Jagmohan S. Saini, Nadine Homeyer, Simone Fulle and Holger … · 2018-11-21 · Jagmohan S. Saini,...

DOI 10.1515/hsz-2013-0188 Biol. Chem. 2013; 394(11): 1529–1541

Jagmohan S. Saini, Nadine Homeyer, Simone Fullea and Holger Gohlke*

Determinants of the species selectivity of oxazolidinone antibiotics targeting the large ribosomal subunit

Abstract: Oxazolidinone antibiotics bind to the highly conserved peptidyl transferase center in the ribosome. For developing selective antibiotics, a profound under-standing of the selectivity determinants is required. We have performed for the first time technically challenging molecular dynamics simulations in combination with molecular mechanics Poisson-Boltzmann surface area (MM-PBSA) free energy calculations of the oxazolidinones linezolid and radezolid bound to the large ribosomal sub-units of the eubacterium Deinococcus radiodurans and the archaeon Haloarcula marismortui. A remarkably good agreement of the computed relative binding free energy with selectivity data available from experiment for lin-ezolid is found. On an atomic level, the analyses reveal an intricate interplay of structural, energetic, and dynamic determinants of the species selectivity of oxazolidinone antibiotics: A structural decomposition of free energy components identifies influences that originate from first and second shell nucleotides of the binding sites and lead to (opposing) contributions from interaction ener-gies, solvation, and entropic factors. These findings add another layer of complexity to the current knowledge on structure-activity relationships of oxazolidinones bind-ing to the ribosome and suggest that selectivity analyses solely based on structural information and qualitative arguments on interactions may not reach far enough. The computational analyses presented here should be of suf-ficient accuracy to fill this gap.

Keywords: entropy; eubacterium; free energy; linezolid; MM-PBSA; molecular dynamics simulations.

aPresent address: BioMed X GmbH, Im Neuenheimer Feld 583, D-69120 Heidelberg, Germany.*Corresponding author: Holger Gohlke, Institute for Pharmaceutical and Medicinal Chemistry, Department of Mathematics and Natural Sciences, Heinrich Heine University, Universitätsstr. 1, D-40225 Düsseldorf, Germany, e-mail: [email protected] S. Saini, Nadine Homeyer and Simone Fulle: Institute for Pharmaceutical and Medicinal Chemistry, Department of Mathematics and Natural Sciences, Heinrich Heine University, Universitätsstr. 1, D-40225 Düsseldorf, Germany

IntroductionRibosomes are complex nanomachines that perform protein synthesis in all living cells. This pivotal role makes the bacterial ribosome a prominent target for antibiotics that exert their antimicrobial effect by interfering with protein synthesis (David-Eden et al., 2010). Several crystal structures of ribosomal subunits bound to different classes of antibiotics have been published over the last decade revealing the binding site and binding mode in atomic detail as well as the structural basis for antibiotic specific-ity (Brodersen et al., 2000; Carter et al., 2000; Bottger et al., 2001; Schlunzen et al., 2001; Hansen et al., 2003; Yonath, 2005a; Ippolito et al., 2008; Wilson et al., 2008; Belousoff et al., 2011). These structural insights are complemented by several computational studies addressing the dynamic and energetic determinants of antibiotics binding to the ribosome (Ma et al., 2002; Trylska et al., 2005; Vaiana et al., 2006; Aleksandrov and Simonson, 2008; Ge and Roux, 2009, 2010; David-Eden et al., 2010; Romanowska et al., 2011). Overall, these studies have provided crucial insights into the mode of action of several classes of antibiotics that should facilitate efforts to design new antibiotics in order to combat the emerging rise in infections due to antibiotic-resistant bacteria and a related mortality rate.

Oxazolidinones represent one of only two new chemi-cal classes of antibiotics that have been introduced in the clinics over the past 40 years. Linezolid, the only member of this class approved by the FDA, shows excellent activity against major Gram-positive bacteria and is very effective in the treatment of infections of the respiratory tract and skin disorders (Moellering, 2003; Leach et al., 2011). The co-crys-tal structures of linezolid with the large ribosomal subunit (50S) of Deinococcus radiodurans (D50S), a eubacterium, (Wilson et al., 2008) and Haloarcula marismortui (H50S), an archaeon, (Ippolito et al., 2008) show that the antibi-otic exerts its action by binding to the A-site of the peptidyl transferase center (PTC) and, thereby, hinders the proper placement of the incoming aminoacyl-tRNA (Figure 1A). Archaeal ribosomes are considered more ‘eukaryotic-like’ with respect to their antibiotic specificity, i.e., they possess typical eukaryotic elements at the principal antibiotic

Bereitgestellt von | Heinrich Heine Universität DüsseldorfAngemeldet | 134.99.237.121

Heruntergeladen am | 14.10.13 10:18

1530 J.S. Saini et al.: Species selectivity of oxazolidinone antibiotics

target sites and require much higher than clinically rel-evant antibiotic concentrations for binding (Mankin and Garrett, 1991; Sanz et al., 1993; Hansen et al., 2002; Hansen et al., 2003; Yonath, 2005b). This holds true also for the oxazolidinone class of antibiotics: for co-crystallizing line-zolid with the eubacterial D50S, a concentration of 5 μm was required, while a 1000-fold higher concentration was required for co-crystallization with the archaeal H50S (Wilson et al., 2008; Wilson, 2011). Furthermore, in a func-tional assay using ribosomes isolated from Staphylococcus aureus, a eubacterium, a translation-inhibitory activity of IC50 = 0.9 μm was measured for linezolid (Skripkin et al., 2008), whereas in the case of H. marismortui, an IC50 of 4.96 μm was found for linezolid (E. M. Duffy, personal com-munication). This also conveys a selectivity of linezolid for the eubacterial ribosome. To the best of our knowledge, no corresponding value for D. radiodurans has been reported in the literature.

The co-crystal structures of linezolid with D50S and H50S provided first insights into the structural basis for the species selectivity of the oxazolidinone family (Figure 1B). As such, U2585 (Escherichia coli numbering) forms a hydrogen bond with the morpholino ring of line-zolid in the D50S (Wilson et al., 2008) but not in the H50S structure (Ippolito et al., 2008). In contrast, the phosphate group of G2505 forms a hydrogen bond with the acetamide NH group of linezolid in the H50S but not in the D50S structure. Otherwise, the overall position of linezolid is similar in both species in terms of ring orientations and interactions (Figure 1B). Obviously, the origin for the dif-ference in the binding affinity of linezolid toward D50S or H50S cannot be deduced from structural data alone.

In order to provide a better understanding of the selec-tivity determinants of the oxazolidinone class of antibiotics, we have performed for the first time molecular dynamics (MD) simulations in combination with molecular mechan-ics Poisson-Boltzmann surface area (MM-PBSA) free energy calculations of linezolid bound to D50S and H50S. Further-more, we investigated the oxazolidinones radezolid and rivaroxaban (Figure 1C). Radezolid is a promising member of the oxazolidinone class of antibiotics, which has completed phase II of clinical trials [(Shaw and Barbachyn, 2011); http://www.rib-x.com, access date: 11th May 2013], requires a 100-times lower concentration than linezolid to inhibit protein synthesis in eubacterial ribosomes (Skripkin et al., 2008), and shows an improved pattern of selectivity to bacterial ribosomes (Zhou et al., 2008a). In contrast, rivaroxaban, an oral anticoagulant, does not bind to the ribosome, although it is structurally related to linezolid and radezolid [ChEMBL (Gaulton et al., 2011); access date: 11th May 2013]. Hence, rivaroxaban was used as a negative control in the course of this study. Overall, our analyses reveal an intricate interplay of structural, energetic, and dynamic determinants of the species selectivity of oxazolidinone antibiotics.

