High-resolution crystal structures of ribosome-bound chloramphenicol and erythromycin provide the ultimate basis for their competition.

The 70S ribosome is a major target for antibacterial drugs. Two of the classical antibiotics, chloramphenicol (CHL) and erythromycin (ERY), competitively bind to adjacent but separate sites on the bacterial ribosome: the catalytic peptidyl transferase center (PTC) and the nascent polypeptide exit tunnel (NPET), respectively. The previously reported competitive binding of CHL and ERY might be due either to a direct collision of the two drugs on the ribosome or due to a drug-induced allosteric effect. Because of the resolution limitations, the available structures of these antibiotics in complex with bacterial ribosomes do not allow us to discriminate between these two possible mechanisms. In this work, we have obtained two crystal structures of CHL and ERY in complex with the Thermus thermophilus 70S ribosome at a higher resolution (2.65 and 2.89 Å, respectively) allowing unambiguous placement of the drugs in the electron density maps. Our structures provide evidence of the direct collision of CHL and ERY on the ribosome, which rationalizes the observed competition between the two drugs.

INTRODUCTION

Protein synthesis is a key step in the gene expression pathway, which is catalyzed by the ribosome. The ribosome is composed of two unequal subunits, small and large (30S and 50S in bacteria), which join together to form a 70S particle. While the 30S subunit is in charge of decoding of the genetic information carried by the messenger RNA (mRNA), the peptidyl transferase center (PTC) of the 50S subunit links amino acids into a nascent polypeptide which is then threaded through the nascent peptide exit tunnel (NPET) that spans the body of the large subunit. PTC and NPET are the sites targeted by the broadest array of inhibitors belonging to several distinct chemical classes, including phenicols, lincosamides, oxazolidinones, pleuromutilins, streptogramins A and B, macrolides, and ketolides (Lin et al. 2018; Polikanov et al. 2018). Some of these inhibitors are among the most successful antimicrobial drugs used in the clinic to treat human bacterial infections (Van Boeckel et al. 2014).

Among the “oldest” and the best studied ribosome-targeting antibiotics are chloramphenicol (CHL, Fig. 1A) and erythromycin (ERY, Fig. 1B), which are also the prototype representatives of their respective chemical classes—phenicols and macrolides. CHL binds in the A site of the PTC and is thought to act as a competitive inhibitor, which prevents binding of an aminoacyl moiety of an incoming aminoacyl-tRNA substrate in the ribosomal A site resulting in inhibition of peptide bond formation (Hansen et al. 2003; Bulkley et al. 2010; Dunkle et al. 2010). Although the binding site of ERY (and all other macrolides) is located in the NPET 10 Å away from the PTC (Tu et al. 2005; Bulkley et al. 2010; Dunkle et al. 2010), CHL and ERY are known to compete with each other (Vazquez 1966). This competition might be due to either a direct collision of the two drugs on the ribosome or a drug-induced allosteric effect. The early X-ray crystal structures of CHL and ERY in complex with the 50S large ribosomal subunit from eubacterium Deinococcus radiodurans (PDB entries 1K01 and 1JZY, respectively [Schlünzen et al. 2001]) suggested that these drugs have nonoverlapping binding sites (Supplemental Fig. S1A) and, therefore, should not directly compete with each other pointing to an allosteric mechanism. Later structures by two independent groups revealed substantially different orientations of CHL and ERY in their binding sites on the ribosome (Supplemental Fig. S1B,C; Bulkley et al. 2010; Dunkle et al. 2010). These structures of CHL and ERY in complex with bacterial 70S ribosomes from either Escherichia coli (Eco) (Dunkle et al. 2010) or Thermus thermophilus (Tth) (Bulkley et al. 2010) provided better quality maps allowing more accurate placement of the drugs in their binding sites. The structures revealed that the dichloroacetic moiety of CHL and the desosamine moiety of ERY, chemical groups that are crucial for binding to the ribosome, are either directly juxtaposed or potentially even overlap (Supplemental Fig. S1B,C; Bulkley et al. 2010; Dunkle et al. 2010). Whether the dichloroacetic moiety of CHL clashes with the desosamine of ERY depends on its particular orientation. However, even with the overall better quality crystallographic data, neither of the newer structures allowed for the unambiguous placing of the dichloroacetic moiety of CHL in the electron density maps. Nevertheless, these structures did not reveal any significant drug-induced rearrangements of the 23S rRNA nucleotides surrounding the PTC and NPET that would affect binding of the competitor drug. Therefore, most likely, the observed competition between the CHL and ERY is due to their steric hindrance.

