Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI-107 Biosynthesis

Article

Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI-107 Biosynthesis Graphical Abstract

Authors Manuel A. Ortega, Yue Hao, Mark C. Walker, Stefano Donadio, Margherita Sosio, Satish K. Nair, Wilfred A. van der Donk

Correspondence [email protected] (S.K.N.), [email protected] (W.A.v.d.D.)

In Brief Ortega et al. characterized the tRNA specificity of a lantibiotic dehydratase and showed that these enzymes from different phyla use glutamyl-tRNAGlu as a source of activated glutamate to glutamylate Ser/Thr residues in a precursor peptide during lantibiotic biosynthesis.

Highlights

Accession Numbers Glu

d

Lantibiotic dehydratases use glutamyl-tRNA activated glutamate

as a source of

d

tRNA acceptor stem bases are important for recognition by the enzyme MibB

d

Structural studies suggests a general lantibiotic dehydration mechanism

d

First heterologous production of an actinomycete class I lantibiotic in E. coli

Ortega et al., 2016, Cell Chemical Biology 23, 1–11 March 17, 2016 ª2016 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.chembiol.2015.11.017

5EHK

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Cell Chemical Biology

Article Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI-107 Biosynthesis Manuel A. Ortega,1,5 Yue Hao,1,5 Mark C. Walker,2 Stefano Donadio,3 Margherita Sosio,3 Satish K. Nair,1,4,* and Wilfred A. van der Donk1,2,* 1Roger Adams Laboratory, Department of Biochemistry, University of Illinois at Urbana-Champaign, 600 South Mathews Avenue, Urbana, IL 61801, USA 2Roger Adams Laboratory, Department of Chemistry, Howard Hughes Medical Institute, University of Illinois at Urbana-Champaign, 600 South Mathews Avenue, Urbana, IL 61801, USA 3NAICONS Srl, Viale Ortles 22/4, 20139 Milan, Italy 4Roger Adams Laboratory, Center for Biophysics and Computational Biology, University of Illinois at Urbana-Champaign, 600 South Mathews Avenue, Urbana, IL 61801, USA 5Co-first author *Correspondence: [email protected] (S.K.N.), [email protected] (W.A.v.d.D.) http://dx.doi.org/10.1016/j.chembiol.2015.11.017

SUMMARY

Class I lantibiotic dehydratases dehydrate selected Ser/Thr residues of a precursor peptide. Recent studies demonstrated the requirement of glutamyltRNAGlu for Ser/Thr activation by one of these enzymes (NisB) from the Firmicute Lactococcus lactis. However, the generality of glutamyl-tRNAGlu usage and the tRNA specificity of lantibiotic dehydratases have not been established. Here we report the 2.7-A˚ resolution crystal structure, along with the glutamyl-tRNAGlu utilization of MibB, a lantibiotic dehydratase from the Actinobacterium Microbispora sp. 107891 involved in the biosynthesis of the clinical candidate NAI-107. Biochemical assays revealed nucleotides A73 and U72 within the tRNAGlu acceptor stem to be important for MibB glutamyltRNAGlu usage. Using this knowledge, an expression system for the production of NAI-107 analogs in Escherichia coli was developed, overcoming the inability of MibB to utilize E. coli tRNAGlu. Our work provides evidence for a common tRNAGludependent dehydration mechanism, paving the way for the characterization of lantibiotics from various phyla. INTRODUCTION Lantibiotics are a class of lanthionine-containing peptides (lanthipeptides) that exhibit antimicrobial activities. A common structural feature of these peptides is the presence of thioether rings established by the bis-amino acids lanthionine (Lan) or methyllanthionine (MeLan) (Chatterjee et al., 2005b) (Figure 1A). They belong to a growing class of natural products known as ribosomally synthesized and post-translationally modified peptides (RiPPs) (Arnison et al., 2013). Increased interest in this class

of peptides has emerged, based on the need for antibiotics capable of overcoming bacterial resistance among pathogens. The characteristic (Me)Lan-containing rings in lanthipeptides are installed enzymatically in a two-step biosynthetic process (Knerr and van der Donk, 2012). Selected Ser/Thr residues within a C-terminal core region of a ribosomally synthesized precursor peptide (LanA) are dehydrated to yield 2,3-didehydroalanine (Dha) and (Z)-2,3-didehydrobutyrine (Dhb) residues, respectively (Figure 1A). Following the dehydrations, Cys thiols carry out Michael-type additions onto the newly formed unsaturated amino acids to generate the (Me)Lan structures (Figure 1A). After installation of the thioether rings, a protease removes an N-terminal leader sequence to generate the final mature natural product. Lanthipeptides are grouped into four different classes (I–IV) based on the biosynthetic machinery responsible for their maturation (Knerr and van der Donk, 2012). To date only lanthipeptides from classes I and II have been shown to possess antimicrobial activities. Class II lantibiotics are synthesized by a bifunctional lanthionine synthetase (LanM) (Knerr and van der Donk, 2012). These enzymes catalyze the ATP-dependent dehydration of Ser/Thr residues via phosphorylation of their respective side chains, and also promote the conjugate addition of Cys residues onto the dehydroamino acids (Chatterjee et al., 2005a; Xie et al., 2004). The dehydrations and cyclizations in class I lantibiotics are catalyzed by two different enzymes, a lantibiotic dehydratase (LanB) and a lantibiotic cyclase (LanC) (Knerr and van der Donk, 2012). Whereas the mechanism of Ser/Thr activation employed by LanM synthetases has been known for more than 10 years, only recent studies have started to shed light on how LanB dehydratases catalyze the corresponding Ser/Thr activation during the dehydration of class I lantibiotics (Garg et al., 2013). Biochemical studies on NisB, the lantibiotic dehydratase involved in the biosynthesis of the food preservative nisin, revealed an unexpected requirement for glutamyl-tRNAGlu during catalysis (Ortega et al., 2015). Biochemical and structural data demonstrated that glutamate is transferred from glutamyltRNAGlu to selected Ser/Thr side chains within the precursor peptide of nisin (NisA), giving rise to a series of glutamylated

Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved 1

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Figure 1. Overview of Lantibiotic Biosynthesis (A) Two-step biosynthetic formation of (Me)Lan rings in lantibiotics. (B) Biosynthesis of the lantibiotic NAI-107. The lantibiotic dehydratase MibB dehydrates seven Ser/Thr residues (green and blue) in MibA followed by cyclization events catalyzed by MibC. Additional enzymes involved in NAI-107 biosynthesis include a flavin-dependent decarboxylase (MibD), flavin-dependent Trp halogenase (MibH), flavin reductase (MibS), and a Pro monooxygenase (MibO) responsible for additional post-translational modifications (Foulston and Bibb, 2010). Lan rings are shown in red. MeLan rings are shown in blue. Negative numbers indicate the position of an amino acid in the leader peptide (underlined) with respect to the core region. The shorthand notations for the chemical structures of chlorinated Trp and bishydroxylated Pro are shown.

