Solid-state NMR characterization of amphomycin effects on peptidoglycan and wall teichoic acid biosyntheses in Staphylococcus aureus

Amphomycin and MX-2401 are cyclic lipopeptides exhibiting bactericidal activities against Gram-positive pathogens. Amphomycin and MX-2401 share structural similarities with daptomycin, but unlike daptomycin they do not target bacterial membrane. In this study, we investigate in vivo modes of action for amphomycin and MX-2401 in intact whole cells of Staphylococcus aureus by measuring the changes of peptidoglycan and wall teichoic acid compositions using solid-state NMR. S. aureus were grown in a defined media containing isotope labels [1-13C]glycine and L-[ε-15N]lysin, L-[1-13C]lysine and D-[15N]alanine, or D-[1-13C]alanine and [15N]glycine, to selectively 13C-15N pair label peptidoglycan bridge-link, stem-link, and cross-link, respectively. 13C{15N} and 15N{13C} rotational-echo double resonance NMR measurements determined that cyclic lipopeptide-treated S. aureus exhibited thinning of the cell wall, accumulation of Park’s nucleotide, inhibition of glycine utilization for purine biosynthesis, reduction of ester-linked D-Ala in teichoic acids, and reduction of peptidoglycan cross-linking. Whole cell NMR analysis also revealed that S. aureus, in presence of amphomycin and MX-2401, maintained the incorporation of D-Ala during peptidoglycan biosynthesis while the incorporation of D-Ala into teichoic acids was inhibited. These effects are consistent with amphomycin’s dual inhibition of both peptidoglycan and wall teichoic acid biosyntheses in S. aureus.

(3) superimposable D-amino acids (or glycine) at positions 2, 5, 7, and 8 of the cyclic-peptide core, and (4) a hydrophobic tail of similar lengths. One key difference is that daptomycin is a depsipeptide with a lactone linkage instead of a peptide bond between the amino acid positions 1 and 10, (Fig. 1a, yellow highlight).
Despite their structural similarities, daptomycin and amphomycin-class of antibiotics exhibit different modes of action. Daptomycin undergoes a large conformational change upon calcium-ion binding to form a 10 to 16-mer pore structure 16 . The hydrophobic tail of daptomycin is thought to facilitate binding to the membrane by inserting itself into the lipid bilayer 17 and the ensuing membrane depolarization through K + leakage is thought to be the killing mechanism of daptomycin. In contrast, friulimicin B 2 , amphomycin, and MBX-240 13 do not depolarize the bacterial membrane even at concentrations ten times the minimum inhibitory concentrations (MIC). Furthermore, cell death induced by amphomycin and MX-2401 preceded the membrane depolarization, suggesting that the membrane depolarization is a consequence of bactericidal activity 13 . Sahl and co-workers, using in vitro assays, have shown that friulimicin B 2 and MX-2401 18 target the membrane lipid transporter bactoprenol-phosphate (C 55 -P). The structure of amphomycin bound to C 55 -P is unknown, but the x-ray crystal structure of tsushimycin suggests that a binding cleft for the phosphate of C 55 -P may be formed at the dimer interface between the cyclic peptide core structures 19 .