FIGURE 1.

Electron density maps of ribosome-bound chloramphenicol and erythromycin. (A,B) Chemical structures of PTC-targeting antibiotic chloramphenicol (A, CHL) and NPET-binding macrolide erythromycin (B, ERY). (C,D) Unbiased Fo−Fc electron difference Fourier maps (green mesh) of CHL (C) and ERY (D) in complex with the T. thermophilus 70S ribosome viewed from two different perspectives. The refined models of CHL or ERY are displayed in their respective electron densities contoured at 2.7σ. Carbon atoms are colored yellow for CHL and red for ERY; nitrogens are blue; oxygens are red for CHL and salmon for ERY. Key chemical moieties of each drug are labeled.

In this work, we have obtained two crystal structures of CHL and ERY in complex with the Tth 70S ribosome at a higher resolution (2.65 and 2.89 Å, respectively) allowing unambiguous placement of the CHL dichloroacetic moiety and the ERY desosamine sugar. Our structures provide evidence of the direct collision of CHL and ERY on the ribosome, which rationalizes the observed competition between the two drugs.

RESULTS AND DISCUSSION

High-resolution structures of the ribosome-bound CHL and ERY

In order to unambiguously determine the exact location of all chemical moieties of CHL and ERY in their binding sites on the bacterial ribosome, we cocrystallized Tth 70S ribosomes in the presence of mRNA, deacylated A-, P-, and E-site tRNAs, and either CHL or ERY and solved the structures of the obtained complexes at 2.65 and 2.89 Å resolution, respectively (Supplemental Table S1). To our knowledge, these are the highest resolution structures of ribosome-bound CHL and ERY reported to date. Our structural data revealed more characteristic features of the drug molecules in the electron density maps (Figs. 1C,D, 2C,F). The better quality electron density maps allowed us to visualize the dichloroacetic moiety of CHL (Fig. 2C) and the desosamine sugar of ERY (Fig. 2F), whose placement in the previous structures was ambiguous (Fig. 2A,B,D,E). We believe that the optimized experimental procedures and inclusion of mRNA and tRNAs in our ribosome complexes provided additional stabilization to the ribosome that, in turn, contributed to the higher resolution, similar to other recent structures of ribosome-bound antibiotics (Almutairi et al. 2017; Metelev et al. 2017; Osterman et al. 2017; Pantel et al. 2018; Tereshchenkov et al. 2018).

FIGURE 2.

Comparison of the electron density maps of ribosome-bound CHL and ERY from different structures. The refined models of CHL (A–C) or ERY (D–F) that were either published previously (Bulkley et al. 2010; Dunkle et al. 2010) or determined in the current study are shown in their respective 2Fo−Fc electron density maps contoured at 1.0σ.

The binding position of the amphenicol moiety (but not the dichloracetic) of CHL molecule in our structure is identical to those observed previously for CHL bound to vacant 70S ribosomes from Eco (Supplemental Fig. S1B,D; Dunkle et al. 2010) or Tth (Supplemental Fig. S1C,D; Bulkley et al. 2010) in the absence of mRNA and tRNAs. This suggests that the presence of the deacylated tRNAs in the A and P sites does not affect the general mode of CHL binding to the ribosome. In our structure, the oxygens of the nitro group in the CHL form hydrogen bonds (H-bonds) with the A76 ribose 3′-hydroxyl of the deacylated A-site tRNA and A76 ribose 2′-hydroxyl of the deacylated P-site tRNA (Supplemental Fig. S2). It is unclear whether such interactions are possible in the translating ribosome when the A- and P-site tRNAs are attached to the aminoacyl and peptidyl groups, respectively. Nevertheless, our structure demonstrates that the amphenicol part of CHL anchors it in the PTC, directing the attached dichloroacetic moiety toward the macrolide binding site in the NPET. The overall quality of the obtained electron density map allowed us to unambiguously place all atoms of the dichloracetic moiety (Figs. 1C, 2C). Although in solution this moiety has a certain degree of rotational freedom, in the structure it adopts a unique conformation due to stabilization provided by the H-bond formed with the N6 atom of nucleotide A2062 of the 23S rRNA, which rotates by ∼160° around its N-glycosidic bond into a position where it forms a Hoogsteen base pair with the residue m2A2503 (Fig. 3; Supplemental Movie S1). Similar reorientation of this nucleotide was previously observed in the 70S-CHL structures with vacant Eco (Dunkle et al. 2010) or Tth (Bulkley et al. 2010) ribosomes.

