Research article Received: 4 March 2015
Revised: 12 May 2015
Accepted: 19 May 2015
Published online in Wiley Online Library: 1 July 2015
(wileyonlinelibrary.com) DOI 10.1002/mrc.4279
Three-dimensional structure of cyclic antibiotic teicoplanin aglycone using NMR distance and dihedral angle restraints in a DMSO solvation model Nina C. Gonnella,a* Nelu Grinberg,b Mark Mcloughlin,a† Om Choudhary,a,b‡ Keith Fandrickb and Shengli Mab The three-dimensional solution conformation of teicoplanin aglycone was determined using NMR spectroscopy. A combination of NOE and dihedral angle restraints in a DMSO solvation model was used to calculate an ensemble of structures having a root mean square deviation of 0.17 Å. The structures were generated using systematic searches of conformational space for optimal satisfaction of distance and dihedral angle restraints. Comparison of the NMR-derived structure of teicoplanin aglycone with the X-ray structure of a teicoplanin aglycone analog revealed a common backbone conformation with deviation of two aromatic side chain substituents. Experimentally determined backbone 13C chemical shifts showed good agreement with those computed at the density functional level of theory, providing a cross validation of the backbone conformation. The flexible portion of the molecule was consistent with the region that changes conformation to accommodate protein binding. The results showed that a hydrogenbonded DMSO molecule in combination with NMR-derived restraints together enabled calculation of structures that satisfied experimental data. Copyright © 2015 John Wiley & Sons, Ltd. Keywords: teicoplanin aglycone; nuclear magnetic resonance; conformation; 3D structure; solvation model; density functional theory
Introduction
Magn. Reson. Chem. 2015, 53, 829–835
* Correspondence to: Nina C. Gonnella, Materials and Analytical Sciences, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT, USA. E-mail: nina. [email protected] † ‡
Present Address: 6 Harbour Town Ct, Frisco, TX, USA Present Address: 5535 Centre Ave, Apt 2, Pittsburgh, PA, USA
a Materials and Analytical Sciences, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT, USA b Chemical Development, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT, USA Abbreviations: WET, experiment selective solvent suppression experiment; COSY, correlation spectroscopy; TOCSY, total correlation spectroscopy; HSQC, heteronuclear single-quantum correlation; HMBC, heteronuclear multiple-bond correlation; HMQC, heteronuclear multiple-quantum correlation; ROESY, rotating frame Overhauser effect spectroscopy; QSINE, sine bell squared function; SINE, shifted sine bell function; GARP, globally optimized alternating phase rectangular pulses; WALTZ, wideband alternating phase low-power technique for zero residual splitting; MMFF94×, Merck molecular force field 94×.
Copyright © 2015 John Wiley & Sons, Ltd.
829
Teicoplanin is a potent antibiotic for the treatment of serious infections caused by Gram-positive bacteria, such as methicillin-resistant Staphylococcus aureus, coagulase-negative Staphylococci, Clostridia, and Enterococci. The compound’s antibiotic mechanism of action has been reported to occur through the inhibition of bacterial cell wall synthesis.[1] These antibiotics are known to have in vivo stability based upon studies with radiolabeled teicoplanin in rats and humans where the drug was shown to undergo renal elimination with very minor (~5%) metabolic transformation.[2] The teicoplanin antibiotic is a mixture of several glycopeptidelike antibiotics, which all share a common core system.[3] These compounds are members of the vancomycin group of glycopeptide antibiotics. A fused ring structure forms the common core of the teicoplanin antibiotic family that is termed teicoplanin aglycone (Sigma Aldrich, St Louis, MO, USA).[4] The aglycone core is prepared from hydrolytic treatment of teicoplanin to remove the sugar and fatty acid moieties.[5] Both the chemical structure and stereochemistry of teicoplanin aglycone have been previously reported and are consistent with the chemical structure of the compound described herein.[5] The chemical structure, stereochemistry, and numbering system of teicoplanin aglycone used in this study are given in Fig. 1. Chemical shift assignments have been reported for teicoplanin aglycone hydrochloride salt in DMSO and other solvents[6]; however, NMR studies of teicoplanin aglycone (nonsalt form) have not been reported, and the three-dimensional solution conformation of teicoplanin aglycone has not been solved.
Here, we report the first experimentally determined solution conformation of teicoplanin aglycone in DMSO. Full proton and carbon-13 chemical shift assignments of the molecule are also included. NMR measurement of teicoplanin aglycone in aqueous solution was not possible because of low compound solubility. The three-dimensional solution structure of this molecule is important because it provides valuable insight into the conformation and structural flexibility of the antibiotic required in biological function.
N. C. Gonnella et al.
Figure 1. Chemical structure of teicoplanin aglycone with (R or S) asymmetric centers displayed. Aromatic side chains are labeled A–F with sequential numbering nomenclature for carbon atoms and attached protons.
The reported structure was calculated from NMR-derived interproton distances and targeted dihedral angle restraints based on measured coupling constants. The results revealed that a constrained search of conformational space in DMSO solvent was a main component in establishing the three-dimensional structure. Comparison of experimentally determined 13C chemical shifts with those predicted at the density functional theory (DFT) level of theory supports for the backbone cis/trans-NH–CO ‘ω’ angle peptide bond orientation. In addition, derived conformers were compared with an X-ray structure of a similar teicoplanin aglycone analog showing that flexibility of the core aromatic side chain is consistent with the region of the molecule involved in protein binding.
Experimental and computational methods Sample preparation Teicoplanin aglycone was obtained from Supelco (Sigma Aldrich/Supelco). The sample was prepared by dissolving 5 mg of teicoplanin aglycone in 600-μl DMSO-d6 (Sigma Aldrich), and 30 μl was transferred to a 1.7-mm microcapillary NMR tube. NMR spectra
830
NMR experiments were performed on a Bruker-Biospin DRX 600 NMR spectrometer (Bruker Corporation, Billerica, MA, USA) operating at 600.2 MHz for 1H and 150.92 MHz for 13C, nonspinning at 300 K, using a 1.7-mm CP TCI Z gradient probe. 1H and 13C chemical shifts were referenced to DMSO-d6 at 2.5 and 39.5 ppm, respectively. One-dimensional (1D) 1H data (16K points) were obtained using a spectral width of 12 019 Hz and a WET 1D experiment with double irradiation of H2O and residual proteo-DMSO. WET solvent suppression was achieved using 90° sine.1 shape pulses with 13C low-power decoupling during WET and acquisition.[7] Data were processed with em and line broadening of 0.05 Hz before Fourier transformation. Gradient-selected COSY spectra, using a shaped pulse for double off-resonance presaturation with continuous-wave decoupling on the 13C channel, were obtained (2048 × 160 data points) with a relaxation delay of 2.0 s, a spectral width of 9614 Hz in both dimensions, and a 90° read pulse. TOCSY data were obtained in phasesensitive mode (TPPI) using 0.68 W for excitation and 0.13 W for spinlock. WET solvent suppression with sine.1 shaped pulse and 13 C decoupling on during WET and data acquisition were used.[8] Data were acquired using 2048 × 256 data points, a relaxation delay of 1.5 s, a spectral width of 8680 Hz in both dimensions, and a 60-ms spinlock. Phase-sensitive ROESY spectra with continuous-wave
wileyonlinelibrary.com/journal/mrc
spinlock and presaturation of the water resonance[9] were obtained (1024 × 256 data points) using a relaxation delay of 2.0 s, a spectral width of 12 004 Hz in both dimensions, and a 500-ms mixing time. 1 H–13C WET-HSQC data was collected using echo/anti-echo-TPPI gradient selection and WET solvent suppression with a sine.1 shape pulse with decoupling during acquisition.[10–12] Spectra were obtained (1024 × 256 data points) with GARP decoupling, a relaxation delay of 2.0 s, and a spectral width of 8680 Hz in f2 and 36 219 Hz in f1. Gradient-selected 1H–13C HMQC data were also obtained (2048 × 256 data points) using GARP decoupling, a relaxation time of 2.0 s per transient, and a spectral width of 8680 Hz in f2 and 36 225 Hz in f1. Gradient-selected 1H–13C gHMBC data were acquired (2048 × 512 data points) using WALTZ-16 decoupling, a relaxation delay of 1.5 s, and a spectral width of 8680 Hz in f2 and 36 225 Hz in f1. Delays were calculated using 1J (13C–1H) of 145 Hz and two-bond/three-bond J (13C–1H) of 4.5 Hz. Two-dimensional spectra were processed with SINE or QSINE apodization in both dimensions before Fourier transformation. Computational methods ROE intensities were converted to distance-restraint ranges, assuming isotropic motion and no spin diffusion. A wide tolerance range (1.8–6.0 Å) was applied to most distance restraints. Restraints showing strong ROE cross peaks were tightened to a range of 1.8–3.5 Å. The lower bound was set at 1.8 Å consistent with van der Waals lower bound restraints. The distance-restraint weighting was set to 300 kcal/mol. A total of 57 distance restraints and six NH–CHα dihedral angle restraints were used in the conformational search with a 10° range around an expected dihedral angle (weight of 500 kcal/mol). Backbone ϕ torsion angles of the core were sampled over a range consistent with the expected trend in couplings from the Karplus relationship 3J = A cos2(ϕ) + B cos(ϕ) + C.[13] The starting structure for teicoplanin aglycone was derived from a dynamics run followed by a rigorous and extensive stochastic conformational search. The lowest-energy conformer had cis/trans-NH–CO (ω) geometry consistent with the X-ray crystal structure of A-40926.[14] A gas-phase model with explicit DMSO molecules was minimized in a wall-restraint box with dimensions 7.8598 × 8.1661 × 8.0499 Å. Three-dimensional structures were generated using distance-constrained molecular dynamics perturbations along low frequency and vibrational modes, as implemented in molecular operating environment software[15] (Chemical Computing Group Inc., 1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada), running on a windows-based PC. A modified MMFF94× force field[16] was used in the conformational search, incorporating ROE distance and ϕ angle restraints. The force field was parameterized for organic molecules keeping conjugated nitrogens planar. DFT calculations Optimized structures were used as input geometries for DFT calculations. The DFT functional was B3LYP,[17,18] and the basis set was 6-311G(d).[19,20] Geometry optimization, harmonic frequency calculations, and the magnetic shielding tensor calculations were performed with the Gaussian09 package[21] using the gaugeindependent atomic orbital[22–26] method. The calculations were performed in vacuo. Excellent agreement between experimentally determined backbone chemical shifts and calculated results (using a DMSO-d6 solvation, conductor–like polarizable continuum model) was found.