In this study, we investigate in vivo mode of action of amphomycin and MX-2401 in intact whole cells of S. aureus pair-labeled with [1-13 C]Gly and L-[ε - 15  Gly using solid-state NMR. Cyclic lipopeptides were added to S. aureus at mid-exponential growth phase at sub-MIC, then the changes of peptidoglycan (PG) and wall teichoic acid (WTA) compositions were determined by 13

Results
MX-2401 does not induce ATP leakage in S. aureus. ATP-leakage assay was performed on overnight culture of S. aureus (Fig. 1b) by adding daptomycin or MX-2401 to final drug concentrations of 0, 0.5, 1, 2, 5, 10, 50, and 100 μ g/mL (Fig. 1b). Daptomycin induced ATP leakage was observed at higher concentrations (50 and 100 μ g/mL), but MX-2401 did not show significant ATP leakage at any concentration. Solid-state NMR samples were prepared from harvested S. aureus at mid-exponential growth phase after treated with amphomycin or MX-2401 to final drug concentrations of 20 or 40 μ g/mL. S. aureus grown in the presence of amphomycin and MX-2401 at both drug concentrations did not exhibit any observable growth perturbation (Fig. 1c) nor ATP leakage (Fig. 1b). (a) Chemical structures of cyclic decapeptide antimicrobial agents: friulimicin, amphomycin, tsushimycin, MX-2401, and daptomycin. All of these cyclic lipopeptides share a ten-membered peptide core structure with hydrophobic tail. The hydrophobic tail group in amphomycin is anteiso-tridecenonyl, and in tsushimycin and friulimicin B it is iso-tetradecenonyl. All cyclic decapeptides exhibit calcium dependent bactericidal activity against Gram-positive bacteria and have a conserved calcium-binding motif sequence Asp 4 -Asp 6 -Gly 7 . Daptomycin is a cyclic-depsipeptide with a lacton linkage highlighted in yellow. (b) ATP-leakage assay performed on overnight culture of S. aureus. Daptomycin induces leaking of ATP from S. aureus, but only at a high drug concentration (100 μ g/mL). MX-2401 did not cause the leakage in S. aureus at any concentrations. Thus, MX-2401 does not disrupt the membrane despite sharing structural similarities with daptomycin. (c) Growth curves of S. aureus treated with amphomycin (left), and MX-2401 (right) as monitored by the optical density (OD) at 660 nm. Cyclic lipopeptides were added to final concentrations of 20 or 40 μ g/mL at mid-exponential growth (OD 660nm 0.7). Addition of cyclic lipopeptides at these concentrations did not significantly perturb the growth. The cells were harvested for NMR analysis after one doubling time (60 minutes) of incubation with drugs at the approximate OD 660nm of 0.9. Park's nucleotide (cytoplasmic PG precursor) as lysyl-ε -amine can be observed as a peak with the chemical shift of 10 ppm due to its open bridge-link. Amphomycin-treated S. aureus show the reduction of lysyl-ε -amide at 95 ppm and increase in lysyl-ε -amine at 10 ppm. The reduction of 95 ppm corresponds to the loss of PG through cell wall thinning, and the increased lysyl-ε -amine at 10 ppm corresponds to the accumulation of Park's nucleotide. 13 C{ 15 N} REDOR spectra of whole cells of S. aureus with and without amphomycin treatment after 1.6 ms of dipolar evolution are shown in Fig. 2b. S 0 spectra (bottom) are normalized to natural abundance peaks from 10 to 30 ppm range. The 170-ppm glycyl peak intensities in S 0 spectra (bottom) are directly proportional to the total amount of [1-13 C]Gly found in S. aureus. The reduced glycyl-carbonyl carbon in the S 0 of amphomycin-treated S. aureus indicates the thinning of cell wall due to inhibition of PG biosynthesis. Covalently bonded 13 C-15 N spin pairs unique to the PG bridge-link are selected in Δ S spectra after 1.6 ms dipolar evolution. Reduced Δ S 170-ppm intensity in amphomycin-treated S. aureus is the result of increased incorporation of endogenous unlabeled L-Lys Amphomycin-treated S. aureus show reduced lysyl-ε -amide peak at 95 ppm due to cell-wall thinning, and increased lysyl-ε -amine peak at 10 ppm due to accumulation of Park's nucleotide. These effects are consistent with amphomycin inhibition of cell wall biosynthesis by targeting steps at or prior to PG transglycosylation. Each spectrum is the result of 20,480 accumulated scans. (b) 13 C{ 15 N} REDOR spectra from whole cells of S. aureus at 1.6 ms of dipolar evolution. S 0 spectra are at the bottom, and Δ S spectra at the top. Spectra are normalized to natural abundance peaks from 10 to 30 ppm range. The 170-ppm glycyl peak intensities in S 0 are directly proportional to the amount of [1-13 C]Gly in S. aureus. The decrease in 170-ppm peak intensity in the S 0 spectrum of amphomycin (40 μ g/mL) treated S. aureus is consistent with thinning of the cell wall. Amphomycin-treated S. aureus also show decrease in the 149-ppm peak intensity in the S 0 indicating reduction of [1-13 C]Gly into purine biosynthesis. In Δ S spectra (top) only covalently bonded 13 Cs from 13 C-15 N spin pairs unique to PG bridge-links are detected at 1.6 ms dipolar evolution. 170-ppm dephasing (Δ S/S 0 ) decreases from 11.5% to 8.9% with addition of amphomycin, which indicates the reduction in the proportion of L-[ε -15 N] Lys due to incorporation of unlabeled L-Lys into PG stem structure. Each spectrum is the result of 10,240 accumulated scans. into PG-stem structure from biosynthesis of cytoplasmic PG precursors. The 149-ppm peak in the S 0 is due to the metabolism of [1-13 C]Gly into purine biosynthesis. The reduction of 149-ppm peak in amphomycin-treated S. aureus indicates decreased amount of [1-13 C]Gly entering purine biosynthesis pathway due to the heavy demand on [1-13 C]Gly by PG biosynthesis pathway.
