| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,1
* Biomembrane Structure Unit, Department of Biochemistry, University of Oxford, Oxford, UK;
Department Biochemistry of Membranes, Institute of Biomembranes Utrecht University, Utrecht, Holland;
Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw, Poland; and
School of Biomedical Sciences, University of Nottingham, Nottingham, UK
1Correspondence: School of Biomedical Sciences, University of Nottingham, Nottingham, NG7 2UH, UK. E-mail: Boyan.Bonev{at}nottingham.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Key Words: lantibiotic • staphylococcal resistance • penicillin binding protein • vancomycin
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
, and enterococcal and staphylococcal resistance to vancomycin have been reported (2
, 3)
. These antibiotics interfere with the biosynthesis of bacterial cell walls by blocking the active sites of enzymes involved in cell wall formation (e.g., penicillin binding proteins) or by binding directly to cell wall precursor molecules (e.g., Lipid II for vancomycin). Another class of bactericidal peptides, the bacterial lantibiotics, also target peptidoglycan formation. So far, lantibiotics have found use mainly in the control of food-borne pathogens, and selective lantibiotic pressure has not been applied systematically to important human pathogens (e.g., S. aureus or S. pneumoniae). This presents them as a new alternative to the conventional cell wall inhibitors, to which resistance has become alarmingly common, and as a model system for the development of new classes of cell wall inhibitors. The cell walls of bacteria consist of long-chain oligosaccharides cross-linked via polypeptide bridges (peptidoglycan). The peptidoglycan network encapsulates the bacterial plasma membranes, gives it structural stability and, in this way, protects bacteria from the osmotic challenges of their environment. The peptidoglycan is a dynamic system in which degradation by autolysins and incorporation of new peptidoglycan monomers are tightly coupled and controlled.
The final part of bacterial cell wall synthesis, incorporation of glycopeptide monomers into the peptidoglycan (Fig. 1
), takes place on the outer leaflet of the bacterial membrane. It is therefore relatively accessible and an excellent target for antibacterial action. The building block of the cell wall, the N-acetylglucosamyl-N-acetylmuramyl pentapeptide (GlcNAc-MurNAc-5Pep), is synthesized in the cellular interior in two steps. In the first step, MurNAc-5Pep-phosphate from UDP-MurNAc-5Pep carrier is attached to undecaprenyl phosphate (11P) located on the inner leaflet of the bacterial membrane (Lipid I). Lipid I is then converted into Lipid II after acquisition of a GlcNAc moiety form, UDP-GlcNAc. Lipid II (Fig. 2
) (4)
is translocated to the outer membrane leaflet via an unknown mechanism. The disaccharide-pentapeptide is then incorporated into the oligosaccharide chains of the peptidoglycan and cleaved from the pyrophosphate. The remaining carrier phospholipid, undecaprenyl pyrophosphate, 11PP, is recycled to the membrane interior and dephosphorylated to undecaprenyl monophosphate (11P), where it re-enters the precursor biosynthesis cycle (5)
. During these final stages of the cell wall synthesis cycle, two pyrophosphate-containing cell wall precursors, Lipid II and 11PP, reside on the outer leaflet of the bacterial membrane.
|
|
A number of antibiotics target the pathway of bacterial cell wall synthesis. Penicillins and cephalosporins inhibit transpeptidation by direct binding to transpeptidases (6)
, vancomycin hinders cross-peptidation by binding to the key in this process C-terminal L-Lys-D-Ala-D-Ala residues of the pentapeptide of Lipid II (7)
, and bacitracin prevents 11PP form re-entering the biosynthetic cycle (8)
.
A widely used food preservative (E234), the peptide lantibiotic nisin (9
, 10)
(Fig. 3
) produced by Lactococcus lactis, also targets bacterial cell wall synthesis in its bactericidal action (11
, 12)
. Nisin is a positively charged peptide that is able to bind to negatively charged plasma membranes (13
14
15)
via nonspecific electrostatic interactions. The antibacterial activity of nisin results from pore formation in the bacterial plasma membrane, which leads to dissipation of the transmembrane potential and vital solute gradients (16
, 17)
. The high efficiency of pore formation is the result of a unique mechanism involving the cell wall precursor Lipid II. Lipid II significantly increases the affinity of nisin for the membrane (11
, 18)
, stabilizes a transmembrane orientation of nisin (19)
, and forms and integral part of the nisin pore (20)
. Due to this high affinity of nisin for Lipid II, nisin variants incapable of permeabilizing the membrane, still exercise bactericidal action by sequestering Lipid II form the cell wall biosynthesis; therefore, nisin has been proposed to possess a dual mode of action (18)
.