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 829–835
NMR-derived conformation of teicoplanin aglycone using solvation model Chemical shieldings σcalc were converted to chemical shifts δcalc using the property δcalc = σref σcalc.[27] This conversion is carried out using a reference value, σref, to best align the calculated and experimental spectra. The reference value was determined by a linear fit between calculated shieldings and experimental chemical shifts for each structure. The average value of σref is 178.1 ppm for 13 C chemical shifts.
Results and discussion Structure assignments The numbering system for teicoplanin aglycone consisted of labeling the aromatic rings with letters A–F similar to that previously described[3] and sequential numbering around the macroscopic ring (Fig. 1). A 1D WET proton spectrum of teicoplanin aglycone is given in Fig. 2. The spectrum shows a single set of relatively sharp signals, good peak dispersion, and spin–spin coupling that enabled unambiguous proton resonance assignment. The NMR spectra showed no evidence of distinct slowly interconverting rotamers in solution. Proton chemical shift assignments were achieved using 1H,1H presaturation-COSY, 1H,1H WET-TOCSY, and 1H,1H presaturation– ROESY experiments. The COSY and TOCSY experiments established vicinal and long range through bond proton connectivity, while the ROESY experiment corroborated assignments from through-space proximity. The 1H,13C WET-HSQC experiment corroborated assignments for the nonequivalent protons of C40 as well as providing supporting evidence for the identification of exchangeable amide (NH) resonances and several hydroxyl (OH) resonance protons. Because of the relatively low concentration of teicoplanin aglycone used in this study (~5 mg/600 μl), the 13C resonances were assigned solely based upon 1H,13C HSQC, 1H,13C HMQC, and 1 13 H, C HMBC indirect detection experiments. The HSQC and HMQC experiments allowed facile assignment of all proton-bearing carbons. The quaternary carbons and carbonyl carbons were assigned using HMBC correlations. The full 1H and 13C chemical shift assignments of teicoplanin aglycone in DMSO-d6 are given in Table 1. The ROE measurements not only enabled chemical shift assignments but also provided through-space proximity to generate conformational studies. ROE data were used to derive distancerestraint ranges that enabled the calculation of an ensemble of low-energy conformers (Table 2).
Computational studies The lowest-energy conformers generated by a stochastic conformational search and in agreement with the X-ray structure of antibiotic A40926 aglycone[14] served as a template for establishing the basic peptide backbone starting structure for teicoplanin aglycone. The NH–CO peptide bonds were all trans with the exception of E16A17, which adopts a cis-conformation. Although both A40926 aglycone and teicoplanin aglycone have major similarities, there were two minor differences between the two structures. The first difference is that A-40926 aglycone places the Cl atom on aromatic ring F instead of aromatic ring C in teicoplanin agclycone. Second, the amino group (NH2; G45 in Fig. 1) is methylated in A-40926. The crystal structure of A-40926 aglycone shows two independent molecules in the asymmetric unit cell, crystallizing in the orthorhombic space group P212121. Although there are relatively small differences between the two X-ray conformers, both conformers afforded proper stereochemistry and cis/trans-NH–CO ‘ω’ angles. The backbone peptide cis/trans-conformation was also achieved by carrying out a random conformational search (no restraints applied) using molecular dynamics. A set of six lowest-energy structures is shown in Fig. 3. Such structures showed evidence of hydrophobic collapse of rings G and F with rings D and E and did not provide definition of the orientation of aromatic rings A and C where these rings can flip with chlorines facing up or down. A solvation model of teicoplanin aglycone was produced by soaking the starting structure in a 1.0-margin box of DMSO and energy minimizing the structure. This resulted in incorporation of a
1
1
H NMR spectrum (600.20 MHz) of teicoplanin aglycone in dimethylsulfoxide-d6. Assignments of H chemical shifts are displayed on the spectrum.
Magn. Reson. Chem. 2015, 53, 829–835
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
831
Figure 2.
A total of 55 positive distance restraints were extracted from the ROESY spectrum. Initial conformational studies produced structures placing NH C42 in close proximity with aromatic protons G51 and F57, which were not supported by NMR data. To compensate for these unsupported interactions, two negative restraints were added between C42 and G51 and C42 and F57 (Table 2), which served to structurally satisfy experimental data. Vicinal coupling constants 3JNCα were also extracted from the 1D proton spectra. All five NH–Hα coupling constants were measured from the 1D proton spectrum, and ϕ angle-restraint ranges were estimated based upon the Karplus relationship.[13] The dihedral restraints were used to further refine the backbone conformation of the macrocyclic core (Table 3).
