Discovery and Characterization of a Class IV Lanthipeptide with a Non-overlapping Ring Pattern

Abstract

Lanthipeptides constitute a major family of ribosomally synthesized and post-translationally modified peptides (RiPPs). They are classified into four subfamilies based on the characteristics of their lanthipeptide synthetases. While over a hundred lanthipeptides have been discovered to date, very few of them are class IV lanthipeptides and the latter are all structurally similar. Here, we identified an uncharacterized group of class IV lanthipeptides using bioinformatics analysis. One representative pathway from Streptomyces sp. NRRL S-1022 was expressed in Escherichia coli, which generated a lanthipeptide with two non-overlapping rings that has not been reported for known class IV lanthipeptides. Further investigation into the biosynthetic mechanism revealed that multiple modification pathways are in operation in which dehydration and cyclization occur in parallel. While peptidases for maturation of class IV lanthipeptides have been elusive, two aminopeptidases encoded in the genome of Streptomyces sp. NRRL S-1022 were shown to process the modified peptide by the dual endopeptidase/aminopeptidase activity. This work opens doors to discover more class IV lanthipeptides with interesting structural features and biological activities.

INTRODUCTION

Lanthipeptides constitute a major family of ribosomally synthesized and post-translationally modified peptides (RiPPs).¹ Since the discovery of nisin in 1928, more than a hundred lanthipeptides have been uncovered with diverse bioactivity such as antibacterial, antifungal and antiviral functions.^{2, 3} Similar to other RiPPs, a lanthipeptide is initially synthesized as a precursor peptide encoded by a structural gene, with its leader peptide region present at the N-terminus and its core peptide region at the C-terminus.³ The leader peptide is recognized by the biosynthetic enzymes and post-translational modifications (PTMs) occur in the core peptide. After completion of the PTMs, the leader peptide is removed by a peptidase, which transforms the modified precursor peptide into the final product. Lanthipeptides are distinguished from other RiPPs by the presence of lanthionines and/or methyllanthionines, which are installed by lanthipeptide synthetases (Figure 1). According to the different features of these lanthipeptide synthetases, lanthipeptides are classified into four subfamilies: class I - class IV.^{1, 3} A typical class IV lanthipeptide synthetase (LanL) consist of three domains: an N-terminal lyase domain, a central kinase domain, and a C-terminal cyclase domain.⁴ The kinase and lyase domains catalyze phosphorylation and phosphate elimination reactions, respectively, which convert Ser/Thr residues to 2,3-didehydroalanines (Dha) and 2,3-didehydrobutyrines (Dhb), respectively. Then a Michael-type addition of the thiol of a Cys side chain to the unsaturated double bond of Dha/Dhb results in lanthionine/methyllanthionine. Depending on the face selectivity of the nucleophilic attack, two diastereomers, DL-(methyl)lanthionine (DL-(Me)Lan) and LL-(methyl)lanthionine (LL-(Me)Lan) can be formed.^{3, 5}

Figure 1:

Overview of lanthipeptide biosynthesis. A) Description of the two-step lanthipeptide biosynthesis. Dehydration of Ser/Thr side chains followed by nucleophilic attack of cysteine thiols on the resulting unsaturated double bonds yields the thioether bonds. A subsequent attack of enolate on another Dha can form labionin, a structure sometimes observed in class III lanthipeptides. In class I, the dehydration is achieved by LanB through glutamylation of Ser/Thr hydroxyl groups and subsequent glutamate elimination. In class II - IV, the hydroxyl groups are phosphorylated by LanM/LanKC/LanL prior to elimination. B) Classification of lanthipeptide synthetases.