Results

Overall stability of the oxazolidinone-50S complexes

We studied the differences in oxazolidinone binding to D50S and H50S by all-atom explicit solvent MD

Figure 1 Binding site of oxazolidinone antibiotics in the 50S ribosomal subunit and investigated oxazolidinone derivatives. (A) Global view of the 50S ribosomal subunit showing the location of the binding site of oxazolidinones within the 23S RNA (red). The ribosomal protein chains and the 5S RNA are shown in gray. (B) Close-up view of the binding site with a superposition of binding site nucleotides and linezolid in the H50S (black; PDB code: 3CPW) and D50S (blue; PDB code: 3DLL) structures. Depicted are nucleotides that are part of the first shell of nucleotides around linezolid. Hydrogen bonds are marked by dotted lines: In the H50S structure, a hydrogen bond is formed between the phosphate group of G2505 and the acetamide NH group of linezolid; in the D50S structure, a hydrogen bond is formed between U2585 and the morpholino ring of linezolid. (C) Chemical structures of oxazolidinones investigated in this study: linezolid, radezolid, and rivaroxaban.



http://www.rib-x.com

http://www.rib-x.com

J.S. Saini et al.: Species selectivity of oxazolidinone antibiotics 1531

simulations of the respective complexes of 50 ns length each. This leads to a total of 300 ns of simulation time for systems of the size of ∼8.5 × 105 atoms. For investigat-ing structural deviations along the trajectories, the overall root mean-square deviation (RMSD) as well as the RMSD of the ‘core’ residues were computed with respect to the starting structures. (The term ‘residue’ is used here for both amino acids and nucleotides.) The ‘core’ residues were selected by regarding only those residues with the 90% lowest root mean-square fluctuations (RMSF) of Cα and phosphate atoms. Both the overall RMSD (Figure S1) and the RMSD of the ‘core’ (Figure 2A) stabilized over the initial 30 ns of simulation time: the RMSD values of the ‘core’ residues reach a plateau of 2–3 Å (4–5 Å) for the linezolid/radezolid-H50S (D50S) complexes (Figure 2A). These deviations are comparable to those found in related simulations (Chen and Lin, 2010; Ge and Roux, 2010) and suggest that for the ‘core’ of the ribosome, stable trajec-tories can be obtained already after a few tens of nano-seconds. The results of the RMSD analyses also point out that the nucleotides surrounding the binding site in the oxazolidinone-50S complexes do not undergo major struc-tural rearrangements. Consequently, the last 20 ns of the MD trajectories were used for all subsequent structural analyses. However, in the energetic analysis, the effective binding energies showed considerably higher fluctuations in the 30- to 40-ns range than in the 40- to 50-ns range (Figure S2). Hence, for the energetic analyses, only the last 10 ns were used. This ensured rather stable effective binding energies computed by the MM-PBSA approach

as revealed from the time series of these values (Figure 2B). The average drifts in the effective binding energies, as determined by the slopes of the linear regression lines, are -0.19 kcal/(mol*ns) [0.50 kcal/(mol*ns)] for linezolid (radezolid) in D50S. While the magnitude of these drifts is comparable to those found for ligands binding to proteins (Metz et al., 2012), the changes in the effective binding energies over the last 10 ns associated with it point at the difficulty of obtaining converged estimates of effective binding energies for the investigated systems.

Structural analysis of the oxazolidinone-50S complexes

The crystal structures of linezolid-D50S and linezolid-H50S complexes both show that linezolid binds to the PTC with the acetamide end being located near the mouth of the ribosomal exit tunnel, the oxazolidinone ring making stacking interactions with U2504, and the morpholino ring approaching U2585 (Figure 1B) (Ippolito et al., 2008; Wilson et al., 2008). Despite their similar overall binding modes, the crystal structures also reveal differences (Figure 1B) (Auerbach et al., 2004; Yonath and Bashan, 2004). These include a rotation around the fluorophenyl ring relative to the oxazolidinone ring and different con-formations of U2506 and U2585.

In order to detect whether such differences in the binding modes also lead to differences in specific inter-actions, we analyzed the network of hydrogen bonds of

8

7

6

5

4

RM

SD

(Å

)

20

15

10

-10

5

-5

-20

-15

0

∆Gef

fect

ive

(kca

l/mol

)

3

2

1

0 5 10 15 20 25 30 35

Linezolid-H50SRadezolid-H50SLinezolid-D50S

A B

Radezolid-D50S

40 40 41 42 43 44 45 46 47 48 49 5050

Time (ns) Time (ns)

45

Figure 2 Time course of root mean square deviations and effective binding energies of oxazolidinone-50S complexes.(A) Root mean-square deviations of Cα and phosphate atoms of the ‘core residues’ along the MD trajectories of oxazolidinone-50S complex structures determined with respect to the starting structure of the production run. The ‘core residues’ were defined to be those residues with the 90% lowest RMSF of the Cα and phosphate atoms. RMSD values for linezolid and radezolid in H50S are depicted in black and red, and for linezolid and radezolid in D50S in blue and green, respectively. (B) Time series of effective binding energies calculated for 500 snapshots extracted in 20-ps intervals from the last 10 ns of MD simulations of linezolid-D50S (blue) and radezolid-D50S complexes (green). The drifts in the effective binding energies, determined from the slopes of the linear regression lines, are -0.19 kcal/(mol*ns) and 0.50 kcal/(mol*ns), respectively.




the antibiotics inside the binding pockets along the MD trajectories. In the linezolid-H50S crystal structure, the acetamide NH of linezolid forms a hydrogen bond with the phosphate group of G2505 (Ippolito et al., 2008) (Figure 1B). In contrast, the crystal structure of linezolid bound to D50S shows the involvement of the oxygen atom of the morpholino ring in a hydrogen bond interaction with N3 of U2585 (Wilson et al., 2008) (Figure 1B). Figure 3 depicts the shortest distances between these respective atoms (and those of the triazolyl-methylamino moiety of radezolid and the carbonyl oxygens’ of U2585 in the case of radezolid) along the MD simulations of the H50S and D50S complexes with linezolid and radezolid, respec-tively. A distance < 3.2 Å between the acceptor and donor atoms and an angle > 120° for the coordinate triple (accep-tor atom, hydrogen, donor atom) are used as criteria for the existence of a hydrogen bond (Gohlke et al., 2003).

In the case of radezolid-H50S, the hydrogen bond between the acetamide NH of the ligand and the phos-phate group of G2505 is present ∼90% of the time (Figure 3C), whereas the respective hydrogen bond is

disrupted throughout the whole MD simulation in the lin-ezolid-H50S case (Figure 3A). Instead, in the latter case, the ligands’ acetamide NH group forms a strong hydro-gen bond with the sugar part of U2504 (occupancy: 75%) (Figure 3A). No hydrogen bond formation is observed for the morpholino and triazolyl-methylamino moieties, however, in both H50S simulations.

In the case of D50S, no hydrogen bond is found involv-ing the acetamide NH of linezolid or radezolid, respectively, and the phosphate group of G2505 (Figure 3B, D). However, in the linezolid-D50S case, the acetamide NH forms a weak interaction with O6 of A2061 (occupancy: 20%) (Figure 3B). Furthermore, a stable interaction between linezolid’s oxygen of the oxazolidinone core and the NH group of the base of U2504 is observed (distance between O and N: 3.08 ± 0.09 Å); this interaction does not qualify as a hydro-gen bond according to our angle criterion, however (occu-pancy: 0%). No hydrogen bond involving the morpholino moiety of linezolid is found. In contrast, in the case of rad-ezolid-D50S, hydrogen bond interactions of the triazolyl-methylamino moiety occur at ∼34 ns and around 37 ns.