N. C. Gonnella et al. Table 1. Chemical shift assignments of teicoplanin aglycone in DMSOd6 at 300 K Atom number
832
D1–CO2 D2 D3 D4 D5 D6 D7 D8 D20–NH E9 E10 E11 E12 E13 E14 E15 E16 E65–NH A17–NH A18 A19 A21 A22 A23 A24 A25 A26 A27 B28 B29 B30 B31 B32 B33 B62–NH B63 B64 C34 C35 C36 C37 C38 C39 C40 C41 C42–NH C60 G43 G44 G45–NH2 G46 G47 G48 G49 G50 G51 F52 F53 F54 F55 F56 F57 F58 F59–NH F61 A21(OH) D5(OH) F54(OH)
1
H δ (ppm)
— 4.33 — 6.22 — 6.34 — — 8.34 — — 6.60 6.64 — 7.06 4.29 — 8.43 6.70 4.09 — 5.04 — 7.39 7.21 — — 7.75 — 5.06 — 5.46 — — 7.61 5.64 — — — 7.17 — 7.64 7.22 2.81, 3.27 4.94 7.38 — — 4.52 n/o — 6.61 — — 6.88 7.07 — 6.28 — 6.28 — 6.34 5.28 7.67 — 5.87 9.35 9.68
wileyonlinelibrary.com/journal/mrc
13
C δ (ppm) 171.7 56.1 135.5 104.8 156.2 101.6 156.3 116.9 — 120.2 154.4 116.0 124.8 124.0 134.7 52.7 168.1 — — 60.6 166.7 70.9 141.3 126.4 122.7 147.4 125.2 126.4 146.6 103.3 126.7 106.4 146.6 133.2 — 53.7 169.4 149.5 126.8 129.9 134.6 129.1 123.7 35.7 53.4 — 168.3 172.6 57.0 — 130.5 116.0 140.5 145.9 117.3 124.1 156.3 103.8 157.5 109.2 139.9 101.6 57.0 — 167.2 — — —
Table 2. Restraints from ROESY of teicoplanin aglycone in DMSO-d6 at 300 K Proton 1
Proton 2
Restraint range
A18 C40 C40b C40a A18 E15 D2 D2 A18 E15 D2 A18 A18 B62 B31 F59 D2 F59 E15 A21 E65 E65 E65 E14 F54(OH) G44 G51 G44 C41 C41 B63 A21 F58 B63 F58 B62 B62 A21(OH) A21(OH) A17 G47 C47 F58 C38 A18 A18 C41 A21 A21 B29 B29 F58 C41 C42 C42 C42 C42
E15 C41 C36 C38 A21 B29 D4 D20 A23 E14 E14 A27 D20 F57 C39 C42 E11 F57 E65 D20 B29 B63 E12 D20 F53 C42 G44 G47 C42 G47 B29 A21(OH) F55 B31 F59 B31 B63 A23 A17 A23 F55 F57 F57 C42 A17 E14 F59 A27 A23 A23 A24 B62 F58 G47 C39 G51 F57
(1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–5.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (4.0–7.0) negative restraint (4.0–7.0) negative restraint
DMSO molecule in a pocket formed by the B, C, F, and G rings establishing hydrogen bonding between the DMSO oxygen and amide NH C42 and NH F59. This interaction with DMSO secured a Karplus ϕ angle relative trend consistent with the experimentally measured NMR coupling constants. Structures incorporating the hydrogen-bonded DMSO molecule and NMR-derived distance and dihedral angle restraints yielded conformers where the Karplus
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 829–835
NMR-derived conformation of teicoplanin aglycone using solvation model Table 3. NH–CHα coupling constants (Hz) and dihedral angles (ϕ) for teicoplanin aglycone in DMSO-d6 Proton
A17 A18 D2 D20 E15 E65 B62 B63 F58 F59 C41 C42
3
JNα (Hz)
11.42 (±0.5) 11.95 (±0.5) 6.05 (±0.5) — 5.33 (±0.5) — 8.25 (±0.07) 8.32 (±0.07) 10.63 (±0.01) 10.64 (±0.01) — 9.65 (+0.5)
Estimated Karplus angle from quadratic fit[13]
Measured range of dihedral angles (ϕ) from conformational search with NMR restraints (six conformers)
~176.8 (10.77 Hz)*
177,
Measured dihedral angles (ϕ) from X-ray structure (two conformers)[14]
176
171.0,
157.6, 150.2
~142.6 (6.88 Hz)*
147.2,
149.5
~135.0 (5.48 Hz)*
135.9,
143.4
~151.5 (8.39 Hz)*
165.2
140,
153.5
153.2, 155.2
152.2,
150
~167.0 (10.27 Hz)*
165.7,
170
177.2, 175.7
~160.0 (9.56 Hz)*
158.2, 164.1
138.2, 144.2
RMSD, root mean square deviation. * Calculated coupling constant for estimated ϕ dihedral angle using Karplus relationship.
Figure 3. Superposition of six lowest-energy conformers of teicoplanin aglycone generated from a random conformational search using molecular dynamics perturbation (no restraints or solvation model was applied).
Magn. Reson. Chem. 2015, 53, 829–835
The backbone cis/trans-conformation was further supported by comparison of 13C chemical shifts computed at the DFT level of theory and the experimentally determined chemical shifts (Table 4). The data show excellent agreement of the average predicted and experimental 13C chemical shifts within ±4.1 ppm, consistent with the ensemble of structures. Examination of the DMSO solvent model shows that DMSO forms hydrogen bonds with teicoplanin aglycone between the sulfoxide oxygen S = O and NH C42 (1.93 Å) and NH F59 (2.38 Å) and with a longer range weak interaction with NH B62 (3.17 Å). Strong hydrophobic interactions between the methyl groups of DMSO and aromatic rings B and C keep this solvent molecule locked in a hydrophobic pocket formed by the macrocyclic ring structure. There appears to be a significant energy barrier to disrupting this interaction because the solvent molecule remained
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
833
relationship was met relative to the experimentally determined coupling constants (Fig. 4). A set of six conformers was generated from the solvation model using the ROE and dihedral restraints in Tables 2 and 3. The orientation of rings A and C was established with the application of observed NOE restraints between H29 and H24 as well as A17 and A23 for ring A and between H42 and H38 for ring C. With unrestrained modeling, rings A and C can flip orientation producing structures where these restraints would be violated. In addition, key distance restraints between G47 and F55 as well as between D2 and E11 were essential for establishing the relative orientation of rings G and F and rings E and D, respectively. The solvation model without NMR restraints produced structures that did not satisfy these ROE data. The resultant ensemble of structures superposed with a backbone root mean square deviation of 0.17 Å as shown in Fig. 5. The low-energy conformers yielded energy values of ~297 kcal/mol.
Figure 4. Solvation model of one low-energy conformer derived from NMR distance and dihedral angle restraints showing a DMSO molecule hydrogen bonded with C42–NH and F59–NH with weak association to B6–NH. Ring G is pushed back to accommodate the solvent molecule. This solvent molecule is also well situated for hydrophobic interaction between the DMSO methyl and teicoplanin aglycone aromatic rings B and C.
N. C. Gonnella et al.
Figure 5. Ensemble of six conformers (backbone root mean square deviation of 0.17 Å) derived from NMR restraints in a DMSO solvation model.
13
Table 4. Comparison of experimental and predicted C NMR chemical shifts for the backbone of teicoplanin aglycone 13
Atom number
Cδ (ppm)
D2 C41 C60 G43 G44 E15 E16 A18 A19 B63 B64 F58 F61
56.1 53.4 168.3 172.6 57 52.7 168.1 60.6 166.7 53.7 169.4 57.0 167.2
13
C δ (ppm) (DFT) calculated† 56.4 52.8 166.9 170.7 58.9 51.4 168.3 59.4 168.5 53.1 173.5 55.2 167.7
13
Δ Cδ (ppm) 0.3 0.6 1.4 1.9 1.9 1.3 0.2 1.2 1.8 0.6 4.1 1.8 0.5
RMSD 0.27 0.61 1.44 2.30 1.89 1.32 0.20 1.18 1.82 0.64 4.05 1.80 0.69
agreement between the two structures in the peptide backbone. The areas where the structures differ most center on the G and F aromatic ring regions. This is illustrated in the superposition of an NMR conformer with the X-ray structure (Fig. 6). In particular for the X-ray structure, aromatic ring G is bent in towards the hydrophobic pocket, whereas in the NMR structure, ring G opens up slightly. A major reason for the difference between the two conformers is that in the NMR structure, a DMSO molecule is hydrogen bonded to three amino NH protons pointing towards the center of the macrocycle cavity (Fig. 4). This opened side chain conformation, which may be attributed to van der Waals repulsions between DMSO and the aromatic side chain, largely contributes to the conformational change that causes aromatic ring G to bend back by approximately 2.1 Å relative to the X-ray structure. Although no solution structure of teicoplanin aglycone in water has been reported, the crystal structure of the teiocoplanin analog does show the molecule to exist in a network of water molecules. In fact, the crystal structure of A40926 aglycone shows hydrogenbonded water molecules in the same cavity and to the same relative amide nitrogens. Hence, even considering the 2.1-Å backward bend for ring G, there are still significant structural similarities for the solution structure in DMSO to what may be considered a more biologically relevant aqueous environment in the crystal structure analog. It is not surprising that the G and F ring portion of teicoplanin aglycone presents the opportunity for flexibility. Crystallography studies of the aglycone binding to a protein have shown the G
† Values are from GIAO [B3LYP-6-311G(d)] DFT level of theory (averaged over all relevant conformers).