Amphomycin-treated S. aureus show reduced PG cross-linking. S. aureus was grown in defined media with D,L-[1-13 C]Ala and [ 15 N]Gly to selectively label 13 C-15 N pair of the PG cross-link in order to determine the effect of amphomycin on transpeptidation step of PG biosynthesis (Fig. 3a). 13 C{ 15 N} REDOR spectra of whole cells of S. aureus grown in presence (40 μ g/mL) and absence of amphomycin with dipolar evolution of 1.6 ms are shown in Fig. 3b. The S 0 spectra are normalized to natural abundance peaks from 10 to 30 ppm range. The S 0 spectrum of amphomycin-treated S. aureus shows reduction of D,L-[1-13 C]Ala incorporation into PG. In Δ S spectra, 175-ppm peak intensity is proportional to the total amount of covalently bonded 15 N-13 C spin pairs found in S. aureus. Although the intensity of Δ S 175-ppm peak includes contributions from [ 15 N]Gly-L-[1-13 C] Ala sequences found in proteins, the primary contribution to the intensity comes from cross-linked PG. Alanine racemase inhibitor alaphosphin has been used to ensure selective labeling of PG with D-[1-13 C]Ala and to remove protein contribution to the Δ S 175-ppm intensity 8,21-28 ; however, alaphosphin was not used to eliminate potential synergic effect of alaphosphin with amphomycin. Amphomycin-treated S. aureus showed reduction in Δ S 175-ppm peak intensity, which is directly proportional to the total number of PG cross-links. This reduced PG cross-linking in amphomycin-treated S. aureus is apparent in the enlarged carbonyl carbon region of the S 0 overlaid with the dephased (S) spectra (Fig. 3c), where the difference between 175-ppm intensities of S 0 and S is directly proportional to the total number of PG cross-links.
Amphomycin and MX-2401 treated S. aureus exhibit reduced ester-linked D-Ala. The lineshape of S 0 175 ppm alanyl-carbonyl carbon peak of amphomycin-treated S. aureus (Fig. 3d, red line) is different from untreated sample (Fig. 3d, black line). Overlaid spectra show that the 175-ppm peak from amphomycin-treated S. aureus has a broadened shoulder due to the increase in 178-ppm contribution from the penultimate D-Ala carbonyl carbon of accumulated Park's nucleotide 8

Discussion
We investigated in vivo mode of action of amphomycin and MX-2401 by measuring the changes in cell wall composition of intact whole-cells of S. aureus using solid-state NMR. It was found that amphomycin-treated S. aureus exhibit cell wall thinning (Figs 1b, 2b and 3b) concomitant with the accumulation of Park's nucleotide (Figs 1a and 3d), which indicates that amphomycin inhibits cell wall biosynthesis by targeting steps at or prior to PG transglycosylation. These observations are consistent with the proposed membrane transporter targeting mode of action for amphomycin 1,29 . Recent in vitro studies using bacterial membrane extracts have shown that amphomycin 18 and friulimicin B 2 form Ca 2+ dependent complexes with bactoprenol-phosphate (C 55 -P) to inhibit the formation of lipids I, II, and III. Lipids I and II are the membrane transporters for PG biosynthesis, and lipid III is for WTA biosynthesis.
