Structural basis and dynamics of multidrug recognition PNAS PLUS in a minimal bacterial multidrug resistance system

Judith Habazettla, Martin Allana,1, Pernille Rose Jensena,2, Hans-Jürgen Sassa, Charles J. Thompsonb, and Stephan Grzesieka,3

aFocal Area Structural Biology and Biophysics, Biozentrum, University of Basel, CH-4056 Basel, Switzerland; and bDepartment of Microbiology and Immunology, Life Sciences Center, University of British Columbia, Vancouver, British Columbia V6T 1Z3, Canada

Edited by Adriaan Bax, National Institutes of Health, Bethesda, MD, and approved November 11, 2014 (received for review June 27, 2014) TipA is a transcriptional regulator found in diverse bacteria. It con- induces folding of a larger, intrinsically unstructured part of stitutes a minimal autoregulated multidrug resistance system the (24). By this mechanism, TipA recognizes a variety of against numerous thiopeptide antibiotics. Here we report the thiopeptide antibiotics such as thiostrepton, nosiheptide, or structures of its drug-binding domain TipAS in complexes with promothiocin A, and confers resistance via up-regulation of the promothiocin A and nosiheptide, and a model of the thiostrepton tipA and possibly other MDR systems (23–25). The thio- complex. Drug binding induces a large transition from a partially peptide antibiotics induce expression of the tipA gene as two unfolded to a globin-like structure. The structures rationalize the alternate in-frame products: a long protein, TipAL mechanism of promiscuous, yet specific, drug recognition: (i)a (253 aa), and a short protein, TipAS (144 aa), which also con- four-ring motif present in all known TipA-inducing antibiotics is stitutes the C-terminal part of TipAL (24, 26). TipAL belongs to recognized specifically by conserved TipAS amino acids; and (ii) the MerR family of stress response regulators, characterized by the variable part of the antibiotic is accommodated within a flex- their homologous N-terminal DNA-recognition domains; how- ible cleft that rigidifies upon drug binding. Remarkably, the iden- ever, its mechanism of promoter activation has distinctive char- tified four-ring motif is also the major interacting part of the acteristics (27). Their highly diverse C-terminal domains (19, 20, antibiotic with the . Hence the TipA multidrug resistance 28) recognize a wide variety of ligands ranging from divalent mechanism is directed against the same chemical motif that inhibits metal ions to large antibiotics (29). The C-terminal TipAS do- protein synthesis. The observed identity of chemical motifs respon- main of the TipA recognizes thiopeptides and defines BIOCHEMISTRY sible for antibiotic function and resistance may be a general princi- a large subfamily within the MerR proteins (24). The tipA gene is ple and could help to better define new leads for antibiotics. often found in Streptomyces strains that do not produce thio- peptides, whereas thiopeptide-producing strains rather carry the thiopeptides | antibiotic recognition | transcriptional regulation | thiostrepton-resistance gene tsr (30), which provides stronger protein dynamics | solution NMR thiostrepton protection via ribosome methylation (31). Cur- rently, the TipAS subfamily in the Pfam database (32) includes ultidrug resistance (MDR) systems that respond to and 1,938 proteins widely distributed over many common pathogenic Minactivate cytotoxic compounds with diverse structures and and environmental bacteria (Fig. S1A). A phylogenetic tree of targets are found in most forms of life (1, 2). These mechanisms these bacteria reveals homologs in Firmicutes (1,321 proteins are usually active against a large variety of chemically and struc- in 969 species), Actinobacteria (252 proteins in 205 species), turally diverse compounds while still being selective for certain substance classes, such as heterocyclic, hydrophobic compounds Significance with specific hydrogen bond acceptor motifs (3). Structural in- formation, which would provide a rationale for this polyspecific Multidrug recognition is an important phenomenon that is not recognition is scarce, as few transporter structures with bound well understood. TipA, a bacterial transcriptional regulator, substrates have been solved with limited resolution (4, 5). constitutes a minimal multidrug resistance system against nu- The most well-known MDR mechanism is active transport by merous thiopeptide antibiotics. We show that motions in the efflux pumps, like ABC (6–11), RND, SMR, and MFS (4) efflux millisecond to microsecond time range form the basis of the pumps, which confer resistance to numerous anticancer (5) and TipA multidrug recognition mechanism. This may be common antimicrobial drugs (12). However, cytoplasmic proteins can also to many multidrug recognition systems. The discovery that the sequester and thereby inactivate toxic compounds such as bleo- structural antibiotic motifs essential for binding to TipA and to mycin, β-lactams, and fusidic acid (13–16) or drugs such as cis- the ribosome are identical makes the multidrug recognition platinum, melphalan, and chlorambucil (17). The expression of mechanism of TipA a useful model for ribosomal thiopeptide MDR transporters is often regulated by transcription factors, binding and current antibiotic drug development. which sense the presence of the substrates by direct interaction and have a similar spectrum of multidrug recognition (18). This over- Author contributions: J.H., C.J.T., and S.G. designed research; J.H., M.A., and P.R.J. per- lap of recognition specificity by regulatory and effector gene part- formed research; H.-J.S. contributed new reagents/analytic tools; J.H., M.A., P.R.J., and S.G. analyzed data; and J.H. and S.G. wrote the paper. ners also favors the evolution of MDR systems. High-resolution structures of such transcription factors as complexes with various The authors declare no conflict of interest. substrates (19–22) reveal structural motifs, which are assumed to This article is a PNAS Direct Submission. be similar to those found in MDR transporters (4, 10). Multidrug Data deposition: The NMR chemical shifts have been deposited in the BioMagResBank, www.bmrb.wisc.edu [accession nos. 19421 (TipAS·promothiocin A) and 19422 (TipAS· recognition is often achieved in a large hydrophobic pocket with nosiheptide)]. The atomic coordinates and structure factors have been deposited in the multiple binding possibilities, but two-site binding is also observed, Protein Data Bank, www.pdb.org [PDB ID codes 2MBZ (TipAS·promothiocin A) and whereby specific recognition of a part of the ligand occurs in a 2MC0 (TipAS·nosiheptide)]. rigid slot connected to a larger vestibule that accommodates the 1Present address: Harvard Medical School, Boston, MA 02115. variable ligand remainder (10). 2Present address: Albeda Research, 1799 Copenhagen, Denmark. The bacterial transcriptional activator TipA (thiostrepton 3To whom correspondence should be addressed. Email: [email protected]. Streptomyces lividans induced protein A) of (23) exhibits a nov- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. el multidrug recognition mechanism, in which drug binding 1073/pnas.1412070111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1412070111 PNAS Early Edition | 1of10 Downloaded by guest on October 8, 2021 Proteobacteria (176 proteins in 145 species), and other phyla and dynamics of TipAS complexes with the antibiotics promo- (Chloroflexi, Bacteroidetes, Cyanobacteria, Acidobacteria, and thiocin A and nosiheptide as well as a high-resolution model of Thermotogae). the TipAS·thiostrepton complex. These data give unique insights Antibiotics belonging to the thiostrepton class react with into multidrug recognition by a complete MDR system: (i) TipAS or TipAL to form a covalent bond between a dehy- specific recognition of a conserved antibiotic chemical motif is droalanine residue of the antibiotic and -214 of the achieved by conserved TipAS amino acids and induces folding of protein (33). Antibiotic binding to TipAL increases the affinity the unstructured apo N terminus; and (ii) the variable part of the to its operator site and stabilizes binding of the RNA polymerase antibiotic is accommodated in the large, flexible hydrophobic to the ptipA promoter (27). TipAL binds DNA as a dimer and cleft, which rigidifies upon binding. This mechanism allows presumably, by analogy to MerR (28, 29, 34), might activate promiscuous, yet specific recognition of the entire class of TipA- transcription of tipA by twisting the DNA. In S. lividans cultures inducing thiostrepton antibiotics. The identified recognition induced with thiostrepton, TipAS is expressed in large molar motif is also the major interacting part of thiostrepton and excess (>20:1) over TipAL (26). The sequestration of the drug by nosiheptide in complexes with the D. radiodurans ribosome (38). TipAS constitutes a resistance mechanism (25), and the drug- Hence, the tipA multidrug resistance mechanism is directed induced TipAL activation provides a feedback loop for tipA against the same motif, which is responsible for the antibiotic expression. Thus, TipAL/TipAS represents a minimal autoregu- inhibition of protein synthesis. This structural determinant might lated MDR system (Fig. 1) that responds to and inactivates mul- serve as a scaffold for the development of new antibiotics. tiple thiopeptides. The wide distribution of TipA homologs suggests that the TipAS family provides similar MDR functions Results and Discussion against ubiquitous, related metabolites in diverse bacterial phyla. NMR Analysis and Structure Determination of TipAS Antibiotic Thiopeptide antibiotics are sulfur-containing, highly modified Complexes. The N-terminal region (residues M110–D164) of macrocyclic peptides (35) that target the bacterial ribosome at apo TipAS is largely unstructured and highly mobile (24). This the interface between the 23S rRNA and the L11 ribosomal leads to broadening of the respective resonances and protein (36–39). This interaction inhibits translation and also crowding in the central, random coil region of the apo 1H-15N- synthesis of the global gene expression regulator (p)ppGpp (40, HSQC (heteronuclear single quantum coherence) spectrum (Fig. 41). Furthermore, thiostrepton inhibits the proteasome in eu- 2). In contrast, binding of thiostrepton (Fig. S2), promothiocin A karyotic cells (42, 43) exerting antimalarial (44, 45) and anti- (Fig. 2), and nosiheptide (Fig. 2) induces large changes toward cancer (42–47) activities. Despite their unique mechanism of a more dispersed spectrum, indicative of a transition to a folded action and their high potency against pathogens such as methi- N-terminal structure. Nevertheless, the TipAS·thiostrepton cillin-resistant Staphylococcus aureus and Enterococcus faecium spectrum still contains many overlapping, broadened reso- or penicillin-resistant Streptococcus pneumonia (48), the clinical nances resulting from conformational heterogeneity, which application of thiopeptides is impeded by their considerable size precluded the determination of a high-resolution structure and low water solubility (35, 48–51). However, recent progress in based on the NMR data. In contrast, heterogeneity is much thiopeptide chemical synthesis (43, 52), biosynthesis (53, 54), and reduced in the promothiocin A and nosiheptide complexes, activity screens (55) may enable the development of new thio- such that a full structure determination was possible. Protein 1H, peptide derivatives with improved pharmaceutical properties. 13C, and 15N resonances of both complexes could be assigned by A previous NMR analysis (24) has shown that apo TipAS standard methods. Because of the low proton density of thio- consists of an unfolded N-terminal region of approximately 50 aa peptides, the assignment of the 1H resonances of the unlabeled and a C-terminal, globin-like α-helical structure with a deep antibiotics was challenging, but finally achieved by a combina- hydrophobic, antibiotic binding cleft [Protein Data Bank (PDB) tion of 15N-/13C-filtered and -edited NOESY and J-correlation ID code 1NY9]. Here we report the NMR solution structures spectra (56). Despite their higher quality, structural heterogeneity is also observed in the spectra of the promothiocin A and nosiheptide TipAS complexes. In both cases, residues M110–S154 at the N TipAS Thiostrepton-like terminus as well as N203 and Y205 in the loop between helices antibiotic TipAS TipAS α10 and α11 that contacts the N terminus (as detailed later) show TipAS a second set of slightly shifted resonances indicative (Fig. 2) of TipAS TipAS a minor (30–50%) population with almost identical structure. This was confirmed by a detailed analysis of NOESY spectra. TipAS TipAS Similarly, the C-terminal region (A248–P253) of both complexes TipAS TipAS has a second conformation of ∼21% population that could be TipAS TipAS traced to the peptide bond cis-trans equilibrium of residue P253. TipAL The formation of a covalent bond between the dehydroalanine of T TH TipAN H H H promothiocin A and the sulfur atom of residue C214 is evident T TH 13 β H H H from a strong shift of the C214 C resonance (26.6 to 36.2 ppm) TH H binds in the TipAS·promothiocin A complex, whereas this resonance was unobservable in the TipAS·nosheptide complex as a result of induces conformational exchange. Transcription of tipA ptipA Structure determinations for both complexes were carried out for their main conformations. The large number of measured Fig. 1. The TipA antibiotic resistance mechanism. Thiostrepton-like antibiotics structural parameters such as 74 (131) intermolecular NOEs induce the expression of the tipA gene as two in-frame translation products: (i) (Fig. S3) and 157 (222) RDCs for the TipAS·promothiocin A TipAL, consisting of the DNA-binding domain TipAN and the antibiotic binding (TipAS·nosiheptide) complex resulted in highly defined structures domain TipAS; and (ii) TipAS alone, produced in 20-fold excess. The antibiotics irreversibly bind to the TipAS domain of both forms by a covalent bond and (Fig. 3 and Table S1). The heavy atom coordinate rmsds are 0.55 Å induce folding of its N-terminal 50 aa. Binding to single-domain TipAS neu- (0.48 Å) for the ordered region of the protein backbone and 0.32 Å tralizes the antibiotics. Antibiotic binding to TipAL induces a conformational (0.44 Å) for the antibiotic in the promothiocin A (nosiheptide) switch that leads to tipA expression via a series of events including increased complex, respectively. The slightly lower definition of the nosi- binding to the promoter and recruitment of RNA polymerase. heptide antibiotic very likely results from the increased number