fixed in all conformational searches, which include all dynamics runs before energy minimization. Other observed interactions between solvent and the cyclic peptide involved electrostatic interactions (~3.5 Å) between the solvent sulfoxide oxygen (S = O) and the chlorine atoms on aromatic rings C and A (Fig. 4). These results are consistent with hydrogen bonding of water to this NH-rich hydrogen donor site that was previously observed in the X-ray structure of a teicoplanin aglycone analog.[14] It should be noted that a conformation for optimal energy without distance or dihedral angle restraints was also generated. Without applied distance restraints and no solvation model, the molecule undergoes hydrophobic self-association in the gas phase bringing side chain rings G and D in close proximity. Such conformers were not consistent with NMR-derived experimental data. The overall results showed that ROE distance restraints and dihedral restraints derived from J coupling constants were best satisfied in the resultant structures when the DMSO solvation model was used in the conformational search. Conformational flexibility
834
Comparison of the calculated NMR structure of teicoplanin aglycone with the X-ray structure of A40926 aglycone shows good
wileyonlinelibrary.com/journal/mrc
Figure 6. Superposition of NMR-derived structure of teicoplanin aglycone (blue) and the X-ray structure of A-40926 (gold)[14] show differences in the F and G ring regions of the molecule.
Figure 7. Superposition of two molecules of teicoplanin aglycone extracted from a complex with Teg12.[28] The figure illustrates the dramatic conformational difference adopted by aromatic rings G and F upon protein binding.
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 829–835
NMR-derived conformation of teicoplanin aglycone using solvation model and F rings to interface at the binding site. The flexibility of this part of the molecule is evident from an X-ray structure of the aglycone bound to sulfotransferase Teg12.[28] Teg12 is a protein known to sulfate teicoplanin-like glycopeptides and is reported to undergo a series of significant conformational rearrangements during glycopeptide recruitment, binding, and catalysis when complexed with teicoplanin aglycone molecules. Likewise, teicoplanin aglycone shows considerable flexibility at rings G and F upon binding to Teg 12. This is dramatically illustrated in the superposition of two conformers extracted from the protein–aglycone complex (Fig. 7).[28] These results are consistent with our findings regarding the flexibility of the G and F portions of teicoplanin aglycone.
Conclusions The solution structure of teicoplanin aglycone in DMSO was solved yielding a well-defined ensemble of conformers. Simulations in a DMSO solvation model converged to produce structures of teicoplanin aglycone that satisfied experimental NMR data. Chemical shift predictions using the DFT level of theory further supported the backbone cis/trans-conformation of the calculated structures. Greater flexibility on the G and F rings was evident from comparison with the X-ray structure of an aglycone analog and corroborated by reported protein-binding structures. Differences between the solution and X-ray structures may be attributed to hydrogen bonding between the NH cluster in the antibiotic cavity and the oxygen of the DMSO solvent where the flexible side chains can move to accommodate solvent. Overall use of a solvation model was important for the best fit of solution conformers with NMR data. Acknowledgement The authors wish to thank Dr David Bell, Sigma Aldrich/Supelco, Bellefonte, Pennsylvania, for the generous gift of teicoplanin aglycone.
References [1] F. Parenti, G. Beretta, M. Berti, V. Arioli. J. Antibiot. 1978, 31, 276–283. doi:10.7164/ antibiotics. 31.276. [2] A. Bernareggi, A. Borghi, M. Borgonovi, L. Cavenaghi, P. Ferrari, K. Vékey, M. Zanol, L. Zerilli. Antimicrob. Agents Chemother. 1992, 36(8), 1744– 1749. doi:10.1128/AAC.36.8.1744. [3] A. H. Hunt, R. M. Molloy, J. L. Occolowitz, G. G. Marconi, M. Debono. J. Am. Chem. Soc. 1984, 106, 4891–4895. doi:10.1021/ja00329a043. [4] J. C. J. Barna, D. H. Williams, D. J. M. Stone, T.-W. C. Bung, D. M. Doddrell. J. Am. Chem. Soc. 1984, 106, 4895–4902. doi:10.1021/ja00329a044. [5] A. Malabarba, P. Strazzolinia, A. Depaol, M. Landi, M. Berti, B. Cavalleri. J. Antibiot. 1984, 37, 988–999. doi:10.7164/antibiotics.37.988.
[6] A. Malabarba, P. Ferrari, G. G. Gallo, J. Kettenring, B. Cavalleri. J. Antibiot. 1986, 1430–1442. doi:10.7164/ antibiotics.39.1430. [7] S. H. Smallcombe, S. L. Patt, P. A. Keifer. J. Magn. Reson. A 1995, 117, 295–303. doi:10.1006/jmra.1995.0759. [8] A. Bax, D. G. Davis. J. Magn. Reson. 1985, 65, 355–360. DOI: 10.1016/ 0022-2364(85)90018-6 [9] A. Bax, D. G. Davis. J. Magn. Reson. 1985, 63, 207–213. DOI: 10.1016/ 0022-2364(85)90171-4 [10] A. G. Palmer III, J. Cavanagh, P. E. Wright, M. Rance. J. Magn. Reson. 1991, 93, 151–170. doi:10.1016/0022-2364(91)90036-S. [11] L. E. Kay, P. Keifer, T. Saarinen. J. Am. Chem. Soc. 1992, 114, 10663–5. doi:10.1021/ja00052a088. [12] J. Schleucher, M. Schwendinger, M. Sattler, P. Schmidt, O. Schedletzky, S. J. Glaser, O. W. Sorensen, C. Griesinger. J. Biomol. NMR 1994, 4, 301–306. doi:10.1007/BF00175254. [13] M. Karplus. J. Chem. Phys. 1959, 30, 11–15. doi:10.1063/1.1729860. [14] M. Schafer, E. Pohl, K. Schmidt-Base, G. M. Sheldrick, R. Hermann, A. Malabarba, M. Nebuloni, M. Merrell, G. Pelizzi. Helv. Chim. Acta 1996, 79, 1916–1924. [15] Molecular operating environment (MOE), 2013.08; Chemical Computing Group Inc., 1010 Sherbooke St West, Suite #910, Montreal, QC, Canada, H3A 2R7, 2013. [16] T. A. Halgren. J. Comp. Chem. 1996, 17, 490–519. doi:10.1002/(SICI) 1096-987X(199604)17:5/63.0.CO;2-P. [17] A. D. Becke. J. Chem. Phys. 1993, 98, 5648–5652. doi:10.1063/1.464913. [18] C. Lee, W. Yang, R. G. Parr. Phys. Rev. B 1988, 37, 785–789. [19] R. Krishnan, J. S. Binkley, R. Seeger, J. A. Pople. J. Chem. Phys. 1980, 72, 650–654. doi:10.1063/1.438955. [20] A. D. McLean, G. S. Chandler. J. Chem. Phys. 1980, 72, 5639–48. doi:10.1063/1.438980. [21] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian 09, Revision C.01, Gaussian, Inc., Wallingford CT, 2009. [22] F. London. J. Phys. Radium 1937, 8, 397–409. [23] R. McWeeny. Phys. Rev. 1962, 126, 1028. [24] R. Ditchfield. Mol. Phys. 1974, 27, 789–807. doi:10.1080/ 00268977400100711. [25] K. Wolinski, J. F. Hilton, P. Pulay. J. Am. Chem. Soc. 1990, 112, 8251–60. doi:10.1021/ja00179a005. [26] J. R. Cheeseman, G. W. Trucks, T. A. Keith, M. J. Frisch. J. Chem. Phys. 1996, 104, 5497–509. doi:10.1063/1.471789. [27] M. Baias, J-N. Dumez, P. H. Svensson, S. Schantz, G. M. Day, L. Emsley. J. Am. Chem. Soc. 2013, 135, 17501–17507. DOI: 10.1021/ja4088874 [28] M. J. Bick, J. J. Banik, S. A. Darst, S. F. Brady. Biochemistry 2010, 49(19), 4159–4168. doi:10.1021/bi100150v.