FIGURE 3.

Structure of CHL in complex with the 70S ribosome and A-, P-, and E-site tRNAs. (A,B) Overview of the CHL binding site (yellow) in the Tth 70S ribosome viewed from the PTC down the tunnel as indicated by the inset (A), or as a cross-cut section through the ribosome (B). The 30S subunit is shown in light yellow, the 50S subunit is in light blue, the mRNA is in magenta, and the A-, P-, and E-site tRNAs are colored green, dark blue, and orange, respectively. (C,D) Close-up views of the CHL bound in the PTC. The E. coli nucleotide numbering is used. Potential H-bond interactions are indicated with dashed lines. Note that by forming an H-bond with the base of nucleotide A2062 of the 23S rRNA (light blue) CHL causes characteristic rotation of this nucleotide by ∼160° to form Hoogsteen base pair with the m2A2503 of the 23S rRNA (red dashed arrow). N6 and N7 atoms of nucleotides A2062 and m2A2503 are highlighted in dark blue. The unrotated conformation of A2062 observed in the absence of the drug is shown in blue (PDB entry 4Y4P [Polikanov et al. 2015]).

In our 70S-ERY structure, the drug molecule is located in the canonical macrolide-binding pocket in the NPET near the PTC, which is identical to those observed previously for ERY bound to vacant 70S ribosomes from Eco (Supplemental Fig. S1B,E; Dunkle et al. 2010) or Tth (Supplemental Fig. S1C,E; Bulkley et al. 2010) in the absence of mRNA and tRNAs. Binding of ERY to the ribosome is largely mediated by the H-bonding between the 23S rRNA residues A2058 and A2059 and the desosamine moiety of the drug, which is pointed in the direction of the PTC (Supplemental Movie S1; Fig. 4). Curiously, in our structure, ERY causes similar reorientation of the nucleotide A2062 of the 23S rRNA (Fig. 4B,C) as in the case with CHL (Fig. 3C,D). However, unlike CHL, which forms H-bond with the A2062 base, ERY does not establish this type of interaction but instead forms van der Waals contact with the A2062/A2503 bases (Supplemental Fig. S3). Moreover, this ERY-induced conformational change of A2062 is observed consistently only in Tth 70S ribosomes (current work and Bulkley et al. 2010), but not in the case with Eco 70S ribosomes (Dunkle et al. 2010), or Haloarcula marismortui 50S ribosomal subunit (Tu et al. 2005). This rearrangement potentially could represent a species-specific response to the binding of a macrolide antibiotic.

FIGURE 4.

Structure of ERY in complex with the 70S ribosome and A- and P-site tRNAs. (A) Overview of the ERY binding site (red) in the T. thermophilus 70S ribosome viewed as a cross-cut through the peptide exit tunnel. The 30S subunit is shown in light yellow, the 50S subunit is in light blue, the mRNA is magenta, and the A-, P-, and E-site tRNAs are colored in green, dark blue, and orange, respectively. (B,C) Close-up views of the ERY binding site shown in panel A. Potential H-bond interactions are indicated with dashed lines. Note that binding of ERY causes the same characteristic rotation of the nucleotide A2062 to form Hoogsteen base pair with the nucleotide m2A2503 as observed in the case of CHL (red dashed arrow).

Structural basis for competition between ERY and CHL

The two chemical groups of CHL and ERY that come into close proximity are the dichloroacetic and dimethylamine moieties. The exact rotational orientation of the dichloroacetic group of CHL, which is crucial for our subsequent analysis, was uncertain from the previous structures (Fig. 2A,B). Due to a higher resolution and overall better quality electron density maps that we observed for CHL and ERY (Fig. 2C,F), we were able to unambiguously determine the placement of the dichloroacetic moiety of CHL and provide definitive structural basis for their competition that was previously observed in biochemical assays (Wolfe and Hahn 1965; Vazquez 1966; Pestka 1974).