Line

zolid

dis

tanc

e (Å

)R

adez

olid

dis

tanc

e (Å

)

18

16

14

12

10

8

6

4

2

18

16

14

12

10

8

6

4

230 35

A B

C D

40 50Time (ns)

45 30 35 40 50Time (ns)

45

30 35 40 5045 30 35 40 5045

H50S D50S

G2505

U2504U2585

G2505

A2061U2585

Figure 3 Distances monitoring hydrogen bond formation for (A) linezolid-H50S, (B) linezolid-D50S, (C) radezolid-H50S, and (D) radezolid-D50S complex simulations.Hydrogen bonds were defined by acceptor donor atom distances of < 3.2 Å and acceptor H-donor angles of > 120°. The distance between the ligands’ acetamide NH group and the oxygens’ of the phosphate group of G2505 is shown in black; the distance between the oxygen of the morpholino ring of linezolid and N3 of U2585 or any donor group of the triazolyl-methylamino moiety of radezolid and the carbonyl oxygens’ of U2585 are shown in red. Only the smallest distance found is plotted in all cases. In addition, the distance between linezolid’s acetamide NH group and O2′ of U2504 is shown for H50S (A, cyan) as is the distance between linezolid’s acetamide NH group and O6 of A2061 for D50S (B, blue).




Next, we analyzed interactions between the oxazolidi-none ring and the nucleobase of U2504 as well as between the fluoro-phenyl ring and the nucleobases of A2451 and C2452, which form the so-called A-site cleft, by measuring the distances between the centers of the respective rings (Figure 4). These ring/nucleobase pairs have been identi-fied in the crystal structures of linezolid-D50S and line-zolid-H50S as making important contacts. We used a ring center-to-ring center distance < 5.0 Å as a criterion to iden-tify such contacts, which allows identifying T-shaped or parallel-displaced ring configurations in addition to per-fectly stacked ones (Arunan and Gutowsky, 1993; Ippolito et al., 2008). The results show that linezolid engages in all such ring/nucleobase contacts (Figure 4A, B, D, E, F) except with A2451 when bound to H50S (Figure 4C).

Radezolid forms all contacts (Figure 4A, B, C, D, E, F). It is obvious, however, that – when established – contacts between the fluoro-phenyl ring and nucleobases of A2451 or C2452 of linezolid and radezolid have a longer distance in the D50S case (Figure 4D, F) than in the H50S case (Figure 4C, E). This suggests that weaker ring/nucleobase interactions are formed in the former case.

As the A-site cleft is wider (Blaha et al., 2008) and more rigid (Fulle and Gohlke, 2009) in apo D50S than in apo H50S, the dynamic behavior of the oxazolidinones inside the binding site could also differ between these eubacterial and archaeal ribosomes. We, thus, analyzed the RMSF of the bound oxazolidinones (Figure S3). The RMSF values demonstrate that radezolid shows a higher mobility in the acetamide regions than linezolid (in D50S

U25

04 d

ista

nce

(Å)

A24

51 d

ista

nce

(Å)

C24

52 d

ista

nce

(Å)

10

9

8

7

6

5

4

3

10

9

8

7

6

5

4

3

10

9

8

7

6

5

4

330 35

A B

C D

E F

40 50Time (ns)

45 30 35 40 50Time (ns)

45

30 35 40 5045 30 35 40 5045

30 35 40 5045 30 35 40 5045

LinezolidRadezolid

H50S D50S

Figure 4 Distances between the centers of mass of the oxazolidinone ring and the nucleobase of U2504 (A, B) as well as between the centers of mass of the fluoro-phenyl ring and the nucleobases of A2451 (C, D) and C2452 (E, F), respectively.(A), (C), (E) Distances for oxazolidinone-H50S complex simulations. (B), (D), (F) Distances for oxazolidinone-D50S complex simulations. Black curves are for linezolid, red curves are for radezolid.




and H50S) and that radezolid’s mobility, itself, is higher when bound to D50S than when bound to H50S. The latter is also confirmed when analyzing structural varia-tions of radezolid over the course of the MD trajectories (Figure S4). The difference of RMSF values in the case of linezolid is less pronounced in the acetamide region. In the region of the oxazolidinone core, the mobility is gener-ally low. In contrast, in the region of the morpholino and triazolyl-methylamino moieties, a generally high mobility is observed.

Finally, as the negatively charged ribonucleic acids need Mg2+ ions to maintain their structural stability (Ge and Roux, 2009), a correct description of the properties and behavior of the Mg2+ ions in the simulations is essen-tial. The RMSF of the Mg2+ ions over the course of the MD simulations reveal that all Mg2+ ions remain fixed to their starting positions (data not shown). Thus, it seemed rea-sonable to consider these ions as part of the receptor in the subsequent effective binding energy calculations. For the D50S complexes, the simulations also showed that one of the Mg2+ ions remains in close proximity to the bound oxazolidiones (Figure S5), as also observed in the crystal structure (Wilson et al., 2008).

Energetic analysis of the oxazolidinone-50S complexes

In order to obtain insights into the energetic determinants of the selectivity of linezolid and radezolid with respect to D50S and H50S, effective binding energies (ΔGeffective), i.e., the sum of gas-phase energies and solvation free energies, were computed by the MM-PBSA method using the single-trajectory approach (Table 1; Table S1; Eqs. 1, 2). Although the alternative three-trajectory MM-PBSA approach (Homeyer and Gohlke, 2012) has the advantage that changes in the receptor and ligand conformation upon complex formation are taken into account, the computa-tionally less demanding single-trajectory approach has proven to be a good approximation in several ligand-bind-ing energy studies (Huo et al., 2002; Hou et al., 2011; Yang et al., 2011). Here, ΔGeffective was computed by averaging over 500 snapshots extracted at 20-ps intervals from the last 10 ns of the MD simulations. The associated standard error in the mean determined according to Gohlke et al. (2003) is ∼0.25 kcal/mol (Table S1). The computed ΔGeffective of linezolid (-6.73 kcal/mol; Table S1) and radezolid (-0.05 kcal/mol; Table S1) bound to D50S are (weakly) negative, indicating that the sum of gas-phase and solvation free energies favor binding of these antibiotics. In contrast, for rivaroxaban, a considerably positive ΔGeffective was obtained

Table 1 Differences in the energy and entropy components for oxazolidinone binding to D50S and H50S.a

Contributionb Linezolid (D50S-H50S)

Radezolid (D50S-H50S)

Meanc σd Meanc σd

ΔΔHelec -4.52 0.04 -8.61 0.10ΔΔHvdW 1.12 0.19 7.05 0.26ΔΔHgas -3.39 0.21 -1.56 0.30ΔΔGPB -3.64 0.20 21.82 0.25ΔΔGnonpolar 0.07 0.09 0.48 0.09ΔΔGeffective -6.93 0.26 20.73 0.31TΔSR, T, V, ligand in complex 3.66 –e 3.15 –e

TΔSV, receptor in complex -4.46 –e 7.29 –e

TΔΔStot -0.80 –e 10.44 –e

ΔΔG -6.16 –e 10.29 –e

aGas phase and solvation energy contributions were determined by the MM-PBSA approach, and entropy contributions were calculated by quasiharmonic analysis (QHA) considering 500 snapshots from the last 10 ns of MD simulations of oxazolidione-50S complexes; T = 300 K.bHelec, electrostatic energy; HvdW, van der Waals energy; Hgas, gas phase energy; GPB, polar solvation free energy; Gnonpolar, nonpolar solvation free energy; Geffective, effective energy; SR, T, V, ligand in complex, translational, rotational, and vibrational entropy of the ligand in the complex; SV, receptor in complex, vibrational entropy of the receptor in the complex.cAverage contributions in kcal/mol.dStandard error in the mean values in kcal/mol.eNo values available.