835
Magn. Reson. Chem. 2015, 53, 829–835
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
Revised: 12 May 2015
Accepted: 19 May 2015
Published online in Wiley Online Library: 1 July 2015
(wileyonlinelibrary.com) DOI 10.1002/mrc.4279
Three-dimensional structure of cyclic antibiotic teicoplanin aglycone using NMR distance and dihedral angle restraints in a DMSO solvation model Nina C. Gonnella,a* Nelu Grinberg,b Mark Mcloughlin,a† Om Choudhary,a,b‡ Keith Fandrickb and Shengli Mab The three-dimensional solution conformation of teicoplanin aglycone was determined using NMR spectroscopy. A combination of NOE and dihedral angle restraints in a DMSO solvation model was used to calculate an ensemble of structures having a root mean square deviation of 0.17 Å. The structures were generated using systematic searches of conformational space for optimal satisfaction of distance and dihedral angle restraints. Comparison of the NMR-derived structure of teicoplanin aglycone with the X-ray structure of a teicoplanin aglycone analog revealed a common backbone conformation with deviation of two aromatic side chain substituents. Experimentally determined backbone 13C chemical shifts showed good agreement with those computed at the density functional level of theory, providing a cross validation of the backbone conformation. The flexible portion of the molecule was consistent with the region that changes conformation to accommodate protein binding. The results showed that a hydrogenbonded DMSO molecule in combination with NMR-derived restraints together enabled calculation of structures that satisfied experimental data. Copyright © 2015 John Wiley & Sons, Ltd. Keywords: teicoplanin aglycone; nuclear magnetic resonance; conformation; 3D structure; solvation model; density functional theory
Introduction
Magn. Reson. Chem. 2015, 53, 829–835
* Correspondence to: Nina C. Gonnella, Materials and Analytical Sciences, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT, USA. E-mail: nina. [email protected] † ‡
Present Address: 6 Harbour Town Ct, Frisco, TX, USA Present Address: 5535 Centre Ave, Apt 2, Pittsburgh, PA, USA
a Materials and Analytical Sciences, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT, USA b Chemical Development, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT, USA Abbreviations: WET, experiment selective solvent suppression experiment; COSY, correlation spectroscopy; TOCSY, total correlation spectroscopy; HSQC, heteronuclear single-quantum correlation; HMBC, heteronuclear multiple-bond correlation; HMQC, heteronuclear multiple-quantum correlation; ROESY, rotating frame Overhauser effect spectroscopy; QSINE, sine bell squared function; SINE, shifted sine bell function; GARP, globally optimized alternating phase rectangular pulses; WALTZ, wideband alternating phase low-power technique for zero residual splitting; MMFF94×, Merck molecular force field 94×.
Copyright © 2015 John Wiley & Sons, Ltd.
829
Teicoplanin is a potent antibiotic for the treatment of serious infections caused by Gram-positive bacteria, such as methicillin-resistant Staphylococcus aureus, coagulase-negative Staphylococci, Clostridia, and Enterococci. The compound’s antibiotic mechanism of action has been reported to occur through the inhibition of bacterial cell wall synthesis.[1] These antibiotics are known to have in vivo stability based upon studies with radiolabeled teicoplanin in rats and humans where the drug was shown to undergo renal elimination with very minor (~5%) metabolic transformation.[2] The teicoplanin antibiotic is a mixture of several glycopeptidelike antibiotics, which all share a common core system.[3] These compounds are members of the vancomycin group of glycopeptide antibiotics. A fused ring structure forms the common core of the teicoplanin antibiotic family that is termed teicoplanin aglycone (Sigma Aldrich, St Louis, MO, USA).[4] The aglycone core is prepared from hydrolytic treatment of teicoplanin to remove the sugar and fatty acid moieties.[5] Both the chemical structure and stereochemistry of teicoplanin aglycone have been previously reported and are consistent with the chemical structure of the compound described herein.[5] The chemical structure, stereochemistry, and numbering system of teicoplanin aglycone used in this study are given in Fig. 1. Chemical shift assignments have been reported for teicoplanin aglycone hydrochloride salt in DMSO and other solvents[6]; however, NMR studies of teicoplanin aglycone (nonsalt form) have not been reported, and the three-dimensional solution conformation of teicoplanin aglycone has not been solved.
Here, we report the first experimentally determined solution conformation of teicoplanin aglycone in DMSO. Full proton and carbon-13 chemical shift assignments of the molecule are also included. NMR measurement of teicoplanin aglycone in aqueous solution was not possible because of low compound solubility. The three-dimensional solution structure of this molecule is important because it provides valuable insight into the conformation and structural flexibility of the antibiotic required in biological function.
N. C. Gonnella et al.
Figure 1. Chemical structure of teicoplanin aglycone with (R or S) asymmetric centers displayed. Aromatic side chains are labeled A–F with sequential numbering nomenclature for carbon atoms and attached protons.
The reported structure was calculated from NMR-derived interproton distances and targeted dihedral angle restraints based on measured coupling constants. The results revealed that a constrained search of conformational space in DMSO solvent was a main component in establishing the three-dimensional structure. Comparison of experimentally determined 13C chemical shifts with those predicted at the density functional theory (DFT) level of theory supports for the backbone cis/trans-NH–CO ‘ω’ angle peptide bond orientation. In addition, derived conformers were compared with an X-ray structure of a similar teicoplanin aglycone analog showing that flexibility of the core aromatic side chain is consistent with the region of the molecule involved in protein binding.
Experimental and computational methods Sample preparation Teicoplanin aglycone was obtained from Supelco (Sigma Aldrich/Supelco). The sample was prepared by dissolving 5 mg of teicoplanin aglycone in 600-μl DMSO-d6 (Sigma Aldrich), and 30 μl was transferred to a 1.7-mm microcapillary NMR tube. NMR spectra
830
NMR experiments were performed on a Bruker-Biospin DRX 600 NMR spectrometer (Bruker Corporation, Billerica, MA, USA) operating at 600.2 MHz for 1H and 150.92 MHz for 13C, nonspinning at 300 K, using a 1.7-mm CP TCI Z gradient probe. 1H and 13C chemical shifts were referenced to DMSO-d6 at 2.5 and 39.5 ppm, respectively. One-dimensional (1D) 1H data (16K points) were obtained using a spectral width of 12 019 Hz and a WET 1D experiment with double irradiation of H2O and residual proteo-DMSO. WET solvent suppression was achieved using 90° sine.1 shape pulses with 13C low-power decoupling during WET and acquisition.[7] Data were processed with em and line broadening of 0.05 Hz before Fourier transformation. Gradient-selected COSY spectra, using a shaped pulse for double off-resonance presaturation with continuous-wave decoupling on the 13C channel, were obtained (2048 × 160 data points) with a relaxation delay of 2.0 s, a spectral width of 9614 Hz in both dimensions, and a 90° read pulse. TOCSY data were obtained in phasesensitive mode (TPPI) using 0.68 W for excitation and 0.13 W for spinlock. WET solvent suppression with sine.1 shaped pulse and 13 C decoupling on during WET and data acquisition were used.[8] Data were acquired using 2048 × 256 data points, a relaxation delay of 1.5 s, a spectral width of 8680 Hz in both dimensions, and a 60-ms spinlock. Phase-sensitive ROESY spectra with continuous-wave
wileyonlinelibrary.com/journal/mrc
spinlock and presaturation of the water resonance[9] were obtained (1024 × 256 data points) using a relaxation delay of 2.0 s, a spectral width of 12 004 Hz in both dimensions, and a 500-ms mixing time. 1 H–13C WET-HSQC data was collected using echo/anti-echo-TPPI gradient selection and WET solvent suppression with a sine.1 shape pulse with decoupling during acquisition.[10–12] Spectra were obtained (1024 × 256 data points) with GARP decoupling, a relaxation delay of 2.0 s, and a spectral width of 8680 Hz in f2 and 36 219 Hz in f1. Gradient-selected 1H–13C HMQC data were also obtained (2048 × 256 data points) using GARP decoupling, a relaxation time of 2.0 s per transient, and a spectral width of 8680 Hz in f2 and 36 225 Hz in f1. Gradient-selected 1H–13C gHMBC data were acquired (2048 × 512 data points) using WALTZ-16 decoupling, a relaxation delay of 1.5 s, and a spectral width of 8680 Hz in f2 and 36 225 Hz in f1. Delays were calculated using 1J (13C–1H) of 145 Hz and two-bond/three-bond J (13C–1H) of 4.5 Hz. Two-dimensional spectra were processed with SINE or QSINE apodization in both dimensions before Fourier transformation. Computational methods ROE intensities were converted to distance-restraint ranges, assuming isotropic motion and no spin diffusion. A wide tolerance range (1.8–6.0 Å) was applied to most distance restraints. Restraints showing strong ROE cross peaks were tightened to a range of 1.8–3.5 Å. The lower bound was set at 1.8 Å consistent with van der Waals lower bound restraints. The distance-restraint weighting was set to 300 kcal/mol. A total of 57 distance restraints and six NH–CHα dihedral angle restraints were used in the conformational search with a 10° range around an expected dihedral angle (weight of 500 kcal/mol). Backbone ϕ torsion angles of the core were sampled over a range consistent with the expected trend in couplings from the Karplus relationship 3J = A cos2(ϕ) + B cos(ϕ) + C.[13] The starting structure for teicoplanin aglycone was derived from a dynamics run followed by a rigorous and extensive stochastic conformational search. The lowest-energy conformer had cis/trans-NH–CO (ω) geometry consistent with the X-ray crystal structure of A-40926.[14] A gas-phase model with explicit DMSO molecules was minimized in a wall-restraint box with dimensions 7.8598 × 8.1661 × 8.0499 Å. Three-dimensional structures were generated using distance-constrained molecular dynamics perturbations along low frequency and vibrational modes, as implemented in molecular operating environment software[15] (Chemical Computing Group Inc., 1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada), running on a windows-based PC. A modified MMFF94× force field[16] was used in the conformational search, incorporating ROE distance and ϕ angle restraints. The force field was parameterized for organic molecules keeping conjugated nitrogens planar. DFT calculations Optimized structures were used as input geometries for DFT calculations. The DFT functional was B3LYP,[17,18] and the basis set was 6-311G(d).[19,20] Geometry optimization, harmonic frequency calculations, and the magnetic shielding tensor calculations were performed with the Gaussian09 package[21] using the gaugeindependent atomic orbital[22–26] method. The calculations were performed in vacuo. Excellent agreement between experimentally determined backbone chemical shifts and calculated results (using a DMSO-d6 solvation, conductor–like polarizable continuum model) was found.