Superposition of our CHL and ERY structures reveals a direct steric clash between the chlorine atoms of the dichloroacetic moiety of CHL and the dimethylamine of the desosamine sugar of ERY (Fig. 5). Both chemical groups in each of the two drugs are essential for their binding to the ribosome. The dichloroacetic moiety of CHL forms H-bond with the N6 atom of A2062 of the 23S rRNA (Fig. 5A; Supplemental Movie S1), and mutation of this base to cytidine leads to CHL resistance (Mankin and Garrett 1991; Kloss et al. 1999). Likewise, inability to form H-bonds between desosamine of ERY and the N1 and N6 atoms of A2058 and A2059 bases, respectively (Fig. 5B; Supplemental Movie S1), when these bases are mutated to guanines, also causes antibiotic resistance (Vester and Garrett 1987; Sander et al. 1997; Vester and Douthwaite 2001). Therefore, even a minor displacement of any of these groups should lead to disruption of the crucial interaction for either CHL or ERY that would be manifested as competitive binding.

FIGURE 5.

Overlap between CHL and ERY binding sites in the 70S ribosome. (A,B) Two close-up views of the superimposed binding sites for CHL and ERY. The structures were aligned based on the domain V of the 23S rRNA. The coloring of the 23S rRNA and the antibiotics is the same as in Figures 3 and 4. Note that dichloroacetic group of CHL clashes with the dimethylamine group of the desosamine sugar of ERY, rationalizing their mutually exclusive binding to the 70S ribosome.

The competition between CHL and ERY was demonstrated biochemically decades ago using ribosomes from several bacterial species, including E. coli (Wolfe and Hahn 1965; Vazquez 1966; Pestka 1974). However, such competition has never been tested for the Gram-negative bacterium T. thermophilus, which we use as a source of ribosomes in our structural studies. To confirm the ability of macrolides to displace prebound CHL from Tth 70S ribosomes, we used competition-binding assay exploiting [14C]-CHL and nonlabeled ERY. In our assays, the macrolide efficiently displaces CHL from both Eco (Supplemental Fig. S4A) and Tth (Supplemental Fig. S4B) ribosomes in a concentration-dependent manner. Therefore, CHL-ERY competition occurs on Tth ribosomes and is likely to be manifested in most of the bacterial species.

Our structural analysis clearly points to a direct steric clash mechanism of competition between CHL and ERY, which was first observed biochemically more than five decades ago. We wonder whether there are any PTC-targeting antibiotics that do not compete (and likely can coexist) with the macrolides on the bacterial ribosome. We believe that finding such inhibitors could potentially lead to the identification of drug pairs exhibiting more efficient and possibly even synergistic antibacterial action.

MATERIALS AND METHODS

Materials for biochemical experiments

Nonlabeled CHL and ERY were obtained from MilliporeSigma. Radioactively labeled [14C]-ERY and [14C]-CHL were obtained from MP Biomedicals and Moravek Biochemicals, respectively. Cellulose nitrate filters (type HA, 24-mm diameter, 0.45-µm pore size) were obtained from MilliporeSigma. Scintillation liquid was obtained from Bio-Rad.

Biochemical preparations

Reassociated Escherichia coli 70S ribosomes were prepared from K12 strain as previously described (Blaha et al. 2000). Purified ribosomes were stored in a buffer containing 20 mM HEPES-KOH (pH 7.6), 50 mM CH3COONH4, 6 mM Mg(CH3COO)2, and 4 mM β-mercaptoethanol. 70S ribosomes were prepared from Thermus thermophilus strain HB8 as described in Polikanov et al. (2014).

Competition-binding assays

Reassociated Escherichia coli 70S ribosomes (0.2 µΜ final concentration) were incubated in buffer A [20 mM HEPES-KOH [pH 7.6], 150 mM NH4(CH3COO), 6 mM Mg(CH3COO)2, 2 mM spermidine, 0.05 mM spermine, and 4 mM β-mercaptoethanol] with [14C]-CHL (150 dpm/pmol) at the appropriate/indicated concentration. After incubation for 10 min at 37°C, the mixture was diluted with 3 mL of cold buffer A and filtered through a 25-mm diameter 0.45-µm pore size nitrocellulose membrane (MilliporeSigma). The filter was immediately washed three times with 3 mL of cold buffer A and bound radioactivity was measured. Next, binding of [14C]-CHL was studied in competition with nonradiolabeled CHL or ERY by maintaining a constant concentration of [14C]-CHL (0.6 µM) and increasing concentrations of a nonradioactive competitor (Karahalios et al. 2006).