2of10 | www.pnas.org/cgi/doi/10.1073/pnas.1412070111 Habazettl et al. Downloaded by guest on October 8, 2021 PNAS PLUS G188 G188 apo promothiocin A G188 nosiheptide 105 V135 E121 D125 A176 T152 V249 M211 R173 G181 D185 G181 D185 T213 T152 G181 D185 D195 E169 I229 T213 S154 T213 A246 A201 S154 G199 G199 G140 V220 T142 V220 G199 V220 110 G140 T142

Y205 N203 Y205 G235 G124 4 G124 N203 G235 G235 G208 V122 S186 S186 G208 S186 G208 T115 W139 G111 R138 T115 T252 G111 Q198 Y131 S221 V122 115 F123 Q198 R138 R138 E217 Q198 Y219 R227 S221 R136 S148 S221 E217 S148 Y219 R227 R202 N228 T156 E158 Y219 A232 E158 R202 R250 N141 R202 K233 T156 T226 A232 K233 M240 T226 E217 H251 K233 M240 T226 R227 R250 N203 M240 F123 N141 Y209 F225 E117 H251 N228 Y209 F225 R250 Q20 15 A232 M190 Y145 W160 E117 R224 N228 N N247 M190 N247 Y209 D206 F225 M190 D206 H251 R224 A238 K119 N247 Y145 V135W160 F120 R136 D127 Q161 R241 A238 E150 K119 T171 R224 F120 R136 Q161 R241 A238 T252 T171 V135 T252 D230 H204 M178 D129 T252 D230 M178 Q147 R172 M178 Q130 Q147 K151 H204 E121 V175 F174 E121 E150 V175 R172 D230 V175 F174 D242 T171 D242 R146 E132 D191 F174 D191 I163 E182 D159 E133 R172 R241 D242 E182 H212 K151 L114 D191 I112 E169 I163 A180 E210 A180 E210 A176 E210 H212 M211 E182 L114 D159 120 H212 V249 D168 A144V249 A180 E137 V249 R162 D168 E166 A193 M211E118 L170 E132 A176 A193 F126 R173 R162 I244 D168 A193 L170 C214 L170 K157 I244 E166 L177 K157 R173 A201 D179 D195 Y239 D179 Y239 D179 Q164 D195 L177 D165 L245 L177 C214 D165 L245 D195 A246 Y155 E166 A246Y239 Y155 E187 I112 E187 Q164 E169 A201 F126 E187 D165 L245 L215 E194 Q164 M218 I112 E194 E118 F126 A201 A231 E194 I229 L215 E134 D222 N113 C214 A144 M218 R197 L215 D222 R197 D143 I229 E134 I200 A153 D222 I200 K149 I200 A231 R197 A167 A231 L114 N113 M218 N134 A167 A192 A192 D143 A248 A167 A153 D125 A248 A192 H196 A248 D125 E223 H196 D127 H196 K149 125 A184 A184 E223 A184 D127 A189 A189 A189 E223 N113 L236 C207 L236 L236 A243 A243 A243 W139 1 x W139 1 G216 G216 130 G216 9.0 8.0 7.0 9.0 8.0 7.0 9.0 8.0 7.0 BIOCHEMISTRY ppm 1H

