Negamycin induces translational stalling andmiscoding by binding to the small subunit headdomain of the Escherichia coli ribosomeNelson B. Oliviera, Roger B. Altmanb, Jonas Noeskec, Gregory S. Basarabd, Erin Codea, Andrew D. Fergusona, Ning Gaoa,Jian Huanga, Manuel F. Juetteb, Stephania Livchaka, Matthew D. Millerd, D. Bryan Princea, Jamie H. D. Catec,Ed T. Buurmane,1, and Scott C. Blanchardb,1

aDiscovery Sciences, AstraZeneca R&D Boston, Waltham, MA 02451; bDepartment of Physiology and Biophysics & the Tri-Institutional PhD Program inChemical Biology, Weill Cornell Medical College, New York, NY 10065; cDepartment of Molecular Cell Biology, California Institute of Quantitative Biosciences,University of California, Berkeley, CA 94720; and Departments of dChemistry and eBiosciences, Infection Innovative Medicines Unit, AstraZeneca R&D Boston,Waltham, MA 02451

Edited by Harry F. Noller, University of California, Santa Cruz, CA, and approved August 19, 2014 (received for review July 29, 2014)

Negamycin is a natural product with broad-spectrum antibacterialactivity and efficacy in animal models of infection. Although itsprecise mechanism of action has yet to be delineated, negamycininhibits cellular protein synthesis and causes cell death. Here, weshow that single point mutations within 16S rRNA that conferresistance to negamycin are in close proximity of the tetracyclinebinding site within helix 34 of the small subunit head domain. Asexpected from its direct interaction with this region of theribosome, negamycin was shown to displace tetracycline. How-ever, in contrast to tetracycline-class antibiotics, which serve toprevent cognate tRNA from entering the translating ribosome,single-molecule fluorescence resonance energy transfer investiga-tions revealed that negamycin specifically stabilizes near-cognateternary complexes within the A site during the normally transientinitial selection process to promote miscoding. The crystal struc-ture of the 70S ribosome in complex with negamycin, determinedat 3.1 Å resolution, sheds light on this finding by showing thatnegamycin occupies a site that partially overlaps that of tetra-cycline-class antibiotics. Collectively, these data suggest that thesmall subunit head domain contributes to the decoding mecha-nism and that small-molecule binding to this domain may eitherprevent or promote tRNA entry by altering the initial selectionmechanism after codon recognition and before GTPase activation.

translation | ribosome | fidelity | antibiotic | negamycin

Negamycin, a natural product originally isolated from cul-tures of Streptomyces purpeofuscus, exhibits broad-spectrum

antibacterial activity against key pathogens for which clinicaltreatment options are dwindling (1, 2). The chemical structure ofnegamycin, [2-[(3R, 5R)-3,6-diamino-5-hydroxyhexanoyl]-1-methyl-hydrazino]acetic acid, and synthetic routes for its synthesis wereelucidated in the early 1970s (Fig. 1A) (3, 4). Early toxicity studies indogs revealed that daily administration of negamycin led to theformation of N-methylhydrazinoacetic acid, an inhibitor of gluta-mate pyruvate transaminase, an outcome that caused reversiblehepatic coma. Consequently, further clinical studies were not pur-sued (5). Although the antimicrobial activity and efficacy of neg-amycin have since been confirmed, analogs exhibiting an improvedtherapeutic window have yet to be found (6, 7). Progress on thisfront has been hampered, at least in part, by the fact that the mo-lecular mechanisms of negamycin-induced cell growth inhibitionhave yet to be discerned conclusively.Negamycin decreases the viability of Escherichia coli by pref-

erentially targeting protein synthesis (8). Early mechanistic in-vestigations proposed that negamycin may act by inhibitingtranslation initiation (8), decreasing translational fidelity duringthe elongation phase of protein synthesis (9, 10), and disruptingproper translation termination (9, 11–13). Uncertainties regardingnegamycin’s inhibition mechanism, its highly polar physicochemical

properties, and the lack of streptomycin cross-resistance ledsome to propose that it may act through multiple binding sites(9). Crystallographic data led to the speculation that negamycinbinds in the peptide exit tunnel of the Haloarcula marismortuilarge ribosomal subunit (14). This site of interaction, however,was difficult to reconcile with its inhibitory activities given itsdistance (∼100 Å) from known functional centers of the ribo-some located at the interface of the large (50S) and small (30S)subunits. Experiments showing that negamycin interacts withmodel oligonucleotide analogs of the decoding site (15) sim-ilarly have fallen short of clarifying its mechanism of action.Although the molecular basis of negamycin’s activities remainselusive, a mode of action involving the process of messenger RNA(mRNA) decoding is supported by studies showing that negamycincan suppress premature translation termination of the dystrophingene, mdx, in a mouse model for Duchenne muscular dystrophy(16). As ∼30% of human disease mutations stem from the in-troduction of stop codons that cause premature translation termi-nation (17), such investigations suggest that a deeper understandingof negamycin’s mode of action may one day be harnessed fortherapeutic purpose (reviewed in ref. 18).

