Micrococcin cysteine-to-thiazole conversion through transient interactions between the scaffolding protein TclI and the modification enzymes TclJ and TclN

ABSTRACT

Ribosomally synthesized and post-translationally modified peptides (RiPPs) are a broad group of compounds mediating microbial competition in nature. Azole/azoline heterocycle formation in the peptide backbone is a key step in the biosynthesis of many RiPPs. Heterocycle formation in RiPP precursors is often carried out by a scaffold protein, an ATP-dependent cyclodehydratase, and an FMN-dependent dehydrogenase. It has generally been assumed that the orchestration of these modifications is carried out by a stable complex including the scaffold, cyclodehydratase, and dehydrogenase. The antimicrobial RiPP micrococcin begins as a precursor peptide (TclE) with a 35-amino acid N-terminal leader and a 14-amino acid C-terminal core containing six Cys residues that are converted to thiazoles. The putative scaffold protein (TclI) presumably presents the TclE substrate to a cyclodehydratase (TclJ) and a dehydrogenase (TclN) to accomplish the two-step installation of the six thiazoles. In this study, we identify a minimal TclE leader region required for thiazole formation, demonstrate complex formation between TclI, TclJ, and TclN, and further define regions of these proteins required for complex formation. Our results point to a mechanism of thiazole installation in which TclI associates with the two enzymes in a mutually exclusive fashion, such that each enzyme competes for access to the peptide substrate in a dynamic equilibrium, thus ensuring complete modification of each Cys residue in the TclE core.

IMPORTANCE

Thiopeptides are a family of antimicrobial peptides characterized for having sulfur-containing heterocycles and for being highly post-translationally modified. Numerous thiopeptides have been identified; almost all of which inhibit protein synthesis in gram-positive bacteria. These intrinsic antimicrobial properties make thiopeptides promising candidates for the development of new antibiotics. The thiopeptide micrococcin is synthesized by the ribosome and undergoes several post-translational modifications to acquire its bioactivity. In this study, we identify key interactions within the enzymatic complex that carries out cysteine to thiazole conversion in the biosynthesis of micrococcin.

INTRODUCTION

Ribosomally synthesized and post-translationally modified peptides (RiPPs) are natural products produced by many bacteria that exhibit diverse biological activities including antimicrobial functions (1–8). RiPP biosynthesis starts with the ribosomal translation of a precursor peptide that is then heavily modified by multiple enzymes. The precursor peptide consists of an N-terminal leader sequence, also known as the recognition sequence, responsible for recruiting enzymes that carry out post-translational modifications (PTMs) (9–11), and a C-terminal core peptide sequence where PTMs occur (12–14). In most cases, RiPP biosynthetic gene clusters encode an E1-ubiquitin activating-like (E1-like) protein that has been implicated in leader peptide binding. The E1 domain often includes a RiPP recognition element (RRE) that adopts a highly conserved winged helix-turn-helix (wHTH) structure with three α-helices and a three-stranded β-sheet. RiPP leader peptides bind to RRE domains by interacting at the interface of the 3α/3β fold acting as a fourth β strand (15–17). After proper substrate recognition, numerous possible modifications take place on the core peptide culminating with the proteolytic removal of the leader from the core yielding a mature RiPP (7, 8).

Thiazole/oxazole-modified microcins (TOMMs) are a class of RiPPs that feature thiazol(in)e and oxazol(in)e heterocycles resulting from intramolecular reactions of cysteine, serine, or threonine residues in the precursor peptide (18). Thiazole/oxazole biosynthesis is a two-step process in which an ATP-dependent cyclodehydratase (member of the YcaO superfamily) yields thiazoline/oxazoline heterocycles that are then oxidized into azoles by an FMN-dependent dehydrogenase. In addition to the cyclodehydratase and optional dehydrogenase, TOMM clusters encode proteins that facilitate the coupling of the precursor peptide with these enzymes, but the different TOMM systems are highly variable in this respect. Most include an E1-like scaffold protein (mentioned above) and/or a second type of protein-protein interaction domain annotated as “Ocin-ThiF-like.” Either or both of these may be fused to an RRE, and the E1-like domain is often fused to the cyclodehydratase (19–25). This structural variability in TOMM complexes is illustrated in Fig. S1, which depicts four examples of increasing complexity, for which studies on the architecture of these enzyme systems have been carried out. During the biosynthesis of the cyanobactin trunkamide, the enzyme TruD catalyzes the formation of azoline heterocycles. The crystal structure of TruD shows a cyclodehydratase (YcaO domain) fused to an N-terminal domain (NTD) that contains an E1-like domain with an RRE (26, 27). The biosynthetic gene cluster for the bacteriocin heterocycloanthracin (HCA) contains a single copy of an Ocin-ThiF-like protein (HcaF), a fused E1-YcaO cyclodehydratase (HcaD), and a dehydrogenase (HcaB). In this system, the HcaF interacts in a 1:1 ratio with the E1-like domain on HcaD to yield azoline heterocycles. These azoline rings are then oxidized into azoles by HcaB; however, studies to characterize protein interactions with HcaB have been unsuccessful (28, 29). For Microcin B17, the E1-like scaffold protein (McbB) interacts with a discrete cyclodehydratase (McbD) and a dehydrogenase (McbC) in a higher-order octameric complex with a ratio of 4:2:2 (24, 30). In a more complicated system, the biosynthetic gene cluster for the thiopeptide sulfomycin (Sul) encodes multiple RREs (SulB and SulF), E1/Ocin-ThiF-like proteins (SulB, SulC, SulE, and SulF), cyclodehydratases (SulC and SulD), and dehydrogenases (SulF and SulG). Combinations of these proteins form three complexes (SulBC, SulEFG, and SulDEFG) that achieve Cys, Thr, and Ser conversion into their corresponding thiazole, methyloxazole, and oxazole derivatives (25, 31). All these TOMM systems share certain biochemical features across a vast evolutionary distance, but they vary in their intersubunit architectures.