As C 55 -P is a central lipid transporter shared by PG and WTA biosyntheses, we investigated the comparative effects of amphomycin and MX-2401 on PG and WTA biosyntheses. Accurate biochemical quantification of PG and WTA are difficult, especially for WTA analysis 30 which must rely on a series of purification steps dependent on the release of WTA from the cell wall. In contrast, solid-state NMR approach provides direct and accurate measurement of D-[ 15 N]Ala incorporation into PG and WTA in intact whole cells. We were able to clearly quantify the amount of D-[ 15 N]Ala incorporated into PG by measuring the amount of PG stem-links using 13 C{ 15 N} REDOR (Fig. 3), and D-[ 15 N]Ala incorporation into ester-linked D-Ala in WTA by 15 N{ 13 C} REDOR NMR (Fig. 4). The NMR analysis of whole cells yielded a surprising result that S. aureus differently utilize D-Ala for PG Amphomycin reduced cross-linking, which is consistent with mild inhibition of transpeptidation. (d) Enlarged carbonyl-carbon regions of the 13 C{ 15 N} REDOR full-echo spectra of amphomycin treated (red) and untreated (black) S. aureus. The lineshape of spectra from amphomycin treated cells shows increased 178 ppm contribution from the carboxyl-carbonyl carbon of D-[1-13 C]Ala from Park's nucleotide unincorporated into PG by crosslinking. The decreased 172 ppm contribution is due to the reduced amount of ester-linked D-[1-13 C]Ala in WTA. and WTA biosyntheses under antibiotic stress. 15 N{ 13 C} REDOR NMR spectra (Fig. 4b,c) 15 N{ 13 C} REDOR spectra of S. aureus treated with amphomycin (0, 20, and 40 μ g/mL) at 1.6 ms dipolar evolution. (c) 15 N{ 13 C} REDOR spectra of whole cells of S. aureus treated with MX-2401 (0, 20, and 40 μ g/mL) at 1.6 ms dipolar evolution. The full-echo spectrum (S 0 ) is at the bottom and the REDOR difference (Δ S) at the top. S 0 spectra are normalized to equal alanyl-amide peak at 95 ppm of Δ S spectra. Addition of amphomycin and MX-2401 resulted in the decrease in alanyl-amine peak at 10 ppm in S 0 spectra, indicating reduction in the amount of ester-linked D-Ala in teichoic acid.
Scientific RepoRts | 6:31757 | DOI: 10.1038/srep31757 into teichoic acids by amphomycin and MX-2401. The reduction in ester-linked D-[ 15 N]Ala can be due to either i) direct inhibition of WTA biosynthesis, and/or ii) inhibition of D-Ala incorporation by targeting of D-alanylation pathway specific to WTA biosynthesis. Although solid-state NMR measurements cannot differentiate between these two mechanisms, it is clear that amphomycin and MX-2401 had a direct and consequential effect on wall teichoic acid composition. Surprisingly, amphomycin and MX-2401 had no visible effect on the incorporation of D-[ 15 N]Ala into PG stem-link (Fig. 4b,c), which suggests at the cost of maintaining PG biosynthesis in amphomycin and MX-2401 treated S. aureus is decreased WTA biosynthesis.
Whole cell analysis by solid-state NMR indicates that in vivo mode of action for amphomycin is complex. For example, glycine metabolism towards purine biosynthesis is suppressed in amphomycin treated S. aureus (Fig. 2b) presumably due to routing of all available glycine to PG biosynthesis pathway in response to the cell wall inhibition. While the downstream effect of purine biosynthesis inhibition by amphomycin is unknown, presumably it would directly alter the overall metabolism of bacteria. Another interesting effect of amphomycin on whole cell is the reduction of PG cross-linking (Fig. 3c). This reduction was unexpected as the targeting of C 55 -P by amphomycin is unlikely to have any direct inhibitory effects on transpeptidase activity. We propose two possible explanations for the amphomycin inhibition of PG cross-linking, although there may be multiple ways these scenarios could manifest themselves. First, amphomycin's inhibitory effect on PG cross-linking is an indirect effect from inhibition of ester-linked D-Ala incorporating into WTA (and potentially inhibiting WTA biosynthesis), which can result in mislocalization of penicillin-binding protein 4 (PBP4). PBP4 is the transpeptidase responsible for high levels of PG cross-linking observed in S. aureus. WTA has been shown to be essential in the recruitment of PBP4 to the septum during the cell division. In WTA deficient tagO-deletion mutant of S. aureus, PBP4 is shown to be dispersed over the cell surface resulting in reduced PG cross-linking 31 . Second, amphomycin can directly interfere with transpeptidase activity through steric interference by directly binding to PG and forming a complex. Antibiotics that target PG with possessing a hydrophobic sidechain such as oritavancin 32 and plusbacin A 3 8 reduce PG cross-linking in S. aureus. Similar mechanism of action can be envisioned for amphomycin, but the binding of amphomycin to PG remains yet to be demonstrated.