While more than a hundred lanthipeptides have been discovered to date, only very few of them are members of class IV. The first known class IV lanthipeptide, venezuelin, was discovered from Streptomyces venezuelae (Figure 2).⁶ Although a few more class IV lanthipeptides were reported in subsequent studies, all are structural homologs of venezuelin. The biosynthetic mechanism for class IV lanthipeptides has been much less studied compared with class I and class II lanthipeptides. The in trans activity of individual lyase and kinase domains from the venezuelin synthetase VenL was reconstituted in vitro and residues crucial for the activity of the lyase domain were identified.^{4, 7} Recently, the kinase domain of a VenL ortholog was shown to recognize a putative α-helical stretch at the center to the N-terminal region of the leader peptide and the dehydration directionality was determined as N-to-C.⁸ However, owing to the relatively complicated pattern of venezuelin, the cyclase domain of VenL has not been characterized to date, leaving the cyclization machinery in class IV lanthipeptide biosynthesis largely unexplored. Moreover, unlike class I and class II lanthipeptides, which usually have designated endopeptidases to remove leader peptides, peptidases are missing in all characterized class IV lanthipeptide biosynthetic gene clusters (BGCs), which renders the mechanism for leader peptide removal for class IV lanthipeptides unclear. Recently, a Zn-dependent protease AplP encoded in the BGC of the class III lanthipeptide NAI-112 was found to remove its leader peptide by an unprecedented consecutive endo/aminopeptidase activity.⁹ Although the authors pointed out that AplP homologs were also present in bacterial strains that produce venezuelin-like lanthipeptides, it remains to be demonstrated whether such a mechanism for leader peptide removal also generally applies in class IV lanthipeptides.

Figure 2:

A) Organization of the sfl biosynthetic gene cluster. B) Precursor peptide sequences of venezuelin and S-1022. C) Structure of venezuelin and the C-terminal fragment of trypsin-digested SflL-modified SflA confirmed by MS/MS. Corresponding b-/y-ions are highlighted in red ([M+H]⁺) and blue ([M+2H]²⁺) in the spectrum. D) GC-MS trace demonstrating the stereochemistry of the thioether crosslinks in modified SflA.

In this work, we expanded the current knowledge for class IV lanthipeptide biosynthesis. A group of class IV lanthipeptide BGCs were identified through bioinformatics analysis and one representative cluster from Streptomyces sp. NRRL S-1022 was refactored and heterologously expressed in Escherichia coli. The product has two non-overlapping rings, a new pattern that distinguishes this group of class IV lanthipeptides from venezuelin and its homologs. In order to further study the mechanism of ring formation, the LanL was expressed and purified for in vitro assays. Analysis of the intermediates by MS/MS demonstrated that multiple biosynthetic routes differing in the order of dehydration and cyclization occur in parallel. In addition, four AplP homologs were identified in the genome of Streptomyces sp. S-1022 and in vitro characterization revealed that only two were able to remove the leader peptide from the modified LanA peptide. Overall, we identified a class IV lanthipeptide with non-overlapping ring pattern and resolved some long-standing questions in class IV lanthipeptide biosynthesis.

RESULTS

Reconstitution of Lanthipeptide Biosynthesis in vivo and in vitro

To identify BGCs encoding class IV lanthipeptides with novel ring topologies, we analyzed 20 reported lanthipeptide gene cluster families (GCFs) that have more than 10 BGCs in each¹⁰ and identified three that belong to class IV lanthipeptides, including one (GCF147) that produces venezuelin homologs.⁸ Among the other two (GCF7 and GCF128), GCF128 is the larger family in the current genomes and was chosen for further investigation.¹¹ Only genes encoding putative LanAs and LanLs were conserved across the BGCs of GCF128, whereas proteases and transporters were not identified (Figure 2A). All LanA sequences in GCF128 appear to be highly conserved and each has many serine and threonine residues in the C-terminal region but only two cysteine residues (Figure 2B).