Significance

The identification of negamycin’s binding site within helix34 of the small subunit head domain and the elucidation of itsmechanism of action during messenger RNA decoding provide aphysical framework for exploring structure–activity relationshipsof this largely unexplored antibiotic class. These findings lay thefoundation for the rational design of improved negamycinanalogs that may one day serve as potent antibacterial agentsin the clinic.

Author contributions: N.B.O., R.B.A., J.H.D.C., E.T.B., and S.C.B. designed research; N.B.O.,R.B.A., J.N., G.S.B., E.C., A.D.F., N.G., J.H., S.L., M.D.M., D.B.P., J.H.D.C., E.T.B., and S.C.B.performed research; N.B.O., R.B.A., G.S.B., E.C., A.D.F., N.G., J.H., M.F.J., S.L., M.D.M., D.B.P.,J.H.D.C., E.T.B., and S.C.B. contributed new reagents/analytic tools; N.B.O., R.B.A., J.N., G.S.B.,E.C., A.D.F., N.G., J.H., M.F.J., S.L., M.D.M., D.B.P., J.H.D.C., E.T.B., and S.C.B. analyzed data;and N.B.O., R.B.A., J.N., A.D.F., J.H.D.C., E.T.B., and S.C.B. wrote the paper.

Conflict of interest statement: As indicated, specific authors of this manuscript are full-time employees of AstraZeneca Corporation. R.B.A. and S.C.B. were partially funded byAstraZeneca to perform these investigations. R.B.A. and S.C.B. have equity interest inLumidyne Technologies.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank (PDB), www.pdb.org (accession nos. 4WAO, 4WAP, 4WAQ and 4WAR).1To whom correspondence may be addressed: Email: scb2005@med.cornell.edu orEd.Buurman@astrazeneca.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1414401111/-/DCSupplemental.

16274–16279 | PNAS | November 18, 2014 | vol. 111 | no. 46 www.pnas.org/cgi/doi/10.1073/pnas.1414401111

Here, through genetic, biophysical, and crystallographic in-vestigations of the bacterial translation apparatus, we showthat negamycin binds and operates through a site distinct from,but proximal to, the decoding site region at the base of the smallsubunit head domain. This site overlaps that of tetracyclineand tigecycline (Fig. 1A), structurally distinct compounds thatspecifically inhibit translation by disrupting the process of ami-noacyl-tRNA (aa-tRNA) selection during decoding (19–21).Through this site, negamycin stabilizes near-cognate aa-tRNAson the ribosome to promote translational stalling and the loss oftranslation fidelity.

ResultsNegamycin (Fig. 1A) was synthesized in a novel sequence (Fig.S1) from an intermediate previously used by Nishiguchi et al.(22). Biochemical and microbiological activity was indistinguish-able from negamycin that was purified as a fermentation product(Table 1). This reagent was used for all subsequent experiments.

Resistance Mutations Map to the Head Domain of the 30S Subunit ofthe Ribosome. In an effort to identify the negamycin binding site onthe E. coli ribosome, mutant cells resistant to negamycin wereisolated from an E. coli strain containing only a single genome-encoded rrn operon (SQ110) (22). Spontaneous resistant mutantswere identified at a frequency of 10−9 when selected at 128 μg/mL,fourfold the minimum inhibitory concentration (MIC) of 32μg/mL (∼128 μM) (Table 1) (23). Whole-genome sequencing ofeight colonies obtained in four independent experiments revealeda unique mutation at position U1052G of 16S rRNA, in imme-diate proximity of helix 34 (h34) (Fig. S2). Similar findings wereobtained when analogous experiments were performed usingE. coli SQ171 (23), a strain lacking genomically encoded rRNAand harboring a single rrnC operon on a low copy number plasmid(pHK-rrnC). Such experiments led to the identification of a sec-ond, distinct resistance mutation, U1060A (Fig. S2) that also isproximal to h34. The U1052G and U1060A mutations conferredan ∼30-fold and ∼16-fold increase in the negamycin MICs, re-spectively (Table 1). Consistent with the significance of this regionto the binding of tetracycline-class antibiotics (24, 25), theU1052G mutation resulted in hypersusceptibility to tetracycline(Table 1), whereas the U1060A mutation caused cross-resistanceto both tetracycline and tigecycline (Table 1).

Collectively, these findings suggest that negamycin resistancemay be conferred by local distortions in the vicinity of h34 thatinfluence the binding of negamycin as well as the tetracycline-class antibiotics (26). In the native, apo-ribosome structure, U1052forms a wobble base pair with G1206 and U1060 forms ca-nonical base-pairing interactions with residue A1197 (Fig. S2)(27). Correspondingly, transversion mutations at either posi-tion may distort helical geometries in this region to affect criticaldrug interactions.

Negamycin Displaces Tetracycline from the Ribosome but Is NotSusceptible to TetM-mediated Resistance. To validate the h34 siteof negamycin binding inferred from these studies, filter bindingdisplacement assays were performed to ascertain whether neg-amycin directly or indirectly competes for radiolabeled tetracy-cline binding to the E. coli 70S ribosome (21, 28). Consistent withthe notion that negamycin and tetracycline-class antibiotics shareoverlapping binding sites, these displacement studies showed thatnegamycin, like tigecycline, could efficiently displace tetracyclinefrom the ribosome (Fig. 1B). In line with their observed bio-chemical and antimicrobial potencies (Table 1), in this assaynegamycin and tigecycline exhibited IC50 values of 64 ± 1.2μM and 1.6 ± 1.1 μM, respectively.