intermediates (Garg et al., 2013; Ortega et al., 2015). Upon glutamate elimination, Ser/Thr residues are converted to their unsaturated counterparts (Garg et al., 2013). While these key insights provided the first information regarding the enzymatic mechanism of class I lantibiotic dehydratases, it raised questions as to whether the use of glutamyl-tRNAGlu was a common feature shared among these enzymes. Class I lantibiotic dehydratases are widely distributed among different bacterial species (Marsh et al., 2010). A phylogenetic analysis of these enzymes revealed that they grouped into clades according to their microbial phyla (Zhang et al., 2012). Thus, the generality of the use of glutamyl-tRNAGlu by lantibiotic dehydratases could be assessed by characterizing representative enzymes from different phyla. Having already characterized the activity of a lantibiotic dehydratase (NisB) from a Firmicute (Lactococcus lactis) (Ortega et al., 2015), herein we characterized and reconstituted the enzymatic activity of MibB, a lantibiotic dehydratase from an Actinobacterium. MibB is involved in the biosynthesis of NAI-107 (also called microbisporicin A1), produced by the actinomycete Microbispora sp. 107891 (Figure 1B) (Castiglione et al., 2008; Foulston and Bibb, 2010). NAI-107 is active against a wide panel of antibiotic-resistant bacterial strains and is currently in late pre-clinical trials for the treatment of multi-drug-resistant Gram-positive bacterial infections (Jabe´s et al., 2011). NAI-107 functions by interrupting cell wall biosynthesis (Mu¨nch et al., 2014). During NAI107 biosynthesis, MibB is predicted to dehydrate seven Ser/Thr residues within its precursor peptide MibA (Figure 1B) (Foulston and Bibb, 2010). In this study we reconstituted the enzymatic activity, characterized the tRNA specificity, and solved the 2.7-A˚ resolution crystal structure of MibB. Biochemical and structural data

demonstrate that the enzyme indeed uses glutamyl-tRNAGlu during dehydration as the source of activated glutamate needed to catalyze the glutamylation of Ser/Thr residues in MibA. This observation provides evidence for a general use of glutamyl-tRNAGlu by class I lantibiotic dehydratases. In addition, our data identify nucleotides within the tRNAGlu acceptor stem that are important for recognition by MibB. The acquired knowledge regarding the tRNA specificity of MibB allowed the development of an expression system in Escherichia coli suitable for the heterologous production of NAI-107 analogs. This methodology provides an alternative route for the study and characterization of class I lantibiotics produced by Actinobacteria, a phylum that has recently emerged with a vast potential for producing previously uncharacterized lantibiotics (Li and O’Sullivan, 2012; Maffioli et al., 2015; Zhang et al., 2015). RESULTS Reconstitution of the Enzymatic Activity of MibB The lantibiotic dehydratase MibB and its substrate peptide MibA were each heterologously expressed in E. coli and purified as N-terminal hexahistidine-tagged constructs. Building on the recent characterization of the dehydratase NisB (Ortega et al., 2015), we sought to reconstitute the in vitro activity of MibB using E. coli tRNAGlu and E. coli glutamyl tRNA synthetase (GluRS). Despite many experimental permutations, MibB was only able to catalyze one or two dehydrations in MibA as observed by MALDI-TOF mass spectrometry (MS), instead of the required seven dehydrations (Figure 1B). Reasoning that the partial activity observed might be a consequence of differences between the tRNAGlu sequence of the NAI107 producer strain and that of E. coli (discussed in more detail below), total RNA was purified from the producer strain Microbispora sp. 107891. In addition, the recently sequenced genome of Microbispora sp. 107891 (Sosio et al., 2014) facilitated cloning, heterologous expression, and purification of Microbispora

2 Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Figure 3. Improvement of Glutamate Elimination upon Thioether Ring Formation Figure 2. Reconstitution of MibB Activity In Vitro (A–E) MALDI-TOF MS analysis of His6-MibA following dehydration reactions performed with Glu, ATP, and His6-MibB in the presence of (A) Microbispora total RNA and Microbispora His6-GluRS, (B) Microbispora tRNAGluCUC and Microbispora His6-GluRS, (C) Microbispora tRNAGluCUC, (D) Microbispora His6-GluRS, or (E) Microbispora tRNAGluCUC and Microbispora His6-GluRS in the absence of MibB. M, unmodified MibA ([M + H]+ m/z 8,014, calc. m/z 8,013); 2, 2-fold dehydrated MibA ([M + H]+ m/z 7,981, calc. m/z 7,977); 3, 3-fold dehydrated MibA ([M + H]+ m/z 7,963, calc. m/z 7,959); 4, 4-fold dehydrated MibA ([M + H]+ m/z 7,943, calc. m/z 7,941); 5, 5-fold dehydrated MibA ([M + H]+ m/z 7,925, calc. m/z 7,923); 6, 6-fold dehydrated MibA ([M + H]+ m/z 7,907, calc. m/z 7,905); 7, 7-fold dehydrated MibA ([M + H]+ m/z 7,889, calc. m/z 7,887); +1glu, monoglutamylated MibA intermediates with various numbers of dehydrations (5-fold dehydrated [M + H]+ m/z 8,054, calc. m/z 8,052), (4-fold dehydrated [M + H]+ m/z 8,072, calc. m/z 8,070), (3-fold dehydrated [M + H]+ m/z 8,089, calc. m/z 8,088). See also Figure S1.

GluRS. Upon incubation of MibB, MibA, glutamate, ATP, Microbispora GluRS, and Microbispora total RNA, MibB dehydrated MibA by up to 4-fold as observed by MALDI-TOF MS (Figure 2A). To further confirm that tRNAGlu was required for MibB activity, we purified in vitro transcribed Microbispora tRNAGluCUC, determined the extent of in vitro aminoacylation by Microbispora GluRS (Figure S1A), and performed in vitro dehydration assays. MALDI-TOF MS analysis revealed the generation of up to 7-fold dehydrated MibA (Figure 2B). Omission of either Microbispora tRNAGluCUC, Microbispora GluRS, or MibB resulted in the abolishment of dehydration activity, suggesting that MibB catalyzes MibA dehydration in a glutamyl-tRNAGlu-dependent manner (Figures 2C–2E). In addition to dehydrated MibA intermediates, peaks whose masses corresponded to monoglutamylated and partially dehydrated MibA were observed by MALDI-TOF MS (Figure 2B). The appearance of these peaks suggests a dehydration mechanism similar to that of NisB. Previous alanine scanning mutagenesis of conserved residues in NisB identified Arg786 to be important for glutamate elimination (Garg et al., 2013). To confirm that MibB-catalyzed dehydrations proceed through a similar mechanism, we mutated the corresponding residue in MibB, Arg870, to Ala. In vitro dehydration assays with MibB-R870A revealed the accumulation of glutamylated MibA peptides, as observed

(A–D) MALDI-TOF MS analysis of His6-MibA after dehydration assays in the presence of glutamate, ATP, Microbispora His6-GluRS, and Microbispora tRNAGluCUC (A) without His6-MibB, (B) with His6-MibB, (C) with His6-MibB and His6-MibC, or (D) with His6-MibB and His6-MibC-H364A. M, unmodified MibA ([M + H]+ m/z 8,013, calc. 8,013); 5, 5-fold dehydrated MibA ([M + H]+ m/z 7,924, calc. m/z 7,923); 7, 7-fold dehydrated MibA ([M + H]+ m/z 7,887, calc. m/z 7,887); +1glu-3, monoglutamylated and 3-fold dehydrated MibA ([M + H]+ m/z 8,088, calc. m/z 8,088). Area highlighted in gray emphasizes the mass range of monoglutamylated and partially dehydrated MibA intermediates. See also Figure S2.

by MALDI-TOF MS (Figures S1B and S1C). This result suggests that similar to NisB, MibB catalyzes dehydration of MibA through glutamylation of Ser/Thr residues followed by glutamate elimination. Intriguingly, the observed monoglutamylated and partially dehydrated MibA intermediates only start to appear once 5-fold, 6-fold, or 7-fold dehydrated MibA accumulate during the dehydration assay (compare Figures 2A and 2B). This observation suggests that glutamate is eliminated faster from early glutamylated intermediates than from those formed at a later stage of dehydration. One possible difference between earlyand late-stage glutamylated intermediates is that the latter may be cyclized under physiological conditions but not under the conditions used for the data shown in Figure 2B. Indeed, recent kinetic characterization of HalM2, a class II lanthionine synthetase involved in the dehydration and cyclization of HalA2, revealed that formation of thioether rings in HalA2 increased the rate of subsequent dehydrations (Thibodeaux et al., 2014). The observed difference in the efficiency of glutamate elimination might therefore be a consequence of MibA having to be partially cyclized for MibB to efficiently eliminate glutamate as the dehydration reaction proceeds. To test this hypothesis, we performed dehydration assays of MibB in the presence of the cyclase MibC. MibC is responsible for catalyzing the formation of five thioether crosslinks in MibA (Figure 1B). MibC was heterologously expressed and purified as an N-terminal hexahistidine-tagged protein. Dehydration assays were then performed in the presence and absence of MibC (Figures 3A–3D). Upon incubation of MibA with MibB and MibC, the monoglutamylated

Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved 3

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

intermediates were no longer observed by MALDI-TOF MS (compare Figures 3B and 3C). To demonstrate that MibA indeed needs to be partially cyclized for MibB to eliminate glutamate more efficiently, rather than MibC causing a conformational change in MibB that increases the efficiency of Glu elimination, we performed dehydration reactions with MibC-H364A, a mutant that disrupts the catalytically critical zinc-binding site (Helfrich et al., 2007; Li and van der Donk, 2007). Dehydration assays performed in the presence of MibC-H364A again resulted in the appearance of monoglutamylated and partially dehydrated MibA intermediates as observed by MALDI-TOF MS analysis (Figure 3D). Collectively, these results are consistent with a model that requires some thioether rings in MibA to be installed for MibB to efficiently eliminate later glutamylated MibA intermediates. Based on this observation, subsequent dehydration assays were performed in the presence of MibC unless otherwise noted. Directionality of Dehydroamino Acid Formation in MibA The reconstitution of MibB activity in vitro allowed investigation of the directionality of dehydration. MibB assays in the presence of MibC were stopped at two different time points, and treated with endoproteinase LysC to remove the leader peptide to improve fragmentation efficiency (for this purpose MibA had a Lys inserted between the leader and core region; see Experimental Procedures). The time points were chosen such that the assay mixtures contained intermediates spanning 1- to 6-fold dehydrated MibA, which were analyzed by liquid chromatography electrospray ionization mass spectrometry (LC/ESI-Q/TOF MS) (Figure S2A). MibA has eight possible sites that could be dehydrated (Thr2, Ser3, Ser5, Thr8, Thr12, Ser13, Ser18, and Ser21). However, in the final natural product Thr12 escapes dehydration (Figure 1B). To eliminate any bias on structural assignments based on the collected fragment ions, we analyzed tandem MS (MS/MS) data using an automated hypothetical structure enumeration and evaluation (HSEE) method (Zhang et al., 2014). HSEE serves as an unbiased way for analyzing MS/MS data by matching experimentally observed fragment ions, with theoretical fragment ions arising from all possible hypothetical predicted structures (HS). The fraction of experimental ions that match to a particular predicted structure returns an HS score. The hypothetical structure with the highest HS score represents the most probable structure that matches the collected MS/MS data. HSEE analysis on 1-fold dehydrated MibA core revealed that the first dehydration occurs on either Thr2 or Ser3 (Figure S2B). The lack of diagnostic ions between Thr2 and Ser3 precludes the assignment of the first dehydration to either site. Dehydration at Ser21 results in the highest HS score value for the second dehydration, demonstrating that MibB does not follow a strict directionality. Inspection of the MS/MS data for the 3-fold dehydrated MibA core intermediate revealed the third dehydration to occur on either Ser18 or Ser13. However, the divergent pathway reconverged since HSEE analysis of the 4-fold dehydrated MibA core intermediate revealed Ser13, Ser18, and Ser21 to be dehydrated in addition to either Thr2 or Ser3 (Figure S2B). The 5-fold dehydrated MibA core is again a mixture carrying the additional dehydration on either Thr12 or Thr8 (Figure S2B). Whereas structures with a dehydrated Thr8 (HS34 and HS49 in Figure S2B)

have higher HS scores, diagnostic ions indicating dehydration on Thr12 were also clearly identified. Intriguingly, Thr12 normally escapes dehydration during microbisporicin biosynthesis, and the observation that this site is partially modified reflects an off-pathway event in vitro. Finally, HSEE analysis on the 6-fold dehydrated MibA core intermediate revealed that the sixth dehydration occurs again at the N terminus on either Thr2 or Ser3, thus skipping Ser5. Our data then imply that Ser5 is dehydrated last. Based on the collected information, MibB dehydrates the precursor peptide MibA without strict directionality (Figure S2C), an observation that has also been reported for other lanthipeptide biosynthetic systems (Jungmann et al., 2014; Mukherjee and van der Donk, 2014). Structural Characterization of MibB To obtain structural insights, we solved the 2.7-A˚ crystal structure of MibB (PDB: 5EHK; for crystallographic details, see Table S1) (Figure 4). The structure of the 120-kDa MibB dehydratase shows a topology and domain organization similar to the structure of NisB in complex with the leader peptide of the NisA substrate (PDB: 4WD9) despite the low sequence identity between the two enzymes (less than 20%) (Figure S3A) (Ortega et al., 2015). The structure of MibB provides the first view of a class I lantibiotic dehydratase without its bound cognate leader peptide. The bifurcated structure contains a central cleft separating the two domains, with residues Asp4 through Ser809 comprising a multi-domain N-terminal region and Pro813 through His1114 forming a single C-terminal domain (Figures 4A and S3B). Residues Glu20 through Asp78 were not modeled due to a lack of electron density. Residues important for glutamylation and glutamate elimination in NisB (Garg et al., 2013) are all conserved in MibB (Figure S4) and map to distinct clusters in the amino- and carboxy-terminal regions in the MibB structure, respectively (Figures 4B and 4C). Nearly all of the residues in MibB that have been shown to be important for glutamylation in NisB (Garg et al., 2013) map to within a 10-A˚ radius in the N-terminal region (Figure 4B), demarcating this 800-amino-terminal region of MibB as the glutamylation domain. Similarly, residues in MibB shown to be important for glutamate elimination in NisB (Garg et al., 2013), including Arg870, cluster to a small site in the C-terminal domain, which can be considered the glutamate elimination domain (Figure 4C). Furthermore, a sequence alignment of select lanthipeptide dehydratases from different phyla mapped to the MibB structure revealed substantial amino acid conservation within the glutamylation and glutamate elimination active sites (Figure 4D). Despite the overall similarity in topology to NisB, the Ca rootmean-square deviation (RMSD) between both structures was 3.8 A˚ over 892 residues. The high RMSD can be attributed to a slightly different spatial arrangement of the C-terminal domain in MibB when compared with NisB with its substrate bound. The observed difference between the structures may be reflective of conformational changes that are induced upon substrate binding. Pairwise alignment of the glutamylation domain of MibB with that of NisB provides some plausible insights into conformational reorganization. In the NisB-NisA cocrystal structure, the FNLD sequence within the NisA leader peptide is buried in a hydrophobic region composed of several residues within an

4 Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Figure 4. Crystal Structure of the Lantibiotic Dehydratase MibB (A) Cartoon representation of the overall structure of the MibB (PDB: 5EHK) homodimer. The glutamylation domain in one monomer is shown in purple, the elimination domain within the same monomer is shown in green, and the second monomer is shown in gray. Residues important for glutamylation and glutamate elimination are shown as spheres. (B and C) Conserved residues among lantibiotic dehydratases that have been shown to be important (Garg et al., 2013) for either (B) glutamylation or (C) glutamate elimination cluster together at the N terminus and C terminus of the structure, respectively. (D) Sequence alignment of select lanthipeptide dehydratases mapped onto the structure of MibB indicating the degree of conservation. For visualization purposes the elimination domain was rotated. The glutamylation domain is not shown for clarity (right). See also Table S1 and Figures S1, S3, S4, and S5.

amphipathic a helix formed by residues Glu205 through Asn220 (Ortega et al., 2015). In the MibB structure, in the absence of substrate the equivalent helix (encompassing Arg268 through Glu280) is shifted away from the peptide-binding site and tilted inward such that the hydrophobic face points toward the protein interior (Figure S5). Furthermore, the elimination domain from the second protomer is moved inward to occupy the cavity created by the shift of this helix. Binding of the leader peptide may involve multiple, compensatory movements necessary to establish the hydrophobic pocket needed to engage the equivalent LDLD sequence within the MibA leader peptide (Figure 1B). Unfortunately, the lack of a MibB/MibA cocrystal structure prevents distinction of whether this conformational difference is due to differences in the protein sequences of MibB and NisB or the absence of a leader peptide in the MibB structure. Structure-based alignment identified a region within the N terminus of MibB that is similar in structure to RNA recogni-

tion motif-binding domains that engage RNA substrates (Ryter and Schultz, 1998). Mapping the electrostatic potential onto the surface of MibB reveals a basic patch within this region (Figure S3C) consistent with a likely involvement in binding glutamyl-tRNAGlu. A second small region within the N-terminal domain spanning residues Gln201 through Leu286 likely harbors the leader-binding domain, as observed in the NisB cocrystal structure (Ortega et al., 2015) (Figure S3D). This structural element is also present in additional RiPP biosynthetic enzymes, and was recently shown to be a prevalent motif important for leader peptide recognition across several classes of RiPPs (Burkhart et al., 2015; Dong et al., 2015; Koehnke et al., 2015; Latham et al., 2015; Wieckowski et al., 2015). Characterization of MibB tRNA Specificity A tRNA scanning analysis of the Microbispora sp. 107891 genome (Sosio et al., 2014) revealed the presence of two additional tRNAGlu gene copies in addition to the one used for the in vitro assays described above (Figure S6A). One of the copies encodes a tRNAGluCUC (CUC-2) isoacceptor similar to the one already described (Figure 5A). The other copy encodes a tRNAGluUUC isoacceptor (Figure 5A). The presence of these three different tRNAGlu sequences prompted us to determine whether MibB displayed any isoacceptor preference. We purified all three isoacceptors from in vitro transcription reactions and analyzed the extent of MibB-catalyzed MibA dehydration using each tRNAGlu. To ensure differences observed in MibA dehydration were not the result of reduced amounts

Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved 5

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Figure 5. MibB tRNAGlu Isoacceptor Specificity (A) Predicted cloverleaf structures of E. coli tRNAGlu and the different tRNAGlu isoacceptors present in the Microbispora sp. 107981 genome. (B–E) MALDI-TOF MS analysis of His6-MibA after dehydration reactions performed in the presence of glutamate, ATP, His6-MibB, His6-MibC, and Microbispora His6-GluRS, with (B) no Microbispora tRNAGlu, (C) Microbispora tRNAGluCUC-1, (D) Microbispora tRNAGluCUC-2, or (E) Microbispora tRNAGluUUC. M, unmodified MibA ([M + H]+ m/z 8,013, calc. m/z 8,013); 3, 3-fold dehydrated MibA ([M + H]+ m/z 7,960, calc. m/z 7,959); 7, 7-fold dehydrated MibA ([M + H]+ m/z 7,888, calc. m/z 7,887). Dehydration assays were performed with 5-fold molar excess of Microbispora tRNAGluUUC over Microbispora tRNAGluCUC to achieve similar levels of glutamyl-tRNAGlu over the course of the reaction. See also Figure S6.

of aminoacylated tRNA, we calculated functional tRNAGlu concentrations from radioactive aminoacylation assays for each isoacceptor (Figure S6B). Despite the small sequence differences between each tRNAGlu isoacceptor, MALDI-TOF MS analysis of dehydration assays performed under continuous aminoacylation conditions revealed a greater extent of MibA dehydration when either one of the Microbispora CUC tRNAGlu isoacceptors was used compared with the use of the UUC isoacceptor (Figures 5B–5E). Our results indicate that MibB is not only able to discriminate between tRNAGlu from different organisms (as observed from the low levels of dehydration when using E. coli tRNAGlu, see below), but also discriminates between tRNAGlu isoacceptors encoded in the genome of Microbispora sp. 107891. Based on the previously proposed tRNA-NisB binding model (Ortega et al., 2015), nucleotides within the tRNAGlu acceptor stem are believed to be major recognition elements. The observed variation in the extent of MibA dehydration when using different tRNAGlu isoacceptors could therefore arise from nucleotide differences within this region (Figure S6A).

Comparing the nucleotide sequence of the acceptor stem of both Microbispora tRNAGluCUC isoacceptors revealed a conserved A at position 73, a conserved U72 forming a G1U72 base pair, and a conserved G2-C71 base pair (Figure 5A), whereas they contain different nucleotide pairs at positions 3–70 and 4–69. Because the use of both isoacceptors resulted in a similar extent of MibA dehydration, the base pairs at these latter positions do not appear to be important for efficient MibB dehydration. The Microbispora tRNAGluUUC isoacceptor also contains the conserved A at position 73 but differs at position 72, resulting in a G1-C72 base pair (Figure 5A). Finally, the single tRNAGlu in E. coli contains a G at position 73, a C at position 72 forming a G1-C72 nucleotide pair similar to the Microbispora tRNAGluUUC, and a U2-A71 base pair (Figures 5A and S6A). The differences in nucleotide sequence at positions 73, 72, 71, and 2 could explain why the E. coli tRNAGlu is not used efficiently by the lantibiotic dehydratase MibB. If indeed these nucleotides are important for recognition by MibB, we reasoned that the E. coli tRNAGlu sequence might be engineered to be a better substrate for MibB.

6 Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Figure 6. Engineering of E. coli tRNAGlu (A) Acceptor stem nucleotide sequence of E. coli tRNAGlu, tRNAGlu G73A, and tRNAGlu G73A C72U. Mutated nucleotides are shown in red. (B–E) MALDI-TOF MS analysis of His6-MibA after dehydration assays performed in the presence of glutamate, ATP, His6-MibB, His6-MibC, E. coli His6-GluRS, with (B) no E. coli tRNAGlu, (C) E. coli tRNAGlu, (D) E. coli tRNAGlu G73A, or (E) E. coli tRNAGlu G73A C72U. M, unmodified MibA ([M + H]+ m/z 8,013, calc. m/z 8,013); 1, 1-fold dehydrated MibA ([M + H]+ m/z 7,995, calc. m/z 7,995); 4, 4-fold dehydrated MibA ([M + H]+ m/z 7,941 calc. m/z 7,941); 6, 6-fold dehydrated MibA ([M + H]+ m/z 7,906, calc. m/z 7,905). See also Figure S6.

Step-by-step mutations were performed on the E. coli tRNAGlu acceptor stem region (Figure 6A). We first introduced a G73A mutation followed by a second mutation, C72U, to mimic the Microbispora sp. 107891 tRNAGluCUC isoacceptors. Dehydration assays were then performed with each in vitro transcribed and purified E. coli tRNAGlu variant, and the extent of MibA dehydration was assessed by MALDI-TOF MS analysis. Interestingly, the more the E. coli tRNAGlu resembled the Microbispora sp. 107981 tRNAGluCUC acceptor stem region, the more efficient was the MibA dehydration observed (Figures 6B–6E). These results are not a consequence of differences in the amounts of glutamyltRNAGlu present for each mutant, since all E. coli tRNAGlu variants were aminoacylated by E. coli GluRS to a similar degree (Figures S6C and S6D). Moreover, mutating the E. coli acceptor stem at positions 2 and 71 to mimic the G2-C71 pair present in Microbispora tRNAGluCUC revealed these nucleotides to be unimportant for efficient MibB recognition, since no improvement on the MibA dehydration extent was seen (Figure S6E). Based on the observed MibB preference for Microbispora sp. 107891 tRNAGluCUC and our mutational analysis on the E. coli tRNAGlu, nucleotides A73 and U72 within the acceptor stem region of the tRNAGluCUC are important elements for recognition by MibB.