We chose one gene cluster from Streptomyces sp. NRRL S-1022 in GCF128 for further investigation. We termed this locus sfl for the second-most abundant class IV (four) lanthipeptide BGC. The substrate LanA (SflA) and synthetase LanL (SflL) were coexpressed in E. coli by using the pathway refactoring method we previously developed.¹² Sequence analysis shows that SflL has all the features of a class IV synthetase including conserved Zn-binding residues (Figure S1). The modified SflA was isolated by immobilized metal affinity chromatography (IMAC) followed by high-performance liquid chromatography (HPLC) and analyzed by matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) as previously described.⁸ A modified precursor peptide was observed that had been dehydrated twice while a derivatization assay with N-ethylmaleimide (NEM, a Cys selective alkylating agent) confirmed that no free thiol groups were present in the product (Figure 2B and Figure S2). To investigate the ring pattern, the modified SflA was digested by trypsin and the structure of the C-terminal fragment was analyzed by MS/MS. The fragmentation pattern revealed that two non-overlapping rings were formed between Cys42 and Thr49, as well as between Cys56 and Thr52, which was supported by site-directed mutagenesis experiments in which Thr49 and Thr52 were mutated (Figure 2C). The ring pattern is clearly distinct from venezuelin not only because of two instead of four rings, but also because of the size of the rings and the direction of cyclization.

Partially due to the instability of SflA in E. coli (Figure S4), the titer of modified SflA produced in vivo was less than 10 μg/L. In order to obtain sufficient amounts of compound for characterizing the stereochemistry of the MeLan residues, efforts were spent to optimize the heterologous expression of SflA and SflL and to reconstitute the PTM activity in vitro. Briefly, the expression of SflA was shortened to 1 hour which significantly increased the titer of SflA to 0.25 mg/L. Moreover, maltose binding protein (MBP) was fused to the N-terminus of SflL to increase its solubility. By following previously established reaction conditions,⁸ the PTM process was fully reconstituted in vitro, which generated modified SflA with the same ring pattern as described above. Over 10 mg of modified SflA was successfully obtained through in vitro biosynthesis, which was then hydrolyzed and derivatized for stereochemistry analysis by gas chromatography-mass spectrometry (GC-MS).⁵ GC-MS analysis in comparison with synthetic standards clearly indicated that both MeLan residues have the DL configuration, which is also the stereochemistry of the thioether crosslinks in venezuelin (Figure 2D).¹⁰

SflL Catalyzes Multiple Biosynthetic Routes in Parallel

The mechanism of substrate recognition and the order of dehydration were previously investigated for LanLs encoded in venezuelin-like BGCs,⁸ but the order of cyclization for class IV lanthipeptide biosynthesis remains unknown. Compared with venezuelin, modified SflA has a much simpler ring pattern facilitating investigation of the cyclization process. Following the setup for in vitro biosynthesis as described above, aliquots from the reaction were quenched at different time points and analyzed by MALDI-TOF MS. Intermediates were observed at 10 min, which were then derivatized by NEM and digested by trypsin. As trypsin was not able to cleave the peptide at Arg53 after the C-terminal ring is formed, intermediates containing either the N-terminal or C-terminal ring can be distinguished (Figure 3A). Surprisingly, ions corresponding to intermediates in which either the N-terminal ring or the C-terminal ring had been formed first were observed, together with peptide fragments containing only dehydrated threonine residue(s). Their structures were confirmed by MS/MS (Figure 3 and Figure S5), further illustrating that dehydration and cyclization reactions occur in parallel. Thus, unlike most lanthipeptide synthetases that have been studied,^{8, 13–18} SflL does not appear to have a defined order of post-translational modifications.

Figure 3:

MS analysis of intermediates in the biosynthesis of modified SflA in vitro. A) MALDI-TOF MS results of the in vitro reaction quenched at 10 min and 60 min. Peptides were derivatized by NEM and digested by trypsin. Peaks corresponding to the molecular weights of fully and partially modified peptides are highlighted in blue and red respectively. B) MS/MS fragmentation patterns of the trypsin-digested fully/partially modified SflA C-terminal fragments. The corresponding MS/MS spectra with annotated b-/y-ions are shown in Figure S5.

