursor in the cytoplasm. Treatment of S. aureus with empedopeptin (10ϫ MIC) led to a significant accumulation of UDP-MurNAc-pp similar to vancomycin-treated controls at corresponding multiples of the MIC (Fig. 3A).
The results with whole cells indicated peptidoglycan synthesis as the target pathway of empedopeptin. However, as the antibiotic structure contains a lipophilic tail of considerable length, we also investigated whether the compound might cause additional damage to the cytoplasmic membrane. Disruption of membrane integrity results in a strong decrease of ⌬⌿. To determine ⌬⌿ in the presence of empedopeptin, we quantified the intracellular versus extracellular concentration of the tritium-labeled lipophilic cation TPP ϩ , which diffuses freely across the bacterial membrane in response to a transnegative ⌬⌿. When growing cells of S. aureus were treated with the antibiotic, only a slight drop of membrane potential by 20 mV was observed followed by a rapid recovery (Fig. 3B). In contrast, the addition of the uncoupling ionophore carbonyl cyanide m-chlorophenylhydrazone resulted in a substantial ⌬⌿ decrease of about 70 mV (Fig. 3B).  (31). This central cell wall building block is characteristically modified in different species. In staphylococci, the peptidyltransferases FemXAB catalyze the attachment of a specific pentaglycine interpeptide bridge to the stem peptide (21,32) before the precursor is translocated across the cytoplasmic membrane (33) and incorporated into the peptidoglycan network. As our studies with intact bacterial cells had suggested an inhibitory activity of empedopeptin on membrane-associated stages of peptidoglycan synthesis, we next investigated the effect of the antibiotic in various in vitro systems capable of the respective reactions.

Impact of Empedopeptin on in Vitro Peptidoglycan
Membrane preparations of M. luteus catalyze the membrane-associated synthesis of lipid II by MraY and MurG in vitro in the presence of defined amounts of the soluble precursors UDP-MurNAc-pp and UDP-GlcNAc as well as the bactoprenol carrier C 55 -P. In the control reaction without antibiotic, C 55 -P was completely converted to lipid II by the sequential activity of MraY and MurG, whereas addition of empedopeptin reduced the amount of lipid II (Fig. 4A). Quantitative analysis using [ 14 C]UDP-GlcNAc to specifically label lipid II revealed a concentration-dependent reduction of the lipid II amount. Addition of equimolar concentrations of empedopeptin with respect to C 55 -P resulted in a 50% decrease in extracted lipid II, and a 2-fold molar excess of antibiotic led to an almost complete absence of lipid II (Fig. 4B).

Target of Empedopeptin and Impact of Ca 2ϩ on Target
Interaction-To narrow down the molecular target of empedopeptin and to elucidate the role of Ca 2ϩ in the inhibitory pro-cess, individual peptidoglycan biosynthesis reactions were analyzed in vitro using the purified recombinant proteins MraY, MurG, FemX, and PBP2 together with their respective purified substrates in the presence and absence of empedopeptin. Testing the MurG-catalyzed conversion of lipid I to lipid II as well as the incorporation of the first glycine residue into lipid II by recombinant FemX and the transglycosylation reaction by PBP2, we found all three reactions to be inhibited in a concentration-dependent manner (Fig. 4, C-F). The addition of Ca 2ϩ considerably enhanced the effect of empedopeptin and led to a complete inhibition of the reactions at a 2:1 stoichiometry of empedopeptin to lipid precursor. Approximately physiological concentrations of calcium (1-1.25 mM) were required for empedopeptin to exert full inhibition as exemplified for the FemX-catalyzed glycine addition (Fig. 4F).
To investigate whether the stimulatory effect was specific for calcium or would likewise be triggered by other divalent cations, the influence of Mg 2ϩ , Mn 2ϩ , Fe 2ϩ , and Zn 2ϩ was explored in several of the in vitro systems described above. For Mg 2ϩ , we can only draw limited conclusions because all in vitro systems depend on the presence of Mg 2ϩ , and omitting it from the assay mixtures prevented enzyme activity. Moreover, the presence of Zn 2ϩ inhibited some enzymes, whereas others were more tolerant to the ion. Because of these experimental constraints, we measured all enzymes in the presence of 10 mM Mg 2ϩ and determined the empedopeptin efficacy after further addition of a 1.25 mM concentration of the particular cation under investigation. Under these conditions, the cations stimulated the activity of empedopeptin in the following order: Ca 2ϩ Ͼ Fe 2ϩ Ͼ Zn 2ϩ Ͼ Mn 2ϩ Ͼ Mg 2ϩ . Calcium takes up a special position among all cations tested because it substantially increased the efficacy of empedopeptin even in the presence of 10 mM Mg 2ϩ . Taking into account the physiological concentrations of the cations in human blood where Fe 2ϩ , Zn 2ϩ , and Mn 2ϩ are only available in traces, it is reasonable to presume that calcium has an important and prominent role in promoting the activity of empedopeptin against bacterial infections in the host.