Table 1. Antimicrobial and biochemical activity of negamycin(NMY), tetracycline (TET), tigecycline (TIG), and gentamicin (GEN)

Strain NMY TET TIG GEN

MRE600 (S30 extracts) 1.9 (1.8) 1.4 0.35 0.01ATCC25922 32 (32) 1 0.25 0.5SQ110 (WT) 32 2 0.5 1SQ110 rrnE (U1052G) 1,024 0.12 0.5 1SQ171 pHK-rrnC (WT) 16 2 2 1SQ171 pHK-rrnC (U1052G) 512 0.5 2 0.5SQ171 pHK-rrnC (U1060A) 256 8 8 1

Biochemical activity (IC50, in micromolar; n ≥ 3) was determined by usinga coupled in vitro transcription–translation assay using S30 extracts of strainMRE600. Antimicrobial activity (MIC, in micrograms per milliliter; n ≥ 3) wasdetermined against a wild-type E. coli strain (ATCC25922) and E. coli strainsSQ110 and SQ171 containing only a single copy of the ribosomal DNA operon,including negamycin-resistant derivatives of these strains. Data obtained withNMY fermentation product are indicated in parentheses.

Fig. 1. Negamycin competes with tetracycline and tigecycline for binding to the small ribosomal subunit. (A) Chemical structures, from left to right, ofnegamycin, tetracycline, and tigecycline. (B) [3H] Tetracycline displacement assay with unlabeled competitors. Tigecycline (red, IC50 = 1.6 μM), negamycin(green, IC50 = 64 μM). (C) Assessment of TetM effect on tetracycline, tigecycline, and negamycin. Results of coupled in vitro transcription–translation assaywith TetM at 0 μM (▪), 0.01 μM (♦), and 0.1 μM (●) in the presence of tetracycline (blue), tigecycline (red), or negamycin (green). Error bars represent the SD ofat least two independent experiments.

Olivier et al. PNAS | November 18, 2014 | vol. 111 | no. 46 | 16275

CHEM

ISTR

YBIOCH

EMISTR

Y

TetM is a ribosome protection protein that confers resistanceto tetracycline by binding to the A site and displacing the drugfrom h34 via a GTP hydrolysis-dependent reaction (29, 30).Although a complete understanding presently is lacking, TetMmust dissociate quickly, allowing ternary complex to bind beforedrug rebinding, thus enabling the translation elongation cycle toproceed. Notably, TetM does not confer resistance to tigecycline(21). Given that both drugs bind and recognize overlappingregions within h34 (27, 30, 31), this finding suggests that TetMmay be unable to displace tigecycline effectively because of stericor affinity considerations. In a coupled in vitro transcription–translation assay using S30 extract, we observed negamycin’sinhibition of translation persisting in the presence of up to0.1 μM TetM (see Table S1 for IC50 values) (Fig. 1C). Givennegamycin’s comparatively high IC50 in tetracycline bindingcompetition assays (64 μM) (Fig. 1B), from which a reducedaffinity interaction is inferred, this finding suggests that neg-amycin binds the ribosome at a site near h34 that is physicallydistinct from that of tetracycline and beyond the reach of theC-terminal extension on domain IV of TetM that is believed toenter the A site to displace the drug from the ribosome (29).

Negamycin Specifically Targets the Process of Near-Cognate tRNADecoding. In line with the biochemical potency estimated usingS30 extracts (Table 1) and the inhibitory concentrations reported

by Umezawa and coworkers (10), coupled in vitro transcription–translation assays using recombinant, purified translation com-ponents from E. coli (Materials and Methods) revealed thatnegamycin inhibited luciferase production with an IC50 of ∼1.5 μM.To ascertain negamycin’s impact on elongation cycle reactions, weset out to investigate its molecular basis of action using purifiedtranslation components and single-molecule fluorescence reso-nance energy transfer (smFRET) methods by (i) imaging ribosomedynamics under equilibrium conditions and (ii) imaging functionduring single-turnover pre–steady-state reactions of aa-tRNA se-lection and translocation (19, 20, 32). Such experiments have thebenefit of affording intuitive structural perspectives that facilitateinsights into the mechanisms of antibiotic action (31, 33, 34).Whereas chemically diverse antibiotics have been shown to sub-stantially alter both tRNA motions between their classical andhybrid positions within the aminoacyl (A) and peptidyl (P) sites(34), as well as the process of subunit rotation, negamycin had nodetectable impact on either of these structural processes. Thisobservation suggests that the site of negamycin binding may beremote from the intersubunit bridge elements that coordinatespontaneous and reversible tRNA motions at the subunit interface(33, 35, 36).Direct single-molecule imaging methods enable the process of

tRNA selection to be tracked by time-dependent changes inFRET (19, 20, 37). In such experiments, wide-field total internal