Micrococcin is a thiopeptide produced by several Gram-positive bacteria, including Bacillus cereus and Macrococcus caseolyticus (32–34). Its biosynthesis involves several PTMs, including thiazole formation, C-terminal decarboxylation, dehydroamino acid formation, and the creation of a pyridine-anchored macrocycle (35). The gene cluster responsible for micrococcin production in M. caseolyticus (Fig. 1A) is located on a plasmid and consists of 12 tcl genes, which is simpler than the 24-gene cluster found in B. cereus (36, 37). Out of these 12 tcl genes, 8 are essential for micrococcin production (33, 35). The roles of these genes are illustrated in Fig. 1B. The precursor peptide for micrococcin, TclE, has an N-terminal leader of 35 amino acids and a C-terminal core of 14 amino acids. Its biosynthesis begins with the conversion of all six cysteine residues in the core to thiazoles (33). Thiazole installation is required for all subsequent modifications (Fig. 1B). The work presented here focuses on the thiazole installation step in micrococcin biosynthesis. Each thiazole conversion is a two-step process requiring three proteins: a putative scaffold (TclI), a cyclodehydratase (TclJ), and a dehydrogenase (TclN; Fig. 1C). We have previously shown that in the absence of TclI, no formation of thiazolines or thiazoles occurs, suggesting that this putative scaffold protein is essential for cys-to-thiazole conversion. When the TclJ cyclodehydratase is absent, there are no detectable thiazolines or thiazoles. In the absence of the TclN dehydrogenase, all six thiazoline heterocycles accumulate, suggesting that each Cys residue does not require complete modification before the next one is processed (33).

Fig 1.

Genes and proteins controlling micrococcin biosynthesis. (A) Map of the native tcl gene cluster from M. caseolyticus. Essential proteins for complete micrococcin production are annotated by colored blocks at the bottom. The gene encoding the precursor peptide (TclE) is colored black, and the genes encoding proteins for thiazole installation (TclI, TclJ, and TclN) are colored blue. (B) Overview of the micrococcin biosynthetic pathway that converts the TclE core peptide into micrococcin. Modifications and corresponding enzymes are color coded. Abbreviations: Tz, thiazolyl; Dc, decarboxyl; Dh, dehydro. (C) Two-step conversion of TclE Cys residues to thiazole by the enzymes TclJ and TclN.

In this study, we investigate how the thiazole installation proteins in the micrococcin biosynthetic pathway interact with the substrate peptide, and with each other, to carry out these modifications. By conducting a truncation analysis on the TclE leader peptide sequence, we determined a 20-amino acid minimal recognition region required for thiazole installation. Furthermore, by using computational modeling and an Escherichia coli-based expression system for mutagenesis and copurification experiments, we demonstrate complex formation between TclI, TclJ, and TclN, and we propose a mechanism for cysteine to thiazole conversion in which the scaffold protein TclI coordinates thiazole installation by presenting the TclE substrate to each modifying enzyme in dynamic equilibrium.

RESULTS

TclE, TclI, TclJ, and TclN can be functionally expressed in E. col i

We engineered a system in which E. coli would express codon-optimized tcl genes encoding TclE, TclI, TclJ, and TclN. Each of these was engineered with affinity tags in a manner that was previously shown to preserve functionality (33). To test the functionality of E. coli-expressed Tcl proteins, we evaluated the in vitro conversion of the six TclE Cys residues to heterocycles by mass spectrometry (Fig. S2). TclE was purified with an N-terminal cleavable glutathione S-transferase (GST) tag, and the three other Tcl proteins were purified as complexes using N-terminally His-tagged TclI (these complexes are further described later in the paper). In the absence of modifying enzymes, TclE purification and proteolytic removal from the GST tag yield a leader-plus-core fragment of the expected molecular weight (m/z = 5,373). When treated with E. coli-produced TclI and TclJ (TclN excluded), TclE resolved to two major peaks of m/z = 5,285 and 5,266, consistent with the appearance of five and six thiazolines, respectively. When treated with E. coli-produced TclI, TclJ, and TclN, the prominent TclE product has m/z = 5,253, consistent with the complete 6-thiazole product with an expected −120 Da change [−6 × (H₂O + 2H)] compared to the unmodified peptide.