It is important to note that empedopeptin inhibited the MurG, FemX, and PBP2 reactions at comparable concentrations, which made a specific interference with one of the enzymes highly unlikely and pointed to an interaction with the peptidoglycan precursors instead. The three enzymes discussed so far utilize undecaprenyl pyrophosphate-containing peptidoglycan precursors that also include a sugar moiety. In contrast, two enzymes that use the structurally simpler C 55 -P as a substrate were unaffected by empedopeptin when applied in the same molar ratio, namely MraY, which catalyzes the synthesis of lipid I (supplemental Fig. S1), and TagO, which catalyzes the formation of lipid III, the first membrane-bound lipid intermediate of teichoic acid biosynthesis (34, 35) (supplemental Fig. S2). Moreover, dephosphorylation of C 55 -PP by UppP (36) was also not markedly inhibited at a 2-fold molar excess of empedopeptin over C 55 -PP (supplemental Fig. S3).
Binding Site of Empedopeptin at Peptidoglycan Precursors-The interaction of empedopeptin with different undecaprenolbound peptidoglycan precursors was directly demonstrated by incubating each lipid precursor with the antibiotic in buffer in various molar ratios followed by analysis via thin layer chromatography (Fig. 5). Free lipid I and lipid II were clearly visible at defined positions in the chromatogram, whereas free empedopeptin was not detectable by the phosphomolybdic acid staining method applied. In complex with empedopeptin, lipid I and lipid II migrated considerably more slowly and remained close to the application spot. In accordance with the results obtained in the enzyme reaction assays described above, a 2-fold molar excess of empedopeptin was sufficient for full sequestration of the peptidoglycan precursors lipid I and lipid II in a stable complex that even withstood solvent treatment during separation by thin layer chromatography. In contrast, even a 4-fold molar excess of the antibiotic could not trap undecaprenyl pyrophosphate in a stable complex under these conditions (Fig. 5). Notwithstanding, empedopeptin had some influence on the migration behavior of undecaprenyl pyrophosphate during thin layer chromatography because the lipid moved as a more focused band in the presence of the antibiotic (Fig. 5C, compare lane 1 with lanes 2-4). To narrow down the region of the lipid precursors that interacts with empedopeptin, we performed antagonization assays with diverse cell wall precursors and truncated variants thereof ( Table 2). A concentration of empedopeptin corresponding to 8ϫ MIC was preincubated with a 0.156 -10-fold molar excess of the potentially antagonizing agents before being tested for growth inhibition of B. subtilis in Ca 2ϩ -supplemented (1.25 mM) culture broth. Lipid II and lipid I showed the strongest antagonizing effect on the antimicrobial activity of empedopeptin, confirming that the antibiotic is effectively trapped in a complex with the purified peptidoglycan precursors. Hence, it is no longer available for interaction with its membrane-standing targets when added to the cells. A lipid I congener with the peptide side chain shortened by the three C-terminal amino acids was only slightly less effective in antagonization. Also lipid III and C 55 -PP still retained substantial antagonizing potential, although their affinity for empedopeptin decreased in that order. In contrast, C 55 -P and the nucleotide-activated peptidoglycan sugars had no effect in the concentration range tested ( Table 2).
Interaction of Empedopeptin with Model Membranes-The target sites of empedopeptin are embedded among the phospholipid head groups of the cytoplasmic membrane. Bearing in mind the structure of empedopeptin with its hydrophobic C 14myristic acid tail, it is highly likely that the antibiotic establishes close contacts with the membrane that might also promote target binding. To determine the affinity of empedopeptin for phospholipids bilayers per se, we used the quartz crystal microbalance biosensor technique. The capacity of the antibiotic to interact with model membranes was investigated on the basis of the mass increase that occurs upon its binding to the lipid bilayer. A mixture of 90 mol % 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine and 10 mol % 1,2-dioleoyl-sn-glycero-3phospho-(1Ј-racemic glycerol) served as a model system for the negatively charged bacterial membranes. Kinetic binding constants, such as k a , k d , and k D , were calculated from the quartz crystal microbalance frequency curves in the presence and absence of the physiological calcium concentration. Empedopeptin demonstrated a high membrane affinity in accordance with its overall hydrophobic nature ( Table 3). Addition of calcium further increased membrane affinity by ϳ2.5-fold, resulting from a higher association rate and a diminished dissociation rate. The rapid association kinetics and the k D values match those of the lantibiotic gallidermin (37). Cyclic voltammetry revealed membrane perturbation upon binding, suggesting a tight contact of the antibiotic with the lipid layer (data not shown).