Fig. 2. Ligand-induced folding of TipAS as evident from 1H-15N HSQC spectra of apo TipAS (24), TipAS·promothiocin A, and TipAS·nosiheptide complexes. Resonances are marked with assignment information. Resonances shown in red are aliased in the 15N dimension. Horizontal dashed lines connect correlations

of side-chain NH2 groups. The spectrum of apo TipAS shows a large number of broad resonances in the central random coil region consistent with its unfolded N terminus. In contrast, spectra of TipAS·promothiocin A and TipAS·nosiheptide are well dispersed, indicative of a folded structure.

of degrees of freedom as a result of the second ring. Although not analogous position as the heme in globins, and, similarly to detectable by chemical shifts in the TipAS·nosiheptide complex, TipAS, heme binding to apo globins induces folding of addi- a covalent bond between the sulfur atom of C214 and the nosi- tional α-helices (59). The high fold similarity is maintained for β heptide dehydroalanine C atom is evident from the distance in the holo structure of TipAS (z-score 4.1 to the Ascaris hemo- the calculated structures without using this bond as a restraint. globin, 9% sequence identity), which includes 50 additional N- This is consistent with the observation that nosiheptide and terminal residues (Fig. S4). This similarity may indicate that the TipAS form an irreversible complex in vivo and in vitro (25). globin fold has emerged by convergent evolution as a motif that is especially suited for the recognition of macrocyclic structures Antibiotics Induce a Fold That Encloses them in a Globin-Like “Sandwich.” such as heme or thiostrepton antibiotics, or that the ancestral The structures of the promothiocin A and nosiheptide complexes antibiotic resistance proteins of the TipA class have recruited (Fig. 3) reveal that antibiotic binding causes the flexible N terminus heme as a ligand, thereby acquiring new functions related to of the apo structure to fold into three new helices (α6–α8) and to protection from oxygen or oxygen transport (24). The latter hy- extend helix α9 of the apo state by two additional N-terminal pothesis is supported by phylogenetic analyses of globins (60) turns. The antibiotic sits in a deep cleft between helices α9 and that point to a bacterial origin of this protein family with hori- α13 already present in the apo state, hereafter referred to as the zontal gene transfer occurring from bacteria to eukaryotes at apo antibiotic binding cleft, and the newly formed N-terminal a later stage. Along these lines, it seems very likely that the helices α7 and α8 cover its opposite face, such that it is almost evolution of an oxygen transport function accompanied the fully enclosed by protein. Despite the varying size of the ligand, emergence of multicellular animals (61). In contrast, the evolu- the structures of the promothiocin A and nosiheptide complexes tion of antibiotic resistance in antibiotic-producing bac- are very similar (1.33 Å heavy atom protein backbone rmsd). teria seems to have taken place parallel to the evolution of the Minor differences appear in the N-terminal region, where, as a corresponding bacteria (62), i.e., much earlier. result of missing resonances from conformational exchange, helix α7 in the TipAS·promothiocin A complex is less well de- Modeling of the Thiostrepton Complex. As the spectra of the fined and shorter than in the complex with the larger nosi- TipAS·thiostrepton complex did not allow a structure determina- heptide. Apparently, in the latter complex, the increased antibiotic tion, we sought to model this important complex based on the contacts stabilize α7 and extend this helix toward the N terminus experimentally most well defined TipAS·nosiheptide structure. (Fig. 3). As a starting structure, nosiheptide was replaced by thiostrepton It was previously observed that the fold of the apo TipAS with coordinates from its X-ray structure (PDB ID code 1E9W) structure has high similarity to the globin family of proteins (63). In a subsequent simulated annealing step, all intraprotein [DALI server (57) z-score 4.2 to the Ascaris hemoglobin; PDB restraints of the TipAS·nosiheptide complex were retained. The ID code 1ASH (58)] despite a low sequence identity (11%) (24). position of thiostrepton was constrained by additional antibiotic- In addition, the antibiotic is located within the TipAS fold in an protein NOEs taken from the TipAS·nosiheptide data involving

Habazettl et al. PNAS Early Edition | 3of10 Downloaded by guest on October 8, 2021 Promothiocin A Nosiheptide Thiostrepton

NH H Ile10 Oxb6 Dha14 Ind9 O Qua9 N O Thz5 O O Cys10 OH H NH S Thz14 NH S N N 2 O B S Thz6 2 N N O S HN OHO NH H N O OH N Ala11 O O N N N N O Ala8 Ala10 S N H O Dha15 B Pyr7 S Dha16 NH NH N Glu8 Thz5 OH Glu8 O NH 2 O S A HO Thz5 O O Thz1 HN N Pyr7 S Dha12 O O O HN S HO NH Oxa4 N O HN HN O O HN OThz1 N NH O N Ala13 HN O N N O NH S O Gyn3 Thz4 H Tha6 N N Dha15 apo TipAS Val2 HO Thr2 A O But3 NH S Thz4 S N O N Thz14 N O Pyr7 S N NH N But3 H Thz1 HO Thr2

C

Fig. 3. Structures of apo TipAS and TipAS antibiotic complexes. (Top) Chemical structures of promothiocin A, nosiheptide, and thiostrepton. (Middle) Lowest- energy NMR structures of apo TipAS (24), TipAS·promothiocin A and TipAS·nosiheptide complexes as well as the model structure of the TipAS·thiostrepton complex in ribbon presentation. Helices already present in the apo form are shown in green. The N-terminal 50 aa, which are unfolded in apo TipAS, but fold to several helices in the antibiotic complexes, are shown in blue. Antibiotics are depicted in pink stick representation. (Bottom) The 10 lowest-energy structures of TipAS·promothiocin A and TipAS·nosiheptide complexes in bundle representation.