In the case of Thermus thermophilus 70S ribosomes, the binding competition between CHL and ERY was assessed in the same conditions as the “fragment reaction” described previously (Fernandez-Munoz et al. 1971). Each 25-µL reaction contained 33% (v/v) ethanol, 33 mM Tris-HCl pH 7.5, 270 mM KCl, 13 mM Mg(CH3COO)2, 4 µM of Tth 70S ribosomes and 3 µM of [14C]-CHL (150 dpm/pmol). Increasing concentrations of nonradioactive CHL or ERY were added to the reactions. After incubation at 0°C for 30 min, the ribosomes were precipitated by centrifugation at 14,000g for 20 min and radioactivity of CHL in ribosome-free supernatants was measured.

Crystallographic structure determination

Ribosome complexes with mRNA and tRNAs were formed as described previously (Polikanov et al. 2015). CHL or ERY were added to the preformed ribosome complexes to a final concentration of 250 µM prior to crystallization. All Tth 70S ribosome complexes were formed in the buffer containing 5 mM HEPES-KOH (pH 7.6), 50 mM KCl, 10 mM NH4Cl, and 10 mM Mg(CH3COO)2, and then crystallized in the buffer containing 100 mM Tris-HCl (pH 7.6), 2.9% (w/v) PEG-20 K, 7%–12% (v/v) MPD, 100–200 mM arginine, 0.5 mM β-mercaptoethanol. Crystals were grown by the vapor diffusion method in sitting drops at 19°C and stabilized as described previously (Polikanov et al. 2015) with CHL or ERY being added to the stabilization buffers (100 µM each). Diffraction data were collected using beamline 24ID-C and 24ID-E at the Advanced Photon Source (Argonne National Laboratory). A complete data set for each ribosome complex was collected using 0.979 Å wavelength at 100 K from multiple regions of the same crystal using 0.3° oscillations. The raw data were integrated and scaled using the XDS software package (Kabsch 2010). All crystals belonged to the primitive orthorhombic space group P212121 with approximate unit cell dimensions of 210 × 450 × 620 Å and contained two copies of the 70S ribosome per asymmetric unit. Each structure was solved by molecular replacement using PHASER from the CCP4 program suite (McCoy et al. 2007). The search model was generated from the previously published structures of T. thermophilus 70S ribosome with bound mRNA and tRNAs (PDB entry 4Y4P from Polikanov et al. [2015]). The initial molecular replacement solutions were refined by rigid body refinement with the ribosome split into multiple domains, followed by positional and individual B-factor refinement using PHENIX (Adams et al. 2010). Noncrystallographic symmetry restraints were applied to four domains of the 30S ribosomal subunit (head, body, spur, helix 44), and four domains of the 50S subunit (body, L1-stalk, L10-stalk, C terminus of the L9 protein).

Atomic models of CHL and ERY were generated from their known chemical structures (Fig. 1A,B) using PRODRG online software (Schüttelkopf and van Aalten 2004), which was also used to generate restraints for energy minimization and refinement based on idealized 3D geometry. Atomic models and restraints were used to fit/refine CHL and ERY into the obtained electron density maps (Fig. 1C,D). The final models of the Tth 70S ribosome in complex with mRNA/tRNAs and CHL or ERY were generated by multiple rounds of model building in COOT (Emsley and Cowtan 2004), followed by refinement in PHENIX (Adams et al. 2010). The statistics of data collection and refinement are compiled in Supplemental Table S1. All figures showing atomic models were generated using the PyMol software (www.pymol.org).

DATA DEPOSITION

Coordinates and structure factors were deposited in the RCSB Protein Data Bank with the following accession codes: 6ND5 for the T. thermophilus 70S ribosome in complex with chloramphenicol, mRNA, A-, P-, and E-site tRNAs; 6ND6 for the T. thermophilus 70S ribosome in complex with erythromycin, mRNA, A-, P-, and E-site tRNAs.

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.