DISCUSSION
The lipodepsipeptide empedopeptin inhibits bacterial growth by blocking peptidoglycan biosynthesis. It forms complexes with undecaprenyl pyrophosphate-containing cell wall precursors, thereby interrupting the lipid cycle and depleting the utilizing enzymes of their substrates. Empedopeptin displayed the highest affinity for both lipid I and lipid II. Several distinct assay systems indicated a 2:1 binding stoichiometry of antibiotic to lipid precursors, and the complex was remarkably stable even in the presence of the butanol/pyridine acetate used to develop the thin layer plates depicted in Fig. 5.
The presence of Ca 2ϩ ions at the concentration at which they occur in human serum (11) was required for full inhibitory activity of empedopeptin in our in vitro peptidoglycan synthesis assays. As several of these assay systems contained purified enzymes and substrates, it can be excluded that this stimulating effect is merely due to facilitated interaction of the antibiotic with the bacterial membrane. Rather, our data suggest that calcium markedly promotes the direct interaction of empedopeptin with its target. Empedopeptin is an amphipathic molecule with a hydrophobic region containing the myristoyl side chain and two proline residues opposite to a decidedly hydrophilic, net negative region that includes the arginine as well as two aspartic acid residues. It is reasonable to conclude that positively charged calcium ions play a crucial role in bridging the antibiotic to the pyrophosphate region of the lipid precursors. Considering the 2:1 stoichiometry of the complex, Ca 2ϩ ions might also facilitate the formation of empedopeptin dimers by bridging two aspartic acid moieties. In addition, our study with target-free artificial phospholipids layers (Table 3) demonstrates that the interaction with negatively charged membranes is also enhanced in the presence of calcium. The strong decrease of the

of peptidoglycan and wall teichoic acid precursors and truncated variants thereof on antimicrobial activity of empedopeptin
Empedopeptin at 8ϫ MIC was exposed to the potentially antagonizing agents in the indicated molar ratios prior to mixture with B. subtilis in culture broth containing 1.25 mM Ca 2ϩ . Results of two independent experiments are summarized. ϩ, antagonization; Ϫ, no antagonization.  Calcium ions are also involved in the interaction of other antibiotics with peptidoglycan precursors. Recently it was demonstrated that the antibiotic activity of friulimicin B is dependent on Ca 2ϩ -mediated complex formation with C 55 -P (38). Calcium was also shown to promote the interaction of particular lantibiotics carrying the mersacidin-type binding motif with their target lipid II, although the antibiotic activity was only slightly enhanced in these cases (39).

UDP-GlcNAc
Our phospholipid binding study demonstrated high affinity of empedopeptin for negatively charged membranes per se. The electric currents detected by cyclic voltammetry on model membranes even pointed to membrane perturbations upon empedopeptin binding. The effects detected in this model membrane system correlated with a slight drop of the membrane potential of bacterial cells after empedopeptin addition. However, the effect on ⌬⌿ was only temporary, and there was no indication for a significant disturbance of membrane integrity. Rather, this effect suggests an intimate interaction of the lipophilic antibiotic with the cytoplasmic membrane in accordance with its membrane-standing targets. Along the same line, the continuation of DNA, RNA, and protein synthesis in the presence of empedopeptin precludes a gross depolarization of the cytoplasmic membrane. These biosynthetic processes are highly energy-consuming and require constant ATP replenishment by oxidative phosphorylation. Also, the fact that UDP-MurNAc-pp accumulated in the cytoplasm indicates a functional membrane barrier because the precursor was retained inside the cells.