TclI_NTD directly interacts with TclE

We hypothesized that TclI functions as an adaptor to recruit TclJ and TclN to TclE. Bioinformatic analysis using the program HHpred (38–40) indicates that TclI is structurally similar to Ocin-ThiF proteins. We obtained a structural model of the TclI:TclE dimer using AlphaFold2 (Fig. 2A). The TclI model contains two distinct domains, with the NTD being a wHTH structure comprised of three α-helices and three β-strands, consistent with an RRE. This model places the TclE leader at the interface between the α-helices and the three-stranded β-sheet with the leader sequence acting as a fourth β strand, similar to what has been shown for crystallographically solved structures from other TOMM systems (22, 24, 26). Key TclE residues in this interaction start with F17 occupying a hydrophobic pocket at the interface of the third α-helix and the third β-strand of TclI. Other predicted key interactions between the TclE leader and the TclI RRE involve three salt bridges: TclE(E21)-TclI(K31), TclE(E22)-TclI(K78), and TclE(E28)-TclI(K22) (Fig. 2A). To further investigate whether the TclI_NTD is an RRE, we tested for TclI_NTD binding to TclE by copurification. We co-expressed His-tagged TclI_NTD (residues 1–85) with GST-tagged TclE in E. coli and carried out copurification with nickel-NTA beads. As shown in Fig. 2B, TclI_NTD pulls down GST-TclE (Fig. 2B, Lane 3), indicating a non-covalent interaction between the two proteins. Furthermore, we observed reproducibly that TclI_NTD is greatly stabilized in the presence of GST-TclE (compare Fig. 2B, Lanes 2 and 3). We attribute this stabilization of TclI_NTD to its physical interaction with the TclE leader region. Consistent with this, TclI_NTD copurifies the TclE leader region when the TclE core region is absent (Fig. S3).

Fig 2.

RRE of TclI interacts with TclE. (A) AlphaFold2 model of the TclE:TclI dimer. TclE is shown in yellow. TclI_NTD is highlighted in red, while TclI_CTD is shown in brown. Residues featuring key protein interactions are labeled as follows: TclI_NTD (residues F7, V30, V31, I74, V74, K78, N27, and K22) in red and TclE (residues F17, E21, E22, and E28) in yellow. (B) SDS-PAGE analysis with coomassie blue staining to detect TclE:TclI_NTD interactions. Protein identities are given on the right side of the gel, and information of samples loaded in each lane is given above the gel.

We next attempted to determine a minimal TclE leader sufficient for interaction with TclI. First, a series of TclE leader truncations were tested in the TclI_NTD pull-down assay described above (see Fig. 2B). We generated a series of leader truncations in which three residues were consecutively removed from the N-terminus (Fig. 3A) for a total of six TclE truncation variants: Δ3, Δ6, Δ9, Δ12, Δ15, and Δ18. Each truncated variant was co-expressed with His-tagged TclI_NTD. Recall that the stability of TclI_NTD appears to be dependent on its interaction with TclE (Fig. 2B). When co-expressed with the series of TclE truncation variants, the abundance of detected TclI_NTD is noticeably decreased in truncations Δ15 and Δ18 (Fig. 3B). Background bands functioning as internal controls suggest that this decrease in TclI_NTD abundance is not a result of inconsistent loading. Taken together, these results show that TclI_NTD binds stably to the TclE leader region, and they suggest that the deletion of 15 or more residues from the TclE N-terminus interferes with this interaction. This is in strong agreement with the structural model shown in Fig. 2A. This model was derived independently of the experimental truncation data shown here.

Fig 3.

TclE leader truncation analysis. (A) Schematic diagram of TclE leader truncations. Native TclE from M. caseolyticus is shown on top followed by the designated truncations on the N-terminus of the TclE leader. (B) Nickel copurification experiment to detect TclI_NTD interactions with truncated variants of The TclE leader. The top panel shows the Tcl proteins that were co-expressed for this experiment. His-tagged TclI_NTD was co-expressed with each TclE leader truncated variant in E. coli and subjected to purification with nickel-NTA beads and SDS-PAGE. Protein identities are given on the right side of the gel. For each GST-TclE truncation variant, the molecular weights are as follows: Δ3 (31.4 kDa), Δ6 (31.0 kDa), Δ9 (30.7 kDa), Δ12 (30.4 kDa), Δ15 (30.1 kDa), and Δ18 (29.7 kDa).

Determination of a functionally minimal TclE leader peptide

We then used the TclE truncation series (above) to investigate leader sequence requirements for thiazole installation in E. coli cells co-expressing TclI, TclJ, and TclN. We used four of the TclE truncated variants (Δ9, Δ12, Δ15, and Δ18) and assessed thiazole installation by Orbitrap liquid chromatography-mass spectrometry (LC-MS) after GST-TclE purification from co-expressing cells (Fig. 4; see Fig. S4 for sample spectra). For this analysis, we define “fully modified peptides” as those in which all Cys residues are converted to thiazoles. For the Δ9 truncation, 100% of detected TclE peptides were fully modified. The Δ12 variant also yielded products consistent with having a fully modified core, while the Δ15 variant produced a slightly reduced yield of fully modified product (mean = 99.8% ± 0.2%; n = 3), with some detected peptides containing intermediates with 4–5 thiazoles. The Δ18 leader truncation yielded multiple intermediates containing a combination of thiazoles, thiazolines, and cysteines. The mean of fully modified TclE with the Δ18 truncation is 71.6% ± 4.8% (n = 3; Fig. 4A and B). We conclude from this that the first 12 N-terminal amino acids of TclE are not required for thiazole installation, and amino acids 12–15 have minimal impact. When amino acids 15–18 are removed, it significantly impairs thiazole installation, potentially due to loss of key TclE-RRE non-covalent interactions (see Fig. 2A). Recall that TclE(F17) was already predicted to mediate an important interaction in a hydrophobic pocket of the modeled TclI_NTD. Previous studies with TOMM systems such as streptolysin and microcin B17 have shown that the leader peptide is primarily engaged through a conserved FXXXB (B = V, I, or L) motif (41, 42). The TclE leader contains a similar motif (FXXXXB) in residues 17–22. The convergence of these empirical findings with the AlphaFold2 structural prediction and these observations in other TOMM systems gives credibility to a model in which the TclI_NTD functions as a genuine RRE in micrococcin biosynthesis.

Fig 4.

Effects of TclE leader truncations on the production of fully modified core peptides. (A) Fully modified TclE is defined by the presence of six thiazoles in the core peptide. Each bar shows the percentage of TclE peptides detected by LC-MS that are fully modified. Each TclE-truncated variant (see Fig. 3A; Δ9, Δ12, Δ15, and Δ18) was co-expressed with TclI, TclJ, and TclN in E. coli and purified for LC-MS. Reactions included tobacco etch virus (TEV) protease to cleave TclE from the GST tag prior to LC-MS analysis. Data are shown as the mean of three independent replicates with SD. Asterisks show statically significant differences (P < 0.05) according to a parametric t test carried out with the Benjamini, Krieger, and Yekutieli method (43). (B) Representative MS/MS data showing the different peptide species observed from the ∆18 variant. The 6-thiazole modification was the most highly represented species, but variable combinations of unmodified cysteines and thiazoles were also observed. Spectra highlighted in green shows the fully modified (m/z = 1058.04) species, highlighted in orange are peptides containing five thiazoles and one cysteine (m/z = 1083.72), and featured in blue are species containing two cysteines with four modifications, which consisted of either four thiazoles (m/z = 1109.40, dashed line) or three thiazoles and one thiazoline (m/z = 1110.05, solid line). See File S1 for detailed values of peptide abundance detected by LC-MS and Fig. S4 for representative mass spectra.

TclI_CTD binds to TclJ and TclN

Given that TclI_NTD engages with TclE, we hypothesized that the C-terminal domain (CTD) of TclI may be primarily involved in recruiting the enzymatic proteins TclJ and TclN. Ocin-ThiF proteins like TclI have previously been shown to mediate interactions with TOMM enzymes (3, 25, 28). To test interactions of full-length TclI to the modifying enzymes, we co-expressed His-tagged TclI with TclJ or with TclN in E. coli and carried out nickel copurification experiments. As shown in Fig. 5 (Lane 3), TclI interacts with both TclJ and TclN when all three proteins are expressed together. When co-expressed with each individual enzyme, TclI also copurifies them (Fig. 5, Lanes 4–5). His-tagged TclI_CTD (residues 85–242) also copurifies TclJ and TclN (Fig. 5, Lanes 7–8), though TclI_CTD:TclN shows a weaker interaction than TclI_CTD:TclJ. These results point to TclI_CTD as being sufficient for binding to both TclJ and TclN.

Fig 5.

Domain analysis of TclI binding to TclJ and TclN. SDS-PAGE analysis to detect Tcl protein expression and copurification. Maps of His-tagged TclI, TclI_CTD, TclJ, and TclN used in this study are shown above the gel. The asterisk shows a weak band corresponding to TclN that does not appear in the vector-only control (Lane 1) but consistently appears in independently replicated gels. The weak nature of this band could be explained by TclN being degraded by proteases or weak interactions with TclI_NTD.

Initially, we interpreted these results to mean that TclI has two unique interaction surfaces on its CTD, simultaneously recruiting TclJ and TclN as a stable enzymatic complex. To test whether the TclIJN proteins form a ternary stable complex, we carried out copurification experiments with varying tagging arrangements. We created nine strains expressing different combinations of His-tagged TclI, TclJ, and TclN (Fig. 6). These experiments show that TclI copurifies TclJ and TclN regardless of whether TclI is N-terminally or C-terminally His-tagged (Fig. 6, Lanes 2–3), although TclI may be more poorly expressed or less stable when C-terminally tagged. Furthermore, TclI binding to TclJ appears to be favored when TclI is C-terminally tagged. N-terminally tagged TclJ does not copurify any detectable amounts of TclI or TclN (Fig. 6, Lane 4); however, when the tag is moved to the TclJ C-terminus, TclJ robustly pulls down TclI, with little evidence of TclN copurifying (Fig. 6, Lane 5). When an N-terminal region of TclJ (residues 1–115) is C-terminally tagged, it very robustly copurifies TclI, and a minor band consistent with TclN can be observed (Fig. 6, Lane 6). TclN strongly copurifies TclI regardless of the location of the tag (Fig. 6, Lanes 8–9), with possibly a minor TclJ copurification product. In all cases where TclI is pulled down by a tagged version of TclJ or TclN, the band intensities indicate a stoichiometric excess of TclI ranging from 2 to 8 (compared to enzyme molecules) based on densitometry that accounts for staining intensity and molecular weight. In all cases where TclI is pulled down by either tagged enzyme, copurification of the untagged other enzyme is absent or barely evident. From these overall results, we hypothesize that TclI engages with TclJ and TclN in a competitive fashion, and there may be a weak interaction between the two enzymes.

Fig 6.

Analysis of TclI, TclJ, and TclN interactions. Coomassie-stained SDS-PAGE of purified TclIJN proteins expressed together in E. coli to detect complex formation. The upper panel shows the combinations of each His-tagged Tcl protein. The difference between Lane 5 and Lane 7 is that in Lane 7, the linker between TclJ and the histidine tag is longer (6-Gly) compared to the normal linker (Gly-Gly-Ser) used in the other lanes. Asterisks denote weak protein bands that do not appear in the vector-only control (Lane 1) but consistently show up in replicated copurification experiments.

A single surface of TclI facilitates binding to both TclJ and TclN

We generated structural models of TclI:TclJ and TclI:TclN dimers using AlphaFold2 (Fig. 7A and B). According to these two models, TclI_NTD folds in a similar conformation for both, while TclI_CTD takes on a slightly different structure in each predicted complex, with Helix 6 (TclI residues 188–213) being a central structure for binding to both enzymes. These models reinforce the notion that the TclI:TclJ and TclI:TclN complexes are alternative and mutually exclusive structures. To investigate whether Helix 6 is the primary interaction surface of TclI for binding to the enzymes, we genetically dissected the region corresponding to Helix 6 by substituting each of its 15 surface-exposed residues with an arginine residue and tested for binding to TclJ and TclN under the same conditions used for the copurification shown in Fig. 6, Lane 2. For this, 15 E. coli strains were constructed that express TclJ, TclN, and His-tagged TclI with its corresponding Helix 6 substitutions (Fig. S6). Copurification experiments show that TclI residues Y189, I198, I202, and T212 are important for the TclI:TclJ interaction since these Arg substitutions abolish binding interactions between the two proteins (Fig. 8A, Lanes 2, 4, 6, and 10). TclI residues H194, N201, S205, F208, and L209 are critical for interaction with TclN, as when these residues are substituted, TclI:TclN interactions are disrupted (Fig. 8A, Lanes 3, 5, 7, 8, and 9). Substitutions T188, L197, C199, E206, Y210, and S213 had no effect on TclI interactions with either TclJ or TclN (see Fig. 8B; Fig. S2). Substitutions T212R and F208R resulted in a weaker TclI band, suggesting that these changes may also have a negative effect on TclI folding or stability. These findings indicate that Helix 6 is a pivotal structure within TclI responsible for recruiting both TclJ and TclN. Key residues of Helix 6 facilitating binding to each enzyme are interspersed within this postulated surface of TclI_CTD, further supporting the notion of two enzymes competing for one TclI docking site.

Fig 7.

Predicted AlphaFold2 models of the TclI:TclJ (A) and TclI:TclN (B) dimers. TclI_NTD is shown in red, and TclI_CTD is shown in brown with the central Helix 6 highlighted in magenta. TclJ is depicted in cyan, and TclN is shown in green.

Fig 8.

Dissection of Helix 6 on TclI to determine key interactions with TclJ and TclN. (A) Nickel copurification experiment with His-tagged TclI with its corresponding Helix 6 substitution co-expressed with TclJ and TclN. Each substitution on TclI is labeled on each lane. (B) Depiction of Helix 6 highlighting key residues for interacting with TclJ in cyan (Y189, I198, I202, and T212) and TclN in green (H194, N201, S205, F208, and L209). Residues highlighted in gray (T188, L197, C199, E206, Y210, and S213) had no effect on TclI interactions with either TclJ or TclN (see Fig. S6).

An independent domain of TclJ facilitates binding to TclI

We wanted to further investigate how the N-terminal region of TclJ (when C-terminally tagged), is able to pull down TclI, while the N-terminally tagged full-length TclJ is unable to (see Fig. 6, Lanes 4–5). We reasoned that the N-terminus of full-length TclJ is important for TclI interaction. We obtained AlphaFold2 models of TclJ and TclN (Fig. S5A and B). The TclJ model features two distinct domains, corresponding to a small NTD (residues 1–105) with the enzyme active site in a larger YcaO-like CTD (residues 115–242). HHpred analysis showed that the NTD of TclJ is a short E1-like domain likely involved in protein-protein interactions. The TclJ C-terminus is embedded in its putative active site, while its N-terminus is predicted to be surface exposed on this E1-like NTD. We hypothesized that TclJ_NTD folds independently and facilitates binding to TclI. To test this, two versions of the TclJ_NTD (resides 1–105 and residues 1–115) were co-expressed with His-tagged TclI for nickel copurification. Both versions of TclJ_NTD copurified with TclI (Fig. 9, Lanes 3–4) indicating that the E1-like domain of TclJ (residues 1–105) folds independently and forms a TclI-binding domain. Tags on the N-terminus of this domain appear to obstruct TclI binding (see Fig. 6, Lane 4).

Fig 9.

E1-like domain analysis of TclJ. (A) Map of TclJ highlighting its E1-like and YcaO domains. Labeled are the two NTD fragments of TclJ used in this study. (B) SDS-PAGE analysis to detect interactions between the two TclJ_NTD fragments (NTD1 and NTD2) and TclI.

Analysis of TclIJN complexes by size-exclusion chromatography

To investigate our model that TclI complexes with either TclJ or TclN but not both simultaneously, we carried out a preliminary size-exclusion chromatography experiment to evaluate possible TclIJN complexes. Where TclI was expressed on its own, we observed a peak consistent with a monomeric molecular weight of ~30 kDa (Fig. S7). However, when TclI, TclJ, and TclN were analyzed as a mixture, a new peak arose with a retention time consistent with a molecular weight of ~120 kDa, which suggests a complex that contains two copies of TclI and one copy of either enzyme (2TclI:1TclJ, expected: 121 kDa or 2TclI:1TclN, expected: 111 kDa). Due to problems with the solubility of TclIJN complexes, follow-up experiments were not performed.

DISCUSSION

In this study, we identify the NTD of TclI as the RRE that directly interacts with the TclE leader, and we define a minimal leader region that is essential for binding to the RRE and allows for complete thiazole installation by the biosynthetic enzymes TclJ and TclN. Reports on streptolysin and microcin B17 have determined that the leader peptide is recognized via a conserved FXXXB (B = V, I, or L) motif (41, 42). Our studies show that interruption of a similar motif (FXXXXB) in the TclE leader (residues 17–22) significantly reduces RRE binding and thiazole installation. This suggests a shared mechanism in leader-RRE interactions among evolutionarily diverse TOMM systems. The specificity between a leader peptide and the recruited enzymatic machinery has been exploited to generate novel thiopeptides. In one study, thiopeptide analogs were created in vitro using a hybrid leader peptide that recruits enzymes from different thiopeptide biosynthetic pathways (44). This work shows that leader and core regions may be used modularly, and different leader regions may be hybridized to recruit enzymes from different systems.

Aside from substrate peptides, TOMM biosynthetic systems exhibit a wide variety of molecular arrangements. In the simplest cases, the solved structures of the TruD and LynD cyclodehydratases include a YcaO domain fused to both an E1-like domain and an RRE (26, 27, 45, 46). The RRE is responsible for leader peptide recruitment. These systems install azolines without subsequent oxidation by a dehydrogenase. The Heterocycloanthracin (Hca) biosynthetic system features an Ocin-ThiF-like protein fused to an RRE (HcaF), a fused E1-YcaO cyclodehydratase (HcaD), and a dehydrogenase (HcaB; see Fig. S1). The adaptor HcaF interacts in a 1:1 ratio with HcaD, but attempts to elucidate interactions between HcaF and HcaB have not been possible because HcaB is heavily proteolyzed when expressed in E. coli (28). Our findings suggest that the Tcl proteins associate in a similar fashion, with TclI binding to the E1-YcaO cyclodehydratase TclJ. Fortuitously, the TclN dehydrogenase behaves well in E. coli, allowing us to examine its interactions with TclI and/or TclJ. We find, rather surprisingly, that TclI binds to either TclJ or TclN in a competitive, mutually exclusive manner (Fig. 10). We are not aware of previous models for thiazole installation in which the cyclodehydratase and dehydrogenase are presented to the substrate peptide based on a dynamic exchange mediated by an adaptor protein. Our data show that TclI contains one domain (the NTD) for substrate peptide capture and a second domain (the CTD) for presenting the two enzymatic activities to the peptide, but not simultaneously. We wonder whether such a dynamic exchange mechanism is operative in other azole installation systems currently under investigation.

Fig 10.

Model for cysteine to thiazole conversion by TclI, TclJ, and TclN on the precursor peptide TclE in the biosynthesis of micrococcin. Schematic model of the two functional complexes (TclI:TclJ and TclI:TclN) that assemble to achieve two-step thiazole installation on the core of TclE.

MATERIALS AND METHODS

Plasmids, strains, and culture conditions

The bacterial strains and plasmids used in this study are summarized in Tables S1 and S2. For full plasmid sequences, refer to the Supplemental Materials file. Plasmids were constructed and maintained in E. coli strain DH5α. The tcl genes were synthesized with codon optimization for expression in E. coli. The sequences of vectors and tcl gene inserts are given in the supplemental information file. For protein expression, plasmids were transformed into strain BL21, Nico 21 (DE3), or DH5α. All bacterial cultures were grown in Luria broth (LB: per liter, 10 g Bacto tryptone, 5 g Bacto yeast extract, 5 g NaCl, and 1 mL 2N NaOH). Antibiotics used were kanamycin (30 µg/mL), ampicillin (100 µg/mL), and chloramphenicol (30 µg/mL). Cultures were induced for protein expression using 0.3 mM isopropyl β-D-1-thiogalactopyranoside (IPTG).

Tcl protein expression and purification

To prepare samples for SDS-PAGE analysis, overnight liquid cultures (4 mL) were grown from single colonies in the presence of appropriate antibiotics. Fifty milliliter of cultures was inoculated with 2 mL of overnight culture, allowed to grow at 30°C for 1 h, followed by induction with IPTG for another 6 h at 30°C. Cells were collected by centrifuging the culture, and cell pellets were frozen at −80°C for a minimum of 1 h. Cell pellets were then processed for protein purification with either Ni-NTA-linked (for the His₆ tag) or glutathione-linked (for the GST tag) resin, as detailed below.

For His purification, cell pellets were thawed on ice and re-suspended in 1 mL of lysis buffer (50 mM HEPES pH 7.8, 300 mM NaCl, 0.2% Triton X-100, 0.5 mg/mL lysozyme, 60 mM imidazole, and 1 mM EDTA). Lysis took place for 1 h at 4°C. Cell lysates were then sonicated 4 × 20 seconds using a probe sonicator to ensure complete lysis and fragmentation of DNA. Samples were centrifuged at 13,000 rpm for 10 min (4°C), and approximately 1 mL of supernatant was transferred to a new microcentrifuge tube. The supernatant was incubated end-over-end with 50 µL of NTA-nickel agarose beads (Qiagen) at 4°C for 30 min. Nickel beads were pelleted at 13,000 rpm for 30 seconds and washed 3 × 1 mL with wash buffer (60 mM imidazole, 300 mM NaCl, and 50 mM HEPES pH 7.8). Purified proteins were then eluted in 50 µL of 2× SDS sample buffer (20% glycerol, 83 mM Tris pH 6.8, 40 mg/mL SDS, 0.01% bromophenol blue, and 0.03 µL/mL 2-mercaptoethanol).

For GST purifications, cell pellets were thawed on ice and re-suspended in 1 mL lysis buffer (50 mM Tris 8.0, 150 mM NaCl, 0.5 mg/mL lysozyme, 2 mM EDTA, and 0.2% Triton X-100). Cells were lysed for 1 h at 4°C, and then dithiothreitol (DTT) was added to a final concentration of 1.5 mM. Samples were sonicated 4 × 20 seconds. Cell lysates were centrifuged at 13,000 rpm at 4°C for 10 min to pellet cell debris. Approximately 1 mL of supernatant was transferred to a new microcentrifuge tube. Fifty microliters of unwashed glutathione-agarose beads was added, and samples were rotated end-over-end for 45 min at 4°C. The slurry was pelleted at 13,000 rpm for 30 seconds. The supernatant was removed, and beads were washed with 3 × 1 mL GST buffer (50 mM Tris 8.0 and 150 mM NaCl). GST buffer was completely removed, and proteins were eluted from resin in 50 µL of 2× SDS sample buffer.

Purified samples were heated at 100°C for 5 min. Unless stated otherwise, 8 µL of supernatant was loaded onto a 12% resolving Laemmli gel with a 4% stacking gel. A vector-only control was loaded on the first lane of each gel, and background bands from this lane were used to qualitatively compare the density of background bands in experimental lanes to assess equivalent loading. Gels were run using 1× Laemmli running buffer, stained overnight in coomassie blue stain, followed by destaining and soaking in water prior to imaging.

Mass spectrometry analysis of TclE processing

For purification of His₆-tagged enzymes for mass spectrometry analysis, 25 mL of overnight cultures was grown from single colonies. These overnight cultures were then used to inoculate 1 L induction cultures (30°C). After 1 h, IPTG was added, and the cultures were grown for an additional 6 h. The cells were harvested by centrifugation, and the cell pellets were frozen at −80°C overnight. For copurification of His₆-TclIJ, His₆-TclIN, or His₆-TclIJN, the cells were then thawed on ice with the addition of lysis buffer (50 mM HEPES, 150 mM NaCl, and pH 7.8). A protease inhibitor tablet (Roche), 0.2% Triton X-100, and 0.5 mg/mL lysozyme were added, and the cells were incubated on ice for 1 h. Complete lysis was achieved by sonication for 2 min on ice using a Branson Sonifier 450, followed by centrifugation for 20 min at 32,539 × g. The supernatant was incubated with 1 mL of Talon resin for 30 min at 4°C. Resin was washed with 3 × 10 mL lysis buffer, followed by elution with lysis buffer plus 75 mM imidazole (4 × 1 mL). The elution fractions containing protein were buffer exchanged back into lysis buffer, concentrated, flash frozen with 10% glycerol, and stored at −80°C.

For the purification of GST-tagged TclE, 30 mL cultures were inoculated with 1 mL overnight culture, grown at 37°C until an OD₆₀₀ = 0.6, then IPTG was added, and the cells were grown for an additional 20 h at 25°C. The cells were harvested by centrifugation, and the cell pellets were frozen at −80°C for at least 30 min. The cells were then thawed and resuspended in 1 mL lysis buffer (50 mM Tris pH 8, 150 mM NaCl, 0.5 mg/mL lysozyme, 2 mM EDTA, and one Roche protease inhibitor tablet per 10 mL). Complete lysis was achieved after a 15 min incubation at room temperature, followed by addition of DTT to 1.5 mM. Lysate was processed with several short sonication pulses with a microtip. Insoluble material was centrifuged at 7,000 × g, and the supernatant was combined with 30 µL of glutathione-agarose resin (slurry) at 4°C for 45 min (rotating). The resin was pelleted, and the beads were washed with GST buffer three times, and the peptide was eluted with 40 µL GST buffer plus 10 mM reduced glutathione. The eluant was either frozen at −80°C for later use or directly treated with tobacco etch virus (TEV) protease and ZipTipped (using the manufacturer’s instructions).

Activity of Tcl enzymes was tested in vitro. Twenty microliters of reactions containing 20 µM GST-TclE, 5 mM DTT, 2 mM ATP, 20 mM MgCl₂, 1 µM enzymes, and 1 µg TEV protease was allowed to react for 40 min at room temperature. Reactions were zip-tipped (using the manufacturer’s instructions) and analyzed by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry. The m/z values obtained from these experiments are for the z = 1 ion.

Orbitrap LC-MS for TclE truncation series

GST-TclE leader-truncated samples were co-expressed with TclIJN as described in the Tcl protein expression and purification section. GST-TclE leader truncations were purified using glutathione beads as described above, and TclE peptides were removed from the GST tag through TEV protease cleavage. Truncated TclE peptide samples were alkylated to cap any reduced cysteines using chloroacetamide at 20 mM. The peptides were then separated and measured via LC-MS on an Easy nLC 1,200 in connection with a Thermo Easy-spray source and an Orbitrap Fusion Lumos. Peptides were pre-concentrated with buffer A (3% acetonitrile and 0.1% formic acid) onto a PepMap Neo Trap Cartridge (particle size 5 µm, inner diameter 300 µm, and length 5 mm) and separated with an EASY-Spray HPLC column (particle size 2 µm, inner diameter 75 µm, and length 25 mm) with increasing buffer B (80% acetonitrile and 0.1% formic acid) gradient.

Samples were eluted using a gradient of 5%–22% B over 85 min (128 min for muscle), 22%–32% B over 15 min (22 min for muscle), with a wash of 32%–95% B over 15 min, which was held at 95% B for 15 min followed by a wash step consisting of two washes going from 95% B to 2% B over 3 min, holding at 2% B for 3 min, returning to 95% over 3 min and holding for 3 min were performed. Sample loading and equilibration were performed using the HPLC’s built-in methods. LC-MS only runs were performed using 2,400 V in the ion source, scan range of 375–1700 m/z, 30% RF Lens, Quadrupole Isolation, 8 *105 AGC Target, and a maximum injection time of 50 ms. The MS-based data-dependent acquisition method was set to a 3-second cycle time. MS1 scans were acquired by the orbitrap at a resolution of 120,000. Precursors with a charge >1 and <6 were selected for MS2 fragmentation. MS2 scans of collision-induced dissociation (CID) precursor fragments were detected with the linear ion trap at a scan rate of 33.333 Da/second with a Dynamic injection time. CID collisions were set to 30% for 10 ms. After three selections, a 60-second dynamic exclusion window was enabled; isotopes and unassigned charge states were excluded.

Data processing for label-free quantitation

Raw files were searched against a FASTA database for the Tcl operon (containing I, J, N, and E entries) with the E. coli proteome as a contaminant (Uniprot Reference UP000000625) using Peaks Studio analysis. The parent mass error tolerance was set to 10 ppm, and the fragment mass error tolerance was set to 0.5 Da. Cysteine carbamidomethylation, thiazoline, and thiazole were set as a variable modification, as well as methionine oxidation and pyro-glu from glutamine were set as variable modifications in the search. Digest mode was set to unspecific, and the peptide length range was set to 6–55 amino acids. The false discovery rate for peptide matches was set to 1%, and protein ID significance was set to −10log(P-value) ≥ 15. Label-free data were normalized using the TIC option in PEAKS, and then the total signal for the modified form was compared as a fraction of the total signal for the peptide of interest (File S1). The m/z values detected from these truncation experiments are for the z = 2.3 ions.

In-vitro activity of E. coli-purified Tcl proteins was detected with MALDI-TOF MS (see Mass spectrometry analysis of TclE processing section), while thiazole installation for leader truncation experiments was detected with Orbitrap LC-MS. Both data sets were carried out in different MS instruments which favored different ion distributions.

Modeling and optimization of Tcl protein structures

The sequences of TclI, TclJ, and TclN, as well as the protein hetero-dimers such as TclE:TclI, TclI:TclJ, and TclI:TclN, were submitted online to the AlphaFold2 Google colab (47, 48) for structural predictions. The top-ranked models were selected for each Tcl protein and dimer. Each model underwent optimization using the FastRelax algorithm (49) within PyRosetta. Energy calculations for each model were carried out using the ref2015 score function (50, 51) in PyRosetta. Binding energies were determined by subtracting the energies of the bound and unbound state models. Additionally, to assess the binding modes, we calculated the shape complementarities (52) of the complex models using Python logic and PyRosetta. These models were visualized using PyMOL, and a list of interacting residues (L187R, T188R, Y189R, H194R, I198R, C199R, N201R, I202R, S205R, E206R, F208R, L209R, Y210R, T212R, and S213R) was identified for use in mutagenesis experiments.

Size exclusion chromatography (SEC) of TclI and TclIJN complex

Purified complex was filtered through 0.2 µm cellulose filter (14,000 g, 2 min). All the samples were vacuum dried, and then resuspended in SEC buffer (100 mM sodium phosphate with pH 6.8 and 0.023% NaN₃) to make the final concentration to 1 µg/µL. The Agilent 1,260 Infinity HPLC System (Agilent, Santa Clara, CA) equipped with quaternary pump, manual injector, thermostatted column compartment, and diode array detector was used to carry out the analytical size exclusion chromatography. A Yarra-1.8 µm × 150 Å, 150 × 4.6 mm HPLC column (00F-4631-E0, Phenomenex, USA) was used for the separation of molecules. The Agilent system and column were equilibrated with 100 mM Sodium Phosphate with pH 6.8 and 0.023% NaN₃ at a flow rate of 0.3 mL/min at 25°C. The molecular weight calibration curve for SEC was obtained by running a protein standard mix containing bovine thyroglobulin (670 kDA), IgA (300 kDa), IgG (150 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and uridine (0.244 kDa; AL0-3042, Phenomenex, CA, USA) to relate the molecule’s size to elution volume. The injection volume for the protein standard and the protein samples was 5 µL, and the elution volume was measured using the UV detector. All the data were collected at 280 nm, with a reference wavelength of 600 nm.

Bioinformatics analysis of E1/YcaO and Ocin-ThiF-like domains

Protein sequences of TclI and TclJ were submitted to HHpred (38–40) in FASTA format using standard parameters, and the PDB70 and TIGRFAMs databases were analyzed for the presence of E1/YcaO/Ocin-ThiF-like domains using pairwise comparison of profile HMM (hidden Markov model). The results of this analysis provided a list of known homologs with well-defined domains, along with multiple sequence alignments that were used to define each domain on the TclI and TclJ proteins.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.00244-24.

File S1. aem.00244-24-s0001.xlsx.

Summary of peptide abundance of cysteine-to-thiazoline and cysteine-to-thiazole conversion analyzed by Orbitrap liquid chromatography mass spectrometry (LC-MS).

Supplemental material. aem.00244-24-s0002.docx.

Tables S1 and S2, Figures S1 to S7, and sequences.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.