Fig. 2. Pre–steady-state tRNA selection experiments. (A) Schematic of the tRNA selection process imaged using smFRET, in which surface-immobilized ini-tiation complexes bearing tRNAfMet in the P site are monitored as the ternary complex of Phe-tRNAPhe enters the A site. 50S subunits (gray), 30S subunits (lightblue), tRNAfMet (orange), Phe-tRNAPhe (green), EF-Tu (blue), Cy3 (green circle), and Cy5 (red circle) are shown. From left to right: a zero FRET state is observedbefore ternary complex binding; during initial selection, a FRET transition occurs between a low-FRET, codon recognition (CR) state and an intermediate FRET,GTPase-activated (GA) state, in which GTP (yellow circle) is hydrolyzed to GDP (maroon). During proofreading, a FRET transition occurs between the in-termediate-FRET, GA state and the fully accommodated AC state. (B) Representative smFRET trace recording (blue) and idealization (red) of the cognate tRNAselection process, in which each colored rectangle highlights the FRET values corresponding to CR, GA, and AC states. Rectangle placements approximatea 2-SD width from the mean FRET value assigned to each state. (C) Postsynchronized FRET histograms representing all molecules observed to undergo tRNAselection under pre–steady-state measurements of cognate (UUC)- (Left) and near-cognate (UCU)-programmed ribosomes (Right). (D) Postsynchronized FREThistograms showing productive tRNA selection events: those reaching a stable (>120 ms) AC state in the presence of 200 μM negamycin for cognate (UUC)-(Left) and near-cognate (UCU)-programmed ribosomes (Right).

16276 | www.pnas.org/cgi/doi/10.1073/pnas.1414401111 Olivier et al.

reflection fluorescence microscopy is used to monitor the bind-ing of acceptor (Cy5)-labeled Phe-tRNAPhe in ternary complexwith elongation factor tu (EF-Tu) and GTP to surface-immobilized70S initiation complexes bearing a donor (Cy3)-labeled initiatortRNAfMet within the P site (Fig. 2A). In this system, changes inFRET reflect ternary complex binding to the A site and reversibleconformational events and thermal fluctuations that ultimatelydrive the incoming aa-tRNA between codon recognition (CR)and GTPase-activated (GA) states during initial selection (a low-to intermediate-FRET transition) and between the GA and ac-commodated (AC) states during proofreading (an intermediate-to high-FRET transition) (Fig. 2B). In the presence of 200 μMnegamycin—i.e., >100-fold the IC50 for translation inhibition—the mechanism of cognate aa-tRNA selection was unaffected interms of both the bimolecular rate constant of ternary complexbinding and the efficiency and manner with which FRET evolvedduring the selection process (Fig. 2 C and D, Left). Thus, unliketetracycline and tigecycline (19, 21), negamycin’s interactions withthe ribosome do not sterically impinge on the entry of cognate aa-tRNA as it enters the A site.To assess negamycin’s impact on the fidelity mechanism, iden-

tical experiments were performed using ribosomes programmedwith a near-cognate mRNA codon in the A site (UCU). Althoughribosomes efficiently reject near-cognate Phe-tRNAPhe ternarycomplex (Fig. 2C, Right), near-cognate ternary complexes wereobserved to enter the A site in the presence of negamycin (Fig.2D, Right), in line with results obtained with synthetic poly-nucleotides (10). Comparison of population histograms repre-senting the subset of productive binding events observed in theabsence and presence of negamycin revealed that near-cognatetRNAs are stabilized on the ribosome by negamycin in the tran-sition between CR and GA states (Fig. 2D). At saturating neg-amycin concentrations (200 μM), the GA state exhibited an ∼170 ±30-ms lifetime, roughly sixfold longer than observed in the ab-sence of drug (Fig. 3A). We therefore infer that the additionaltime that near-cognate ternary complex persists on the ribosomeserves to increase the probability that productive GA states maybe achieved and irreversible GTP hydrolysis may occur (19, 20).In so doing, the miscoding frequency is increased commensu-rately (Fig. 3B). Here, the effective concentrations at which a 50%amplitude was observed (EC50s) for both effects on the selection

mechanism, estimated by fitting the observed data to Hill functions,were ∼10 μM (Fig. 3). By contrast, tRNA dynamics within pre-translocation ribosome complexes (both cognate and near-cognate)were indistinguishable in both the absence and presence of neg-amycin (data not shown). As expected from our resistance studies(Table 1), negamycin’s impact on the tRNA selection process wassignificantly suppressed by the U1052G resistance mutation (Fig.3 A and B). Hence, these data suggest that negamycin specificallyaffects the tRNA selection mechanism on the E. coli ribosome,acting to steer the near-cognate ternary complex toward the pro-ductive GTPase-activation corridor and preventing off-pathwayexcursions that normally lead to the rejection of near-cognate ter-nary complexes during initial selection. The combined effects oftranslational stalling and miscoding at each step of the elongationcycle likely explain the increased potency of negamycin observed incoupled in vitro transcription–translation reactions (Table 1). Ininvestigations using a Ser199 (TCT) to Phe199 (TTT) luciferasemutant with the goal of showing gain of function caused bynegamycin-induced miscoding, we have found no evidence of sucheffect. We interpret the absence of gain-of-function effects in ourstudies to suggest that both translational stalling-induced inhibitionof luciferase expression and concomitant miscoding of bona fidecodons elsewhere in the transcript are functionally dominant to thenegamycin-induced miscoding effects at the single, mutated codon.Consistent with the notion that negamycin, tetracycline, and

tigecycline share structurally related sites of interaction with theribosome, both tetracycline and tigecycline operate by preventingcognate and near-cognate tRNA from making the CR-to-GAtransition during the tRNA selection process (19, 21). Thesmaller tetracycline molecule has been shown to be somewhatless effective at blocking this step of tRNA entry while modestlyincreasing the lifetime of both cognate and near-cognate tRNA

Fig. 3. Negamycin-induced translational stalling and miscoding. (A) Thelifetime of the CR and GA states observed for wild-type (blue ▲ and red ●,respectively) and U1052G-mutant E. coli ribosomes (gray ▼ and black ▪,respectively) observed when tRNA selection experiments were performed onnear-cognate (UCU) programmed ribosomes in the presence of a Cy5-labeledPhe-tRNAPhe ternary complex (10 nM) over a range of negamycin concen-trations. (B) The fraction of wild-type (red ▪) and U1052G-mutant ribosomesprogrammed with the near-cognate (UCU) codon in the A site observed tosuccessfully complete the selection process after incubation of surface-immobilized 70S initiation complexes bearing Cy3-labeled tRNAfMet in the Psite with 10 nM Cy5-labeled Phe-tRNAPhe ternary complex for 2 min overa range of negamycin concentrations. Data were fit using the Hill equation,where n, the number of binding sites, was set to unity. Error bars representthe SDs obtained from three independent experiments.

Fig. 4. Negamycin binds the small subunit at the base of helix 34. (A)Difference electron density (Fobs-Fcalc) map contoured at 3σ to negamycinin the 30S ribosomal subunit of E. coli. (B) Binding site of negamycin nearhelix 34 of the 30S ribosome of E. coli. The mesh represents a 2Fobs-Fcalcelectron density map contoured at 1σ. (C ) Global view of the 30S subunitof the ribosome showing negamycin’s site of interaction immediatelyproximal to h34 at the base of the head domain. (D) Schematic modelshowing the position of negamycin with respect to the mRNA codon–tRNAanticodon interaction within the A site. The model was created by su-perposition of the negamycin-bound ribosome with Protein Data Bank(PDB) ID code 2wrq, which contains ternary complex stalled in an in-termediate (A/T) state of tRNA selection by the nonhydrolyzable GTP an-alog, GDPNP. The anticodon stem loop of the A/T tRNA (yellow), the A andP site codons of mRNA (blue), and the decoding site residues A1492, A1493within 16S rRNA (red) are shown.

Olivier et al. PNAS | November 18, 2014 | vol. 111 | no. 46 | 16277

CHEM

ISTR

YBIOCH

EMISTR

Y

within the CR state (19). Tigecycline’s strict prevention of aa-tRNA progression beyond the low-FRET, CR state has beenattributed to its extended, ring D structure, which serves to ste-rically occlude tRNA entry and formation of the GA state (21).Thus, although negamycin likely occupies a site that partiallyoverlaps that of tetracycline and tigecycline, where it directly orindirectly influences the fidelity of initial selection, the obser-vation that later events in the selection process appear un-affected by negamycin suggests that its position is compatiblewith A site occupancy. In line with this notion, no perceptibleimpact was observed on tRNA motions or subunit rotationprocesses within the pretranslocation complex or the process ofelongation factor G (EF-G)-catalyzed translocation in the pres-ence of 200 μM negamycin (data not shown). Collectively, thesedata suggest that negamycin acts to specifically influence structuraland/or conformational processes within the head domain of thesmall subunit that facilitate productive tRNA selection, whichnormally are achieved inefficiently when the A site is occupied bya near-cognate ternary complex.

Crystal Structure of E. coli 70S Ribosome in Complex with Negamycin.To obtain direct structural insight on the nature of the negamycin–ribosome interaction, crystals containing two apo E. coli 70Sribosomes per asymmetric unit (27) were transferred sequentiallyin stabilization solutions with 1 mM negamycin. Diffraction datafrom these crystals were collected to ∼3.1 Å, and the structure wasdetermined by molecular replacement (Table S2). Globally, thearchitecture of both 70S ribosomes within the asymmetric unit wasunaltered by the presence of negamycin (27, 38).In contrast to the structure of the archaeal 50S subunit in

complex with negamycin (14), no evidence was found for dif-ference electron density in the vicinity of the exit tunnel, even ata signal-to-noise cutoff of 2.5σ. Positive difference density,however, was identified through a broader search of the “clas-sical”-state ribosome in the crystallographic asymmetric unit(Fig. 4A) (27). This unexplained difference density was posi-tioned next to the phosphate backbone of nucleotides C1054,A1197, and G1198 proximal to h34 (Fig. 4B) at the base of thesmall subunit head domain (Fig. 4C). This binding site directlyoverlaps the previously described binding sites for tetracyclineand tigecycline (Fig. S3) (21, 26, 39, 40). Although weak positivedifference density was observed in the “intermediate”-state ri-bosome in the crystallographic asymmetric unit (27), negamycincould be placed only in the classical-state ribosome.Given the present resolution of our crystallographic data, un-

ambiguous placement of negamycin into the electron density mapwas not possible, and a variety of binding modes were consideredcarefully. Placement was complicated by the lower quality of theelectron density map of the classical-state ribosome in this crystalform compared with the other ribosome in the asymmetric unit,which is in an intermediate state of rotation (27). The chemicalstructure of negamycin inherently allows this molecule to assumea large degree of rotational freedom. However, in the mostplausible model of negamycin binding based on structure–activityrelationships (Table S3), the presence of two Mg2+ ions withinthe binding site restricts the bound conformation of negamycinand serves to increase the number of interactions with thephosphate backbone of the ribosome (Fig. 4A). The carboxylatemoiety of negamycin is positioned near the 2′ hydroxyl of A1196and a phosphate oxygen of C1054, thereby placing the base ofC1054 above the carboxylate group. The C6 carbonyl group isdirected to a bound Mg2+ ion, and the C4 amino group extendstoward the phosphate backbone of A1197. The C2 hydroxyl ispointed toward the phosphate backbone of G966, and the ter-minal amino group of negamycin is positioned in the solvent. Thisbinding mode is supported by the structure–activity relationshipsof negamycin derivatives assessed through coupled in vitro tran-scription–translation assays, which showed that modification of

the carboxylate moiety completely abolishes activity, whereas theaddition of a methyl, ethyl, or benzyl group to the amino sub-stituent has no impact on activity (Table S3).

DiscussionThe data presented reveal that the broad-spectrum antibioticnegamycin interacts with the h34 element of 16S rRNA withinthe head domain of the small ribosomal subunit. This site ofinteraction partially overlaps the established tetracycline andtigecycline binding sites. Negamycin’s mode of action on thetranslation mechanism through this site, verified by both re-sistance mutations in this same region and tetracycline bind-ing competition assays (Fig. 1), was shown through functionalinvestigations to specifically affect the fidelity of tRNA selection.Here, negamycin was observed to stabilize near-cognate ternarycomplex within the A site during the initial selection process.This prolonged period of occupation increased the frequency ofmiscoding by approximately sixfold (Fig. 3B). These findingsshed light on negamycin’s reported propensities to promote ri-bosome stalling, polysome stabilization, and miscoding, resultingin rapid cell death (8–13). Together with the known effects ofboth tetracycline and tigecycline, these findings provide compellingevidence that the corridor between the head domain of the smallsubunit and the aa-tRNA anticodon contribute to the steering ofternary complex into the A site during initial selection (Fig. 4D).These insights reveal the chimeric nature of the h34 binding

site for negamycin and the tetracycline-class antibiotics on theribosome. Although substantial improvements in tetracyclinehave been achieved in this class that increase binding affinity andreduce the effectiveness of resistance mechanisms such as TetM,presently little is known regarding the now-apparent potential ofsmall peptides to affect the fidelity mechanism. As negamycindoes not appear to affect the process of cognate tRNA selection,the inhibitory potency of negamycin derivatives or other com-pounds operating through this site might be improved sub-stantially. The development of negamycin derivatives also mayprovide a therapeutic avenue for specifically targeting tRNAmisincorporation at stop codons on the human ribosome. Thecrystallographic system described here, in conjunction withmechanistic smFRET studies, will allow monitoring of the es-tablishment of new small-molecule–ribosome interactions andthus support an iterative, rational chemistry program.

Materials and MethodsRibosome Purification and Crystallization. E. coli 70S ribosomes lacking S1protein, used in all crystallization experiments, were derived from MRE600and purified as previously described (41) with the exception of a combina-tion of 2 mM DTT and 2 mM tris(2-carboxyethyl)phosphine (TCEP) instead of4 mM 2-mercaptoethanol as the reducing agent. Ribosomes were crystallized aspreviously described (27). Trays of ribosome crystals were transferred to 4 °C forincubation overnight, with all subsequent harvesting steps conducted at 4 °C.Ribosome crystals then were stabilized by brief soaks in crystallization solutionssupplemented with 6–18% (vol/vol) PEG 400 in a total of seven steps. To formthe negamycin-bound complex, an additional soaking step was added by usingcrystallization solution supplemented with 24% (vol/vol) PEG 400, 0.5–1 mMZnCl2, and 1–2 mM negamycin, and the crystals remained in this solution for16–72 h before rapid cooling by direct immersion into liquid nitrogen.

X-Ray Data Collection, Reduction, Structure Determination, and Refinement.Diffraction data were measured at the Advanced Photon Source onbeamline 17-ID (IMCA-CAT) using the Pilatus 6M detector from vitrifiedcrystals under cryogenic conditions using 0.1° oscillations and exposuretimes of 0.15 s per frame with beam attenuation of 40%. Data wereindexed with XDS (42) and scaled using Aimless (43) as scripted in autoPROCfrom Global Phasing (44). Structures were determined by molecular re-placement using coordinates of the E. coli 70S ribosome in the apo state(27) as the starting model. Model building and compound placement wereconducted with Coot (45), and structure refinement was carried out withautoBUSTER (46) or PHENIX (47). See SI Materials and Methods for a moredetailed description.

16278 | www.pnas.org/cgi/doi/10.1073/pnas.1414401111 Olivier et al.

ACKNOWLEDGMENTS. The authors thank Dr. Vonrhein, Dr. Andrew Sharff,and Dr. Bricogne, as well as the entire staff of Global Phasing for their support.We also thank Drs. Inoue, Igarashi, and Takahashi (Institute of MicrobialChemistry, Tokyo) for their donation of negamycin fermentation product andtesting of our material against their panel of bacterial strains. We thank Dr. ZhouZhou, supported by GM098859 (to S.C.B.), and Dr. David Warren, director of theMilstein Chemistry Core Facility at Weill Cornell, for the synthesis of photo-stabilized, soluble fluorophore derivatives used in single-molecule experiments.

Negamycin was synthesized chemically in collaboration with the PharmaronCorporation. N.B.O., G.S.B., E.C., A.D.F., N.G., J.H., S.L., M.D.M., D.B.P., and E.T.B.are employed by AstraZeneca Pharmaceuticals. R.B.A. was supported by fundingfrom AstraZeneca. J.N. was funded by a Human Frontiers in Science ProgramLong-Term Postdoctoral Fellowship. J.H.D.C. was supported by NIH GrantGM65050. S.C.B. was supported by NIH Grants GM079238 and GM098859.M.F.J. was supported by a postdoctoral fellowship from the GermanAcademic Exchange Service (DAAD) and NIH Grant GM079238.

1. Hamada M, Takeuchi T, Kondo S, Ikeda Y, Naganawa H (1970) A new antibiotic,negamycin. J Antibiot (Tokyo) 23(3):170–171.

2. Rice LB (2008) Federal funding for the study of antimicrobial resistance in nosocomialpathogens: no ESKAPE. J Infect Dis 197(8):1079–1081.

3. Kondo S, Shibahara S, Takahashi S, Maeda K, Umezawa H (1971) Negamycin, a novelhydrazide antibiotic. J Am Chem Soc 93(23):6305–6306.

4. Shibahara S, Kondo S, Maeda K, Umezawa H, Ono M (1972) The total syntheses ofnegamycin and the antipode. J Am Chem Soc 94(12):4353–4354.

5. Maeda KO, Ohno M (1979) Chemical studies on antibiotics and other bioactive mi-crobial products by Prof. Hamao Umezawa. Heterocycles 13(1):49–78.

6. Raju B, et al. (2003) N- and C-terminal modifications of negamycin. Bioorg Med ChemLett 13(14):2413–2418.

7. Raju B, et al. (2004) Conformationally restricted analogs of deoxynegamycin. BioorgMed Chem Lett 14(12):3103–3107.

8. Mizuno S, Nitta K, Umezawa H (1970) Mechanism of action of negamycin in Escher-ichia coli K12. I. Inhibition of initiation of protein synthesis. J Antibiot (Tokyo) 23(12):581–588.

9. Uehara Y, Kondo S, Umezawa H, Suzukake K, Hori M (1972) Negamycin, a miscodingantibiotic with a unique structure. J Antibiot (Tokyo) 25(11):685–688.

10. Mizuno S, Nitta K, Umezawa H (1970) Mechanism of action of negamycin in Escher-ichia coli K12. II. Miscoding activity in polypeptide synthesis directed by snytheticpolynucleotide. J Antibiot (Tokyo) 23(12):589–594.

11. Uehara Y, Hori M, Umezawa H (1974) Negamycin inhibits termination of proteinsynthesis directed by phage f2 RNA in vitro. Biochim Biophys Acta 374(1):82–95.

12. Uehara Y, Hori M, Umezawa H (1976) Inhibitory effect of negamycin on polysomalribosomes of Escherichia coli. Biochim Biophys Acta 447(4):406–412.

13. Uehara Y, Hori M, Umezawa H (1976) Specific inhibition of the termination process ofprotein synthesis by negamycin. Biochim Biophys Acta 442(2):251–262.

14. Schroeder SJ, Blaha G, Moore PB (2007) Negamycin binds to the wall of the nascentchain exit tunnel of the 50S ribosomal subunit. Antimicrob Agents Chemother 51(12):4462–4465.

15. Xie Y, Dix AV, Tor Y (2010) Antibiotic selectivity for prokaryotic vs. eukaryoticdecoding sites. Chem Commun (Camb) 46(30):5542–5544.

16. Arakawa M, et al. (2003) Negamycin restores dystrophin expression in skeletal andcardiac muscles of mdx mice. J Biochem 134(5):751–758.

17. Mendell JT, Dietz HC (2001) When the message goes awry: Disease-producing mu-tations that influence mRNA content and performance. Cell 107(4):411–414.

18. Lee HL, Dougherty JP (2012) Pharmaceutical therapies to recode nonsense mutationsin inherited diseases. Pharmacol Ther 136(2):227–266.