(6.96 kcal/mol; Table S1), which is in line with experimen-tal findings that rivaroxaban does not bind to the ribosome [ChEMBL (Gaulton et al., 2011); access date: 11th May 2013]. The analysis of the MD simulation of the rivaroxaban-D50S complex revealed that due to a displacement from its initial position, rivaroxaban loses important interactions inside the binding site that are present in the linezolid-D50S complex crystal structure (data not shown).

The difference in the effective binding energy (ΔΔGeffective) for linezolid bound to D50S vs. H50S is -6.93 ± 0.26 kcal/mol (Table 1), thus, demonstrating that binding of the antibiotic to D50S is favored. At T = 300 K, this difference is equivalent to a 105-fold larger association constant for linezolid-D50S than for linezolid-H50S. This result agrees well with experiment according to which a 103-fold higher concentration of the antibiotic is required for a successful co-crystallization in H50S compared to D50S (Wilson, 2011). It is also in line with results from functional assays on S. aureus and H. marismortui ribo-somes, where a selectivity of linezolid toward the eubacte-rical ribosome is found (Skripkin et al., 2008) (E. M. Duffy, personal communication). The decomposition of ΔΔGeffective




with respect to the energy terms contributing to it reveals that the electrostatic (ΔΔHelec = -4.52 kcal/mol) and polar solvation free energy (ΔΔGPB = -3.64 kcal/mol) contributions favor binding to D50S over H50S, whereas van der Waals interactions (ΔΔHvdW = 1.12 kcal/mol) disfavor it.

In contrast, for radezolid, ΔΔGeffective = 20.73 kcal/mol was computed (Table 1), which would suggest that, based only on the sum of gas phase and solvation free ener-gies, binding to H50S is favored over D50S. This result is at variance with the experiment: while to the best of our knowledge no specific experimental data of radezolid binding to D50S or H50S has been reported, it has been stated that radezolid is > 100-fold selective for bacterial ribosomes over those of rabbit reticulocyte (Zhou et al., 2008a). Regarding the decomposition of ΔΔGeffective with respect to the energy terms contributing to it, we find that the polar solvation free energy (ΔΔGPB = 21.82 kcal/mol) highly favors binding to H50S compared to D50S. The van der Waals interactions also considerably contribute to the more favorable effective binding energy found for H50S (ΔΔHvdW = 7.05 kcal/mol), whereas the electrostatic energy (ΔΔHelec = -8.61 kcal/mol), as in the case of linezolid, shows the opposite trend, i.e., is more favorable in the D50S case.

While the continuum solvation model provides estimates of the free energy of solvation and includes entropic contributions of the solvent, changes in the entropy of the solute molecules upon binding have been neglected so far. However, such changes can signifi-cantly influence the binding affinity (Brady and Sharp, 1997; Singh and Warshel, 2010; Hou et al., 2011). Hence, we estimated differences in the changes in the con-figurational entropy (TΔΔS at T = 300 K) upon binding of linezolid (or radezolid) to D50S and H50S based on motions of the ligands and the nucleotides forming the first shell of the binding site by quasiharmonic analy-sis (QHA) (Table 1; Tables S1 and S2; Eqs. S2a–d; Figure S6). The librational/vibrational entropy of the ligand inside the binding site (TSR,T,V, ligand in complex) was found to be higher in the D50S complex than in the H50S complex by 3–4 kcal/mol for both ligands (Table 1). In contrast, ligand binding has converse effects on the vibrational entropy of the first shell nucleotides (TSV, receptor in complex): linezolid (rade zolid) binding leads to ∼4 kcal/mol lower (∼7 kcal/mol higher) values in the case of D50S than H50S. Taken together, the difference in the changes in the configu-rational entropy is close to zero for linezolid binding to D50S vs. H50S; in contrast, the changes in the configura-tional entropy favor radezolid binding to D50S over H50S by ∼10 kcal/mol (Table 1).

To gain insights into the origin of the selectivity of oxa-zolidinone binding on an atomic level, we decomposed

the effective binding energy in terms of contributions from structural subunits of D50S and H50S. Such structural decompositions have been successfully applied before in the context of protein-protein interactions (Gohlke et al., 2003; Wichmann et al., 2010; Metz et al., 2012) and protein-peptide interactions (Fulle et al., 2013) by us. The standard error in the mean ΔGeffective value for one residue is assumed to be of the same order of magnitude than the one for the overall effective binding energy (∼0.25 kcal/mol). Note that the mean ΔGeffective values do not include contributions due to changes in the configurational entropy of the receptor residues upon ligand binding. Such contributions could be of significant influence, however, according to the TSV, receptor in complex computations above. As expected, the structural decomposition reveals that first shell nucleotides contribute most to the effec-tive binding energy (between -13.06 and -10.03 kcal/mol) (Figure 5; Tables S3, S4). Still, second shell nucleotides show contributions of up to 1.76 kcal/mol, i.e., 15%. As a single-trajectory alternative is used for the computations of effective binding energies, long-range electrostatic influences are most likely responsible for the contribution by the second shell nucleotides (Selzer et al., 2000).

Regarding linezolid, the most pronounced difference is observed for A2451 in the first shell (Tables S3–S5): this nucleotide contributes strongly (-3.61 kcal/mol) to the binding of the antibiotic in D50S (Table S3) but shows an almost neutral influence (-0.12 kcal/mol) in the case of H50S (Table S4). The difference in the contribution of the highly favorable C2452-ligand interaction is not as pronounced (-4.43 kcal/mol in D50S vs. -5.54 kcal/mol in H50S), but amounts up to ∼1 kcal/mol in favor of H50S. Notably, also small differences in the effective binding energies of second shell nucleotides are observed: nucleo-tides 2055 (2572) contribute favorably (neutrally) to binding in the case of D50S [-0.40 kcal/mol (0.15 kcal/mol)], whereas they disfavor binding to H50S [0.29 kcal/mol (0.80 kcal/mol)].

As to radezolid, again A2451 and C2452 show signifi-cant and opposing differences in their contributions to the effective binding energy (Table S5): both nucleotides contribute highly favorably to binding in D50S and H50S, but A2451 shows a higher effective binding energy in D50S (-4.16 kcal/mol) than in H50S (-3.21 kcal/mol), whereas C2452 exhibits a lower effective binding energy in D50S (-2.21 kcal/mol) than in H50S (-3.44 kcal/mol). For U2504, which shows the second largest difference, the same trend as for A2451 is observed: this nucleotide’s contribution is more favorable in the D50S case (-3.54 kcal/mol) than in the H50S case (-2.22 kcal/mol). Finally, G2505 and U2506 contribute to the effective binding energy by




Figure 5 Per-residue contributions to the effective binding energy as calculated by a structural decomposition via MM-PBSA using ensem-bles from MD simulations of (A) linezolid-H50S, (B) linezolid-D50S, (C) radezolid-H50S, and (D) radezolid-D50S complex structures.Each sphere represents the center of mass of the nucleobase of the respective nucleotide; nucleotides of the first shell are labeled with straight letters, those of the second shell in italics. Ligands are shown as sticks. The per-residue contributions are mapped using a color code with a linear scale (red: ≤ -3 kcal/mol; white: 0 kcal/mol; blue: ≥ +3 kcal/mol). Hydrogen bonds observed during the MD simulations between oxazolidinones and nucleotides (Figure 3) are indicated by dotted green lines; ring/nucleobase contacts observed during the MD simulations (Figure 4) are shown as dotted black lines.

