Portal de Periódicos da CAPES

Antibiotic Action and Peptidoglycan Formation on Tethered Lipid Bilayer Membranes

2006; Wiley; Volume: 45; Issue: 13 Linguagem: Inglês

10.1002/anie.200504035

1521-3773

Michael Spencelayh, Yaling Cheng, Richard J. Bushby, Timothy D. H. Bugg, Jianjun Li, Peter J. F. Henderson, John O’Reilly, Stephen D. Evans,

Bacteriophages and microbial interactions

At the end of its tether: Solid-supported tethered lipid bilayers that present native versions of the peptidoglycan precursors lipid I and lipid II are used to compare the binding of vancomycin and ramoplanin. Through the use of E. coli inner membranes, the resulting "proteolipid layer" contains the full range of intrinsic and extrinsic proteins found in vivo, thus allowing the in vitro biosynthesis of a peptidoglycan cell wall. E=enzyme. The inner-bilayer membranes of Gram-positive bacteria are surrounded by a layer of peptidoglycan, in the form of a polymer net, which serves to provide mechanical support against lysis due to osmotic stress. Prevention of peptidoglycan formation makes bacterial cells prone to lysis and represents one of the three main mechanisms of antibiotic action; the others being inhibition of protein synthesis and the inhibition of RNA/DNA synthesis.1 Vancomycin has been a mainstay of the glycopeptide antibiotics and is active against a wide range of Gram-positive bacteria. However, the emergence of widespread resistance amongst pathogenic bacteria has highlighted the need for new and effective antibacterial compounds, and consequently there has been a significant effort devoted to understanding the mechanisms of antibiotic action and resistance.2a–2e Most studies performed, to date, on understanding the structure–function relationship involved in antibiotic interaction with peptidoglycan precursors have focused on the use of micelles, vesicles, and soluble precursors. Although important information has been gained from such studies, there are drawbacks in that either they do not represent a close mimic of the in vivo situation or, in the case of vesicles, they are not amenable to study by using the array of surface-science techniques currently available. In 1997, Williams et al. formed a "hybrid" bilayer that consisted of a lipid monolayer adsorbed onto a hydrophobic self-assembled monolayer (SAM).3a The lipid monolayer was decorated with N-α-Docosanoyl-ε-acetylLysD-AlaD-Ala (Doc-KAA) and was shown to bind vancomycin and other glycopeptide antibiotics. The dissociation constant for vancomycin was determined to be ≈0.7 μM. In a subsequent study in which the terminal D-Ala residue was replaced by a D-lactate residue, the degree of binding was reduced significantly, yielding a Kd value of ≈1000 μM.3b Solid-supported bilayer lipid membranes (sBLMs) and tethered-bilayer lipid membranes (tBLMs) have been used widely in recent years for the investigation of ion-channel proteins and the development of biosensors.4a–4f These bilayers are fluid and represent a reasonable mimic, for many purposes, of the cell membrane.4e,4f Herein, we show that tBLMs can provide a suitable platform for addressing questions related to antibiotic resistance. We started by introducing two precursors to peptidoglycan formation—namely, the full-length "native" versions of lipid I and lipid II. The incorporation of functional lipid I or lipid II is demonstrated by specific binding of vancomycin and ramoplanin. In a further step, we showed that native E. coli membranes can be used to form tBLMs (and sBLMs) and that these contain the necessary functional proteins for the in vitro synthesis of the peptidoglycan cell wall. This system possibly represents the closest mimic of the in vivo situation yet. Figure 1 shows the major steps involved in in vivo cell-wall biosynthesis. Soluble UDPMurNAcpentapeptide (UDP=undecaprenyl phosphate, MurNAc=N-acetylmuramic acid) is localized at the membrane surface by the MraY enzyme and attached to the lipid carrier UDP. The resulting "lipid I" molecule is modified by the enzyme MurG, with the addition of UDPGlcNAc (GlcNAc=N-acetylglucosamine), forming the "lipid II" molecule. Lipid II "flips" from the cytoplasm to the outside of the cell, where it is cross-linked, by the transglycosylase and transpeptidase enzymes, to form the peptidoglycan cell wall. The undecaprenyl lipid carrier is then recycled.5, 6a,6b, 7 The glycopeptide antibiotic, vancomycin, and the lipodepsipeptide antibiotic, ramoplanin, are believed to inhibit peptidoglycan biosynthesis by complexation with the peptidoglycan precursor (lipid II).2a, 5, 6a,6b, 8 Vancomycin prevents transglycosylation by binding to the C-terminal D-AlaD-Ala groups. The emergence of vancomycin resistance, due to mutation of the terminal D-AlaD-Ala group to D-AlaD-lactate has highlighted a need for alternative antibiotics.1 Ramoplanin is a potential candidate, as it has been shown to inhibit the transglycosylasation step by binding to the MurNAcL-Ala portion of lipid II, that is, close to the membrane interface.9a–9f Schematic diagram showing the principal stages involved in in vivo cell-wall biosynthesis. UMP=uridine-5′monophosphate. The tethered lipid bilayers (Figure 2 A) were formed by immersing a "mixed", cholesteryl containing, self-assembled-monolayer (SAM) modified support into a solution containing egg-phosphatidylcholine (PC) vesicles.4b The vesicles contained lipid I or lipid II (20 % by weight). Bilayer formation was traced by using surface plasmon resonance (SPR) and impedance spectroscopy. The fully formed bilayers gave capacitance values of 0.69 μF cm−2, which is consistent with the determined thickness (from SPR) and indicates the formation of continuous bilayers with few defects. A) Schematic diagram showing the binding of glycopeptide antibiotics to cell-wall precursors in a tethered bilayer containing lipids I or II. B) Vancomycin binding to tethered bilayers containing lipid I. The Figure shows the increase in adsorbed-layer thickness, Δd, as a function of antibiotic concentration, determined using SPR. C) Vancomycin binding to tethered bilayers containing lipid II. Error bars indicate the deviation observed between multiple experiments. The binding of vancomycin to lipid I (and lipid II) was measured through the incubation of the bilayers with solutions containing increasing concentrations of antibiotic. The amount of adsorbed material was determined by using SPR and plotted against the vancomycin concentration (Figure 2 B, C2). The data presented were corrected for nonspecific binding by subtraction of the change in thickness observed in the control experiments from that of pure egg-PC bilayers, that is, with no lipid I (or II) present, and tested over the same range of antibiotic concentrations. In all cases, the degree of nonspecific binding was less than 20 % of that observed in the presence of lipid I (or II). The data were fitted by using the Hill equation to obtain estimates of the Hill coefficient, nH, and the dissociation constant Kd (Table 1). The Hill coefficient relates to the degree of cooperativity in the binding event, with nH 1 indicating positive cooperativity. In both cases, we found that nH>1 (for lipid I and lipid II), suggesting that the binding of vancomycin to lipid II (or I) lowers the energy barrier for the binding of a subsequent vancomycin molecule (or dimer). Our results also indicate that vancomycin binding to lipid II is stronger (lower Kd) and has a greater degree of cooperativity (higher nH) than to lipid I. The difference in Kd values (for lipid I and lipid II) implies that interactions between the precursor and sugar groups, attached to the peptide backbone of vancomycin, may also influence the nature of the interaction.10a,10b It should be noted, however, that the Kd values for vancomycin are approximately an order of magnitude larger than those for the surface-bound peptide (KAA) analogues.3a–3d To our knowledge, there has only been one study of vancomycin binding to lipid II, by Vollmerhaus and co-workers, in which a water soluble lipid II analogue was compared to the more widely studied KAA model peptide.3e The dissociation constant for the lipid II analogue was less than half that for the model peptide and similair to those reported by Williams, Whitesides, and co-workers.3a–3d Although the absolute values determined from our systems do differ, the observed trends are similar despite that our systems are different (using native precursors embedded in a lipid bilayer). Bilayer/Antibiotic nH[a] Kd [μM][b] lipid I/vancomycin 1.7±0.3 91±8 lipid II/vancomycin 3.5±0.1 59±0.5 lipid I/ramoplanin 1.3±0.1 7.8±0.4 lipid II/ramoplanin 2.1±0.5 1.6±0.2 Figure 3 A shows a real-time SPR trace that tracks the binding of ramoplanin to a bilayer, incorporating lipid I, over a range of antibiotic concentrations. Figure 3 B and C3 show the change in adsorbed layer thickness, as a function of ramoplanin concentration, upon incubation with bilayers that contain lipid I and lipid II, respectively. These data fits give Hill coefficients greater than one, suggesting the presence of a cooperative phenomenon. Further, the value of nH for binding to lipid II is nearly twice that of lipid I, whereas the dissociation constant is approximately four-times smaller (Table 1). Walker, McCafferty, and co-workers have independently shown that ramoplanin undergoes aggregation resulting in formation of insoluble fibrils in the presence of the soluble analogues of lipid I and II.9a–9f The detailed 1H NMR spectroscopy studies performed by Walker, McCafferty, and co-workers provided a major advance in the identification of the binding interface and provided estimates of the dissociation constants for the soluble precursors citronellyl–lipid I (Kd<100 μM)9d and citronellyl–lipid II (Kd≈50 nM).9c More recently, Walker and co-workers have reported that ramoplanin binds to analogues of lipid II between five and ten-times more strongly than to analogues of lipid I.9g As ramoplanin binding involves the MurNAcGlcNAc moiety of lipid II, which resides close to the membrane interface, one might expect steric effects to affect the magnitude of the Kd values observed in our studies (compared to those found by using the water-soluble analogues cited above). Notwithstanding this, it is clear that our membrane consists of "native" peptidoglycan precursors that behave in a qualitatively similar manner to those reported by Walker and co-workers, and supports the notion that lipid II is the primary target (and that inhibition of the transglycosylase step is the primary mode of antibacterial action).9g A) SPR measurements showing ramoplanin binding (change in thickness; Δd) to a tethered bilayer containing lipid I as a function of time. The vertical dashed lines indicate points at which the concentration of the antibiotic was increased. B) Change in thickness, Δd, of the adsorbed layer on a bilayer containing lipid I as a function of ramoplanin concentration. C) Change in thickness, Δd, of the adsorbed layer on a bilayer containing lipid II as a function of ramoplanin concentration. Error bars indicate the standard deviation observed between multiple experiments; note that experiments on lipid II containing bilayers were not repeated due to limited stocks of ramoplanin. Ram=ramoplanin. The native E. coli inner membrane contains the full range of intrinsic and extrinsic proteins necessary for bacterial cell-wall formation. Vesicles of this membrane, created by using the French-press technique, were used to form sBLMs and tBLMs of the native inner membrane. Figure 4 shows the kinetics of bilayer formation, as determined from SPR, and indicates that the bilayers are formed within 20 min and are stable to rinsing with buffer solution (see Experimental Section). The average thickness of the adsorbed lipid film was 3.8±0.2 nm (obtained from fitting the "before" and "after" curves shown in the inset). If one takes into account the insertion of anchor groups into the lower leaflet of the membrane, we would expect a thickness change of 3.6 nm (taking the E. coli membrane thickness to be 4.5 nm),11 that is, close to that found experimentally. SPR trace showing the kinetics of adsorption (change in thickness, Δd, versus time). Insert: SPR curves showing the angular shift in the reflectance, R, minimum following adsorption of a 3.5-nm tethered proteolipid layer on a mixed SAM. Figure 5 A shows an SPR trace showing the change in average thickness during peptidoglycan biosynthesis onto a preformed E. coli tBLM. At t=0, the soluble precursors UDPMurNAcpentapeptide and UDPGlcNAc (each at a concentration of 0.05 mg mL−1) were introduced to the bathing solution. The resulting increase in thickness can be attributed to the attachment, modification, and cross-linking of the pentapeptide to form a peptidoglycan cell wall.6b, 7 In vivo, the first steps of peptidoglycan formation occur on the inner leaflet of the membrane before "flipping" to the outer leaflet for the later stages. It is likely that the in vivo bacterial inner-membrane proteins are arranged asymmetrically in the bilayer. The French-press method of vesicle preparation used herein forms both the inside-out and right-side-out vesicles. As a result, the "flipping" stage is not required for peptidoglycan synthesis on a tethered-membrane bilayer. The average thickness of the cell wall was 1.6±0.2 nm. After a rinsing with clean buffer solution to remove any adventitiously bound material, this value fell to 1.1±0.1 nm. The thickness of a fully formed E. coli cell wall in vivo is typically close to 3 nm, thus our data correspond to about 35 % of a full peptidoglycan layer.11 This compares favorably to previously reported values of in vitro cell-wall biosyntheses (measured in vesicles by the incorporation of radiolabeled precursors), which achieved 2 % of the in vivo level.6b Traces 5 B–E show a number of control experiments that were designed to test the hypothesis outlined above. Trace 5 B shows the same experiment as that of trace 5 A, but in the absence of adenosinetriphosphate (ATP) and MgCl2. ATP is required for the recycling of UDP molecules and Mg2+ ions that are cofactors for the proteins of cell-wall biosynthesis. In this experiment, no increase in thickness was observed, indicating the necessity for recycling of UDP to create sufficient quantities of lipid I in the bilayer. In the case of trace 5 C, the UDPGlcNAc molecule is missing from the experiment, and as a result, cell-wall biosynthesis can only proceed as far as the attachment of UDPMurNAcpentapeptide to the surface. Although lipid I to lipid II modification cannot occur as there is no substrate for the enzyme, a small change in thickness is observed ≈0.1 nm. This is possibly attributable to the localization of unmodified pentapeptide at the surface and is within the uncertainty of our measurements. The fourth trace, 5 D, shows an experiment wherein both UDP-MurNAcpentapeptide and UDPGlcNAc were omitted. The bilayer was rinsed with buffer solution containing only ATP and MgCl2, and resulted in no increase in the thickness of the lipid layer. Trace 5 E was obtained under the same conditions as trace 5 A except that a pure egg-PC bilayer was used rather than the E. coli inner membrane. Once again, there was no evidence of any cell-wall biosynthesis. Finally, cell-wall biosynthesis was attempted in the presence of 5 mM Vancomycin, a glycopeptide antibiotic that disrupts cell walls by binding to lipid-linked precursors prior to transglycosylation.2a–2e The resulting adsorbed film (see the inset in Figure 5) was readily removed by a single rinse with clean buffer solution, suggesting (at best) the formation of a weakened cell wall in the presence of the antibiotic. SPR kinetics traces showing changes in adsorbed film thickness, Δd, as a function of time. Trace A shows the formation of a peptidoglycan layer on a tethered E. coli membrane. Traces B–E show the results obtained in a number of control experiments (offset for clarity). Trace B shows the absence of ATP/Mg2+, and Trace C shows the absence of UDPGlcNAc. Trace D shows the absence of both UDPGlcNAc and UDPMurNAcpentapeptide; and in trace E, the conditions are as for trace A except that the bilayer is formed from pure egg-PC vesicles. Inset: shows bilayer formation, followed by an attempt at peptidoglycan biosynthesis after the introduction of Vancomycin (VAN; 5 μm). These controls demonstrate that the increase in thickness observed requires; 1) the presence of both soluble cell-wall precursors (lipid I and II), 2) the presence of ATP and Mg2+, 3) the use E. coli inner membranes, and 4) that it can be disrupted by glycopeptide antibiotics. These data strongly suggest that protein-mediated peptidoglycan cell-wall biosynthesis is occurring at the membrane surface, demonstrating that the proteins involved in bacterial cell-wall synthesis have been maintained in a functional form. Herein we describe the first demonstration of a tethered bacterial membrane that is capable of the biosynthesis of a peptidoglycan cell wall. This system readily lends itself to a range of applications including the search for novel antibiotics and fundamental studies of cell-wall formation, for example, it is not known whether ramoplanin–lipid II complexes polymerize in the membrane as they do in solution. The planar nature of the tBLMs not only allows investigation with a wide range of analytical techniques, but also is compatible with array formation for high-throughput screening of novel bactericidal agents. The mixed SAMs were comprised of "anchor" molecules (HS(CH2CH2O)5 CH2CH2NHCHO2-Cholesteryl) that were 50 % by mole fraction on the surface and were designed to tether the lipid bilayer to the solid support. "Spacer" molecules (HS(CH2)2OH) were used to promote a hydrophilic region beneath the lipid bilayer. These SAMs were fully characterised by using X-ray photoelectron spectroscopy (XPS), wetting, and Ellipsometry. The synthesis of lipids I and II was carried out according to the method of Brandish et al. with the exception that they were extracted by using 1-butanol.12 E. coli strain JM1100 (pPER3) was grown, and the membranes were prepared by using a French press to make the inner-membrane vesicles (outer membrane not formed with this method).13 All bilayers and cell walls were formed and rinsed in a buffer solution that contained Tris chloride (20 mm; pH 7.5; Tris=tris(hydroxymethyl)aminomethane), ATP (1 mm), and MgCl2 (1 mm). Soluble UDPMurNAcpentapeptide was isolated from the Bacillus subtilis strain W23, HCINB 11824 (17-18). UDPGlcNAc was obtained from Sigma (UK). SPR experiments were performed in the Kretschmann configuration.4b A 50-nm layer of gold was thermally evaporated onto a high-refractive-index glass prism, n=1.847. Resonance curves were obtained by recording the reflected intensity, R, that was normalized with respect to the incident intensity, Ro, as a function of the angle of incidence. Resonance curves were fitted by using a Levenberg–Marquardt algorithm, assuming each layer to be optically isotropic. With abrupt interfaces, a scattering-matrix method was used. The aqueous ambient was assumed to have a refractive index of n=1.333, whereas that of the lipid, antibiotics, and peptidoglycan were assumed to be 1.5. Supporting information for this article is available on the WWW under http://www.wiley-vch.de/contents/jc_2002/2006/z504035_s.pdf or from the author. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.