19. Blanchard SC, Gonzalez RL, Kim HD, Chu S, Puglisi JD (2004) tRNA selection and ki-netic proofreading in translation. Nat Struct Mol Biol 11(10):1008–1014.

20. Geggier P, et al. (2010) Conformational sampling of aminoacyl-tRNA during selectionon the bacterial ribosome. J Mol Biol 399(4):576–595.

21. Jenner L, et al. (2013) Structural basis for potent inhibitory activity of the antibiotictigecycline during protein synthesis. Proc Natl Acad Sci USA 110(10):3812–3816.

22. Nishiguchi S, et al. (2010) Total synthesis of (+)-negamycin and its 5-epi-derivative.Tetrahedron 66(1):314–320.

23. Orelle C, et al. (2013) Tools for characterizing bacterial protein synthesis inhibitors.Antimicrob Agents Chemother 57(12):5994–6004.

24. Moazed D, Noller HF (1987) Interaction of antibiotics with functional sites in 16Sribosomal RNA. Nature 327(6121):389–394.

25. Connell SR, et al. (2002) The tetracycline resistance protein Tet(o) perturbs the con-formation of the ribosomal decoding centre. Mol Microbiol 45(6):1463–1472.

26. Brodersen DE, et al. (2000) The structural basis for the action of the antibioticstetracycline, pactamycin, and hygromycin B on the 30S ribosomal subunit. Cell 103(7):1143–1154.

27. Zhang W, Dunkle JA, Cate JH (2009) Structures of the ribosome in intermediate statesof ratcheting. Science 325(5943):1014–1017.

28. Grossman TH, et al. (2012) Target- and resistance-based mechanistic studies withTP-434, a novel fluorocycline antibiotic. Antimicrob Agents Chemother 56(5):2559–2564.

29. Dönhöfer A, et al. (2012) Structural basis for TetM-mediated tetracycline resistance.Proc Natl Acad Sci USA 109(42):16900–16905.

30. Burdett V (1986) Streptococcal tetracycline resistance mediated at the level of proteinsynthesis. J Bacteriol 165(2):564–569.

31. Wang L, et al. (2012) Allosteric control of the ribosome by small-molecule antibiotics.Nat Struct Mol Biol 19(9):957–963.

32. Munro JB, Wasserman MR, Altman RB, Wang L, Blanchard SC (2010) Correlatedconformational events in EF-G and the ribosome regulate translocation. Nat StructMol Biol 17(12):1470–1477.

33. Wang L, Wasserman MR, Feldman MB, Altman RB, Blanchard SC (2011) Mechanisticinsights into antibiotic action on the ribosome through single-molecule fluorescenceimaging. Ann N Y Acad Sci 1241:E1–E16.

34. Feldman MB, Terry DS, Altman RB, Blanchard SC (2010) Aminoglycoside activity ob-served on single pre-translocation ribosome complexes. Nat Chem Biol 6(1):54–62.

35. Moore PB (2012) How should we think about the ribosome? Annu Rev Biophys 41:1–19.

36. Chen J, Tsai A, O’Leary SE, Petrov A, Puglisi JD (2012) Unraveling the dynamics ofribosome translocation. Curr Opin Struct Biol 22(6):804–814.

37. Blanchard SC, Kim HD, Gonzalez RL, Jr, Puglisi JD, Chu S (2004) tRNA dynamics on theribosome during translation. Proc Natl Acad Sci USA 101(35):12893–12898.

38. Dunkle JA, Xiong L, Mankin AS, Cate JH (2010) Structures of the Escherichia coli ri-bosome with antibiotics bound near the peptidyl transferase center explain spectra ofdrug action. Proc Natl Acad Sci USA 107(40):17152–17157.

39. Pioletti M, et al. (2001) Crystal structures of complexes of the small ribosomal subunitwith tetracycline, edeine and IF3. EMBO J 20(8):1829–1839.

40. Murray JB, et al. (2006) Interactions of designer antibiotics and the bacterial ribo-somal aminoacyl-tRNA site. Chem Biol 13(2):129–138.

41. Blaha G, et al. (2000) Preparation of functional ribosomal complexes and effect ofbuffer conditions on tRNA positions observed by cryoelectron microscopy. MethodsEnzymol 317:292–309.

42. Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66(Pt 2):125–132.43. Evans PR, Murshudov GN (2013) How good are my data and what is the resolution?

Acta Crystallogr D Biol Crystallogr 69(Pt 7):1204–1214.44. Vonrhein C, et al. (2011) Data processing and analysis with the autoPROC toolbox.

Acta Crystallogr D Biol Crystallogr 67(Pt 4):293–302.45. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot.

Acta Crystallogr D Biol Crystallogr 66(Pt 4):486–501.46. Bricogne G, et al. (2011) Buster V2.11.5 (Global Phasing Ltd., Cambridge, UK).47. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macro-

molecular structure solution. Acta Crystallogr D Biol Crystallogr 66(Pt 2):213–221.

Olivier et al. PNAS | November 18, 2014 | vol. 111 | no. 46 | 16279

CHEM

ISTR

YBIOCH

EMISTR

Y