-1.57 and -2.09 kcal/mol in H50S, whereas their contribu-tions remain almost neutral in the case of D50S. As to the contributions from second shell nucleotides, again dif-ferences are found between D50S and H50S: nucleotides 2055 and 2572 both disfavor binding to H50S, whereas their contribution to binding in D50S is neutral.

In summary, the energetic analyses of the oxazolid-ione-50S complexes reveal significant differences in the effective binding energies, also in terms of structural

contributions, and configurational entropies with respect to the binding of linezolid or radezolid to D50S and H50S.

DiscussionThe highly conserved functional sites in the ribo-some make the task of developing selective antibiotics




challenging. In the present study, we aimed at identifying on an atomic level the determinants of selectivity of lin-ezolid and radezolid, binding to the eubacterial D50S and the archaeal H50S subunits, by means of MD simulations and free energy computations. Although structure-activ-ity relationships of oxazolidinones have been extensively reported (Gregory et al., 1989; Gregory et al., 1990; Park et al., 1992; Brickner et al., 1996; Barbachyn and Ford, 2003; Brickner et al., 2008; Zhou et al., 2008a,b; Locke et al., 2010), none of these studies has considered aspects of structure, energetics, and dynamics to selectivity simul-taneously. Furthermore, to the best of our knowledge, this is the first study to investigate the selectivity of oxazolidi-none binding to 50S subunits by MD simulations and free energy calculations. Preliminary results addressing this question have been reported by Franceschi et al. as a con-ference contribution using MCPRO (MCPRO) for a residue-by-residue analysis of interaction energies (Franceschi et al., 2010).

From a global point of view, it is remarkable that the MM-PBSA analyses including estimates of changes in the configurational entropy agree well with results from experiment on the selectivity of linezolid. These computa-tions furthermore reveal that linezolid’s selectivity toward D50S over H50S is favored by the effective energy, whereas there is only a vanishing contribution by the configura-tional entropy (Table 1). A converse picture emerges for the global selectivity determinants of radezolid: the effec-tive energy strongly favors binding to H50S over D50S now, but differences in the changes of the configurational entropy oppose this effect (Table 1). Still, in sum, binding of radezolid to H50S remains favorable. Note, however, that the contributions of the configurational entropy only consider the effects of the first shell of binding site nucleotides due to the lack of convergence when includ-ing second (and, likely, higher) shell nucleotides (see Materials and methods, and Figure S6) such that entropy contributions from residues further away are neglected. Furthermore, when extending the computation of con-tributions of the configurational entropy to the last 20 ns of the MD trajectories, the differences in the changes of the configurational entropy even more favor radezolid binding to D50S such that, in sum, binding of radezolid to D50S would become favorable (Table S2). The last two points call for performing longer MD simulations in related studies in the future.

These observations on the energetics on a global level are consistent with structural and mobility analyses of oxazolidinone-50S complexes from the MD trajectories, and structural analyses of apo and holo 50S crystal struc-tures. As such, a wider A-site cleft in the case of apo D50S

compared to apo H50S (Blaha et al., 2008) may contribute to the observed higher mobility of radezolid particularly in the acetamide region (Figures S3) and, consequently, to the higher vibrational/librational entropy (Table 1) when bound to D50S compared to H50S. In turn, rigidity analy-ses on 50S subunits revealed that the glycosidic bond of A2451 is flexible in H50S but rigid in D50S (Fulle and Gohlke, 2009). Accordingly, binding of the antibiotics can rigidify this crevice only in the case of H50S, even more so if strong aromatic interactions are formed between the fluoro-phenyl ring and the cleft’s nucleobases. The latter is particularly observed for radezolid binding to H50S (Table 1; Figure 5; Table S4). Thus, it does not come as a surprise that vibrational entropy contributions of first shell nucleotides of the binding sites favor radezolid binding to D50S over H50S (Table 1).

The structural analysis of the MD trajectories reveals that linezolid does not form strong hydrogen bonds between its acetamide group and the phosphate group of G2505, neither when it is bound to D50S nor to H50S. Regarding the acetamide group, this observation agrees with observations from the linezolid-D50S crystal struc-ture (Wilson et al., 2008) but disagrees with what has been found for the linezolid-H50S structure (Ippolito et al., 2008). Rather, linezolid’s acetamide groups form hydrogen bonds with the sugar part of U2504 in H50S, and with O6 of A2061 in D50S. In the latter case, an additional stable interaction between linezolid’s oxazolidinone core and the base of U2504 was observed, too.

Regarding the morpholino moiety, according to the structural analysis of the MD trajectories, linezolid neither forms hydrogen bonds when bound to D50S nor to H50S. In the linezolid-D50S crystal structure, only a relatively weak hydrogen bond has been found (Wilson et al., 2008), which stabilizes an otherwise highly flexible U2585. Thus, the observed breaking of this hydrogen bond during the MD simulations at 300 K (Figure 3A, B) is not unex-pected. Interactions of linezolid’s fluoro-phenyl ring are observed to C2452 in both the D50S and H50S simulations (Figure 4E, F), in agreement with crystal structure analy-ses (Ippolito et al., 2008; Wilson et al., 2008), with them being tighter in the H50S case. In contrast, A2451 forms additional interactions with the fluoro-phenyl ring of lin-ezolid only in D50S, which explains to a large extent why the effective binding energy of linezolid is more favorable for D50S than for H50S (Table 1).

Radezolid forms a strong hydrogen bond with its acetamide group to G2505 in H50S only. Furthermore, although it forms tighter interactions with its oxazolidi-none ring with U2504 in D50S, it does so too between its fluoro-phenyl ring and A2451 as well as C2452 in H50S.




Taken together, these differences in the interactions can explain to a large extent why the effective binding energy of radezolid is computed to be more favorable in the H50S case than in the D50S case (Table 1).

The results from the structural decomposition of the effective binding energies confirm essentially these analy-ses of the influence of interactions of first shell nucleotides of the binding sites on oxazolidinone selectivity. As an advantage, the structural decomposition provides a spa-tially resolved picture of the binding energetics (Figure 5), making the identification of selectivity ‘hot spots’ or ‘cold spots’ easier. This is particularly helpful for investigat-ing influences of the second (and higher) shell residues, where often a direct link to interactions with the ligand is not obvious. In the present case, the structural decomposi-tion reveals the most pronounced influence of second shell residues on the selectivity of oxazolidinones for nucleotides 2055 and 2572 (Figure 5; Table S5). This finding is in line with previous reports according to which both nucleotides are involved in restraining the conformational space of U2504 in eubacteria vs. archaea/eucaryotes and, hence, the selectiv-ity of antibiotics binding to the A-site cleft (Davidovich et al., 2008; Gürel et al., 2009): in general, in eubacteria, U2504 is sterically fostered by C2055 and A2572 to adopt a conforma-tion that favors interactions with antibiotics; in contrast, in archaea/eucaryotes, U2504 adopts a conformation unsuita-ble for interactions with antibiotics due to A2055 and U2572. Oxazolidinone binding to the A-site cleft is now peculiar in that U2504 in H50S adopts a eubacterial conformation. As a consequence, the induced fit should incur a cost in free energy. In line with this, the structural decomposition iden-tifies U2504 to disfavor oxazolidinone binding to H50S over D50S. Notably, the conformational strain is also relayed to nucleotides 2055 and 2572 as these disfavor oxazolidinone binding to H50S over D50S, too (Figure 5; Table S5). This finding provides direct evidence for the energetic involve-ment of second shell residues in oxazolidinone selectivity.