residues Thz1, Thr2, But3, and Thz4, which are conserved between orientation of the antibiotics. Despite its small size, promo- nosiheptide and thiostrepton. The resulting 10 lowest-energy thiocin A forms numerous hydrophobic contacts with the structures of this TipAS·thiostrepton model are included as a protein. One half of its ring structure (residues Ala8, Thz5, Dataset S1. The TipAS·thiostrepton complex (Fig. 3) has a very Ala10, Oxb6) inserts into the apo antibiotic binding cleft and similar overall appearance as the promothiocin A and nosiheptide makes contacts with the side chains of residues I200, H204, complexes, with thiostrepton sandwiched between apo helices and Y205, L215, M218, Y219, F225, and I229 (Fig. 4 and Fig. S5). antibiotic-induced N-terminal helices. Also, this model structure is Only part of the cleft is filled by the antibiotic, whereas well defined (Table S1) with a very low overall energy, namely, the a volume of 372 Å3 remains as an internal void presumably insertion of the larger thiostrepton into the antibiotic cavity did not filled by water molecules. The other half of the promothiocin cause any steric clashes. As shown earlier (25), residue C214 of A ring (Pyr7, Thz1, Val2, Gyn3, Oxa4) contacts residues F126, TipAS forms a covalent bond with one of the two thiostrepton Y131, V135, W139, and Y145 located on the N-terminal dehydroalanine residues, Dha15 or Dha16. The calculated struc- α-helices α7andα8 and their adjacent linkers. Apparently, ture of TipAS·thiostrepton is consistent only with a bond to Dha15, these contacts induce the folding of the N terminus. The N- and repeating the structure calculations with this bond included did terminal structure is then further stabilized by hydrophobic not lead to perturbations. In contrast, calculations with a bond clusters between helices α6, α9, and α8 formed by residues I112, between Dha16 and C214 resulted in strongly violated restraints. L114, K119, V122, F123, W160, I163, T152, and Y155. Possibly, such a bond may form in vitro in a subpopulation, causing The larger antibiotic nosiheptide binds with its first ring A in the heterogeneous appearance of the TipAS·thiostrepton NMR almost the same way to TipAS as the single ring of the smaller spectra (Fig. S2). promothiocin A (Fig. 4). One half of this ring again inserts into the apo antibiotic binding cleft, whereas the other side con- Multidrug Recognition Occurs in a Large Hydrophobic Cavity Covered tacts the N-terminal helices and apparently causes their fold- by a Ligand-Induced Lid. All three antibiotic complexes show ing. In contrast, the second ring B is directed into the internal a very similar overall organization of the binding interface and void that had been left in the promothiocin A complex, thereby

4of10 | www.pnas.org/cgi/doi/10.1073/pnas.1412070111 Habazettl et al. Downloaded by guest on October 8, 2021 PNAS PLUS top Y131 F126 V135

W139

Y145

Y205 bottom Thz1

Pyr7

Oxb6

Oxa4 L215

F225

Promothiocin A Nosiheptide Thiostrepton

Fig. 4. Specific recognition of promothiocin A, nosiheptide, and thiostrepton antibiotics by TipAS. Antibiotics and TipAS are shown in stick and space-filling BIOCHEMISTRY representations, respectively. The identified antibiotic recognition motif (see text) is marked in magenta, and specific TipAS recognition residues are shown in yellow. (Bottom) View into the antibiotic binding cleft of the three complexes. This cleft is already present in the apo form (green). (Top) View from inside the antibiotic binding cleft toward the newly formed N-terminal structure (blue).

reducing its volume to 85 Å3. As a result, additional hydro- evident from the suppression of the conformational exchange phobic contacts are established between nosiheptide residue broadening. Ind9 and protein residues I112, F123, and W160 (Fig. S5). Although thiostrepton has an even larger volume, the model Compared with the promothiocin A complex, ring A is shifted structure of its complex reveals that thiostrepton binds in a very slightly toward helix α7. This shift leads to stronger contacts similar way to TipAS as the other two antibiotics. Its first ring A with residues P128, Y131, and R138 and stabilizes helix α7, as is also positioned between the apo cavity and the N-terminal

nosiheptide promo- thiocin A apo 10 ) -1 (s 5 ex R

0 1 Fig. 5. Information on backbone dynamics from 15 1 15 relaxation data. N T1, T2, and { H}- N NOE re- 0.8 laxation data (Fig. S6) were fitted by the Lipari– Szabo model-free approach (84). Exchange con-

0.6 tributions Rex from micro- to millisecond flexibility 2 2

S and subnanosecond order parameters S are shown 0.4 for apo TipAS (green), TipAS·promothiocin A (cyan), and TipAS·nosiheptide (blue). Residues, which are in 0.2 van der Waals contact with the ligand, are high- 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 lighted in yellow on the Rex data panel. (Top)Sec- ondary structures of apo TipAS and the two complexes. residue Solid cylinders represent α-helices.

Habazettl et al. PNAS Early Edition | 5of10 Downloaded by guest on October 8, 2021 X = S/O Thiostrepton A: R = X2NH2 Nosiheptide Promothiocin A: R = X1NH2 X B: R = NH NH B: R = X HN 2 O 3 2 I H O MO: R = OCH N 3 S NH N O S HN 2 MN: R = NH HO O S S 2 N O NH OH N N H OH O N N O N O N H N 6 S II N O S OH O N O S O NH N III NH HN N O N S 7 IV O O S A R O N HO O N 1 X H S O HN O 4 O N HO HN O N O H N N R N N N H H N N N N O H O S O HO O O NH N H S Geninthiocin NH Berninamycin A A10255G O N S O N N 7 bonds O O OH O HN S HN NH NH NH N O O N N N O HO O H O H O OHN O O HO HN HN O HN S HN N O NH O N X2NH2 N O N O NH S S N NH O N H O O N N X NH N H O O N N 2 2 N ... X4OH O N H O S HO Xn = NH S O HN O NH H N O HN OH O N O H N 1 n O O N N N O HN H OH OH HO O N O NH O O NH O O OH HO Radamycin O S HN O O O HO N X22NH HN NH N H Methylsulfomycin Micrococcin P1 Promoinducin N O N N O O O S O N HN O NH N OH S S S N NH N NH HN NH S HN O O O N HN N HN N O S N O N O S O HN S N O N O HN N S O O HN O N S O X2NH2 NH S N S N O NH N N S N S H O O N X3NH2 N X OH N O N N 4 O O N HO Siomycin A O NH NH S S HN H S O HN OH O N OH NH O N O N O O N O O N H O O HN H O HN N N NH O HO O HO OH O NH OH OH O O N O N NH O O N NH NH O O HN N NH O O O HN O HN HO S HN S O HN HN O O O HO N X2OCH3 N O HN N S N O N O H O S N H N N O N N O S H O X3OH S O N O NH N N X3NH2 R HO NH S N NH H S NH S N O S O O N O S N S N NH O O N O H OH N O O HN N N N HN H S O O O N NH O O N O N O Thioactin: R = X1NH2 H OH H NH Sulfomycin I OH HO Thiopeptin A1a Thiotipin HO Thioxamycin: R = X4OH

Fig. 6. Chemical structures of all known tipA inducers (23, 25, 35, 66, 67, 85–87). The identified structural motif for recognition of thiostrepton-like antibiotics by TipAS is highlighted in magenta. The motif consists of one pyridyl or piperidyl and three thia- or oxazole rings separated by a fixed number of intervening bonds(Inset).