|
Here we investigate the mechanism of molecular target recognition by nisin during its bactericidal action and, more specifically, the molecular association of the peptide with pyrophosphate-containing cell wall intermediates on the outer leaflet of bacterial membranes. We propose a new method for assaying intermolecular association within lipid bilayers based on the use of solid-state NMR spectroscopy. We used the new assay in combination with a carboxyfluorescein leakage assay to identify 11PP as a target for nisin in addition to Lipid II and an additional mode of action for nisin. The use of NMR and leakage assays made possible the distinction between binary molecular complex formation from lytic pore formation. We suggest that nisin recognizes Lipid II via a two-site docking mechanism. We found that the presence of a pyrophosphate was essential to the affinity of nisin for the cell wall precursors. The binding site within nisin responsible for the interaction of nisin with the pyrophosphate was shown to be the N-terminal Ile1-amine. We suggest that a secondary binding site for nisin exists on Lipid II that is different from the pyrophosphate group.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
; [15N]-nisin was isolated and purified similarly from nisin producing bacteria grown on rich medium with 15N as the sole nitrogen source. Nisin was permethylated by reductive methylation of the 
-amino group of the lysine residues as well as the amino group of the N terminus (22)
, following the same procedure described (13)
, with the exception that in order to obtain complete methylation, a large (20-fold) molar excess of formaldehyde over nisin was used. The presence of eight methyl groups in the permethylated nisin was confirmed by mass spectrometry. Lipid I and Lipid II were produced following a semisynthetic method (20)
from 11-P and 11-PP. The latter two precursors were obtained from phosphorylation of undecaprenol isolated from Laurus nobilis according to ref 23
. Synthetic dioleoyl phosphatidylcholine (DOPC) and dioleoyl phosphatidylglycerol (DOPG) were purchased from Avanti Polar Lipids at >98% purity and used without further purification. Samples for NMR experiments were prepared by codissolving DOPC, DOPG (3:1 molar ratio) and 1 molar % of the corresponding cell wall precursor in a 1:1 chloroform/methanol mixture. The solvent was removed under vacuum and the dry lipid films were hydrated in nisin-containing buffer (10 mM MES, 100 mM NaCl, 1 mM EDTA, pH 6.0). Multilamellar vesicles were obtained by 5-fold freeze-thawing between liquid nitrogen and a 50°C water bath. The samples were pelleted by 1 h centrifugation at 400,000 g and the pellet was loaded into 6 mm magic angle spinning (MAS) NMR zirconia rotors (Chemagnetics/Varian, Palo Alto, CA, USA).
Carboxyfluorescein leakage measurements were performed as described (18)
using large unilamellar DOPG/DOPC (1:3 molar ratio) vesicles containing 1 mol % peptidoglycan precursor.
High-resolution cross-polarization (CP) MAS NMR (24
, 25)
experiments were performed on a CMX-Infinity 400 spectrometer equipped with a 6 mm MAS HX Apex probe (Chemagnetics/Varian). A single 62.5 kHz pulse was used to excite the entire 31P nuclear population and the FID was acquired under 37 kHz continuous wave proton decoupling with a 5 s interpulse delay. Cross-polarization MAS experiments, during which 92.5 kHz proton excitation was followed by 1 ms of 65 kHz contact field and 55 kHz TPPM proton decoupling (26)
during acquisition and 2 s interpulse delay, were used to excite only relatively immobile 31P nuclei. Heteronuclear dipolar recoupling experiments, REDOR (27)
, were performed on a CMX-Infinity 500 spectrometer equipped with a 6 mm MAS HXY T3 probe (Chemagnetics/Varian). Proton excitation was achieved using a 111 kHz pulse, followed by 1 ms of 58 kHz contact field and 52 kHz TPPM decoupling during acquisition. A 62.5 kHz field was used for refocusing 15N pulse, and the strength of the (24 to 48) 31P dephasing pulses was 83 kHz. Between 65536 and 131072 transients were averaged with an interpulse delay of 2.5 s and a spectral width of 50 kHz. Alternating measurements were taken and averaged independently for the dephased and nondephased experiments. Each FID consisted of 1024 data points at 20 µs dwell time. Phosphorus-31 spectra were referenced externally to 0 ppm for phosphoric acid (85%). The 15N spectra were referenced externally to 15NH3Cl and chemical shifts are reported relative to TMS.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
40 ppm. Under 5 kHz MAS conditions, this is seen as the presence of single, low intensity rotational sidebands separated from the principal high-resolution lines by the magnitude of the rotation speed. The absence of rotational sidebands from any of the membrane-incorporated precursors even at very low rotational speeds (<800 Hz; not shown) reveals the presence of additional motional freedom in the pyrophosphate/phosphate region (Fig. 4
A). Such motional freedom facilitates the complete averaging of the residual CSA and only a central high-resolution resonance is observed from each precursor phosphate. The dipolar couplings between the phosphates and the protons located in their vicinity are averaged, and the pyrophosphates are not observed after cross-polarization (Fig. 4B
).