To further confirm the independency of ring formation, SflA mutants in which the two Thr residues involved in cyclization were mutated to Ala (T49A and T52A) were reacted with SflL following the standard reaction condition used for wild type SflA. T49A was converted to a peptide with only a C-terminal ring (Figure 4). However, T52A was only partially converted to a peptide with the N-terminal ring and intermediates with one or two-fold dehydration were also observed. Of note, in the intermediate with two dehydrations, Thr51 was dehydrated, which was not observed when the wild type SflA was used (Figure 4). Based on the results from in vitro characterization of SflL, we propose the post-translational modifications take place mainly with C-to-N-terminal directionality while alternative biosynthetic routes also exist (Figure 4C).

Figure 4:

Independency of ring formation confirmed by point mutation in SflA. A) MALDI-TOF MS results of the in vitro reaction for SflA mutants (T49A and T52A) quenched at 60 min. The peptides were processed and analyzed following the same procedure as described for Figure 3. Fully/partially modified peptides are highlighted in red. B) MS/MS fragmentation patterns of the C-terminal fragments of SflL-modified SflA mutants. The corresponding MS/MS spectra with annotated b-/y-ions are shown in Figure S5. C) Proposed biosynthetic mechanism of modified SflA. The major and alternative biosynthetic routes are indicated by green arrows and dotted arrows, respectively.

Previous studies have shown that dehydration reactions at different positions can occur in parallel for different classes of lanthipeptides, but cyclizations are usually thought to have a strict order.^{13–15, 18, 19} Jungmann and coworkers characterized the biosynthesis of curvopeptin, a class III lanthipeptide which also contains two rings.¹³ Through site-directed mutagenesis experiments, the authors demonstrated that the order of cyclization strictly followed a C-to-N directionality. Such tight control over the cyclization order also exists in the biosynthesis of catenulipeptin, another class III lanthipeptide which contains two labionin crosslinks.¹⁶ However, our data clearly indicate that SflL can catalyze the formation of the N-terminal ring and C-terminal ring independently, though some kinetic preference for the C-terminal ring is observed. To the best of our knowledge this represents the first example showing cyclization of different rings can occur in parallel for lanthipeptide biosynthesis.

Zn-dependent Aminopeptidases Encoded in the Genome are Responsible for Leader Peptide Removal

Similar to all characterized class IV lanthipeptides, no genes encoding peptidases were identified in the BGC of Streptomyces sp. NRRL S-1022. Aminopeptidases are assumed to be responsible for leader peptide removal for class IV lanthipeptides, and the aminopeptidase Apl was recently shown to be involved in the biosynthesis of the class III lanthipeptide NAI-112.¹¹ Therefore, four AplP homologs (StrS1022P1-StrS1022P4) that are encoded in the genome of Streptomyces sp. NRRL S-1022 were selected for further characterization. To determine which aminopeptidase(s) might be involved in leader peptide removal from SflA, the genes encoding all four putative aminopeptidases were cloned into pET28a and expressed in E. coli BL21(DE3). Both unmodified and modified SflA were used as substrates and time-course analysis of digested peptides was performed by MALDI-TOF MS (Figure 5). Three out of the four aminopeptidases (P2-P4) were observed to digest unmodified SflA. However, only StrS1022P2 and StrS1022P4 exhibited activity towards modified SflA while the activity of StrS1022P3 was completely abolished, possibly because of interference by the C-terminal post-translational modifications (Figure 5A). By assigning observed ions to peptide fragments, we identified one initially formed N-terminal fragment (18N) from the StrS1022P2-digested modified SflA peptide and two initially formed N-terminal fragments (18N and 20N) from the StrS1022P4-digested peptide, which confirmed that these enzymes have endopeptidase activity (Figure 5B and 5C). Upon longer reaction times, the initially formed C-terminal fragments were converted to shorter peptides through aminopeptidase activity (Figure 5).

Figure 5:

A) Summary of the activity of StrS1022P1-StrS1022P4 towards both modified SflA and unmodified SflA mutants. B and C) Peptide fragments of SflL-modified SflA generated by StrS1022P2 and StrS1022P4 at various time points. D) Activity of StrS1022P2 towards SflA mutant PDLLK.

To identify regions for peptidase recognition, the conserved residues of SflA and its homologs were identified through alignment using the basic local alignment search tool (BLAST) (Figure S6A).²⁰ A series of SflA variants were prepared accordingly (Figure S6). Most of these point mutations had little impact when incubated with StrS1022P2, although decreased activity was observed in some cases, such as with the V5A and G18E variants (Figure S6B). Because Lys10 was identified as the endopeptide cleavage site (see above), we next analyzed if Lys10 and its flanking residues might serve as recognition sequence for StrS1022P2. Therefore, Lys10 was mutated both individually to Ala and together with four residues located immediately N-terminal to Lys10 (P6A/D7A/L8A/L9A/K10A, termed the PDLL mutant from here onwards). While little effect on cleavage was observed when K10A was incubated with StrS1022P2, cleavage was abolished when the PDLL-mutant was used as the substrate (Figure 5). Thus, the PDLLK region might serve as a recognition site for StrS1022P2. In addition, these results also suggest that AplP-like aminopeptidases strictly follow a two-step process when removing leader peptides, in which the endopeptidase activity occurs first and determines if the follow-up aminopeptidase activity can occur efficiently.

While AplP-like aminopeptidases are metalloenzymes with conserved Zn-binding sites, sequence alignment revealed that the metal-binding residues are missing in StrS1022P4 (Figure S7). Therefore, we analyzed both StrS1022P2 and StrS1022P4 for the presence of putative metals including Mg, Mn, Fe, Co, Cu and Zn by inductively coupled plasma mass spectrometry (ICP-MS). The data revealed that StrS1022P4 is a Zn-binding aminopeptidase but appears to have lower binding affinity compared with StrS1022P2, which has the same Zn-binding site as AplP (Figure 6A). To further support that the activity of StrS1022P4 is Zn-dependent, StrS1022P4 was dialyzed against EDTA-containing buffer to yield its apoenzyme form. After coincubation with metal cations (Mg²⁺, Mn²⁺, Fe³⁺, Co²⁺, Cu²⁺ and Zn²⁺) and SflA, enzyme activity was only restored by Zn²⁺ (Figure 6B). Thus, StrS1022P4 is a Zn--dependent aminopeptidase.

Figure 6:

The activity dependence of StrS1022P4 on Zn²⁺. A) ICP-MS analysis of StrS1022P2 and StrS1022P4. The protein concentrations are 108 mg/mL and 17 mg/mL for StrS1022P2 and StrS1022P4, respectively. The detected metal concentrations are in ppb (μg/L). The equivalent of Zn retained in StrS1022P2 and StrS1022P4 are 0.1 and 0.02 respectively. B) Activity of Apo-StrS1022P4 to SflA supplied with Zn²⁺. IMAC purified StrS1022P4 was dialyzed in 20 mM potassium phosphate buffer with 500 mM KCl and 10 mM EDTA, pH 7.6 at 4 °C overnight, which was then concentrated by Amicon® Ultra Centrifugal Filters. The reaction was set up by following the condition used for testing the holoenzyme form StrS1022P4, while ZnCl₂ was added to a final concentration of 5 mM.

Another interesting feature for StrS1022P2 and StrS1022P4 is that both enzymes exhibited no detectable activity in the first 24 h while processing most of the substrate in the next 24 h (Figure 5). We therefore wondered whether the enzyme went through a self-activation process. By incubating StrS1022P2 without any peptide substrate in the first 24 h and supplying substrate afterwards, we observed that the substrate was digested immediately in the next 24 h, suggesting some unknown changes happened in the first 24 h to StrS1022P2. Since most metalloproteases are zymogens and are activated by a self-cleavage event, a maltose binding protein (MBP) tag was fused to the N-terminus of StrS1022P2 to monitor possible self-cleavage events on SDS-PAGE. In addition, time course analysis of the activity of MBP-StrS1022P2 with both amino acid-pNA and unmodified SflA was performed alongside. Although selfcleavage was observed after 72 h incubation, it has little correlation with the activity change (Figure S8). Thus, the mechanism by which StrS1022P2 is activated is not clear.

In addition to the successful reconstitution of the protease activity in vitro, the modified SflA with its leader peptide removed was tested for its antimicrobial potential. A range of Gram-positive and Gram-negative bacteria, as well as fungi, were used as indicator strains and the bioactivity was screened by standard diffusion assay (Table S1). No significant antimicrobial activity was observed, suggesting that either these compounds have non-antimicrobial activities or they have activities against organisms that were not tested.

CONCLUSION

Lanthipeptides are classified into four categories based on different features of the corresponding lanthipeptide synthetases. Class IV lanthipeptides are the least characterized class and currently known members are all structural homologs to venezuelin, the prototype class IV lanthipeptide from Streptomyces venezuelae.⁶ In this study, we have characterized a class IV lanthipeptide BGC identified from Streptomyces sp. NRRL S-1022 by genome mining. A previous bioinformatics study revealed that this BGC belongs to the second-most abundant class IV lanthipeptide BGC and the BGC was therefore termed herein as sfl.¹¹ Heterologous expression of the putative precursor peptide (SflA) and the lanthipeptide synthetase (SflL) in E. coli produced modified SflA and structural characterization by MS/MS demonstrated two non-overlapping MeLan crosslinks, a different ring pattern to all characterized class IV lanthipeptides. Further characterization by GC-MS elucidated both MeLan residues have the DL-configuration, as found for venezuelin.

Previous sequence analysis of GCF128 showed that only sflA and sflL are conserved across all gene clusters in this family, which implies that no other biosynthetic genes are involved in the modification of SflA.¹⁰ However, the SflL-modified C-terminal part of SflA exhibited no antimicrobial activity to the indicator strains in the standard diffusion assay (Table S1). Both the wild type native producer and its Δsfl counterpart were cultured in multiple mediums, but we were unable to detect any metabolites that are related to sfl cluster from the cell extracts by MALDI-TOF MS. Although more comprehensive bioactivity screening might be helpful for elucidating the function of SflL-modified SflA, we cannot rule out the possibility that genes encoded outside the sfl cluster might also be involved in the post-translational modification of SflA. However, there are no such examples from prior studies and more likely is the possibility that the product has a different function. Notably, lack of antimicrobial activity has been reported for most class III and IV lanthipeptides.³

Co-expression of SflA and SflL in E. coli only produced the modified peptide with a low titer, which is likely caused by the unwanted expression of an aminopeptidase encoded in the E. coli genome that is homologous to AplP.⁹ In vitro reconstitution of the SflA modification not only avoided the substrate/product instability in vivo, this strategy also significantly increased the titer, and provided the opportunity to characterize the order of the multiple biosynthetic reactions catalyzed by SflL. Through quenching the reaction at multiple time points and analyzing the structures of intermediates by MS/MS, we propose that the cyclization of SflA occurs via multiple parallel pathways, a hypothesis that was further supported by using substrates with residues involved in cyclization mutated. Although several studies have shown that lanthipeptide synthetases can catalyze dehydration reactions at different positions in parallel, to the best of our knowledge, this study provides the first evidence that cyclization can also take place via multiple parallel pathways in lanthipeptide biosynthesis. The non-overlapping ring pattern likely aids independent formation of each ring, but other lanthipeptides with non-overlapping rings are still formed by ordered cyclization processes.²¹