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 829–835
NMR-derived conformation of teicoplanin aglycone using solvation model Chemical shieldings σcalc were converted to chemical shifts δcalc using the property δcalc = σref σcalc.[27] This conversion is carried out using a reference value, σref, to best align the calculated and experimental spectra. The reference value was determined by a linear fit between calculated shieldings and experimental chemical shifts for each structure. The average value of σref is 178.1 ppm for 13 C chemical shifts.
Results and discussion Structure assignments The numbering system for teicoplanin aglycone consisted of labeling the aromatic rings with letters A–F similar to that previously described[3] and sequential numbering around the macroscopic ring (Fig. 1). A 1D WET proton spectrum of teicoplanin aglycone is given in Fig. 2. The spectrum shows a single set of relatively sharp signals, good peak dispersion, and spin–spin coupling that enabled unambiguous proton resonance assignment. The NMR spectra showed no evidence of distinct slowly interconverting rotamers in solution. Proton chemical shift assignments were achieved using 1H,1H presaturation-COSY, 1H,1H WET-TOCSY, and 1H,1H presaturation– ROESY experiments. The COSY and TOCSY experiments established vicinal and long range through bond proton connectivity, while the ROESY experiment corroborated assignments from through-space proximity. The 1H,13C WET-HSQC experiment corroborated assignments for the nonequivalent protons of C40 as well as providing supporting evidence for the identification of exchangeable amide (NH) resonances and several hydroxyl (OH) resonance protons. Because of the relatively low concentration of teicoplanin aglycone used in this study (~5 mg/600 μl), the 13C resonances were assigned solely based upon 1H,13C HSQC, 1H,13C HMQC, and 1 13 H, C HMBC indirect detection experiments. The HSQC and HMQC experiments allowed facile assignment of all proton-bearing carbons. The quaternary carbons and carbonyl carbons were assigned using HMBC correlations. The full 1H and 13C chemical shift assignments of teicoplanin aglycone in DMSO-d6 are given in Table 1. The ROE measurements not only enabled chemical shift assignments but also provided through-space proximity to generate conformational studies. ROE data were used to derive distancerestraint ranges that enabled the calculation of an ensemble of low-energy conformers (Table 2).
Computational studies The lowest-energy conformers generated by a stochastic conformational search and in agreement with the X-ray structure of antibiotic A40926 aglycone[14] served as a template for establishing the basic peptide backbone starting structure for teicoplanin aglycone. The NH–CO peptide bonds were all trans with the exception of E16A17, which adopts a cis-conformation. Although both A40926 aglycone and teicoplanin aglycone have major similarities, there were two minor differences between the two structures. The first difference is that A-40926 aglycone places the Cl atom on aromatic ring F instead of aromatic ring C in teicoplanin agclycone. Second, the amino group (NH2; G45 in Fig. 1) is methylated in A-40926. The crystal structure of A-40926 aglycone shows two independent molecules in the asymmetric unit cell, crystallizing in the orthorhombic space group P212121. Although there are relatively small differences between the two X-ray conformers, both conformers afforded proper stereochemistry and cis/trans-NH–CO ‘ω’ angles. The backbone peptide cis/trans-conformation was also achieved by carrying out a random conformational search (no restraints applied) using molecular dynamics. A set of six lowest-energy structures is shown in Fig. 3. Such structures showed evidence of hydrophobic collapse of rings G and F with rings D and E and did not provide definition of the orientation of aromatic rings A and C where these rings can flip with chlorines facing up or down. A solvation model of teicoplanin aglycone was produced by soaking the starting structure in a 1.0-margin box of DMSO and energy minimizing the structure. This resulted in incorporation of a
1
1
H NMR spectrum (600.20 MHz) of teicoplanin aglycone in dimethylsulfoxide-d6. Assignments of H chemical shifts are displayed on the spectrum.
Magn. Reson. Chem. 2015, 53, 829–835
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
831
Figure 2.
A total of 55 positive distance restraints were extracted from the ROESY spectrum. Initial conformational studies produced structures placing NH C42 in close proximity with aromatic protons G51 and F57, which were not supported by NMR data. To compensate for these unsupported interactions, two negative restraints were added between C42 and G51 and C42 and F57 (Table 2), which served to structurally satisfy experimental data. Vicinal coupling constants 3JNCα were also extracted from the 1D proton spectra. All five NH–Hα coupling constants were measured from the 1D proton spectrum, and ϕ angle-restraint ranges were estimated based upon the Karplus relationship.[13] The dihedral restraints were used to further refine the backbone conformation of the macrocyclic core (Table 3).