|
Individual phosphates from the pyrophosphate groups in Lipid II and Lipid I were well resolved at –8 and –10 ppm. The two phosphates in 11PP produce overlapping lines at –10 ppm. The longitudinal relaxation times of the lipid matrix and the precursor were within 1/5 of the oversampling delay time and a single pulse excitation, followed by proton decoupling, was used for acquisition. The isotropic chemical shift values of the principal resonances from each phospholipid are summarized in Table 1
.
|
Nisin was added in consecutive experiments to DOPC/DOPG bilayers containing 1% of Lipid II, Lipid I, 11PP, or 11P at 2/1 molar excess to each precursor. The 31P MAS NMR spectra, acquired from Lipid II-containing bilayers after a single pulse excitation, showed only matrix DOPC and DOPG resonances (Fig. 4C
). The resonance lines arising from the Lipid II phosphates were saturated under the 5 s pulse repetition due to an increase in their 31P longitudinal relaxation time.
The Lipid II phosphates were well resolved in a short mixing time (1 ms) cross-polarization experiment (Fig. 4D
), where excitation of the matrix DOPC and DOPG phosphates was less efficient (optimal mixing at 10–15 ms). Four orders of sidebands, bracketing the pyrophosphates’ center bands, were resolved under 5 kHz MAS. The overall width of the phosphate spectrum was 200 ppm, which corresponds to the typical CSA of immobile phosphates. This dramatic increase in the observed CSA, as well as the increase in 31P T1, reflects the presence of severe motional restrictions imposed on the pyrophosphates by binding of Lipid II to nisin. Such loss of molecular mobility allows for a high efficiency of 31P and 1H cross-polarization during the short mixing time, which contributes to the relatively high intensity of the pyrophosphates observed in comparison to the DOPC/DOPG lipid matrix. The bilayer matrix lipids DOPC and DOPG underwent only subtle changes in response to the addition of nisin (cf. ref 15
), which shows the strong affinity of nisin for Lipid II. This selective immobilization of Lipid II by nisin is a clear evidence of the formation of nisin/Lipid II molecular complexes of monomeric or oligomeric nature. Besides the dramatic increase in effective CSA, the addition of nisin caused a change in the chemical shift of the two phosphates of 2 ppm downfield. This suggests the presence of a positive charge in the immediate vicinity of the pyrophosphate moiety (28)
. Similar results were obtained when Lipid II was replaced by Lipid I, indicating that the GlcNAc moiety of Lipid II does not play a role in the interaction of nisin with Lipid II.
Under conditions used to observe the immobilization of Lipids I and II, we tested for the presence of pores in the model membranes using carboxyfluorescein leakage experiments (Table 2
). In both cases, addition of nisin to 1 mol % Lipid I- or II-containing vesicles (DOPG/DOPC 1:3 molar ratio) resulted in the rapid efflux of carboxyfluorescein (CF) and a sharp increase in fluorescence intensity. No difference could be observed between the two systems (not shown). Therefore, nisin/Lipid I as well as nisin/Lipid II complexes observed by high-resolution 31P CP MAS NMR permeabilize the lipid membranes via the formation of sufficiently large nonselective oligomeric pores. The observed pore formation by nisin in the presence of either lipid, and the indistinguishable spectroscopic feature of the two precursors suggest that the GlcNAc moiety of Lipid II is not important for pore formation. To test the importance of the MurNAc-pentapeptide moiety for the pore-forming activity of nisin, CF leakage experiments were performed with 1% 11-PP containing vesicles (DOPG/DOPC 1:3 molar ratio). No leakage of CF could be detected even in excess of nisin (not shown). Thus, the MurNAc-pentapeptide moiety is absolutely required for the pore-forming activity of nisin.