In summary, our analyses reveal an intricate inter-play of structural, energetic, and dynamic determinants of the species selectivity of oxazolidinone antibiotics. Even for the structurally rather similar members linezolid and radezolid investigated here, significant differences in the (opposing) contributions from interaction energies, solvation, and entropic factors have been identified, as have been influences of first and second shell nucleotides detected. These findings add another layer of complexity to the current knowledge on structure-activity relation-ships of oxazolidinones and suggest that selectivity analy-ses solely based on static crystal structures and qualitative arguments on interactions may not reach far enough. Com-putational analyses may aid in filling this gap, particularly

when their accuracy is sufficiently high, as demonstrated here for the species-selectivity of linezolid. However, such analyses are computationally highly demanding. It may, thus, be worth to invest in the development of small RNA model systems of antibiotic binding sites, like those that have been extraordinarily useful to study the A-site of bac-terial 16S rRNA in the 30S subunit (Hermann, 2006).

Materials and methods

Molecular dynamics simulationsAll MD simulations were performed using the AMBER 10 suite of programs (Case et al., 2005) together with the ff99SB force field (Cor-nell et al., 1995; Hornak et al., 2006) for proteins and RNAs, and the general amber force field (GAFF) (Wang et al., 2004) for the ligands linezolid, radezolid, and rivaroxaban. Details on how the oxazolidi-one-50S systems were set up for these MD simulations are described in the Supplemental Material as are details of the MD simulation pro-cedures for the D50S systems. The MD simulations of the H50S sys-tems have been executed analogously. A detailed description of the protocol used in these MD simulations will be published elsewhere (Saini, Fulle, Homeyer, Gohlke, unpublished results). Details of the structural analysis of the MD trajectories are given in the Supplemen-tal Material, too.

Calculation of effective binding energiesWe applied the MM-PBSA method (Homeyer and Gohlke, 2012) as implemented in the mm_pbsa.pl module of the AMBER 10 (Case et al., 2005) suite of programs. MM-PBSA estimates the free energy of a molecule x as the sum of its gas phase energy ( gas

xH ), solvation free energy ( solv

xG ), and configurational entropy (TS x) (Eq. 1).

gas solv( ) ( ) ( )- ( )x x x xG i H i G i TS i= + (1)

The effective binding energy (Gohlke and Case, 2004) was com-puted as the mean difference of the effective energies (Geffective = Hgas+Gsolv) of the complex and the receptor and ligand (Eq. 2). < · > denotes an average over snapshots i taken from MD trajectories.

complex receptor ligand

e�ective e�ective e�ective e�ective( )- ( )- ( )G G i G i G i∆ =< > (2)

Details of the calculations of effective binding energies are described in the Supplemental Material.

Structural decomposition of the effective binding energiesThe contributions to the effective binding energies calculated accord-ing to Eqs. (1 and 2) (but without considering the configurational entropy) were decomposed at the per-residue level. Residue-wise gas




phase energies were determined using the decomposition scheme implemented in the sander and mm_pbsa.pl modules of AMBER 10 (Case et al., 2005). Per-residue contributions of the polar part of the solvation free energy were computed on an atomic level as described for the total polar solvation contribution in the Supplemental Mate-rial; then, the energy contributions of all atoms belonging to a resi-due were summed up. Per-residue contributions of the nonpolar part of the solvation free energy were obtained with the ICOSA method (Case et al., 2005). Finally, the effective binding energy per residue was calculated from the gas phase and solvation free energy contri-butions using the mm_pbsa_statistics.pl script of AMBER 10 (Case et al., 2005).

Approximation of entropy componentsThe configurational entropy of a molecule consists of translational, rotational, and vibrational contributions. Obtaining accurate esti-mates of vibrational entropy contributions is difficult and compu-tationally demanding (Gohlke and Case, 2004). Determining such contributions for a system of the size of the ribosome did not seem feasible to us, neither by normal mode analysis (NMA) (Case, 1994) nor by QHA (Teeter and Case, 1990; Gohlke and Case, 2004; Baron et al., 2009). Thus, we decided to estimate changes in the entropy

due to ligand binding considering only the ligand and ribosomal residues of the ligand-binding site by QHA. We showed recently that changes of the vibrational entropy due to ligand binding originate mostly from changes in the degrees of freedom of residues to which the ligand forms direct interactions in the complex (Kopitz et al., 2012). Details on how individual entropy contributions were calcu-lated are described in the Supplemental Material.

Acknowledgements: We gratefully acknowledge sup-port (and training) from the International NRW Research School BioStruct, granted by the Ministry of Innovation, Science and Research of the State North Rhine-West-phalia, the Heinrich-Heine-University of Düsseldorf, and the Entrepreneur Foundation at the Heinrich-Heine-Uni-versity of Düsseldorf. We are grateful to the Jülich Super-computing Center (NIC project 5921) and to the ‘Zentrum für Informations- und Medientechnologie’ (ZIM) at the HHU for providing computational support.

Received May 20, 2013; accepted September 1, 2013; previously pub-lished online September 3, 2013

ReferencesAleksandrov, A. and Simonson, T. (2008). Molecular dynamics

simulations of the 30S ribosomal subunit reveal a preferred tetracycline binding site. J. Am. Chem. Soc. 130, 1114–1115.

Arunan, E. and Gutowsky, H.S. (1993). The rotational spectrum, structure and dynamics of a benzene dimer. J. Chem. Phys. 98, 4294–4296.

Auerbach, T., Bashan, A., and Yonath, A. (2004). Ribosomal antibiotics: structural basis for resistance, synergism and selectivity. Trends Biotechnol. 22, 570–576.

Barbachyn, M.R. and Ford, C.W. (2003). Oxazolidinone structure-activity relationships leading to linezolid. Angew. Chem. Int. Ed. 42, 2010–2023.

Baron, R., Hünenberger, P.H., and McCammon, J.A. (2009). Absolute single-molecule entropies from quasi-harmonic analysis of microsecond molecular dynamics: correction terms and convergence properties. J. Chem. Theory Comput. 5, 3150–3160.

Belousoff, M.J., Shapira, T., Bashan, A., Zimmerman, E., Rozenberg, H., Arakawa, K., Kinashi, H., and Yonath, A. (2011). Crystal structure of the synergistic antibiotic pair, lankamycin and lankacidin, in complex with the large ribosomal subunit. Proc. Natl. Acad. Sci. USA 108, 2717–2722.

Blaha, G., Gürel, G., Schroeder, S.J., Moore, P.B., and Steitz, T.A. (2008). Mutations outside the anisomycin-binding site can make ribosomes drug-resistant. J. Mol. Biol. 379, 505–519.

Bottger, E.C., Springer, B., Prammananan, T., Kidan, Y., and Sander, P. (2001). Structural basis for selectivity and toxicity of ribosomal antibiotics. EMBO Rep 2, 318–323.

Brady, G.P. and Sharp, K.A. (1997). Entropy in protein folding and in protein-protein interactions. Curr. Opin. Struct. Biol. 7, 215–221.

Brickner, S.J., Hutchinson, D.K., Barbachyn, M.R., Manninen, P.R., Ulanowicz, D.A., Garmon, S.A., Grega, K.C., Hendges, S.K., Toops, D.S., Ford, C.W., et al. (1996). Synthesis and antibacterial activity of U-100592 and U-100766, two oxazolidinone antibacterial agents for the potential treatment of multidrug-resistant gram-positive bacterial infections. J. Med. Chem. 39, 673–679.

Brickner, S.J., Barbachyn, M.R., Hutchinson, D.K., and Manninen, P.R. (2008). Linezolid (ZYVOX), the first member of a completely new class of antibacterial agents for treatment of serious gram-positive infections. J. Med. Chem. 51, 1981–1990.

Brodersen, D.E., Clemons, W.M. Jr., Carter, A.P., Morgan-Warren, R.J., Wimberly, B.T., and Ramakrishnan, V. (2000). The structural basis for the action of the antibiotics tetracycline, pactamycin, and hygromycin B on the 30S ribosomal subunit. Cell 103, 1143–1154.

Carter, A.P., Clemons, W.M., Brodersen, D.E., Morgan-Warren, R.J., Wimberly, B.T., and Ramakrishnan, V. (2000). Functional insights from the structure of the 30S ribosomal subunit and its interactions with antibiotics. Nature 407, 340–348.

Case, D.A. (1994). Normal mode analysis of protein dynamics. Curr. Opin. Struct. Biol. 4, 285–290.

Case, D.A., Cheatham, T.E. 3rd, Darden, T., Gohlke, H., Luo, R., Merz, K.M. Jr., Onufriev, A., Simmerling, C., Wang, B., and Woods, R.J. (2005). The Amber biomolecular simulation programs. J. Comput. Chem. 26, 1668–1688.

Chen, S.Y. and Lin, T.H. (2010). A molecular dynamics study on binding recognition between several 4,5 and 4,6-linked aminoglycosides with A-site RNA. J. Mol. Recognit. 23, 423–434.




Cornell, W.D., Cieplak, P., Bayly, C.I., Gould, I.R., Merz, K.M., Ferguson, D.M., Spellmeyer, D.C., Fox, T., Caldwell, J.W., and Kollman, P.A. (1995). A second generation force field for the simulation of proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc. 117, 5179–5197.

David-Eden, H., Mankin, A.S., and Mandel-Gutfreund, Y. (2010). Structural signatures of antibiotic binding sites on the ribosome. Nucleic Acids Res. 38, 5982–5994.

Davidovich, C., Bashan, A., and Yonath, A. (2008). Structural basis for cross-resistance to ribosomal PTC antibiotics. Proc. Natl. Acad. Sci. USA 105, 20665–20670.

Franceschi, F., DeVito, J., Dalton, J., Remy, J., Skripkin, E., Ippolito, J.A., and Duffy, E. (2010). Radezolid Overcomes CFR-Mediated Linezolid-Resistance by Efficiently Inhibiting Protein Synthesis of CFR-Methylated Ribosomes. Paper presented at: Interscience Conference on Antimicrobial Agents and Chemotherapy (Boston, Massachusetts, USA).

Fulle, S. and Gohlke, H. (2009). Statics of the ribosomal exit tunnel: implications for cotranslational peptide folding, elongation regulation, and antibiotics binding. J. Mol. Biol. 387, 502–517.

Fulle, S., Withers-Martinez, C., Blackman, M.J., Morris, G.M., and Finn, P.W. (2013). Molecular determinants of binding to the plasmodium subtilisin-like protease 1. J. Chem. Inf. Model 53, 573–583.

Gaulton, A., Bellis, L.J., Bento, A.P., Chambers, J., Davies, M., Hersey, A., Light, Y., McGlinchey, S., Michalovich, D., Al-Lazikani, B., et al. (2011). ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res. 40, D1100–D1107.

Ge, X. and Roux, B. (2009). Calculation of the standard binding free energy of sparsomycin to the ribosomal peptidyl-transferase P-site using molecular dynamics simulations with restraining potentials. J. Mol. Recognit. 23, 128–141.

Ge, X. and Roux, B. (2010). Absolute binding free energy calculations of sparsomycin analogs to the bacterial ribosome. J. Phys. Chem. B 114, 9525–9539.

Gohlke, H. and Case, D.A. (2004). Converging free energy estimates: MM-PB(GB)SA studies on the protein-protein complex Ras-Raf. J. Comput. Chem. 25, 238–250.

Gohlke, H., Kiel, C., and Case, D.A. (2003). Insights into protein-protein binding by binding free energy calculation and free energy decomposition for the Ras-Raf and Ras-RalGDS complexes. J. Mol. Biol. 330, 891–913.

Gregory, W.A., Brittelli, D.R., Wang, C.L., Wuonola, M.A., McRipley, R.J., Eustice, D.C., Eberly, V.S., Bartholomew, P.T., Slee, A.M., and Forbes, M. (1989). Antibacterials. Synthesis and structure-activity studies of 3-aryl-2-oxooxazolidines. 1. The “B” group. J. Med. Chem. 32, 1673–1681.

Gregory, W.A., Brittelli, D.R., Wang, C.L., Kezar, H.S. 3rd, Carlson, R.K., Park, C.H., Corless, P.F., Miller, S.J., Rajagopalan, P., Wuonola, M.A., et al. (1990). Antibacterials. Synthesis and structure-activity studies of 3-aryl-2-oxooxazo-lidines. 2. The “A” group. J. Med. Chem. 33, 2569–2578.

Gürel, G., Blaha, G., Moore, P.B., and Steitz, T.A. (2009). U2504 determines the species specificity of the A-site cleft antibiotics: the structures of tiamulin, homoharringtonine, and bruceantin bound to the ribosome. J. Mol. Biol. 389, 146–156.

Hansen, J.L., Ippolito, J.A., Ban, N., Nissen, P., Moore, P.B., and Steitz, T.A. (2002). The structures of four macrolide antibiotics bound to the large ribosomal subunit. Mol. Cell 10, 117–128.

Hansen, J.L., Moore, P.B., and Steitz, T.A. (2003). Structures of five antibiotics bound at the peptidyl transferase center of the large ribosomal subunit. J. Mol. Biol. 330, 1061–1075.

Hermann, T. (2006). A-site model RNAs. Biochimie 88, 1021–1026.Homeyer, N. and Gohlke, H. (2012). Free energy calculations by

the molecular mechanics Poisson–Boltzmann surface area method. Mol. Inf. 31, 114–122.

Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg, A., and Simmerling, C. (2006). Comparison of multiple Amber force fields and development of improved protein backbone parameters. Proteins Struct. Funct. Bioinf. 65, 712–725.

Huo, S., Wang, J., Cieplak, P., Kollman, P.A., and Kuntz, I.D. (2002). Molecular dynamics and free energy analyses of cathepsin D-inhibitor interactions: insight into structure-based ligand design. J. Med. Chem. 45, 1412–1419.

Hou, T., Wang, J., Li, Y., and Wang, W. (2011). Assessing the performance of the MM/PBSA and MM/GBSA methods. 1. The accuracy of binding free energy calculations based on molecular dynamics simulations. J. Chem. Inf. Model 51, 69–82.