Heterologous Expression of NAI-107 Analogs in E. coli Recently, several lantibiotic production systems have been developed in E. coli by coexpressing precursor peptides together with the corresponding lantibiotic biosynthetic machinery affording fully modified lantibiotics (Kuthning et al., 2015; Lin et al., 2011; Nagao et al., 2007; Shi et al., 2011). This methodology provides a tool for the characterization and engineering of previously uncharacterized lantibiotics, circumventing several problems that can be associated with such studies when using native producer strains (Garg et al., 2012; Shi et al., 2012). However, for several class I lantibiotics, the E. coli lantibiotic expression system is only capable of delivering precursor peptides carrying an incomplete number of dehydrations (e.g. Vela´squez et al. [2011] and the observations reported here for NAI-107). The selectivity exhibited by MibB toward the tRNA acceptor stem prompted attempts to improve the E. coli coexpression system. A plasmid compatible with the T7 expression system in E. coli was designed using the pACYCDuet-1 vector backbone. Four copies of the Microbispora sp. 107891 tRNAGluCUC-1 isoacceptor were cloned into the multiple cloning site 1 (MCS-1) of pACYCDuet-1. Each tRNA copy was flanked with a T7 promoter sequence at the 50 end and an rrnC terminator sequence at the 30 end (Figure 7A). To ensure that sufficient levels of Microbispora glutamyl-tRNAGlu were available during coexpression experiments, we cloned the Microbispora sp. GluRS into the MCS-2 of the same pACYCDuet-1 plasmid. Heterologous production experiments were then performed by expressing MibB, MibC, Microbispora tRNAGlu, and Microbispora GluRS with hexahistidine-tagged MibA in E. coli. Using this expression system we were able to generate NAI-107 analogs containing up to seven dehydrations, as observed by MALDI-TOF MS analysis after immobilized metal affinity chromatography. This expression platform resulted in a significant improvement over the levels of MibA dehydration observed without the expression of Microbispora tRNAGlu and Microbispora GluRS (Figure 7B). DISCUSSION Recent biochemical and structural studies on the lantibiotic dehydratase NisB from the Firmicute Lactococcus lactis demonstrated that the enzyme uses glutamyl-tRNAGlu as a cosubstrate to catalyze Ser/Thr dehydration (Ortega et al., 2015). In this study we provide evidence that this is a general mechanism by characterizing the enzymatic activity of a representative class I lantibiotic dehydratase from an Actinobacterium, MibB. MibB is not only produced by a bacterium from a different phylum than the producer of NisB, the two enzymes also fall into different clades based on their amino acid sequences (Zhang et al., 2012). These findings suggest a common use of glutamyl-tRNAGlu by class I lantibiotic dehydratases. This hypothesis is also supported by the observed conservation of residues among selected class I lanthipeptide dehydratases from Actinobacteria, Firmicutes, and Bacteroidetes that map to the glutamylation and glutamate elimination active sites of MibB (Figure 4D). The 2.7-A˚ resolution crystal structure of MibB shows an overall conservation in topology with that of the NisB-NisA complex, but closer inspection reveals conformational rearrangements

Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved 7

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Figure 7. Heterologous Expression of NAI-107 Analogs in E. coli (A) Representation of the plasmids used for coexpression experiments in E. coli. (B) MALDI-TOF MS analysis of purified His6-MibA after E. coli coexpression of MibB and MibC (top) or MibB, MibC, Microbispora tRNAGlu and Microbispora GluRS (bottom). 1, 1-fold dehydrated MibA ([M + H]+ m/z 7,229, calc. m/z 7,231); 7, 7-fold dehydrated MibA ([M + H]+ m/z 7,125, calc. m/z 7,123). The difference in mass when compared with previous mass spectra is due to differences in the sequence of the hexahistidine tag between pET-28 and pETDuet-1. Plasmid map not drawn to scale.

near the leader peptide-binding site. The hydrophobic region that engages the FNLD sequence of NisA in the NisB/NisA cocrystal structure is buried in the unliganded MibB structure, resulting in compaction of the binding site. Upon MibA leader peptide binding, the core region is thought to reach the glutamylation domain in MibB, where glutamate is transferred from glutamyltRNAGlu to selected Ser/Thr side chains on the precursor peptide. After each glutamylation the core region moves to the elimination domain, where glutamylated Ser/Thr residues are converted to dehydroamino acids. Based on the orientation of the glutamylation and glutamate elimination domains in NisB and MibB, dimerization of lanthipeptide dehydratases might be a requirement for catalysis. Distance constraints also suggest that glutamylation of the core region occurs within the glutamylation domain of one monomer while glutamate elimination occurs within the glutamate elimination domain of the other monomer.

Including this study, the individual activities of two class I lantibiotic dehydratases have been reconstituted in vitro from purified components. However, it is noteworthy that in 2007 Saris and coworkers described the biosynthesis of nisin in vitro by expressing the genes nisA, nisB, and nisC using a cell-free translation system (Cheng et al., 2007), which includes tRNAs, tRNA transferases, and all amino acids. The successful in vitro biosynthesis was puzzling at the time and suggested that perhaps NisA dehydration was cotranslational. The demonstration that lantibiotic dehydratases are glutamyl-tRNAGlu-dependent enzymes explains why nisin biosynthesis could then be reconstituted under the conditions used. Previous yeast two-hybrid assays have shown that NisB forms a multi-oligomeric complex with additional lantibiotic biosynthetic enzymes including the cyclase NisC and the transporter NisT (Siegers et al., 1996). This observation as well as mutants of the nisin producer strain supports the notion that thioether ring formation during lantibiotic biosynthesis is a process that alternates between dehydration and cyclization (Kuipers et al., 2008; Lubelski et al., 2009). The observed dependency of the glutamate elimination catalyzed by MibB on thioether ring formation in the current study provides additional evidence supporting such an alternating mechanism, and suggests that cyclized MibA might adopt a critical structural conformation needed for efficient glutamate elimination catalyzed by the glutamate elimination active site. The absence of some partners of the complete multi-enzyme complex in our in vitro assays and the heterologous production system in E. coli may explain why incompletely processed intermediates are generated that are not observed in the producing organism. The requirement for the cyclase MibC for complete dehydration by MibB is different from the observations with NisB, which can fully dehydrate its substrate NisA in the absence of NisC. We also observed other differences between the two enzymes. NisB dehydrates NisA in a mostly N- to C-terminal fashion (Lubelski et al., 2009; Zhang et al., 2014), whereas MibB processed MibA following a mostly C- to N-terminal directionality after the first dehydration occurred at the N terminus. A similar order of post-translational modifications that are not strictly directional has been reported for other RiPP biosynthetic enzymes (Jungmann et al., 2014; Melby et al., 2012). Interestingly, during the initial steps of microbisporicin biosynthesis, the C terminus of MibA must undergo oxidative decarboxylation presumably catalyzed by the enzyme MibD (Foulston and Bibb, 2010). This decarboxylation must precede cyclization of the C-terminal ring system (Kupke et al., 1995). Perhaps the evolved overall C- to N-terminal directionality of MibB resulted from crosstalk of MibB and MibC with additional microbisporicin biosynthetic enzymes such as MibD and/or the transporter pair MibTU. Despite MibB and NisB both catalyzing tRNAGlu-dependent dehydration reactions and having similar structural topology, MibB exhibited a distinct preference for Microbispora tRNAGlu. This observation suggests that lantibiotic dehydratases may have coevolved with their cognate aminoacylated tRNAs. In addition, the fact that dehydration activity proceeded with in vitro transcribed tRNAs indicates that tRNA post-transcriptional modifications are not needed for recognition by class I lanthipeptide dehydratases.

8 Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Overall, MibB activity was enhanced when tRNAGlu sequences containing nucleotides A73 and U72 in their acceptor stem region were used as cosubstrates. Previous work on the tRNA specificity of tRNA-dependent enzymes, including aminoacyl-tRNA synthetases and cyclodipeptide synthetases, have shown nucleotides within the acceptor stem to be important elements for efficient recognition by their respective enzymes (Fung et al., 2014; Hou, 1997; Moutiez et al., 2014). The presence of specific nucleotides within this region is believed to facilitate the 30 CCA end of the tRNA to adopt a conformation suitable for catalysis (Lee et al., 1993; Rould et al., 1989). Based on this precedent, the nucleotides A73 and U72 within the Microbispora sp. 107891 tRNAGlu are likely responsible for helping the acceptor stem achieve a suitable conformation for binding to and/or catalysis by MibB. Presumably, the presence of specific amino acid residues governs nucleotide recognition within the tRNAGlu binding pocket of LanB enzymes. Differences within these residues between NisB and MibB could account for their different tRNAGlu specificities. Interestingly, codon frequency analysis of all the putative open reading frames in Microbispora sp. 107891 genome shows GAG as the most frequent Glu codon (frequency per thousand: GAG, 50.9; GAA, 4.2). The observed tRNA specificity seen in MibB for tRNAGluCUC (decodes GAG) over tRNAGluUUC (decodes GAA) could serve as a strategy to avoid the depletion of the presumably limited pool of aminoacylated tRNAGluUUC by tRNA-dependent processes other than protein biosynthesis. As such, nucleotides in the acceptor stem region could be used as recognition elements by MibB to discriminate between the different tRNAGlu isoacceptors. A similar tRNA specificity was recently reported for E. coli L/F transferase. These enzymes are responsible for mediating protein turnover following the N-rule degradation by attaching either Leu or Phe in a tRNAdependent manner at the N terminus of select proteins (Dougan et al., 2010). Similarly to MibB, E. coli L/F transferase showed substrate selectivity towards the E. coli tRNALeu isoacceptor that decodes the most abundant codon for Leu (CUG) (Fung et al., 2014). Recent genome-mining efforts have identified Actinobacteria as a vast reservoir of previously uncharacterized lanthipeptides (Zhang et al., 2015). The recent characterization of a new family of class I lantibiotics from Actinobacteria exemplify the opportunities for the discovery of novel compounds in this phylum (Maffioli et al., 2015). However, only a limited number of lantibiotics from Actinobacteria have been identified, possibly because they are not expressed under typical growth conditions. Coexpression strategies in E. coli that produce fully modified lantibiotics in principle can overcome lack of production in the native producer (Garg et al., 2012; Kuthning et al., 2015; Nagao et al., 2007; Shi et al., 2011). Yet for class I lantibiotics, the heterologous expression system often fails to generate fully modified products. Our improved system that coexpresses a lantibiotic dehydratase with its cognate tRNAGlu and GluRS allowed for the first time the production of a class I lantibiotic analog from Actinobacteria carrying a complete number of dehydrations. The methodology described herein thus offers an opportunity to characterize unexplored chemistry associated with the biosynthesis of these compounds in addition to increasing our repertoire of possible antibiotic candidates. We envision this

methodology to be useful for the future characterization of not only previously uncharacterized class I lantibiotics produced by Actinobacteria but also of other RiPPs predicted to be biosynthesized in a tRNA-dependent manner (Ortega et al., 2015), overcoming challenges that may be associated with the native producer strains. In conclusion, through the biochemical characterization of the lantibiotic dehydratase MibB we were able to provide additional evidence of a common use of glutamyltRNAGlu by these unusual catalysts, as well as a new methodology for future genome-mining efforts. SIGNIFICANCE New antibiotics are needed to alleviate the current antibiotic resistance problem. Lantibiotics are polycyclic antimicrobial peptides that appear to be relatively unaffected by the appearance of resistant bacteria. Genome-mining efforts have identified Actinobacteria as a vast reservoir of previously uncharacterized lantibiotics, but only a small number of lantibiotics from this phylum have been characterized. Our studies on the biochemical and structural characterization of the actinobacterial lantibiotic dehydratase MibB allowed the development of an expression system in E. coli suitable for the production of NAI-107 analogs by coexpressing MibB with its cognate tRNA substrate. This methodology can serve as a tool for the characterization of not only lantibiotics predicted to be produced by Actinobacteria but also of other RiPPs biosynthesized in a tRNA-dependent manner, thus increasing the potential of finding novel antimicrobial candidates. EXPERIMENTAL PROCEDURES For in vitro transcription and purification of tRNAGlu, tRNAGlu aminoacylation assays, structural characterization of MibB, and heterologous production of NAI-107 in E. coli, please see Supplemental Experimental Procedures. Cloning Expression and Purification The NAI-107 biosynthetic gene cluster was provided by NAICONS. The E. coli codon-optimized mibA gene (GeneArt, for sequence, see Table S2) was cloned into pET28a. The cloning, expression, and purification of His6-MibB, His6-MibB-R870A, His6-MibC, His6-MibC-H364A, Microbispora His6-GluRS, E. coli His6-GluRS, and His6-MibA is described in detail in Supplemental Experimental Procedures. In brief, E. coli BL21 (DE3) cells were transformed with the appropriate plasmid via electroporation. Cells were plated on LuriaBertani agar supplemented with appropriate selection markers, and a single colony was used to inoculate an initial culture to be used for expression. Protein expression was induced when cells reached an OD600 of 0.6 with an appropriate amount of isopropyl b-D-1-thiogalactopyranoside (0.1–0.2 mM) and overexpression was performed for 20 hr at 18 C. Peptide expression was performed similarly but at 37 C for 3 hr. Purification was performed as described in Supplemental Experimental Procedures. RNA Extraction from Microbispora sp. 107891 Microbispora sp. 107891 cells were grown in 2 l of Seed M medium (30 g l1 trypticase soy broth, 17.5 g l1 dextrose, 2 g l1 yeast extract, and 1 g l1 CaCO3) for 6 days at 30 C. Cells were washed three times with S30 buffer (10 mM Tris-acetate buffer [pH 8.2], 14 mM magnesium acetate, 60 mM potassium acetate, 1 mM DTT, 0.3 mM EDTA, 0.3 mM MgCl2) and cell extract was prepared as described previously (Garg et al., 2013). RNA was extracted from the cell extract by acidic phenol extraction using a previously described method (Ortega et al., 2015).

Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved 9

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

In Vitro Dehydration Assays The following reaction conditions were used for dehydration assays: 100 mM HEPES (pH 7.5), 1 mM TCEP, 10 mM L-glutamate, 1–200 mM MibA, 10 mM MgCl2, 10 mM KCl, 0.1–5 mM MibB, 0.02 U ml1 thermostable inorganic pyrophosphatase, 1 mg ml1 Microbispora sp. 107891 total RNA, 4–10 mM Microbispora GluRS, and 5 mM ATP in a final volume of 30 ml. When noted, total RNA was substituted for 2–20 mM Microbispora tRNAGlu. In addition, when specified, 1–5 mM MibC and 1–5 mM ZnCl2 were also added to the reaction. The assay was incubated at 30 C for 3–5 hr, centrifuged to remove insoluble material (14,000 3 g, 5 min, 25 C), and desalted using a C-18 ZipTip concentrator (Millipore). The sample was mixed in a 1:1 ratio with 2,5-dihydroxybenzoic acid matrix, spotted on a Bruker MALDI plate, and analyzed by MALDI-TOF MS. For dehydration assays using the Escherichia coli tRNAGlu variants the same conditions described above were used except the Microbispora GluRS and tRNAGlu were substituted for 10 mM E. coli GluRS and 20 mM corresponding E. coli tRNAGlu. All biochemical assays were performed with hexahistidinetagged substrates and enzymes. Directionality of Dehydroamino Acid Formation in MibA Dehydration assays were performed using Microbispora tRNAGluCUC-1, Microbispora GluRS, MibB, MibA, and MibC using similar conditions as described above in a final volume of 150 ml. Aliquots (75 ml) were taken at 5 min and 300 min and quenched by freezing them in liquid nitrogen. Samples were thawed and digested for 6 hr at 37 C using LysC at a final concentration of 0.002 U ml1 to remove the leader peptide from MibA. An aliquot (10 ml) from each time point was injected into an Acquity UPLC Phenomenex Jupiter C18 column (5 mm, 0.1 3 150 mm) coupled to an electrospray ionization mass spectrometer (Q-TOF Synapt-G1 Waters, in positive ion scan mode using the manufacturer’s conditions). The sample was fractionated using a linear gradient from 2% (v/v) solvent A (0.1% [v/v] formic acid in acetonitrile) in 98% (v/v) solvent B (0.1% [v/v] formic acid in water) to 98% (v/v) solvent A over 38 min. Dehydrated intermediates were then subjected to MS/MS characterization and HSEE analysis using a previously described method (Zhang et al., 2014).