structure, whereas its large ring B still fits very well into the in- result of motions in the micro- to millisecond regime. From res- ternal void without perturbing the antibiotic-induced fold of the idue Y145 toward the N terminus, the few observable resonances 15 N terminus or other parts of the structure (Fig. 4 and Fig. S5). In show a continuous increase in N T2 and T1 and a strong de- all three complexes, the protein tightly encloses the antibiotic crease in {1H}-15N NOEs, indicating that the entire region such that only 4–7% of its surface area remains accessible to becomes more and more flexible toward the N terminus in the solvent (Table S1). nanosecond time range. The folded part of apo TipAS has more uniform 15N relaxation behavior, corresponding to an isotropic The Drug Recognition Interface Is Dynamic. To understand the dy- diffusion tensor with Djj/D⊥ of 1.4, an average rotational corre- namical aspects of the multidrug recognition, the TipAS apo and lation time of 9.0 ns, and nanosecond timescale order parame- holo forms were analyzed by 15N relaxation experiments (Fig. 5 ters S2 around 0.9. It is remarkable that the rotational correlation and Fig. S6). Most residues of the TipAS·promothiocin A and time of apo TipAS is larger than that of the complexes despite 15 TipAS·nosiheptide complexes have very uniform N T1, T2 re- its approximately 1 kDa smaller mass. Presumably, the slower laxation times and heteronuclear {1H}-15N NOEs, correspond- diffusion of the apo form is caused by the viscous drag from ing to a well defined structure. A numerical analysis revealed the large unstructured N terminus. Significant exchange con- a slightly anisotropic rotational diffusion tensor [Djj/D⊥ = 1.2 tributions Rex are observed for a number of residues in the folded (1.3)] with an average rotational correlation time of 8.3 ns (8.5 C-terminal part of apo TipAS (Fig. 5). These mobile residues of ns) for the promothiocin A (nosiheptide) complex, nanosecond the apo form are located in a region different from the observed timescale order parameters S2 around 0.9 for most N-H bond heterogeneity in the N- and C-terminal parts of antibiotic com- vectors, and basically no observable conformational exchange. plexes. Notably, most of these residues line the antibiotic binding Minor exceptions are the micro- to millisecond exchange in helix pocket and make van der Waals contacts with the antibiotic in α7 of the promothiocin A complex, which rendered many reso- the complex, where this exchange is abrogated. This is particu- nances invisible in this region, as well as reduced nanosecond larly pronounced for residues H196 to R202 of helix α10, and order (S2) in loops between helices, e.g., α6/α7, α8/α9, α10/α11, helices α11 and α12. Thus, the backbone of the binding pocket is and α12/α13. flexible on the micro- to millisecond time scale before antibiotic In contrast, the dynamical behavior of the apo form is much binding, but becomes locked when the pocket is filled in the more varied. In its unfolded N-terminal region (G111–Q164), complex. The motions in the microsecond to millisecond time a large number of 1H-15N resonances are not observable as a scale of residues located in the cleft of apo TipAS favor

6of10 | www.pnas.org/cgi/doi/10.1073/pnas.1412070111 Habazettl et al. Downloaded by guest on October 8, 2021 PNAS PLUS S Oxb6 H Promothiocin A N O N I O N O O NH N S II R N III IV O H N O N Pyr7 N Thz1 Oxa4 H O

L215 Oxb6

Pyr7 Oxa4 Y145 V135 Y205 W139 Thz1 F126 F225 TipAS Y131

Fig. 7. Mechanism of TipAS promiscuous antibiotic recognition via the identified thiopeptide motif (Fig. 6). The hydrophobic cavity of apo TipAS is shown in green. Ring A of the thiopeptide (promothiocin A shown) inserts into the cavity and makes specific contacts from Oxb6(I), Pyr7(II), Thz1(III), and Oxa4(IV) to residues L215, Y205, F225 in the cavity. This induces the folding of the unstructured N terminus (blue) into a helical structure on top of the antibiotic, which recognizes the thiopeptide motif (aromatic rings I–IV) via specific contacts from residues F126, Y131, V135, W139, and Y145.

conformational selection (64, 65) over an induced fit mechanism structures: a thia-/oxazole(I)–pyridyl/piperidyl(II)–(III) triple for the antibiotic recognition. Accordingly, helices α11 and α12 ring system and a fourth thia- or oxazole ring (IV), which is of the TipAS·antibiotic complexes exhibit the largest displace- separated from thiazole(III) by a dipeptide linker of seven bonds ments upon ligand binding (Fig. 3). (Fig. 6, magenta). We have searched all thiopeptides described in the literature (35) and the THIOBASE database (68) for a

Identifying the Specific Thiopeptide Recognition Motif. The tran- BIOCHEMISTRY tipA S. lividans match of this motif. Besides the known 15 chemical scaffolds of scription of the gene in is induced by many cyclic tipA inducers, this analysis identified 13 additional thiopeptides thiopeptides of highly variable primary structure and size (Fig. 6 as putative tipA inducers (Table S2). In contrast, five thiopep- and Table S2) (23, 25, 66, 67). Many of these inducers contain tides in this data set did not match the recognition motif. Three dehydroalanine moieties that may form a covalent bond with of these (amythiamicin, cyclothiazomycin, and GE 2270A) are residue C214. However, the presence of dehydroalanine is not tipA a necessary condition for tipA induction (25). Thus, recognition known not to induce , whereas the two others (thiomuracin, clearly involves an additional chemical motif. The following com- GE37468) are putative new noninducers. parison of the promothiocin A, nosiheptide, and thiostrepton TipA Recognizes the Same Thiopeptide Motif That Binds to the complex structures with the chemical structures of all known tipA inducers reveals this common thiopeptide chemical motif rec- Ribosome and Inhibits Translation. Most thiopeptide antibiotics in- ognized by the TipA proteins. hibit protein synthesis by targeting the GTPase-associated center One half of the antibiotic ring A is more strongly conserved of the ribosome at the site of contact between the ribosomal among the tipA inducers than the rest of their chemical structure protein L11 and the 23S rRNA (69). Whereas NMR studies had (Fig. 6). This half, ranging from a thia- or oxazole (position 6, previously given insights into dynamics and domain orientations promothiocin A numbering) over a pyridyl or piperidyl (position of thiostrepton complexes with L11 and 23S rRNA fragments (70, 7), a thiazole (position 1), to a thia- or oxazole ring (position 4), 71), crystallographic structures of complexes of the entire 50S has almost identical contacts and orientation in all three TipAS subunit of the D. radiodurans ribosome with thiostrepton or complexes (Fig. 4). It is positioned at the top of the antibiotic nosiheptide are now available (38). Thiostrepton and nosiheptide cleft of apo TipAS and oriented toward the newly formed N- (Fig. 8) bind in a very similar mode into a hydrophobic cleft be- terminal helices. The thia-/oxazole, pyridyl/piperidyl, and thia- tween L11 and helices H43 and H44 of the 23S rRNA such that zole rings in positions 6, 7, and 1 are recognized from one side by the side of ring A with the identified TipA recognition motif hydrophobic and stacking interactions, respectively, with the side inserts deepest into this pocket. Thiazole rings III and IV make chains of L215 and Y205 at the opening of the cleft (Figs. 4 and intimate contacts with the bases of A1106 and A1078, respectively, 7). From the other side, this triple ring system is covered by the whereas thiazole and pyridyl/piperidyl ring I+II contact the L11 aromatic rings of the N-terminal residues W139 and Y145. residue P25 (D. radiodurans numbering). In particular, bases Similarly, the thia- or oxazole ring at position 4 stacks onto the A1106 and A1078 have been identified as strong determinants of aromatic ring of F225 at the top of the cleft, whereas its upper thiostrepton binding by single point mutations (72) and the fact face makes extensive contacts with the side chains of the N- that 2′O-methylation of A1078 renders thiostrepton-producing terminal residues F126, Y131, and V135. Apparently, the con- streptomycetes self-resistant (31, 72). In contrast, the nonconserved tacts to these specific recognition residues induce the fold of the N terminus into a defined helical bundle structure on top of the part of the chemical structure of thiostrepton and nosiheptide does apo antibiotic binding cleft. The relevance of the (mostly aro- not form contacts, as it is directed toward the open bulk solvent. matic) recognition residues Y131, W139, Y145, L215, and F225 Therefore, the specific recognition of TipA for thiopeptide anti- is further corroborated by their high conservation across the biotics is directed against the same chemical motif that is rec- currently available 1938 homologous TipAS sequences (Fig. ognized by the ribosome and inhibits protein biosynthesis. In S1B), making it plausible that very similar antibiotic recognition both cases, the variable part of the antibiotic is accommodated by and drug-induced folding mechanisms occur in a large variety a nonselective void. The specific TipA recognition of the same of bacteria. antibiotic motif, which is harmful to bacteria, provides further Without exception, all known tipA inducers (Fig. 6) contain the strong evidence that indeed the TipA family of proteins consti- specific antibiotic motif recognized in the three TipAS complex tute a resistance mechanism against thiopeptide antibiotics.