In antagonization assays, very low amounts of externally added lipid I or lipid II were sufficient to fully complex free empedopeptin and to prevent its interaction with bacterial cells ( Table 2). Among the different peptidoglycan precursors tested, lipid II and lipid I were equally well bound by the antibiotic, indicating that the second sugar GlcNAc is not crucial for the interaction. The C-terminal region of the stem peptide containing the L-lysyl-D-alanyl-D-alanine moiety also does not significantly contribute to binding as evident from the comparison of lipid I with lipid I(dipeptide). This result highlights that the target site of empedopeptin differs from that of the glycopeptide antibiotics (40) and is in accordance with the activity of empedopeptin against vancomycin-resistant bacterial isolates. However, if the cell wall precursor lacks the whole stem peptide and the lactyl moiety of MurNAc as it occurs in lipid III, its potential for complex formation is 4-fold lower than that of lipid I or lipid II. Undecaprenyl pyrophosphate as such is also recognized by empedopeptin, but the absence of the GlcNAc moiety further reduced the affinity by a factor of 2. It is also noteworthy that in a similar antagonization study with tripropeptin C 20 -PP showed stronger antagonizing potential than FIGURE 6. Overview of peptidoglycan biosynthesis in S. aureus and schematic model for mechanism of action of empedopeptin. Starting from UDP-GlcNAc, the ultimate soluble precursor UDP-MurNAc is synthesized in the cytoplasm by the sequential action of MurA to MurF enzymes. Lipid I is formed in the first membrane-associated step at the inner face of the membrane by MraY, which transfers the soluble sugar-peptide moiety to the lipid carrier C 55 -P. The transferase MurG subsequently links a GlcNAc moiety to the muramoyl portion of lipid I yielding lipid II. In S. aureus, lipid II is further decorated by the addition of five glycine residues, catalyzed by FemXAB enzymes. Finally, lipid II is translocated across the membrane, and the peptidoglycan-monomer is incorporated into the growing peptidoglycan network by transglycosylation and transpeptidation. Once the lipid II monomer appears at the outside of the cytoplasmic membrane, empedopeptin binds to it in a Ca 2ϩ -dependent manner at a region involving the pyrophosphate group, the first sugar, and the proximal parts of the stem peptide and the undecaprenyl chain. Dimerization of empedopeptin may occur in the course of this process. C 55 -PP is also bound at the outer leaflet of the cytoplasmic membrane but with lower affinity than lipid II. Size, shape, and positioning of all components are schematic and not based on structural data. PEP, phosphoenolpyruvate. C 15 -PP (9), suggesting that the undecaprenyl chain also takes part in the interaction. In summary, the interaction of empedopeptin with lipid II appears to involve the pyrophosphate group, the first sugar, and the first one or two amino acids of the stem peptide as well as the proximal part of the undecaprenyl chain (Fig. 6).
Because of its comparably large size, empedopeptin will probably not enter the bacterial cell. Whether it can flip across the membrane to interact with lipid I that resides within the inner phospholipid leaflet is questionable and remains to be studied. Therefore, the surface-exposed lipid II appears to be the likely physiological target of empedopeptin. Because lipid II is bound with high affinity, its sequestration at low antibiotic concentrations and concomitant inhibition of peptidoglycan biosynthesis are probably the primary killing event. As shown for a variety of other antibacterial agents among them lipopeptides, lantibiotics, and defensins (25,41), lipid II binding is sufficient to prevent the synthesis of novel peptidoglycan followed by bacterial cell lysis and death. Bacterial cells exposed to a lipid II binder do not merely suffer from inhibition of the transglycosylation step in peptidoglycan synthesis but they also suffer from interruption of the lipid cycle and general depletion of the undecaprenyl pool. Depletion of free bactoprenol inhibits several reactions in peptidoglycan synthesis and the membraneassociated stages of teichoic acid and capsule biosynthesis, which also rely on C 55 -P as a carrier.
However, empedopeptin differs from several other lipid II binders described to date (e.g. mersacidin and plectasin) (18, 25) by its affinity for additional bactoprenol-containing precursors that appear outside the cell. C 55 -PP can be sequestered as probably can the lipid III structural motif within extracellular teichoic acid precursors (e.g. within C 55 -PP-GlcNAc-N-acetylmannosamine-(glycerol phosphate) 2 -(ribitol phosphate) 10 -40 in S. aureus) (42). Among these structures, empedopeptin has the highest affinity for lipid II, which we consider its primary and main interaction site. However, if bacterial cells are exposed to sufficiently high empedopeptin concentrations, the antibiotic will probably recognize the other structures as well, thereby trapping the bactoprenol carrier at more than one stage in parallel. As interaction with multiple targets is beneficial in the light of reduced resistance development (43,44), this characteristic is an advantage of empedopeptin.
While this work was in progress, a report on the mechanism of action of tripropeptin C was published, proposing C 55 -PP as the primary target and its dephosphorylation as the target reaction (9). The authors based their conclusion on biosynthesis reactions with bacterial membrane fractions and a MIC-type antagonization assay with externally added C 55 -PP. Similar to our results, they found that the activity of tripropeptin C was antagonized by a comparably high molar excess of C 55 -PP. However, neither purified lipid I, II, or III nor purified peptidoglycan biosynthetic enzymes were used in this study. As a consequence, the assay systems applied did not allow measurement of the affinity of tripropeptin C for the full spectrum of peptidoglycan precursors or investigation of individual enzyme reactions separately in the absence of other interfering precursors and enzymes. On the basis of the clear results obtained in our defined systems and in light of the strong structural similarity between empedopeptin, the tripropeptins, and the plus-bacins (Fig. 1), we propose that complex formation with lipid II as the primary target and with C 55 -PP and the lipid III moiety as secondary targets represents the mechanism of antibacterial action for this entire class of compounds.