supported by the NIGMS-NIH Chemistry-Biology Interface Training Grant (5T32-GM070421) and by the Ford Foundation. Y.H. was supported partially by a Lowell P. Hager fellowship from the Department of Biochemistry. A Bruker UltrafleXtreme MALDI TOF/TOF mass spectrometer was purchased in part with a grant from the NIH (S10 RR027109 A). The contents of this work are solely the responsibility of the authors and do not necessarily represent the official views of the NIH, Ford Foundation, or the European Commission. The authors declare a conflict of interest. S.D. and M.S. are employees of NAICONS. Received: October 4, 2015 Revised: November 23, 2015 Accepted: November 25, 2015 Published: February 11, 2016 REFERENCES Arnison, P.G., Bibb, M.J., Bierbaum, G., Bowers, A.A., Bugni, T.S., Bulaj, G., Camarero, J.A., Campopiano, D.J., Challis, G.L., Clardy, J., et al. (2013). Ribosomally synthesized and post-translationally modified peptide natural products: overview and recommendations for a universal nomenclature. Nat. Prod. Rep. 30, 108–160. Burkhart, B.J., Hudson, G.A., Dunbar, K.L., and Mitchell, D.A. (2015). A prevalent peptide-binding domain guides ribosomal natural product biosynthesis. Nat. Chem. Biol. 11, 564–570. Castiglione, F., Lazzarini, A., Carrano, L., Corti, E., Ciciliato, I., Gastaldo, L., Candiani, P., Losi, D., Marinelli, F., Selva, E., et al. (2008). Determining the structure and mode of action of microbisporicin, a potent lantibiotic active against multiresistant pathogens. Chem. Biol. 15, 22–31. Chatterjee, C., Miller, L.M., Leung, Y.L., Xie, L., Yi, M., Kelleher, N.L., and van der Donk, W.A. (2005a). Lacticin 481 synthetase phosphorylates its substrate during lantibiotic production. J. Am. Chem. Soc. 127, 15332–15333. Chatterjee, C., Paul, M., Xie, L., and van der Donk, W.A. (2005b). Biosynthesis and mode of action of lantibiotics. Chem. Rev. 105, 633–684.

ACCESSION NUMBERS

Cheng, F., Takala, T.M., and Saris, P.E. (2007). Nisin biosynthesis in vitro. J. Mol. Microbiol. Biotechnol. 13, 248–254.

Atomic coordinates and structure factors for the reported crystal structure have been deposited in the PDB (http://www.rcsb.org) under accession code PDB: 5EHK.

Dong, S.H., Tang, W., Lukk, T., Yu, Y., Nair, S.K., and van der Donk, W.A. (2015). The enterococcal cytolysin synthetase has an unanticipated lipid kinase fold. ELife 4, e07607.

SUPPLEMENTAL INFORMATION

Dougan, D.A., Truscott, K.N., and Zeth, K. (2010). The bacterial N-end rule pathway: expect the unexpected. Mol. Microbiol. 76, 545–558.

Supplemental Information includes Supplemental Experimental Procedures, six figures, and two tables and can be found with this article online at http:// dx.doi.org/10.1016/j.chembiol.2015.11.017. AUTHOR CONTRIBUTIONS M.A.O. performed all biochemical assays, which were designed by M.A.O., W.A.v.d.D., and M.C.W. M.A.O and W.A.v.d.D. analyzed biochemical data. Y.H. and S.K.N. performed and interpreted all structural studies. S.D. and M.S. provided Microbispora sp. 107891 and genome data. M.C.W. performed tRNA bioinformatics analysis. M.A.O., S.K.N., M.C.W., and W.A.v.d.D. wrote the manuscript. M.A.O. and Y.H. contributed equally to this study. ACKNOWLEDGMENTS The authors thank Dr. Neha Garg and Dr. Bo Li for initial cloning of mib genes. We thank Dr. Tiit Lukk for data collection at C.H.E.S.S. (Ithaca, NY) and Keith Brister and colleagues for facilitating data collection at LS-CAT (Argonne National Labs, IL). We also thank Christopher J. Thibodeaux, Gabrielle N. Thibodeaux, and Madeline Lo´pez for helpful discussions, as well as members of the van der Donk, Martinis, and Silverman laboratories at UIUC. This work was supported by a grant from the National Institutes of Health (R01 GM 058822 to W.A.v.d.D and R01 GM 079038 to S.K.N.), and by a grant from the European Commission (contract no. 245066 for FP7-KBBE-2009-3) to M.S. M.A.O. was

Foulston, L.C., and Bibb, M.J. (2010). Microbisporicin gene cluster reveals unusual features of lantibiotic biosynthesis in actinomycetes. Proc. Natl. Acad. Sci. USA 107, 13461–13466. Fung, A.W., Leung, C.C., and Fahlman, R.P. (2014). The determination of tRNALeu recognition nucleotides for Escherichia coli L/F transferase. RNA 20, 1210–1222. Garg, N., Tang, W., Goto, Y., and van der Donk, W.A. (2012). Geobacillins: lantibiotics from Geobacillus thermodenitrificans. Proc. Natl. Acad. Sci. USA 109, 5241–5246. Garg, N., Salazar-Ocampo, L.M., and van der Donk, W.A. (2013). In vitro activity of the nisin dehydratase NisB. Proc. Natl. Acad. Sci. USA 110, 7258–7263. Helfrich, M., Entian, K.D., and Stein, T. (2007). Structure-function relationships of the lanthionine cyclase SpaC involved in biosynthesis of the Bacillus subtilis peptide antibiotic subtilin. Biochemistry 46, 3224–3233. Hou, Y.M. (1997). Discriminating among the discriminator bases of tRNAs. Chem. Biol. 4, 93–96. Jabe´s, D., Brunati, C., Candiani, G., Riva, S., Romano´, G., and Donadio, S. (2011). Efficacy of the new lantibiotic NAI-107 in experimental infections induced by multidrug-resistant Gram-positive pathogens. Antimicrob. Agents Chemother. 55, 1671–1676. Jungmann, N.A., Krawczyk, B., Tietzmann, M., Ensle, P., and Su¨ssmuth, R.D. (2014). Dissecting reactions of nonlinear precursor peptide processing of the class III lanthipeptide curvopeptin. J. Am. Chem. Soc. 136, 15222–15228.

10 Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved

Please cite this article in press as: Ortega et al., Structure and tRNA Specificity of MibB, a Lantibiotic Dehydratase from Actinobacteria Involved in NAI107 Biosynthesis, Cell Chemical Biology (2016), http://dx.doi.org/10.1016/j.chembiol.2015.11.017

Knerr, P.J., and van der Donk, W.A. (2012). Discovery, biosynthesis, and engineering of lantipeptides. Annu. Rev. Biochem. 81, 479–505.

otic-synthesizing enzymes for peptide engineering. J. Mol. Microbiol. Biotechnol. 13, 235–242.

Koehnke, J., Mann, G., Bent, A.F., Ludewig, H., Shirran, S., Botting, C., Lebl, T., Houssen, W.E., Jaspars, M., and Naismith, J.H. (2015). Structural analysis of leader peptide binding enables leader-free cyanobactin processing. Nat. Chem. Biol. 11, 558–563.

Ortega, M.A., Hao, Y., Zhang, Q., Walker, M.C., van der Donk, W.A., and Nair, S.K. (2015). Structure and mechanism of the tRNA-dependent lantibiotic dehydratase NisB. Nature 517, 509–512.

Kuipers, A., Meijer-Wierenga, J., Rink, R., Kluskens, L.D., and Moll, G.N. (2008). Mechanistic dissection of the enzyme complexes involved in biosynthesis of lacticin 3147 and nisin. Appl. Environ. Microbiol. 74, 6591–6597. Kupke, T., Kempter, C., Jung, G., and Go¨tz, F. (1995). Oxidative decarboxylation of peptides catalyzed by flavoprotein EpiD. Determination of substrate specificity using peptide libraries and neutral loss mass spectrometry. J. Biol. Chem. 270, 11282–11289.

Rould, M.A., Perona, J.J., Soll, D., and Steitz, T.A. (1989). Structure of E. coli glutaminyl-tRNA synthetase complexed with tRNA(Gln) and ATP at 2.8 A˚ resolution. Science 246, 1135–1142. Ryter, J.M., and Schultz, S.C. (1998). Molecular basis of double-stranded RNA-protein interactions: structure of a dsRNA-binding domain complexed with dsRNA. EMBO J. 17, 7505–7513. Shi, Y., Yang, X., Garg, N., and van der Donk, W.A. (2011). Production of lantipeptides in Escherichia coli. J. Am. Chem. Soc. 133, 2338–2341.

Kuthning, A., Mo¨sker, E., and Su¨ssmuth, R.D. (2015). Engineering the heterologous expression of lanthipeptides in Escherichia coli by multigene assembly. Appl. Microbiol. Biotechnol. 99, 6351–6361.

Shi, Y., Bueno, A., and van der Donk, W.A. (2012). Heterologous production of the lantibiotic Ala(0)actagardine in Escherichia coli. Chem. Commun. 48, 10966–10968.

Latham, J.A., Iavarone, A.T., Barr, I., Juthani, P.V., and Klinman, J.P. (2015). PqqD is a novel peptide chaperone that forms a ternary complex with the radical S-adenosylmethionine protein PqqE in the pyrroloquinoline quinone biosynthetic pathway. J. Biol. Chem. 290, 12908–12918.

Siegers, K., Heinzmann, S., and Entian, K.-D. (1996). Biosynthesis of lantibiotic nisin. Posttranslational modification of its prepeptide occurs at a multimeric membrane-associated lanthionine synthetase complex. J. Biol. Chem. 271, 12294–12301.

Lee, C.P., Mandal, N., Dyson, M.R., and RajBhandary, U.L. (1993). The discriminator base influences tRNA structure at the end of the acceptor stem and possibly its interaction with proteins. Proc. Natl. Acad. Sci. USA 90, 7149–7152.

Sosio, M., Gallo, G., Pozzi, R., Serina, S., Monciardini, P., Bera, A., Stegmann, E., and Weber, T. (2014). Draft genome sequence of the Microbispora sp. strain ATCC-PTA-5024, producing the lantibiotic NAI-107. Genome Announc. 2, http://dx.doi.org/10.1128/genomeA.01198-13.

Li, X., and O’Sullivan, D.J. (2012). Contribution of the Actinobacteria to the growing diversity of lantibiotics. Biotechnol. Lett. 34, 2133–2145.

Thibodeaux, C.J., Ha, T., and van der Donk, W.A. (2014). A price to pay for relaxed substrate specificity: a comparative kinetic analysis of the class II lanthipeptide synthetases ProcM and HalM2. J. Am. Chem. Soc. 136, 17513– 17529.

Li, B., and van der Donk, W.A. (2007). Identification of essential catalytic residues of the cyclase NisC involved in the biosynthesis of nisin. J. Biol. Chem. 282, 21169–21175. Lin, Y., Teng, K., Huan, L., and Zhong, J. (2011). Dissection of the bridging pattern of bovicin HJ50, a lantibiotic containing a characteristic disulfide bridge. Microbiol. Res. 166, 146–154. Lubelski, J., Khusainov, R., and Kuipers, O.P. (2009). Directionality and coordination of dehydration and ring formation during biosynthesis of the lantibiotic nisin. J. Biol. Chem. 284, 25962–25972. Maffioli, S.I., Monciardini, P., Catacchio, B., Mazzetti, C., Munch, D., Brunati, C., Sahl, H.G., and Donadio, S. (2015). Family of class I lantibiotics from actinomycetes and improvement of their antibacterial activities. ACS Chem. Biol. 10, 1034–1042. Marsh, A.J., O’Sullivan, O., Ross, R.P., Cotter, P.D., and Hill, C. (2010). In silico analysis highlights the frequency and diversity of type 1 lantibiotic gene clusters in genome sequenced bacteria. BMC Genomics 11, 679. Melby, J.O., Dunbar, K.L., Trinh, N.Q., and Mitchell, D.A. (2012). Selectivity, directionality, and promiscuity in peptide processing from a Bacillus sp. Al Hakam cyclodehydratase. J. Am. Chem. Soc. 134, 5309–5316. Moutiez, M., Seguin, J., Fonvielle, M., Belin, P., Jacques, I.B., Favry, E., Arthur, M., and Gondry, M. (2014). Specificity determinants for the two tRNA substrates of the cyclodipeptide synthase AlbC from Streptomyces noursei. Nucleic Acids Res. 42, 7247–7258. Mukherjee, S., and van der Donk, W.A. (2014). Mechanistic studies on the substrate-tolerant lanthipeptide synthetase ProcM. J. Am. Chem. Soc. 136, 10450–10459. Mu¨nch, D., Mu¨ller, A., Schneider, T., Kohl, B., Wenzel, M., Bandow, J.E., Maffioli, S., Sosio, M., Donadio, S., Wimmer, R., et al. (2014). The lantibiotic NAI-107 binds to bactoprenol-bound cell wall precursors and impairs membrane functions. J. Biol. Chem. 289, 12063–12076. Nagao, J., Aso, Y., Shioya, K., Nakayama, J., and Sonomoto, K. (2007). Lantibiotic engineering: molecular characterization and exploitation of lantibi-

Vela´squez, J.E., Zhang, X., and van der Donk, W.A. (2011). Biosynthesis of the antimicrobial peptide epilancin 15X and its unusual N-terminal lactate moiety. Chem. Biol. 18, 857–867. Wieckowski, B.M., Hegemann, J.D., Mielcarek, A., Boss, L., Burghaus, O., and Marahiel, M.A. (2015). The PqqD homologous domain of the radical SAM enzyme ThnB is required for thioether bond formation during thurincin H maturation. FEBS Lett. 589, 1802–1806. Xie, L., Miller, L.M., Chatterjee, C., Averin, O., Kelleher, N.L., and van der Donk, W.A. (2004). Lacticin 481: in vitro reconstitution of lantibiotic synthetase activity. Science 303, 679–681. Zhang, Q., Yu, Y., Vela´squez, J.E., and van der Donk, W.A. (2012). Evolution of lanthipeptide synthetases. Proc. Natl. Acad. Sci. USA 109, 18361–18366. Zhang, Q., Ortega, M., Shi, Y., Wang, H., Melby, J.O., Tang, W., Mitchell, D.A., and van der Donk, W.A. (2014). Structural investigation of ribosomally synthesized natural products by hypothetical structure enumeration and evaluation using tandem MS. Proc. Natl. Acad. Sci. USA 111, 12031–12036. Zhang, Q., Doroghazi, J.R., Zhao, X., Walker, M.C., and van der Donk, W.A. (2015). Expanded natural product diversity revealed by analysis of lanthipeptide-like gene clusters in Actinobacteria. Appl. Environ. Microbiol. 81, 4339– 4350. Note Added in Proof Since this manuscript was accepted, support for glutamylation activity has also been reported for split LanBs involved in the biosynthesis of goadsporin and the thiopeptide thiomuracin: Ozaki, T., Kurokawa, Y., Hayashi, S., Oku, N., Asamizu, S., Igarashi, Y., and Onaka, H. Insights into the biosynthesis of dehydroalanines in goadsporin. ChemBioChem http://dx.doi.org/10.1002/cbic.201500541. Hudson, G.A., Zhang Z., Tietz J.I., Mitchell D.A., and van der Donk, W.A. (2015) In vitro biosynthesis of the core scaffold of the thiopeptide thiomuracin. J. Am. Chem. Soc. 137, 16012–16015.

Cell Chemical Biology 23, 1–11, March 17, 2016 ª2016 Elsevier Ltd All rights reserved 11