Methods
Growth and labeling of whole cells. Starter culture of S. aureus (ATCC 6538P) grown overnight in 5 ml of trypticase soy broth in 37 °C at 250 rpm was added (1% final volume) to two one-liter flasks, each containing 250 mL of S. aureus Standard Medium (SASM) supplemented with calcium chloride to final concentration of 50 μ g/mL 28 . The detailed protocol for the defined medium SASM has been described previously [33][34][35] . Bridge-links and cross-links of PG are selectively 13 C-15 N isotope pair labeled (Fig. 2)  is the solid-state NMR method that recouples heteronuclear dipolar interactions under magic-angle spinning 20 , which is used to determine dipolar couplings dependent on inter-nuclear distances. REDOR is a difference experiment in which two spectra are collected, one in the absence of heteronuclear dipolar coupling (full echo, S 0 spectrum), and the other in the presence of the coupling (dephased echo, S spectrum). In the S 0 spectrum, dipolar dephasing is refocused over a single rotor period due to the spatial averaging by the motion of rotor under magic-angle spinning. In the S spectrum, spin part of the dipolar interaction is manipulated by the application of rotor-synchronized dephasing π -pulses that prevent full refocusing. Dipolar evolution over a rotor period in the S spectrum results in reduced peak intensity for dipolar coupled spin pairs. The difference in signal intensity (REDOR difference, Δ S = S 0 − S) for the observed spin ( 13 C or 15 N) in two parts of the REDOR experiment is directly related to the heteronuclear dipolar coupling from which the corresponding distance to the dephasing spin is determined. The difference spectra are typically collected as a function of N rotor periods to map out the dipolar evolution.
Solid-state NMR Spectrometer. REDOR experiments were performed on whole cells at 7.0 T (300 MHz for 1 H, 75 MHz for 13 C, and 30 MHz for 15 N) provided by 89-mm bore Oxford (Cambridge, U.K.) superconducting solenoids. The four-frequency transmission-line probe used in the 7.0-T spectrometer had a 14-mm long, 9-mm inner-diameter sample coil, while that used in the 4.7-T spectrometer had a 17-mm long, 8.6-mm inner-diameter sample coil. Both probes were equipped with Chemagnetics/Varian magic-angle spinning ceramic stator, and samples were spun at room temperature at 5 kHz (maintained within ± 2 Hz). Radio-frequency pulses were produced by 1-kW Kalmus, ENI, and American Microwave Technology power amplifiers, each under active control; π -pulse lengths were 10 μ s for both 13 C and 15 N. Proton-carbon and proton-nitrogen matched cross-polarization transfers were at 50 kHz for 2 ms. Proton dipolar decoupling during signal acquisition was 105 kHz. Standard XY-8 phase cycling 37 was used for all refocusing and dephasing pulses. The re-cycle delay period was 2 seconds during which each amplifier produced a 300-μ sec test pulse. Resulting diode-detected voltages were compared to the reference voltage previously calibrated. Differences were used to correct the drives of power amplifiers for the next repetition of the REDOR pulse sequence. Combination of active control of the amplifiers 38 and alternate-scan data acquisition for each pair of REDOR spectra (S and S 0 ) eliminated long-term drifts in the performance of the spectrometer. The normalized REDOR difference (Δ S/S 0 ) is a direct measure of dipolar coupling, and this quantity was calculated using modified Bessel function expressions given by Mueller et al. 38 and de la Caillerie and Fretigny 39 for an IS spin-1 / 2 pair. ATP-leakage assay. ATP-leakage assay was performed on overnight cultures of S. aureus grown in TSB harvested at OD 600 1.5. Cells were first pelleted, then resuspended in phosphate buffered saline (PBS) supplemented with 20 mM Ca 2+ . Cyclic lipopeptide was added to the suspension to final drug concentrations of 0, 0.5, 1, 2, 5, 10, 50, and 100 μ g/mL and the cells incubated for 20 min at 37 °C. After the incubation, bacteria were pelleted and removed from the suspension by centrifuging. The amount of leaked ATP in the supernatant was quantified by directly adding 100 μ L of CellTiter-Glo ® 2.0 reagent (Promega, Madison WI) to equal volume of supernatant.
After 10 minutes of equilibration, luminescence given off by the mixture was measured using Fluoroskan Ascent FL Luminometer (Thermo Scientific) with the integration time of 200 ms.