Habazettl et al. PNAS Early Edition | 7of10 Downloaded by guest on October 8, 2021 top side P25 thiostrepton

A1078

A1106

nosiheptide

Fig. 8. Specific recognition of thiostrepton antibiotics by the ribosome occurs via the same chemical motif as in TipA proteins. (Top) details of the structure of the D. radiodurans 50S ribosomal subunit in complex with thiostrepton (PDB ID code 3CF5) (38). The ribosomal RNA (yellow) and the protein L11 (green) are depicted in space-filling representation. (Left) Top view of only ring A of the antibiotic thiostrepton is shown in stick representation. (Right) Side view of the full antibiotic. The specific chemical recognition motif is highlighted in magenta. The recognizing ribosomal residues A1078, A1106, and L11-P25 are shown in CPK colors using light blue for carbon. (Bottom) Details of the structure of the D. radiodurans 50S ribosomal subunit in complex with nosiheptide (PDB ID code 2ZJP) (38) in the same views and representations as for the thiostrepton ribosome complex.

The Constitutive Elements of the Thiopeptide-Specific TipA Multidrug constitutes the mechanical signal by which antibiotic binding Recognition. Our study identifies the main elements of the pro- induces expression of tipA and possibly of other MDR genes miscuous, yet specific, recognition of thiopeptide antibiotics by (24, 25). The conformational selection of the unfolded N terminus TipA. (i) The variable part of the antibiotics inserts into a void to a well-defined, almost identical structure in all three TipAS that is large enough to accommodate their differing sizes. The complexes would then position the DNA binding domain in an walls of this void are lined by hydrophobic residues, which are appropriate way to bind promoter DNA with higher affinity and flexible in the absence of antibiotic and lock into a single or very recruit RNA polymerase to the complex (27). This specific po- few conformations upon ligand binding, thereby optimizing the sitioning of the DNA binding domain and the subsequent ii contacts. ( ) The specific recognition of the thiopeptide motif is induction of transcription needs to emerge from the specific achieved from one side by residues at the entrance of this cavity thiopeptide recognition motif. It could not be achieved reliably and from the other side by residues of the flexible N terminus by an induced-fit accommodation of the nonconserved antibiotic that locks on top of the cavity upon ligand binding. This N ter- part, as the latter would lead to variable N-terminal helix minus is not entirely flexible in the apo form. The considerable orientations. exchange broadening in this region, which varies according to the sequence position of the holo form helices, indicates that minor Conclusion populations of these helices connected by flexible linkers may In summary, we have identified the mechanism of multidrug already be present without antibiotic. These prestructured ele- recognition in the autonomously regulated thiopeptide MDR ments would reduce the entropic cost of the ligand-induced ri- system TipA. The polyspecific recognition is based on a delicate gidification of the N terminus and help in the specific recognition mixture of binding the variable part of the antibiotic in a flexible of the thiopeptide motif. (iii) The binding pose induced by the hydrophobic internal cavity combined with specific binding of specific recognition also orients the dehydroalanine-bearing tail, a conserved thiopeptide TipA inducer motif that leads to folding which is present in many but not all TipA inducers, such that it of the TipAS N terminus by conformational selection. The high can form a covalent bond with residue C214. The TipA thiopeptide recognition mechanism contains some sequence homology within the TipA subfamily of MerR proteins of the features previously identified in multidrug recognition of indicates that this mechanism may be common to numerous MDR gene regulators (10), i.e., a large hydrophobic pocket with bacteria. The identified thiopeptide TipA inducer motif is also unspecific, multiple contact possibilities on a flexible surface and the site of the antibiotic that forms the essential contact to the second site binding for specific recognition. However, the TipA ribosome to suppress protein translation. This structural motif recognition mechanism also contains the additional element of may be a valuable lead in designing new thiopeptide antibiotics folding upon binding (73) of a flexible, yet not completely un- against clinically relevant pathogens. Moreover, the observed structured, part of the molecule. Remarkably, this feature is part congruence of chemical motifs responsible for antibiotic function of the specific recognition. Thus, the recognition is a mixture of and antibiotic resistance may be a general principle to identify induced fit for the unspecific part of the antibiotic and confor- new antibiotic leads also in other MDR systems. By using mational selection for its specific part. As the N-terminal part of a concept of linguistics, we might use the knowledge of diachrony TipAS connects to the DNA binding domain in the transcrip- in synchrony (74), the evolutionary wisdom of bacterial re- tional regulator TipAL, its folding upon binding very likely also sistance mechanisms in drug development today.

8of10 | www.pnas.org/cgi/doi/10.1073/pnas.1412070111 Habazettl et al. Downloaded by guest on October 8, 2021 Materials and Methods Structure calculations were performed with the program XPLOR-NIH (80) PNAS PLUS Protein Preparation and NMR Samples. TipAS and its complexes with pro- by using consecutive simulated annealing protocols (81): calculations started from an extended strand with the antibiotic bound to the protein via the mothiocin A or nosiheptide were prepared as described previously (24, 25). covalent bond between the dehydroalanine of the antibiotic tail and C214 NMR samples of 300 μL volume (Shigemi NMR microtubes) contained using intraprotein restraints only. After the protein and the ligand had 1 mMol 15N/13C-labeled protein in 1:1 complex with antibiotic in 50 mMol condensed into a folded structure, the intermolecular covalent bond was potassium phosphate, 0.02% (wt/vol) NaN , 95% H O/5% D O or 100% D O, 3 2 2 2 switched off, and the structure calculation continued with all intra- and pH 5.9. Partially aligned samples for measurement of RDCs contained 0.8 intermolecular experimental restraints. The results showed that the dehy- mMol TipAS–antibiotic complex, 10 mMol potassium phosphate, 10 mg/mL droalanine was in bonding distance to C214, and the bond was then rein- Pf1 phage (Asla Biotech), and 95% H O/5% D O, pH 5.9. 2 2 troduced as a restraint in a final step. Parameterization of the antibiotics was created by using the Cambridge Structural Database (82). A total of 100 NMR Spectroscopy, Resonance Assignments, and Structure Calculations. NMR structures per complex were calculated, and the 10 lowest-energy structures, spectra were recorded on TipAS complexes at 298 K on Bruker DRX 600 and chemical shifts, other NMR data, and details of experiments were deposited DRX 800 NMR spectrometers equipped with TXI and TCI probe heads, re- in the respective databanks (TipAS·promothiocin A, PDB ID code 2MBZ, spectively. Standard NMR experiments similar to the ones described pre- BMRB accession no. 19421; TipAS·nosiheptide, PDB ID code 2MC0, BMRB viously (75) were performed for protein assignment, structure information, accession no. 19422). Statistics for the complex structures are given in Table 15 and assessment of backbone dynamics by N relaxation. Antibiotic proton res- S1. Structure figures were generated using PyMOL (83). onances were assigned by 2D NOESY and HOHAHA (homonuclear Hartmann– 1 13 15 Hahn) spectra filtered against Hboundto Cor N (56). NMR data were ACKNOWLEDGMENTS. We thank Haruo Seto for providing us with promo- processed by using the NMRPipe suite of programs (76), and spectra were thiocin A, Rhone-Poulenc for nosiheptide, and Jan Kahmann for acquiring displayed and analyzed with the programs SPARKY (77) and PIPP (78). Analysis TipAS·thiostrepton spectra. This work was supported by Swiss National Sci- of relaxation data were carried out with the program Modelfree 4.15 (79). ence Foundation Grants 31-109712 and 31-132857 (to S.G.).

1. Neyfakh AA (2002) Mystery of multidrug transporters: The answer can be simple. Mol 27. Chiu ML, Viollier PH, Katoh T, Ramsden JJ, Thompson CJ (2001) Ligand-induced Microbiol 44(5):1123–1130. changes in the Streptomyces lividans TipAL protein imply an alternative mechanism 2. Nikaido H (2009) Multidrug resistance in bacteria. Annu Rev Biochem 78:119–146. of transcriptional activation for MerR-like proteins. Biochemistry 40(43):12950–12958. 3. Seelig A (1998) A general pattern for substrate recognition by P-glycoprotein. Eur 28. Newberry KJ, Brennan RG (2004) The structural mechanism for transcription activa- J Biochem 251(1-2):252–261. tion by MerR family member multidrug transporter activation, N terminus. J Biol 4. Higgins CF (2007) Multiple molecular mechanisms for multidrug resistance trans- Chem 279(19):20356–20362. – porters. Nature 446(7137):749 757. 29. Brown NL, Stoyanov JV, Kidd SP, Hobman JL (2003) The MerR family of transcriptional 5. Gutmann DAP, Ward A, Urbatsch IL, Chang G, van Veen HW (2010) Understanding regulators. FEMS Microbiol Rev 27(2-3):145–163. BIOCHEMISTRY polyspecificity of multidrug ABC transporters: Closing in on the gaps in ABCB1. Trends 30. Yun B-S, Hidaka T, Kuzuyama T, Seto H (2001) Thiopeptide non-producing Strepto- Biochem Sci 35(1):36–42. myces species carry the tipA gene: A clue to its function. J Antibiot (Tokyo) 54(4): 6. Ambudkar SV, Kimchi-Sarfaty C, Sauna ZE, Gottesman MM (2003) P-glycoprotein: 375–378. From genomics to mechanism. Oncogene 22(47):7468–7485. 31. Cundliffe E (1978) Mechanism of resistance to thiostrepton in the producing-organ- 7. Dawson RJP, Locher KP (2006) Structure of a bacterial multidrug ABC transporter. ism Streptomyces azureus. Nature 272(5656):792–795. Nature 443(7108):180–185. 32. Punta M, et al. (2012) The Pfam protein families database. Nucleic Acids Res 8. Aller SG, et al. (2009) Structure of P-glycoprotein reveals a molecular basis for poly- 40(database issue):D290–D301. specific drug binding. Science 323(5922):1718–1722. 33. Chiu ML, et al. (1996) Characterization of the covalent binding of thiostrepton to 9. Pos KM (2009) Drug transport mechanism of the AcrB efflux pump. Biochim Biophys Acta 1794(5):782–793. a thiostrepton-induced protein from Streptomyces lividans. Biochemistry 35(7): – 10. Wade H (2010) MD recognition by MDR gene regulators. Curr Opin Struct Biol 20(4): 2332 2341. 489–496. 34. Summers AO (1992) Untwist and shout: A heavy metal-responsive transcriptional – 11. Jin MS, Oldham ML, Zhang Q, Chen J (2012) Crystal structure of the multidrug regulator. J Bacteriol 174(10):3097 3101. transporter P-glycoprotein from Caenorhabditis elegans. Nature 490(7421):566–569. 35. Bagley MC, Dale JW, Merritt EA, Xiong X (2005) Thiopeptide antibiotics. Chem Rev 12. Grkovic S, Brown MH, Skurray RA (2002) Regulation of bacterial drug export systems. 105(2):685–714. Microbiol Mol Biol Rev 66(4):671–701. 36. Porse BT, Leviev I, Mankin AS, Garrett RA (1998) The antibiotic thiostrepton inhibits 13. Davies J (1994) Inactivation of antibiotics and the dissemination of resistance genes. a functional transition within protein L11 at the ribosomal GTPase centre. J Mol Biol Science 264(5157):375–382. 276(2):391–404. 14. Glauner B, Keck W, Schwarz U (1984) Analysis of penicillin-binding sites with a micro 37. Poehlsgaard J, Douthwaite S (2005) The bacterial ribosome as a target for antibiotics. method — the binding site of PBP 5 from Escherichia coli. FEMS Microbiol Lett 23: Nat Rev Microbiol 3(11):870–881. 151–155. 38. Harms JM, et al. (2008) Translational regulation via L11: Molecular switches on the 15. Hayes JD, Wolf CR (1990) Molecular mechanisms of drug resistance. Biochem J 272(2): ribosome turned on and off by thiostrepton and micrococcin. Mol Cell 30(1):26–38. 281–295. 39. Walter JD, Hunter M, Cobb M, Traeger G, Spiegel PC (2012) Thiostrepton inhibits 16. Nicholas RA, Ishino F, Park W, Matsuhashi M, Strominger JL (1985) Purification and stable 70S ribosome binding and ribosome-dependent GTPase activation of elonga- sequencing of the active site tryptic peptide from penicillin-binding protein 5 from tion factor G and elongation factor 4. Nucleic Acids Res 40(1):360–370. the dacA mutant strain of Escherichia coli (TMRL 1222). J Biol Chem 260(10): 40. Haseltine WA, Block R, Gilbert W, Weber K (1972) MSI and MSII made on ribosome in – 6394 6397. idling step of protein synthesis. Nature 238(5364):381–384. 17. Kelley SL, et al. (1988) Overexpression of metallothionein confers resistance to anti- 41. Sy J, Ogawa Y, Lipmann F (1973) Nonribosomal synthesis of guanosine 5′,3′-poly- – cancer drugs. Science 241(4874):1813 1815. phosphates by the ribosomal wash of stringent Escherichia coli. Proc Natl Acad Sci 18. Paulsen IT (2003) Multidrug efflux pumps and resistance: Regulation and evolution. USA 70(7):2145–2148. Curr Opin Microbiol 6(5):446–451. 42. Bhat UG, Halasi M, Gartel AL (2009) Thiazole antibiotics target FoxM1 and induce 19. Zheleznova EE, Markham PN, Neyfakh AA, Brennan RG (1999) Structural basis of apoptosis in human cancer cells. PLoS ONE 4(5):e5592. multidrug recognition by BmrR, a transcription activator of a multidrug transporter. 43. Schoof S, et al. (2010) Antiplasmodial thiostrepton derivatives: Proteasome inhibitors Cell 96(3):353–362. with a dual mode of action. Angew Chem Int Ed Engl 49(19):3317–3321. 20. Zheleznova EE, et al. (2000) A structure-based mechanism for drug binding by mul- 44. Goodman CD, Su V, McFadden GI (2007) The effects of anti-bacterials on the malaria tidrug transporters. Trends Biochem Sci 25(2):39–43. parasite Plasmodium falciparum. Mol Biochem Parasitol 152(2):181–191. 21. Schumacher MA, et al. (2001) Structural mechanisms of QacR induction and multidrug 45. Aminake MN, et al. (2011) Thiostrepton and derivatives exhibit antimalarial and recognition. Science 294(5549):2158–2163. 22. Schumacher MA, Brennan RG (2002) Structural mechanisms of multidrug recognition gametocytocidal activity by dually targeting parasite proteasome and apicoplast. – and regulation by bacterial multidrug transcription factors. Mol Microbiol 45(4): Antimicrob Agents Chemother 55(4):1338 1348. 885–893. 46. Kwok JM-M, et al. (2008) Thiostrepton selectively targets breast cancer cells through – 23. Murakami T, Holt TG, Thompson CJ (1989) Thiostrepton-induced gene expression in inhibition of forkhead box M1 expression. Mol Cancer Ther 7(7):2022 2032. Streptomyces lividans. J Bacteriol 171(3):1459–1466. 47. Gartel AL (2011) Thiostrepton, proteasome inhibitors and FOXM1. Cell Cycle 10(24): 24. Kahmann JD, et al. (2003) Structural basis for antibiotic recognition by the TipA class 4341–4342. of multidrug-resistance transcriptional regulators. EMBO J 22(8):1824–1834. 48. Pucci MJ, et al. (2004) Antimicrobial evaluation of nocathiacins, a thiazole peptide 25. Chiu ML, et al. (1999) Broad spectrum thiopeptide recognition specificity of the class of antibiotics. Antimicrob Agents Chemother 48(10):3697–3701. Streptomyces lividans TipAL protein and its role in regulating gene expression. J Biol 49. Singh SB, et al. (2007) Antibacterial evaluations of thiazomycin- a potent thiazolyl Chem 274(29):20578–20586. peptide antibiotic from Amycolatopsis fastidiosa. J Antibiot (Tokyo) 60(9):565–571. 26. Holmes DJ, Caso JL, Thompson CJ (1993) Autogenous transcriptional activation of 50. Schoof S, Baumann S, Ellinger B, Arndt H-D (2009) A fluorescent probe for the 70 S- a thiostrepton-induced gene in Streptomyces lividans. EMBO J 12(8):3183–3191. ribosomal GTPase-associated center. ChemBioChem 10(2):242–245.

Habazettl et al. PNAS Early Edition | 9of10 Downloaded by guest on October 8, 2021 51. Haste NM, et al. (2012) Activity of the thiopeptide antibiotic nosiheptide against 70. Jonker HRA, Ilin S, Grimm SK, Wöhnert J, Schwalbe H (2007) L11 domain rearrange- contemporary strains of methicillin-resistant Staphylococcus aureus. J Antibiot (To- ment upon binding to RNA and thiostrepton studied by NMR spectroscopy. Nucleic kyo) 65(12):593–598. Acids Res 35(2):441–454. 52. Ayida BK, Simonsen KB, Vourloumis D, Hermann T (2005) Synthesis of dehydroalanine 71. Lee D, et al. (2007) The structure of free L11 and functional dynamics of L11 in free, fragments as thiostrepton side chain mimetics. Bioorg Med Chem Lett 15(10): L11-rRNA(58 nt) binary and L11-rRNA(58 nt)-thiostrepton ternary complexes. J Mol – 2457 2460. Biol 367(4):1007–1022. 53. Zhang Q, Liu W (2013) Biosynthesis of thiopeptide antibiotics and their pathway 72. Baumann S, et al. (2010) Molecular determinants of microbial resistance to thio- engineering. Nat Prod Rep 30(2):218–226. peptide antibiotics. J Am Chem Soc 132(20):6973–6981. 54. Li C, Kelly WL (2010) Recent advances in thiopeptide antibiotic biosynthesis. Nat Prod 73. Wright PE, Dyson HJ (2009) Linking folding and binding. Curr Opin Struct Biol 19(1): Rep 27(2):153–164. 31–38. 55. Starosta AL, et al. (2009) Identification of distinct thiopeptide-antibiotic precursor 74. de Saussure F (1916) Cours de Linguistique Générale (Payot, Paris). lead compounds using translation machinery assays. Chem Biol 16(10):1087–1096. 75. Grzesiek S, et al. (1997) Refined solution structure and backbone dynamics of HIV-1 56. Ikura M, Bax A (1992) Isotope-filtered 2D NMR of a protein-peptide complex: Study of Nef. Protein Sci 6(6):1248–1263. a skeletal muscle myosin light chain kinase fragment bound to calmodulin. JAm 76. Delaglio F, et al. (1995) NMRPipe: A multidimensional spectral processing system Chem Soc 114:2433–2440. – 57. Holm L, Sander C (1993) Protein structure comparison by alignment of distance ma- based on UNIX pipes. J Biomol NMR 6(3):277 293. trices. J Mol Biol 233(1):123–138. 77. Goddard T, Kneller D (2008) SPARKY 3 (University of California, San Francisco). 58. Kloek AP, Yang J, Mathews FS, Goldberg DE (1993) Expression, characterization, and 78. Garrett D, Powers R, Gronenborn A, Clore G (1991) A common sense approach to crystallization of oxygen-avid Ascaris hemoglobin domains. J Biol Chem 268(24): peak picking in two-, three-, and four dimensional spectra using automatic computer 17669–17671. analysis of contour diagrams. J Magn Reson 95:214–220. 59. Eliezer D, Wright PE (1996) Is apomyoglobin a molten globule? Structural charac- 79. Mandel AM, Akke M, Palmer AG, 3rd (1995) Backbone dynamics of Escherichia coli terization by NMR. J Mol Biol 263(4):531–538. ribonuclease HI: Correlations with structure and function in an active . J Mol 60. Vinogradov SN, et al. (2007) A model of globin evolution. Gene 398(1-2):132–142. Biol 246(1):144–163. 61. Vinogradov SN, et al. (2006) A phylogenomic profile of globins. BMC Evol Biol 6:31. 80. Schwieters CD, Kuszewski JJ, Tjandra N, Clore GM (2003) The Xplor-NIH NMR mo- 62. Woo PCY, Lau SK, Huang Y, Yuen KY (2006) Genomic evidence for antibiotic re- lecular structure determination package. J Magn Reson 160(1):65–73. sistance genes of actinomycetes as origins of antibiotic resistance genes in pathogenic 81. Nilges M, Clore GM, Gronenborn AM (1988) Determination of three-dimensional bacteria simply because actinomycetes are more ancestral than pathogenic bacteria. structures of proteins from interproton distance data by hybrid distance geometry- Med Hypotheses 67(6):1297–1304. dynamical simulated annealing calculations. FEBS Lett 229(2):317–324. 63. Bond CS, Shaw MP, Alphey MS, Hunter WN (2001) Structure of the macrocycle thio- 82. Allen FH (2002) The Cambridge Structural Database: A quarter of a million crystal strepton solved using the anomalous dispersion contribution of sulfur. Acta Crys- structures and rising. Acta Crystallogr B 58(pt 3 pt 1):380–388. – tallogr D Biol Crystallogr 57(pt 5):755 758. 83. Schrodinger LLC (2010) The PyMOL Molecular Graphics System, version 1.3r1 (Schro- 64. Zhang Q, Stelzer AC, Fisher CK, Al-Hashimi HM (2007) Visualizing spatially correlated dinger LLC, Portland, OR). – dynamics that directs RNA conformational transitions. Nature 450(7173):1263 1267. 84. Lipari G, Szabo A (1982) Model-free approach to the interpretation of nuclear 65. Lange OF, et al. (2008) Recognition dynamics up to microseconds revealed from an magnetic resonance relaxation in macromolecules. 2. Analysis of experimental results. RDC-derived ubiquitin ensemble in solution. Science 320(5882):1471–1475. J Am Chem Soc 104:4559–4570. 66. Yun BS, Hidaka T, Furihata K, Seto H (1994) Microbial metabolites with tipA promoter 85. Yun BS, Hidaka T, Furihata K, Seto H (1994) Microbial metabolites with tipA promoter inducing activity. II. Geninthiocin, a novel thiopeptide produced by Streptomyces sp. inducing activity. III. Thioxamycin and its novel derivative, thioactin, two thiopeptides DD84. J Antibiot (Tokyo) 47(9):969–975. – 67. Yun BS, Hidaka T, Furihata K, Seto H (1994) Promothiocins A and B novel thiopeptides produced by Streptomyces sp. DP94. J Antibiot (Tokyo) 47(12):1541 1545. with a tipA promoter inducing activity produced by Streptomyces sp. SF2741. J Anti- 86. Yun B-S, Hidaka T, Furihata K, Seto H (1994) Thiotipin, a novel thiopeptide with a tipA biot (Tokyo) 47(4):510–514. promoter inducing activity produced by Streptomyces sp. DT31. Tetrahedron 50: 68. Li J, et al. (2012) ThioFinder: A web-based tool for the identification of thiopeptide 11659–11664. gene clusters in DNA sequences. PLoS ONE 7(9):e45878. 87. González Holgado G, et al. (2002) Radamycin, a novel thiopeptide produced by 69. Wilson DN (2009) The A-Z of bacterial translation inhibitors. Crit Rev Biochem Mol streptomyces sp. RSP9. I. Taxonomy, fermentation, isolation and biological activities. Biol 44(6):393–433. J Antibiot (Tokyo) 55(4):383–390.

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1412070111 Habazettl et al. Downloaded by guest on October 8, 2021