N. C. Gonnella et al. Table 1. Chemical shift assignments of teicoplanin aglycone in DMSOd6 at 300 K Atom number
832
D1–CO2 D2 D3 D4 D5 D6 D7 D8 D20–NH E9 E10 E11 E12 E13 E14 E15 E16 E65–NH A17–NH A18 A19 A21 A22 A23 A24 A25 A26 A27 B28 B29 B30 B31 B32 B33 B62–NH B63 B64 C34 C35 C36 C37 C38 C39 C40 C41 C42–NH C60 G43 G44 G45–NH2 G46 G47 G48 G49 G50 G51 F52 F53 F54 F55 F56 F57 F58 F59–NH F61 A21(OH) D5(OH) F54(OH)
1
H δ (ppm)
— 4.33 — 6.22 — 6.34 — — 8.34 — — 6.60 6.64 — 7.06 4.29 — 8.43 6.70 4.09 — 5.04 — 7.39 7.21 — — 7.75 — 5.06 — 5.46 — — 7.61 5.64 — — — 7.17 — 7.64 7.22 2.81, 3.27 4.94 7.38 — — 4.52 n/o — 6.61 — — 6.88 7.07 — 6.28 — 6.28 — 6.34 5.28 7.67 — 5.87 9.35 9.68
wileyonlinelibrary.com/journal/mrc
13
C δ (ppm) 171.7 56.1 135.5 104.8 156.2 101.6 156.3 116.9 — 120.2 154.4 116.0 124.8 124.0 134.7 52.7 168.1 — — 60.6 166.7 70.9 141.3 126.4 122.7 147.4 125.2 126.4 146.6 103.3 126.7 106.4 146.6 133.2 — 53.7 169.4 149.5 126.8 129.9 134.6 129.1 123.7 35.7 53.4 — 168.3 172.6 57.0 — 130.5 116.0 140.5 145.9 117.3 124.1 156.3 103.8 157.5 109.2 139.9 101.6 57.0 — 167.2 — — —
Table 2. Restraints from ROESY of teicoplanin aglycone in DMSO-d6 at 300 K Proton 1
Proton 2
Restraint range
A18 C40 C40b C40a A18 E15 D2 D2 A18 E15 D2 A18 A18 B62 B31 F59 D2 F59 E15 A21 E65 E65 E65 E14 F54(OH) G44 G51 G44 C41 C41 B63 A21 F58 B63 F58 B62 B62 A21(OH) A21(OH) A17 G47 C47 F58 C38 A18 A18 C41 A21 A21 B29 B29 F58 C41 C42 C42 C42 C42
E15 C41 C36 C38 A21 B29 D4 D20 A23 E14 E14 A27 D20 F57 C39 C42 E11 F57 E65 D20 B29 B63 E12 D20 F53 C42 G44 G47 C42 G47 B29 A21(OH) F55 B31 F59 B31 B63 A23 A17 A23 F55 F57 F57 C42 A17 E14 F59 A27 A23 A23 A24 B62 F58 G47 C39 G51 F57
(1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–5.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–3.5) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (1.8–6.0) (4.0–7.0) negative restraint (4.0–7.0) negative restraint
DMSO molecule in a pocket formed by the B, C, F, and G rings establishing hydrogen bonding between the DMSO oxygen and amide NH C42 and NH F59. This interaction with DMSO secured a Karplus ϕ angle relative trend consistent with the experimentally measured NMR coupling constants. Structures incorporating the hydrogen-bonded DMSO molecule and NMR-derived distance and dihedral angle restraints yielded conformers where the Karplus
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 829–835
NMR-derived conformation of teicoplanin aglycone using solvation model Table 3. NH–CHα coupling constants (Hz) and dihedral angles (ϕ) for teicoplanin aglycone in DMSO-d6 Proton
A17 A18 D2 D20 E15 E65 B62 B63 F58 F59 C41 C42
3
JNα (Hz)
11.42 (±0.5) 11.95 (±0.5) 6.05 (±0.5) — 5.33 (±0.5) — 8.25 (±0.07) 8.32 (±0.07) 10.63 (±0.01) 10.64 (±0.01) — 9.65 (+0.5)
Estimated Karplus angle from quadratic fit[13]
Measured range of dihedral angles (ϕ) from conformational search with NMR restraints (six conformers)
~176.8 (10.77 Hz)*
177,
Measured dihedral angles (ϕ) from X-ray structure (two conformers)[14]
176
171.0,
157.6, 150.2
~142.6 (6.88 Hz)*
147.2,
149.5
~135.0 (5.48 Hz)*
135.9,
143.4
~151.5 (8.39 Hz)*
165.2
140,
153.5
153.2, 155.2
152.2,
150
~167.0 (10.27 Hz)*
165.7,
170
177.2, 175.7
~160.0 (9.56 Hz)*
158.2, 164.1
138.2, 144.2
RMSD, root mean square deviation. * Calculated coupling constant for estimated ϕ dihedral angle using Karplus relationship.
Figure 3. Superposition of six lowest-energy conformers of teicoplanin aglycone generated from a random conformational search using molecular dynamics perturbation (no restraints or solvation model was applied).
Magn. Reson. Chem. 2015, 53, 829–835
The backbone cis/trans-conformation was further supported by comparison of 13C chemical shifts computed at the DFT level of theory and the experimentally determined chemical shifts (Table 4). The data show excellent agreement of the average predicted and experimental 13C chemical shifts within ±4.1 ppm, consistent with the ensemble of structures. Examination of the DMSO solvent model shows that DMSO forms hydrogen bonds with teicoplanin aglycone between the sulfoxide oxygen S = O and NH C42 (1.93 Å) and NH F59 (2.38 Å) and with a longer range weak interaction with NH B62 (3.17 Å). Strong hydrophobic interactions between the methyl groups of DMSO and aromatic rings B and C keep this solvent molecule locked in a hydrophobic pocket formed by the macrocyclic ring structure. There appears to be a significant energy barrier to disrupting this interaction because the solvent molecule remained
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
833
relationship was met relative to the experimentally determined coupling constants (Fig. 4). A set of six conformers was generated from the solvation model using the ROE and dihedral restraints in Tables 2 and 3. The orientation of rings A and C was established with the application of observed NOE restraints between H29 and H24 as well as A17 and A23 for ring A and between H42 and H38 for ring C. With unrestrained modeling, rings A and C can flip orientation producing structures where these restraints would be violated. In addition, key distance restraints between G47 and F55 as well as between D2 and E11 were essential for establishing the relative orientation of rings G and F and rings E and D, respectively. The solvation model without NMR restraints produced structures that did not satisfy these ROE data. The resultant ensemble of structures superposed with a backbone root mean square deviation of 0.17 Å as shown in Fig. 5. The low-energy conformers yielded energy values of ~297 kcal/mol.
Figure 4. Solvation model of one low-energy conformer derived from NMR distance and dihedral angle restraints showing a DMSO molecule hydrogen bonded with C42–NH and F59–NH with weak association to B6–NH. Ring G is pushed back to accommodate the solvent molecule. This solvent molecule is also well situated for hydrophobic interaction between the DMSO methyl and teicoplanin aglycone aromatic rings B and C.
N. C. Gonnella et al.
Figure 5. Ensemble of six conformers (backbone root mean square deviation of 0.17 Å) derived from NMR restraints in a DMSO solvation model.
13
Table 4. Comparison of experimental and predicted C NMR chemical shifts for the backbone of teicoplanin aglycone 13
Atom number
Cδ (ppm)
D2 C41 C60 G43 G44 E15 E16 A18 A19 B63 B64 F58 F61
56.1 53.4 168.3 172.6 57 52.7 168.1 60.6 166.7 53.7 169.4 57.0 167.2
13
C δ (ppm) (DFT) calculated† 56.4 52.8 166.9 170.7 58.9 51.4 168.3 59.4 168.5 53.1 173.5 55.2 167.7
13
Δ Cδ (ppm) 0.3 0.6 1.4 1.9 1.9 1.3 0.2 1.2 1.8 0.6 4.1 1.8 0.5
RMSD 0.27 0.61 1.44 2.30 1.89 1.32 0.20 1.18 1.82 0.64 4.05 1.80 0.69
agreement between the two structures in the peptide backbone. The areas where the structures differ most center on the G and F aromatic ring regions. This is illustrated in the superposition of an NMR conformer with the X-ray structure (Fig. 6). In particular for the X-ray structure, aromatic ring G is bent in towards the hydrophobic pocket, whereas in the NMR structure, ring G opens up slightly. A major reason for the difference between the two conformers is that in the NMR structure, a DMSO molecule is hydrogen bonded to three amino NH protons pointing towards the center of the macrocycle cavity (Fig. 4). This opened side chain conformation, which may be attributed to van der Waals repulsions between DMSO and the aromatic side chain, largely contributes to the conformational change that causes aromatic ring G to bend back by approximately 2.1 Å relative to the X-ray structure. Although no solution structure of teicoplanin aglycone in water has been reported, the crystal structure of the teiocoplanin analog does show the molecule to exist in a network of water molecules. In fact, the crystal structure of A40926 aglycone shows hydrogenbonded water molecules in the same cavity and to the same relative amide nitrogens. Hence, even considering the 2.1-Å backward bend for ring G, there are still significant structural similarities for the solution structure in DMSO to what may be considered a more biologically relevant aqueous environment in the crystal structure analog. It is not surprising that the G and F ring portion of teicoplanin aglycone presents the opportunity for flexibility. Crystallography studies of the aglycone binding to a protein have shown the G
† Values are from GIAO [B3LYP-6-311G(d)] DFT level of theory (averaged over all relevant conformers).