|
The lack of pore formation by nisin in 11PP-containing membranes pointed to this compound as a negative control in the complex formation assays, shown for Lipid I and II. Single pulse excitation 31P MAS NMR spectra from 1% 11PP in DOPC/DOPG bilayers showed a single resonance at –10 ppm from both phosphates in 11PP (Fig. 5
a). This differs from the 31P solution spectrum of 11PP at pH 8, where the terminal phosphate is observed at –6.7 and the alkylated phosphate at –9.7 ppm, possibly as a result of the substantial difference between the phosphate hydration in solution and the lipid bilayer. Nisin was added at a 2:1 molar excess to DOPC/DOPG bilayers, containing 1% 11PP. The 31P CP MAS spectrum from the pyrophosphate showed a single broad component at –10 ppm, with FWHH of almost 5 ppm and three orders of spinning sidebands (Fig. 5b
). The presence of a high number of spinning sidebands clearly indicates, contrary to our expectation, the formation of stable nisin/11PP complexes. The large line width must reflect the lack of a unique binding location on nisin for the pyrophosphate group or of a unique orientation of a putative hydrogen bond with nisin.
|
The above experiments suggest a specific interaction of nisin with the pyrophosphate. We tested this specificity by examining the interaction of nisin with 11P-containing membranes. The bilayer-reconstituted 11P produced under a single pulse excitation MAS NMR a single resonance, overlapped by the matrix phospholipids, at a chemical shift characteristic of free inorganic phosphate (Fig. 5c
). No changes in this behavior were observed in the presence of nisin either under a single pulse excitation or after CP (Fig. 5d
). Therefore, removal of one of the phosphates completely abolishes the ability of nisin to engage 11-P into membrane-bound complexes. The lack of nisin induced CF leakage from 11-P doped vesicles confirmed this observation (not shown).
The large downfield shift of the Lipid II phosphates after the addition of nisin is caused by the N-terminal amino group of nisin. We confirmed this after observing by REDOR solid-state MAS NMR a direct internuclear 15N-31P dipolar coupling between the N terminus of uniformly 15N-labeled nisin and the pyrophosphate of Lipid II in the complex (Fig. 6
). Nitrogen-15 MAS NMR spectra from uniformly labeled nisin in its membrane-associated complexes with Lipid II were acquired in the presence and in the absence of 31P-dephasing pulses. Reduction in the spectral intensity was observed after application of 31P-dephasing pulses only for those resonances, which arose from nitrogens directly coupled to phosphates and located within a range of up to 6 Å from 31P nuclei. The backbone nitrogens produced a largely overlapping set of resonances at 120 ppm and could not be assigned in this experiment. The -nitrogens from lysines 12, 22 and 34 appear at 34 ppm and the N-terminal amino group of Ile1 nisin is clearly resolved at 40 ppm with 1/3 of the intensity observed from the overlapping Lys-12, 22 and 34 (Fig. 6)
. After a train of 24 31P-dephasing pulses, the intensity of the terminal amino group and not that of the lysines appeared substantially reduced, which indicates that the terminal amino group of Ile1 is directly involved in the docking of nisin onto the pyrophosphates of Lipid II. In addition, dephasing was observed in residues within the backbone region, which could not be assigned in this experiment.
|
Direct involvement of the terminal amino group of nisin in the complex with Lipid II and the required presence of an accessible pyrophosphate in the target molecule suggest the possibility of formation of a hydrogen bond between PP on Lipid II and NH2-Ile1 of nisin. We examined this by doubly methylating the four amino groups of nisin. While the importance of the lysines in nisin has been attributed mainly to their contribution to the overall molecular charge in binding to lipid bilayers, methylation of the N terminus would prevent its involvement in the formation of hydrogen bonds with Lipid II. Indeed, 31P MAS NMR showed that Met-nisin does not have the ability to form complexes with 11PP. Its activity against Gram-positive organisms is severely reduced and does not permeabilize vesicles containing any of the precursors investigated here (not shown). Our 31P MAS NMR experiments showed that Met-nisin does engage Lipid II in stable complexes. The CP MAS spectra (Fig. 7
) show the presence of a single broad resonance at –14 ppm with a line width of over 5 ppm, bracketed by sidebands revealing spectral width of 200 ppm. From this we conclude that nisin engages Lipid II in a second binding site.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
, but information on important sites on Lipid II has been lacking. In this report we have identified an important docking motif for nisin on Lipid II.
Our 31P MAS NMR results show in a direct way the formation of a complex between Lipid II and nisin. This is seen as a loss in mobility in the region of the phosphates of Lipid II after addition of nisin to lipid membranes containing the precursor and points to the formation of either monomeric or oligomeric nisin/Lipid II complexes. Since the formation of nisin/Lipid II complexes was shown in our carboxyfluorescein leakage assays to permeabilize lipid membranes, the existence of oligomeric proteolipid pores emerges as the likely mode of nisin action in the presence of the precursor, in agreement with recent results (20)
. In contrast, complexes between nisin and 11PP observed in our 31P MAS NMR experiments did not appear to induce membranes leakage, which indicates the existence of monomeric membrane-incorporated nisin/11PP complexes. In our 31P MAS NMR and leakage assays, we did not observe differences in the behavior of Lipid II and Lipid I in the presence of nisin, which means that the terminal GlcNAc sugar of Lipid II is not likely to be involved in the interaction with nisin.
The change of the chemical shifts of the pyrophosphate upon binding to nisin and the heteronuclear dipolar recoupling MAS NMR experiments have highlighted the pivotal role of the terminal amino group of nisin in pyrophosphate recognition of Lipid II and in pore formation. Other experiments have emphasized the importance of the amino-terminal region of nisin in that extension of the N terminus by addition of four residues (ITPQ) has abolished the bactericidal activity of nisin (30)
.
The strong reduction of nisin’s antimicrobial activity, which follows N-terminal methylation, points to a mechanism of Lipid II recognition that involves the formation of a hydrogen bond between NH2-nisin and PP-Lipid II. This is confirmed by the dipolar dephasing MAS NMR estimate of 6 Å for the 15N-31P internuclear distance between 15HN4Cl3 and P2O7.
Two independent binding sites for nisin exist on Lipid II. One is identified here as the pyrophosphate group; the other must reside on the MurNAc-pentapeptide moiety. The equivalence on Lipid I and Lipid II as nisin targets from an NMR standpoint and as targets for membrane lysis in vitro has ruled out GlcNAc as an association site. Earlier experiments suggest lack of involvement of the terminal D-Ala in nisin binding as well (11)
. The existence of the second binding site was deduced after we observed a strong association of methylated nisin with Lipid II but not with 11PP. This second site of association alone is not sufficient for pore formation; both sites of contact must be engaged by nisin for the formation of lytic transmembrane pores to occur.
The engagement of pyrophosphate-containing cell wall precursors into stable intermolecular complexes by nisin that takes place on the outer membrane leaflet suggests an additional molecular mechanism of action of nisin whereby the precursor is withdrawn from the biosynthetic pathway by the antibiotic. The existence of such activity has been inferred from mutational analysis, which showed that nisin variants can lose their capability to form pores while retaining bactericidal activity by inhibiting bacterial cell wall synthesis (18)
.
A third mode of activity is proposed here whereby nisin interferes with peptidoglycan formation by withdrawing 11PP from the biosynthetic pathway into membrane-associated complexes (Fig. 8
). The formation of hydrogen bonds makes the association of nisin with Lipid II far stronger than that responsible for the action of the highly toxic antibacterial agent bacitracin (8)
, which interacts with 11PP only via electrostatic effects. Indeed, we observed that bacitracin did not restrict 11PP pyrophosphate mobility under similar conditions used for nisin but with added magnesium ions (not shown).
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
A new mode of attack by nisin on the bacterial cell wall biosynthesis is outlined. In the proposed mechanism, nisin engages cell wall intermediates Lipid II and 11PP into stable detergent-insoluble complexes, which are withdrawn from the cell wall biosynthesis. Pore formation and cell wall inhibition involve at molecular level N-terminal binding of nisin to the pyrophosphate moiety of cell wall precursor molecules, located on the outer leaflet of bacterial membranes. The importance of the present findings lies in the demonstration of an attack on extracellular pyrophosphates, which are unique to the bacterial kingdom and whose chemical functions cannot be substituted by simple mutations. Therefore, resistance via target mutation to antibiotics, which use pyrophosphates for target recognition, is unlikely to occur.
|
ACKNOWLEDGMENTS |
|---|
, 3
, and 8
. Received for publication May 14, 2004. Accepted for publication August 5, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  B. Bonev, J. Hooper, and J. Parisot Principles of assessing bacterial susceptibility to antibiotics using the agar diffusion method J. Antimicrob. Chemother., June 1, 2008; 61(6): 1295 - 1301. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Parisot, S. Carey, E. Breukink, W. C. Chan, A. Narbad, and B. Bonev Molecular Mechanism of Target Recognition by Subtilin, a Class I Lanthionine Antibiotic Antimicrob. Agents Chemother., February 1, 2008; 52(2): 612 - 618. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. J. Hyde, J. Parisot, A. McNichol, and B. B. Bonev Nisin-induced changes in Bacillus morphology suggest a paradigm of antibiotic action PNAS, December 26, 2006; 103(52): 19896 - 19901. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. R. Bonelli, T. Schneider, H.-G. Sahl, and I. Wiedemann Insights into In Vivo Activities of Lantibiotics from Gallidermin and Epidermin Mode-of-Action Studies. Antimicrob. Agents Chemother., April 1, 2006; 50(4): 1449 - 1457. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