Ippolito, J.A., Kanyo, Z.F., Wang, D., Franceschi, F.J., Moore, P.B., Steitz, T.A., and Duffy, E.M. (2008). Crystal structure of the oxazolidinone antibiotic linezolid bound to the 50S ribosomal subunit. J. Med. Chem. 51, 3353–3356.

Kopitz, H., Cashman, D.A., Pfeiffer-Marek, S., and Gohlke, H. (2012). Influence of the solvent representation on vibrational entropy calculations: generalized born versus distance-dependent dielectric model. J. Comput. Chem. 33, 1004–1013.

Leach, K.L., Brickner, S.J., Noe, M.C., and Miller, P.F. (2011). Linezolid, the first oxazolidinone antibacterial agent. Ann. NY Acad. Sci. 1222, 49–54.

Locke, J.B., Finn, J., Hilgers, M., Morales, G., Rahawi, S., Kedar, G.C., Picazo, J.J., Im, W., Shaw, K.J., and Stein, J.L. (2010). Structure-activity relationships of diverse oxazoli-dinones for linezolid-resistant Staphylococcus aureus strains possessing the CFR methyltransferase gene or ribosomal mutations. Antimicrob. Agents Chemother. 54, 5337–5343.

Ma, C., Baker, N.A., Joseph, S., and McCammon, J.A. (2002). Binding of aminoglycoside antibiotics to the small ribosomal subunit: a continuum electrostatics investigation. J. Am. Chem. Soc. 124, 1438–1442.

Mankin, A.S. and Garrett, R.A. (1991). Chloramphenicol resistance mutations in the single 23S rRNA gene of the archaeon Halobacterium halobium. J. Bacteriol. 173, 3559–3563.

MCPRO (2009), Schrödinger, LLC, New York, NY, USA.Metz, A., Pfleger, C., Kopitz, H., Pfeiffer-Marek, S., Baringhaus, K.H.,

and Gohlke, H. (2012). Hot spots and transient pockets: predicting the determinants of small-molecule binding to a protein-protein interface. J. Chem. Inf. Model 52, 120–133.

Moellering, R.C. (2003). Linezolid: the first oxazolidinone antimi-crobial. Ann. Intern. Med. 138, 135–142.

Park, C.H., Brittelli, D.R., Wang, C.L., Marsh, F.D., Gregory, W.A., Wuonola, M.A., McRipley, R.J., Eberly, V.S., Slee, A.M., and Forbes, M. (1992). Antibacterials. synthesis and structure-activity studies of 3-aryl-2-oxooxazolidines. 4. Multiply-substituted aryl derivatives. J. Med. Chem. 35, 1156–1165.

Romanowska, J., McCammon, J.A., and Trylska, J. (2011). Understanding the origins of bacterial resistance to aminogly-




cosides through molecular dynamics mutational study of the ribosomal A-site. PLoS Comput. Biol. 7, 1–6.

Sanz, J.L., Marín, I., Ureña, D., and Amils, R. (1993). Functional analysis of seven ribosomal systems from extremely halophilic archaea. Can. J. Microbiol. 39, 311–317.

Schlunzen, F., Zarivach, R., Harms, J., Bashan, A., Tocilj, A., Albrecht, R., Yonath, A., and Franceschi, F. (2001). Structural basis for the interaction of antibiotics with the peptidyl transferase centre in eubacteria. Nature 413, 814–821.

Selzer, T., Albeck, S., and Schreiber, G. (2000). Rational design of faster associating and tighter binding protein complexes. Nat. Struct. Biol. 7, 537–541.

Shaw, K.J. and Barbachyn, M.R. (2011). The oxazolidinones: past, present, and future. Ann. NY Acad. Sci. 1241, 48–70.

Singh, N. and Warshel, A. (2010). A comprehensive examination of the contributions to the binding entropy of protein-ligand complexes. Proteins Struct. Funct. Bioinf. 78, 1724–1735.

Skripkin, E., McConnell, T.S., DeVito, J., Lawrence, L., Ippolito, J.A., Duffy, E.M., Sutcliffe, J., and Franceschi, F. (2008). Rχ-01, a new family of oxazolidinones that overcome ribosome-based linezolid resistance. Antimicrob. Agents Chemother. 52, 3550–3557.

Teeter, M.M. and Case, D.A. (1990). Harmonic and quasiharmonic descriptions of crambin. J. Phys. Chem. 94, 8091–8097.

Trylska, J., McCammon, J.A., and Brooks III, C.L. (2005). Exploring assembly energetics of the 30S ribosomal subunit using an implicit solvent approach. J. Am. Chem. Soc. 127, 11125–11133.

Vaiana, A.C., Westhof, E., and Auffinger, P. (2006). A molecular dynamics simulation study of an aminoglycoside/A-site RNA complex: conformational and hydration patterns. Biochimie 88, 1061–1073.

Wang, J., Wolf, R.M., Caldwell, J.W., Kollman, P.A., and Case, D.A. (2004). Development and testing of a general amber force field. J. Comput. Chem. 25, 1157–1174.

Wichmann, C., Becker, Y., Chen-Wichmann, L., Vogel, V., Vojtkova, A., Herglotz, J., Moore, S., Koch, J., Lausen, J., Mantele, W., et al. (2010). Dimer-tetramer transition controls RUNX1/ETO leukemogenic activity. Blood 116, 603–613.

Wilson, D.N. (2011). On the specificity of antibiotics targeting the large ribosomal subunit. Ann. NY Acad. Sci. 1241, 1–16.

Wilson, D.N., Schluenzen, F., Harms, J.M., Starosta, A.L., Connell, S.R., and Fucini, P. (2008). The oxazolidinone antibiotics perturb the ribosomal peptidyl-transferase center and effect tRNA positioning. Proc. Natl. Acad. Sci. USA 105, 13339–13344.

Yang, T., Wu, J.C., Yan, C., Wang, Y., Luo, R., Gonzales, M.B., Dalby, K.N., and Ren, P. (2011). Virtual screening using molecular simulations. Proteins Struct. Funct. Bioinf. 79, 1940–1951.

Yonath, A. (2005a). Antibiotics targeting ribosomes: resistance, selectivity, synergism and cellular regulation. Annu. Rev. Biochem. 74, 649–679.

Yonath, A. (2005b). Ribosomal crystallography: peptide bond formation, chaperone assistance and antibiotics activity. Mol. Cells 20, 1–16.

Yonath, A. and Bashan, A. (2004). Ribosomal crystallography: initiation, peptide bond formation, and amino acid polymer-ization are hampered by antibiotics. Annu. Rev. Microbiol. 58, 233–251.

Zhou, J., Bhattacharjee, A., Chen, S., Chen, Y., Duffy, E., Farmer, J., Goldberg, J., Hanselmann, R., Ippolito, J.A., Lou, R., et al. (2008a). Design at the atomic level: design of biaryloxazoli-dinones as potent orally active antibiotics. Bioorg. Med. Chem. Lett. 18, 6175–6178.

Zhou, J., Bhattacharjee, A., Chen, S., Chen, Y., Duffy, E., Farmer, J., Goldberg, J., Hanselmann, R., Ippolito, J.A., Lou, R., et al. (2008b). Design at the atomic level: generation of novel hybrid biaryloxazolidinones as promising new antibiotics. Bioorg. Med. Chem. Lett. 18, 6179–6183.



Jagmohan S. Saini, Nadine Homeyer, Simone Fulle and Holger … · 2018-11-21 · Jagmohan S. Saini,...

Documents

Transcript of Jagmohan S. Saini, Nadine Homeyer, Simone Fulle and Holger … · 2018-11-21 · Jagmohan S. Saini,...