fixed in all conformational searches, which include all dynamics runs before energy minimization. Other observed interactions between solvent and the cyclic peptide involved electrostatic interactions (~3.5 Å) between the solvent sulfoxide oxygen (S = O) and the chlorine atoms on aromatic rings C and A (Fig. 4). These results are consistent with hydrogen bonding of water to this NH-rich hydrogen donor site that was previously observed in the X-ray structure of a teicoplanin aglycone analog.[14] It should be noted that a conformation for optimal energy without distance or dihedral angle restraints was also generated. Without applied distance restraints and no solvation model, the molecule undergoes hydrophobic self-association in the gas phase bringing side chain rings G and D in close proximity. Such conformers were not consistent with NMR-derived experimental data. The overall results showed that ROE distance restraints and dihedral restraints derived from J coupling constants were best satisfied in the resultant structures when the DMSO solvation model was used in the conformational search. Conformational flexibility
834
Comparison of the calculated NMR structure of teicoplanin aglycone with the X-ray structure of A40926 aglycone shows good
wileyonlinelibrary.com/journal/mrc
Figure 6. Superposition of NMR-derived structure of teicoplanin aglycone (blue) and the X-ray structure of A-40926 (gold)[14] show differences in the F and G ring regions of the molecule.
Figure 7. Superposition of two molecules of teicoplanin aglycone extracted from a complex with Teg12.[28] The figure illustrates the dramatic conformational difference adopted by aromatic rings G and F upon protein binding.
Copyright © 2015 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2015, 53, 829–835
NMR-derived conformation of teicoplanin aglycone using solvation model and F rings to interface at the binding site. The flexibility of this part of the molecule is evident from an X-ray structure of the aglycone bound to sulfotransferase Teg12.[28] Teg12 is a protein known to sulfate teicoplanin-like glycopeptides and is reported to undergo a series of significant conformational rearrangements during glycopeptide recruitment, binding, and catalysis when complexed with teicoplanin aglycone molecules. Likewise, teicoplanin aglycone shows considerable flexibility at rings G and F upon binding to Teg 12. This is dramatically illustrated in the superposition of two conformers extracted from the protein–aglycone complex (Fig. 7).[28] These results are consistent with our findings regarding the flexibility of the G and F portions of teicoplanin aglycone.
Conclusions The solution structure of teicoplanin aglycone in DMSO was solved yielding a well-defined ensemble of conformers. Simulations in a DMSO solvation model converged to produce structures of teicoplanin aglycone that satisfied experimental NMR data. Chemical shift predictions using the DFT level of theory further supported the backbone cis/trans-conformation of the calculated structures. Greater flexibility on the G and F rings was evident from comparison with the X-ray structure of an aglycone analog and corroborated by reported protein-binding structures. Differences between the solution and X-ray structures may be attributed to hydrogen bonding between the NH cluster in the antibiotic cavity and the oxygen of the DMSO solvent where the flexible side chains can move to accommodate solvent. Overall use of a solvation model was important for the best fit of solution conformers with NMR data. Acknowledgement The authors wish to thank Dr David Bell, Sigma Aldrich/Supelco, Bellefonte, Pennsylvania, for the generous gift of teicoplanin aglycone.
References [1] F. Parenti, G. Beretta, M. Berti, V. Arioli. J. Antibiot. 1978, 31, 276–283. doi:10.7164/ antibiotics. 31.276. [2] A. Bernareggi, A. Borghi, M. Borgonovi, L. Cavenaghi, P. Ferrari, K. Vékey, M. Zanol, L. Zerilli. Antimicrob. Agents Chemother. 1992, 36(8), 1744– 1749. doi:10.1128/AAC.36.8.1744. [3] A. H. Hunt, R. M. Molloy, J. L. Occolowitz, G. G. Marconi, M. Debono. J. Am. Chem. Soc. 1984, 106, 4891–4895. doi:10.1021/ja00329a043. [4] J. C. J. Barna, D. H. Williams, D. J. M. Stone, T.-W. C. Bung, D. M. Doddrell. J. Am. Chem. Soc. 1984, 106, 4895–4902. doi:10.1021/ja00329a044. [5] A. Malabarba, P. Strazzolinia, A. Depaol, M. Landi, M. Berti, B. Cavalleri. J. Antibiot. 1984, 37, 988–999. doi:10.7164/antibiotics.37.988.
[6] A. Malabarba, P. Ferrari, G. G. Gallo, J. Kettenring, B. Cavalleri. J. Antibiot. 1986, 1430–1442. doi:10.7164/ antibiotics.39.1430. [7] S. H. Smallcombe, S. L. Patt, P. A. Keifer. J. Magn. Reson. A 1995, 117, 295–303. doi:10.1006/jmra.1995.0759. [8] A. Bax, D. G. Davis. J. Magn. Reson. 1985, 65, 355–360. DOI: 10.1016/ 0022-2364(85)90018-6 [9] A. Bax, D. G. Davis. J. Magn. Reson. 1985, 63, 207–213. DOI: 10.1016/ 0022-2364(85)90171-4 [10] A. G. Palmer III, J. Cavanagh, P. E. Wright, M. Rance. J. Magn. Reson. 1991, 93, 151–170. doi:10.1016/0022-2364(91)90036-S. [11] L. E. Kay, P. Keifer, T. Saarinen. J. Am. Chem. Soc. 1992, 114, 10663–5. doi:10.1021/ja00052a088. [12] J. Schleucher, M. Schwendinger, M. Sattler, P. Schmidt, O. Schedletzky, S. J. Glaser, O. W. Sorensen, C. Griesinger. J. Biomol. NMR 1994, 4, 301–306. doi:10.1007/BF00175254. [13] M. Karplus. J. Chem. Phys. 1959, 30, 11–15. doi:10.1063/1.1729860. [14] M. Schafer, E. Pohl, K. Schmidt-Base, G. M. Sheldrick, R. Hermann, A. Malabarba, M. Nebuloni, M. Merrell, G. Pelizzi. Helv. Chim. Acta 1996, 79, 1916–1924. [15] Molecular operating environment (MOE), 2013.08; Chemical Computing Group Inc., 1010 Sherbooke St West, Suite #910, Montreal, QC, Canada, H3A 2R7, 2013. [16] T. A. Halgren. J. Comp. Chem. 1996, 17, 490–519. doi:10.1002/(SICI) 1096-987X(199604)17:5/63.0.CO;2-P. [17] A. D. Becke. J. Chem. Phys. 1993, 98, 5648–5652. doi:10.1063/1.464913. [18] C. Lee, W. Yang, R. G. Parr. Phys. Rev. B 1988, 37, 785–789. [19] R. Krishnan, J. S. Binkley, R. Seeger, J. A. Pople. J. Chem. Phys. 1980, 72, 650–654. doi:10.1063/1.438955. [20] A. D. McLean, G. S. Chandler. J. Chem. Phys. 1980, 72, 5639–48. doi:10.1063/1.438980. [21] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian 09, Revision C.01, Gaussian, Inc., Wallingford CT, 2009. [22] F. London. J. Phys. Radium 1937, 8, 397–409. [23] R. McWeeny. Phys. Rev. 1962, 126, 1028. [24] R. Ditchfield. Mol. Phys. 1974, 27, 789–807. doi:10.1080/ 00268977400100711. [25] K. Wolinski, J. F. Hilton, P. Pulay. J. Am. Chem. Soc. 1990, 112, 8251–60. doi:10.1021/ja00179a005. [26] J. R. Cheeseman, G. W. Trucks, T. A. Keith, M. J. Frisch. J. Chem. Phys. 1996, 104, 5497–509. doi:10.1063/1.471789. [27] M. Baias, J-N. Dumez, P. H. Svensson, S. Schantz, G. M. Day, L. Emsley. J. Am. Chem. Soc. 2013, 135, 17501–17507. DOI: 10.1021/ja4088874 [28] M. J. Bick, J. J. Banik, S. A. Darst, S. F. Brady. Biochemistry 2010, 49(19), 4159–4168. doi:10.1021/bi100150v.
835
Magn. Reson. Chem. 2015, 53, 829–835
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc
