Cryo-EM structure of the tetracycline resistance protein TetM in complex with a translating ribosome at 3.9-Å resolution

Significance

The ribosome, the protein-synthesizing machine in the cell, is a major target for antibiotics, such as tetracyclines. The widespread usage of tetracyclines has led to an increase in tetracycline resistance amongst medically relevant pathogenic bacteria, limiting their utility. Many bacteria obtain tetracycline resistance via ribosome protection proteins, such as TetM and TetO, that bind to the ribosome and chase tetracycline from its binding site. We have determined a structure of TetM bound to a translating ribosome at 3.9 Å, providing molecular insight into how TetM interacts with the ribosome to dislodge the drug from its binding site.

Abstract

Ribosome protection proteins (RPPs) confer resistance to tetracycline by binding to the ribosome and chasing the drug from its binding site. Current models for RPP action are derived from 7.2- to 16-Å resolution structures of RPPs bound to vacant or nontranslating ribosomes. Here we present a cryo-electron microscopy reconstruction of the RPP TetM in complex with a translating ribosome at 3.9-Å resolution. The structure reveals the contacts of TetM with the ribosome, including interaction between the conserved and functionally critical C-terminal extension of TetM with a unique splayed conformation of nucleotides A1492 and A1493 at the decoding center of the small subunit. The resolution enables us to unambiguously model the side chains of the amino acid residues comprising loop III in domain IV of TetM, revealing that the tyrosine residues Y506 and Y507 are not responsible for drug-release as suggested previously but rather for intrafactor contacts that appear to stabilize the conformation of loop III. Instead, Pro509 at the tip of loop III is located directly within the tetracycline binding site where it interacts with nucleotide C1054 of the 16S rRNA, such that RPP action uses Pro509, rather than Y506/Y507, to directly dislodge and release tetracycline from the ribosome.

The ribosome is one of the major targets for antibiotics within the bacterial cell (1, 2). A well-characterized class of broad-spectrum antibiotics in clinical use are the tetracyclines, which bind to elongating ribosomes and inhibit delivery of the EF-Tu•GTP•aa-tRNA ternary complex to the A-site (1, 3). X-ray crystal structures of ribosomal particles in complex with tetracycline have revealed that the primary drug binding site is located in helix 34 (h34) of the 16S rRNA, overlapping the binding position of the anticodon-stem loop of an A-site tRNA (4–6). The widespread use of tetracyclines has led to an increase in tetracycline resistance among clinically relevant pathogenic bacteria, thus limiting the medical utility of many members of this class (7). Drug efflux and ribosome protection are the most common tetracycline resistance mechanisms acquired by bacteria (8) and have led to the development of the third generation of tetracycline derivatives, such as tigecycline, which display enhanced antimicrobial activity and overcome both the efflux and ribosome protection resistance mechanisms (6, 9–11).

To date, there are 12 distinct classes of ribosome protection proteins (RPPs) that confer resistance to tetracycline, with the most prevalent and best characterized being TetO and TetM (3, 8, 12). The different classes of RPPs have high homology with one another; for example, Campylobacter jejuni TetO displays >75% identity (>85% similarity) with Enterococcus faecalis TetM. Based on the presence of conserved nucleotide binding motifs, RPPs are grouped together within the translation factor superfamily of GTPases (13). Accordingly, TetO and TetM catalyze the release of tetracycline from the ribosome in a GTP-dependent manner (14, 15). Biochemical studies indicate that, although GTPase activity is necessary for multiturnover of RPPs, GTP hydrolysis is not strictly required to dislodge tetracycline because the drug is also released when nonhydrolysable GTP analogs are used (14, 15).

Nonhydrolysable GTP analogs have been used to trap RPPs on the ribosome for structural analysis by cryo-EM. The first structure of an RPP-ribosome complex was a cryo-EM reconstruction of a TetO•70S complex at 16-Å resolution. This structure revealed that TetO binds analogously to the ribosome as translation elongation factor EF-G (16), consistent with the significant homology (∼25/35% identity/similarity) between RPPs and EF-G (17). Because the electron density for TetO did not come within 6 Å of the tetracycline-binding site (16), TetO was suggested to chase the drug from the ribosome by inducing conformational changes within h34 (12, 16, 18). In contrast, two subsequent structures at higher resolution, a TetM•70S complex at 7.2 Å (19) and a TetO•70S complex at 9.6 Å (20), revealed electron density for the RPPs directly overlapping with the tetracycline binding site. Based on the homology with EF-G, molecular models for the RPPs were generated and docked into the cryo-EM maps, suggesting that residues within loop III of domain IV of TetM/TetO come into direct contact with the tetracycline molecule (19, 20). Consistently, mutagenesis studies identified specific residues within loop III that are critical for RPP activity (19–21), in particular the conserved tyrosine residues Y506 and Y507 (19, 20). However, the exact role of these tyrosine residues and a detailed molecular understanding of the mechanism by which RPPs dislodge tetracycline from its binding site was not possible at the reported resolutions.

Here we present a cryo-EM structure of TetM in complex with a translating ribosome at an average resolution of 3.9 Å. Local resolution calculations indicate that the majority of the core of the ribosome and domain IV of TetM extends toward 3.5 Å, enabling bulky side chains to be modeled. We provide a detailed account of the interactions between TetM and the ribosome, in particular revealing a complex network of interactions of the C-terminal helix and domain IV of TetM with the ribosomal decoding site and intersubunit bridge B2a. The structure reveals that Pro509 at the tip of loop III, rather than the previously identified tyrosine Y506 and Y507, overlaps the binding site of tetracycline and is therefore directly involved in releasing tetracycline from the ribosome.

Results and Discussion

Cryo-EM Structure and Molecular Model of a TetM•RNC.

In our previous study (19), the TetM•70S complex was formed using vacant 70S ribosomes, which led to high heterogeneity because the vacant ribosomes adopted both rotated and nonrotated states. Moreover, because TetM only interacts with the nonrotated ribosomes, the heterogeneity reduced the overall occupancy of TetM on the ribosome. A further reduction in occupancy resulted from the presence of tigecycline, the binding of which (contrary to initial expectations; refs. 16 and 19) was mutually exclusive with TetM binding (19). As a result, the final reconstruction of the TetM•70S complex was derived from only 52,701 (12%) of the initial 406,687 particles and yielded a resolution of 7.2 Å (19). To reduce sample heterogeneity and increase the TetM occupancy, we omitted tigecycline and formed a complex between TetM and a translating, rather than vacant, 70S ribosome. We have previously prepared and determined cryo-EM structures of 70S ribosomes stalled during translation of Erm leader peptides by the presence of the macrolide antibiotic erythromycin (22–24). These studies revealed that the ErmCL-stalled ribosome is an ideal substrate for TetM binding because the ribosome adopts a nonrotated conformation with a peptidyl-tRNA in the P-site and a vacant A-site (22, 24). Therefore, the ErmCL-stalled ribosomes were bound with TetM in presence of the nonhydrolysable GTP analog, GDPCP, and the resulting sample was subjected to multiparticle cryo-EM (Materials and Methods).

Data were collected on a Titan Krios transmission electron microscope, fitted with a Falcon II direct electron detector, and processed with the SPIDER software package (25). After removal of nonaligning and edge particles, in silico sorting revealed the presence of two subpopulations of ribosomes bearing peptidyl-tRNA in the P-site, and either a vacant A-site (25%) or an A-site occupied by TetM (75%) (SI Appendix, Fig. S1A). The latter volume, which we term the TetM•ribosome nascent chain complex (TetM•RNC), contained 78,186 particles and was refined further to produce a final cryo-EM map of the TetM•RNC (Fig. 1A) with an average resolution of 3.9 Å (based on the Fourier shell correlation cutoff at 0.143, SI Appendix, Fig. S1B). Similar to our recent cryo-EM structure of the ErmCL-RNC (22), local resolution calculations indicate that the ribosomal core of the TetM•RNC extends to 3.5 Å (SI Appendix, Fig. S1 C and D). The resolution of domains I-V of TetM was predominantly between 3.5–4.5 Å (Fig. 1B), but with some regions extending to >5 Å, indicating flexibility as observed recently for other ribosome-bound ligands (26–29). Strand separation in β-sheets and the pitch of helices is observed, allowing a more accurate and complete backbone model to be presented for all 639 residues in domains I–V of TetM (Fig. 1C).

Fig. 1.

Cryo-EM reconstruction of TetM•RNC. (A) Cryo-EM density map of the TetM•RNC, with TetM (orange), 30S (yellow), 50S (gray), and P-tRNA (green). (B and C) Extracted cryo-EM density for TetM colored according to local resolution (B) and with fitted polyalanine model into the density (gray mesh) for domains I (G domain), G′ subdomain, II, III, IV, V, and C-terminal extension (CTE) (C). (D) Overview of the TetM•RNC showing cryo-EM density with fitted models for 30S (yellow) and 50S (gray) subunits, and TetM (orange). Ribosomal proteins contacting TetM are colored (S12, brown; L6, cyan; L7, yellow; L10, red; L11, violet). (E) Model for the TetM•RNC with rRNA helices that interact with TetM colored (h34, green; h44, pink; H43/H44, dark green; H69, light blue; H95, dark blue). (F) Side-view of E. (G) Side view of E with zoom onto 16S rRNA helices h34 (green) and h44 (pink) and 23S rRNA helix H69 (blue) that directly interact with domain IV and the CTH of TetM (orange).

Moreover, the high resolution of the ribosome enabled us to more precisely map the sites of interaction with TetM (Fig. 1D and SI Appendix, Table S1) compared with previous reports (16, 19, 20). Overall, the interactions of TetM are similar to those for translation GTPases, such as EF-G (30), such that on the 50S subunit, the G domain of TetM contacts the sarcin-ricin loop (SRL, H95 of the 23S rRNA) and ribosomal protein L6, whereas the G′ subdomain interacts with one of the C-terminal domains of L7 (Fig. 1 D and E and SI Appendix, Fig. S2 A–F). Domain V of TetM inserts into the cleft formed by H43/H44 of the 23 rRNA and L11, overlapping the binding site of thiostrepton (31) (SI Appendix, Fig. S2 A–F) and explaining the inhibition of TetM by this antibiotic (32–34). On the small subunit, domain III of TetM contacts ribosomal protein S12 (Fig. 1 D and E), whereas domain IV of TetM wedges between the head and body of the 30S, reaching into the decoding center where contacts with h34 (head) and h44 (body) of the 16S rRNA are observed, as well as between the C-terminal extension of TetM and H69 of the 23S rRNA (Fig. 1 F and G).

Interaction of Domain IV of TetM with the 30S Subunit.

Domain IV of TetM comprises a four-stranded β-sheet and two α-helices, with an overall βββαβα topology. Three loops (termed Loop I, II, and III) protrude from the distal end of domain IV of TetM (Fig. 2A). The proline-rich loop I, located between β2₄ and β3₄, was modeled differently in the recent TetM- and TetO-bound ribosome structures (3, 19, 20). In our structure, loop I adopts a bent conformation to establish interactions with the C-terminal helix αA_CTE of TetM (Fig. 2A), similar to that predicted previously for TetM (19), but quite unlike the extended conformation suggested for TetO (20) (SI Appendix, Fig. S3 A–D). We believe that a bent conformation of loop I of TetO would be more consistent with the electron density for the TetO•70S complex as well as with the high sequence conservation between TetO and TetM (SI Appendix, Fig. S3 A–F). Moreover, the extended conformation modeled for the TetO•70S structure is incompatible with the presence of mRNA (SI Appendix, Fig. S3A), suggesting that loop I is unlikely to form part of a corridor that tetracycline navigates during its release from the ribosome (20).

Fig. 2.

Interactions of domain IV and the CTH of TetM. (A) Overview of domain IV (orange) and the C-terminal extension (cyan) of TetM, indicating interaction of loops I-III with rRNA helices h34, h44 and H69 as well as loop I with the C-terminal helix (CTH) of TetM. (B) Proximity of loop II residues (Cα atoms shown as yellow spheres) to the nucleotides C1209, C1051 and C1214 of h34 of the 16S rRNA. (C and D) Interaction of the CTH of TetM (orange) with nucleotide A1913 of H69 of 23S rRNA (deep blue). In D, the positions of A1913 with ribosome lacking A-tRNA (green, PDB 4G5U; ref. 6) or containing A-tRNA (blue, PDB 3TVE; ref. 36) or A/T-tRNA (pink, PDB 2XQE; ref. 35) are shown. (E and F) Flipped-out conformations of nucleotides A1492 and A1493 of h44 of the 16S rRNA (blue) upon TetM (orange) binding to the ribosome. In F, the positions of A1492 and A1493 with ribosome lacking A-tRNA (green, PDB 4G5T; ref. 6) or containing A-tRNA (blue, PDB 3TVF; ref. 36) or A/T-tRNA (pink, PDB 2XQD; ref. 35) are shown.

The density for Loop II between β4₄ and αA₄ is poorly ordered, however interaction with helix 34 of the 16S rRNA is apparent, with residues Ser465 and Leu466 of TetM coming into close proximity with the backbone of C1209 and the nucleobase of C1214 (Fig. 2B). This finding is in agreement with the protection of C1214 from DMS modification upon TetO binding to the ribosome (18, 32). With the exception of Gly467, the residues of loop II are not highly conserved and mutagenesis of these residues exhibited only moderate affects on TetM activity (SI Appendix, Fig. S3 G and H). We note, however, that shortening of the loop by removal of two amino acids was previously shown to completely inactivate TetO (20).

Interaction of TetM at the Ribosomal Decoding Site.

The C-terminal extension (CTE) of TetM comprises a short 11-aa α-helix (residues 627–637) connected to domain V by a flexible linker (Fig. 2C), consistent with previous reports (19). Sequence alignments, secondary structure predictions, as well as the electron density for the TetO•70S complex (SI Appendix, Fig. S4 A–F) lead us to suggest that the topology of the CTE observed here for TetM is a conserved feature of all RPPs. The C-terminal helix (CTH) is likely to stabilize domain IV of TetM on the ribosome, as we observe contact between the CTH and A1913 located at the tip of H69 of the 23S rRNA (Fig. 2C). A1913 adopts a very defined position, similar to that observed when A-tRNA or A/T-tRNA (in complex with EF-Tu) is bound to the ribosome (35, 36) (Fig. 2D and SI Appendix, Fig. S4 G and H), but distinct from the conformation observed in the absence of A-tRNA where A1913 inserts into h44 of the 16S rRNA (6) (Fig. 2D). Although nucleotides A1492 and A1493 of h44 exhibit some flexibility (SI Appendix, Fig. S4 I and J), both nucleotides clearly adopt preferred conformations when TetM is bound, such that both nucleotides are flipped-out of h44 and extend toward Loop I and the CTH, respectively, of TetM (Fig. 2E). The flipping of A1492 and A1493 by TetM binding was suggested previously at 7.2 Å (19) to resemble the conformation observed during decoding of the mRNA–tRNA duplex (35–37) (Fig. 2F). At higher resolution, however, it is evident that the exact conformations of A1492 and A1493 are distinct and the nucleotides adopt an unusual splayed conformation (Fig. 2 E and F), which to our knowledge has not been observed before. The most similar conformation for A1493 was observed in the P-tRNA bound ribosome with a vacant A-site (Fig. 2F); however, in this structure, A1492 remains buried within h44. Although the resolution of the previous TetO•70S structures (16, 20) was insufficient to unambiguously assign the conformational state of A1492 and A1493, biochemical studies suggest that binding of TetO to the ribosome also flips A1493 from h44, as indicated by exposure of A1408 of the 16S rRNA to DMS modification (18, 19). Because removal of the CTH by truncation of 17 amino acids inactivates TetM (19), it is likely that the interaction of TetM, and presumably TetO, with A1492 and A1493 is critical for stabilization of the RPP on the ribosome.

Pro509 of Loop III of TetM Directly Encroaches Upon the Tetracycline-Binding Site.

Loop III of TetM linking β5₄ to helix αB₄ is the best resolved part of the TetM structure with a local resolution predominantly around 3.5 Å, which enabled the bulky aromatic sidechains, such as tyrosines and phenylalanines, to be modeled (Fig. 3 A and B). In contrast to the previous TetM/O•70S reconstructions at lower resolution (19, 20), where the density for Loop III was ambiguous (SI Appendix, Fig. S5 A and B), we are confident of the register of the amino acids within Loop III of TetM as well as the orientation of the side chains in most cases (Fig. 3 A and B and SI Appendix, Fig. S5C). Based on this model, Pro509 at the tip of loop III stacks against C1054 within h34 of the 16S rRNA (Fig. 3C), explaining the protection of C1054 from DMS modification observed upon TetO binding to the ribosome (18). C1054 comprises part of the primary tetracycline-binding site and establishes stacking interactions with ring D of tetracycline (4–6) (Fig. 3D). Our structure indicates that Pro509 of Loop III of TetM clashes with tetracycline and is therefore directly responsible for dislodging the drug from the ribosome (Fig. 3D). This contrasts with previous suggestions that the two conserved tyrosines, Y506 and Y507, within loop III of TetM are directly involved in tetracycline release (19, 20). It is worth noting that although Pro509 is identical in all available RPP sequences, Y506 and Y507 are substituted with Phe/Val and Ser/Phe/Arg, respectively, in some RPPs (SI Appendix, Fig. S5D). Our structure would also suggest that shortening loop III would remove the overlap with tetracycline, consistent with the lack of activity of TetO mutants where two residues were deleted from loop III (20). Although tigecycline exhibits an even greater overlap with TetM (Fig. 3E), we believe that, in addition to the increased affinity of tigecycline compared with tetracycline (10, 11, 38), the C9-glycyl substituent of tigecycline hinders access of the loop III residues to C1054 and thus contributes to tigecyclines ability to overcome TetM-mediated resistance (6, 11).

Fig. 3.

The role of loop III in TetM in tetracycline resistance. (A and B) Extracted Cryo-EM density of loop III of domain IV in TetM (gray mesh) with molecular model for loop III (A) and colored according to local resolution (B). (C) Stacking interaction of P509 at the tip of loop III (orange) with nucleotide C1054 of h34 of the 16S rRNA (blue). (D and E) Comparison of the binding positions of loop III of TetM domain IV (orange) with (D) tetracycline (Tet) and (E) tigecycline (Tig; ref. 6).

Stabilization of Loop III Is Critical for TetM Activity.

Given that Ser508 and Pro509 located at the tip of Loop III are invariant in all available RPP sequences (SI Appendix, Fig. S5D), it is somewhat surprising that these and neighboring residues can be mutated to alanine with little or no effect on RPP activity (3, 19). Similarly, the double SP508-509/AA and triple SPV508-510/AAA mutants of TetM were also shown to retain tetracycline resistance activity (19). In silico mutagenesis based on our refined model indicated that if loop III of the triple SPV508-510/AAA mutant adopts the same conformation as the wildtype TetM then the backbone of Ala509 would maintain a steric clash with tetracycline (Fig. 4A), providing a possible explanation for the retention in activity of the mutant. In contrast, mutation of Y506/Y507 completely inactivates TetM/TetO (3, 19, 20), indicating an important role for these tyrosine residues. Indeed, in our structure, both tyrosines are involved in intradomain interactions linking loop III with loops I and II (Fig. 4B). Specifically, Y507 comes within 3.5 Å of E435 within loop I and the side chain OH of Y506 is within hydrogen bonding distance to the carbonyl of G467 in loop II, as well as to the ribose 2’ OH of C1051 in h34 of the 16S rRNA (Fig. 4B). Collectively, these results suggest that the role of Y506 and Y507 within loop III is to stabilize the conformation of loop III.

Fig. 4.

Stabilization of loop III in TetM via intra-TetM interactions ensures TetM activity. (A) Relative binding position of TetM triple mutant SPV508-510AAA (orange) and tetracycline (Tet, red; ref. 6). (B) Tyrosine residues Y506 and Y507 of loop III of TetM domain IV (orange) stabilize the conformation of loop III via interactions with G467 of loop II, 16S rRNA residue C1051 and residue E435 of loop I, respectively. (C) Localization of TetM residue F516 within the hydrophobic pocket formed by loop III. (D) Growth curves of wildtype E. coli strain BL21 (black) in the presence of increasing concentrations of tetracycline (0-128 µg/mL) compared with the wildtype strain harboring a plasmid encoding wildtype TetM (red) and TetM single mutants F516A (green) and F516D (brown). (E) Interaction between the sidechain V510 of loop III of TetM with the invariant tryptophan (W442) located in loop I. (F) as in D but with TetM mutant W442A (orange) and the double mutants W442A/Y506A (brown), W442A/Y507A (green), W442A/S508A (olive), W442A/P509A (blue) and W442A/V510A (violet). In D and F, the error bars represent the SD from the mean for three independent experiments.

To further investigate the importance of the stabilization of loop III for TetM activity, we analyzed the activity of two additional TetM mutants: The first mutations were introduced at position F516. F516 is invariant in all RPP sequences (SI Appendix, Fig. S5D), and the phenylalanine side chain is well resolved within the hydrophobic core of loop III, where it clamps the proximal end of helix αB₄ to the distal end of strand β5₄ (Figs. 3A and 4C). To monitor activity of TetM, the growth of wild-type Escherichia coli strain BL21 (−TetM) in the presence of increasing concentrations of tetracycline (0–128 µg/mL) was compared with the same strain bearing a plasmid overexpressing either Enterococcus faecalis TetM (+TetM) or one of the TetM variants (Fig. 4D). In the absence of TetM, the wild-type Escherichia coli strain (black circles) is sensitive to tetracycline with a minimal inhibitory concentration (MIC₅₀) of ∼0.6 µg/mL, whereas as before (19), overexpression of Enterococcus faecalis TetM (red circles) raises the MIC₅₀ by 14-fold to ∼10 µg/mL (Fig. 4D). Although mutation of F516 to alanine (F516A) had a modest affect on TetM activity (MIC₅₀ ∼3 µg/mL), mutation of F516 to the negatively charged Asp (F516D) led to a complete loss of activity (Fig. 4D), consistent with the importance of F516 for providing a hydrophobic environment to maintain the defined conformation of loop III necessary for tetracycline release. Another possible source of stabilization of Loop III is the interaction between the side chain of V510 and an invariant tryptophan (W442) located within loop I (Fig. 4E). Although mutation of W442 to alanine (W442A) alone did not affect the activity of TetM, the presence of the W442A mutation made loop III sensitive to secondary mutations. In particular, mutations of S508 or P509 to alanine in the context of W442A abolished TetM activity (Fig. 4F), whereas wild-type activity was observed for TetM with single S508A or P509A mutations (19). Collectively, these results illustrate the importance of the structural integrity of loop III in the positioning residues S508 and P509 located at the tip of loop III, which is necessary for efficient tetracycline resistance.

Conclusion

In conclusion, our structure enables a molecular model to be presented for how TetM confers resistance to tetracycline by dislodging the drug from its binding site on the ribosome (Fig. 5 A and B): Specifically, Pro509 within loop III of domain IV of TetM directly overlaps in position with ring D of tetracycline and thus dislodges the drug from the ribosome. TetM is proposed to prevent rebinding of tetracycline by altering the conformation of nucleotides such as C1054 within the drug binding site that persist following TetM dissociation (12, 18, 19, 32). Within the constraints of the current resolution, TetM does not appear to alter the conformation of C1054 to prevent drug rebinding (SI Appendix, Fig. S5E), however, we cannot rule out that such an alteration occurs during GTP hydrolysis and dissociation of TetM from the ribosome. Previous studies identified two conserved tyrosines within loop III of TetM as being important for tetracycline resistance (3, 19, 20). Here we show that these tyrosine residues are not directly involved in displacing the drug from its binding site, but rather act like clamps (termed C1 and C2) that stabilize the loop III of domain IV of TetM by establishing intradomain interactions with loop I and II of TetM (Fig. 5A). We also identify an additional clamp C3 between loop I and III that is important for stabilization of loop III. Additionally, domain IV of TetM is positioned on the ribosome for tetracycline displacement via interaction of loop I and the CTH with residues located within intersubunit bridge B2a, namely, A1913 of H69 of the 23S rRNA and a splayed conformation of decoding site nucleotides A1492 and A1493 (Fig. 5B). We believe that the molecular details and mechanism of action reported here for TetM will be conserved for other ribosome protection proteins, such as TetO, that also confer resistance to tetracycline.

Fig. 5.

Schematic model for TetM-mediated tetracycline resistance. (A and B) Upon TetM binding to tetracycline bound ribosomes, the proline residue P509 located at the tip of loop III of domain IV is directly responsible for chasing the drug off the ribosome by interacting with its binding site nucleotide C1054 of the 16S rRNA. (B) TetM binding to the ribosome leads to interaction of the C-terminal helix (CTH) with 23S rRNA nucleotide A1913 (dark blue) and induces 16S rRNA decoding nucleotides A1492 and A1493 (blue) to flip out of helix 44 (h44) of the 16S rRNA. Intramolecular interactions that stabilize the conformation of loop III are represented as green clamps with C1 illustrating the interaction Y506/G467, C2 for Y507/E435 and C3 for V510/W442.

Materials and Methods

Enterococcus faecalis TetM was purified as described (34) and bound to ErmCL-RNC (22). Cryo-EM data were collected using the EPU software on a Titan Krios TEM (FEI) and processed using the SPIDER software package (25). The backbone model of Enterococcus faecalis TetM was initially generated using HHPred (39), then manually fitted using Chimera (40) and refined using Coot (41) and PHENIX (42). A structure of the Escherichia coli 70S ribosome (43) was fitted the cryo-EM density using Chimera (40), manually adjusted and then refined with Coot (41). Site-specific mutations were introduced into the tetM gene using KOD Xtreme Hot Start Polymerase according to the manufacturers instructions and minimal inhibitory concentrations were determined as described (11, 19). Detailed materials and methods can be found in the SI Appendix, SI Materials and Methods.