To resolve the long-standing question of leader peptide removal in class IV lanthipeptide biosynthesis, we identified four aminopeptidases (StrS1022P1-P4) encoded in the genome of Streptomyces sp. NRRL S-1022 that are homologous to AplP, a recently characterized peptidase involved in the biosynthesis of a Class III lanthipeptide - NAI-112.⁹ In vitro characterization using SflA and its mutants as substrates revealed that StrS1022P2 and StrS1022P4 were able to process the modified SflA by their endopeptidase/aminopeptidase activity, a mechanism of action similar to AplP. In addition, the Zn-dependency of StrS1022P4 that has a non-canonical Zn-binding site was confirmed by ICP-MS analysis as well as activity assay after metal cofactor substitution. Interestingly, while StrS1022P2 and StrS1022P4 have no significant preference to the modified and unmodified SflA, the activity of StrS1022P3 was completely abolished when using the SflL-modified SflA as substrate. This implies diverse biological roles of AplP-like aminopeptidases and may possibly explain why multiple copies of AplP-like aminopeptidases are present in class IV lanthipeptide producer strains.

MATERIALS AND METHODS

Bacterial Strains and Materials

E. coli DH5α cells were used for cloning and mutagenesis. E. coli BL21(DE3) cells were used for heterologous expression. The Streptomyces sp. NRRL S-1022 strain used in this study was obtained from the USDA ARS collection (Peoria, IL). Oligonucleotide primers were synthesized by Integrated DNA Technologies (Coralville, IA). Enzymes for cloning were obtained from New England Biolabs (Ipswich, MA), and kits for plasmid miniprep and gel extraction from QIAGEN (Germantown, MD). Plasmids were verified via sequencing carried out by ACGT, Inc. (Wheeling, IL). Antibiotics, mediums for cell cultivation and other chemicals were purchased from Thermo Fisher Scientific (Waltham, MA), and trypsin from Worthington Biochemical Corp. (Lakewood, NJ). Millipore C18 Ziptips were purchased from MilliporeSigma (Burlington, MA). HisTrap HP column was purchased from GE Healthcare (Chicago, IL) and reverse phase HPLC columns from Phenomenex (Torrance, CA).

Heterologous Production of Lanthipeptide S-1022 in E. coli BL21(DE3)

E. coli BL21(DE3) transformed by plasmid coexpressing SflA and SflL was initially grown in Lysogeny Broth (LB) with 50 μg/mL kanamycin at 37 °C. 10 mL of the overnight culture was then used for inoculating 1 L Terrific Broth (TB) medium for expression. Cells were grown in TB at 37 °C till OD600 reached 0.6-0.8, then IPTG was added to a final concentration of 0.1 mM. The cell culture was then cooled down to 18 °C and continued to grow for an extra 18-20 h. Cells were then harvested by centrifugation and processed as described elsewhere.⁵ Briefly, cells were resuspended in LanA start buffer (20 mM NaH₂PO₄, pH 7.5 at 25 °C, 500 mM NaCl, 0.5 mM imidazole, 20% glycerol) and lysed by sonication. The cell lysate was then centrifuged at 23,700×g for 30 min at 4 °C and the supernatant was purified by a Ni-NTA column. Eluates were desalted by C18 solid phase extraction (SPE) columns and further purified by HPLC using a C5 reversed phase semiprep column. H₂O and acetonitrile all suppled with 0.1% TFA (v/v) were used as mobile phases A and B respectively and the HPLC was performed with the following gradient at 3 mL/min: linear increase from 5% to 80% over 30 min, followed by a linear increase from 80% to 100% over 5 min and holding at 100% for another 3 min. In this way, approximately 10 μg modified SflA could be isolated.

In vitro Reconstitution of Lanthipeptide Synthesis and Analysis of Intermediates

The reaction for synthesizing modified SflA was set up as follows: 20 μL 2x buffer (300 mM NaCl, 50 mM HEPES, pH7.5), 2 μL TCEP stock solution (20 mM in H₂O), 2 μL of MgCl₂ stock solution (200 mM in H₂O), 2 μL of ATP stock solution (50 mM in H₂O), 1.6 μL unmodified SflA stock solution (10 mg/mL in 0.1% TFA, 50 μM as the final concentration in the reaction) and MBP-SflL added to a final concentration of 2.5 μM. The reaction was run at room temperature and after 10 min and 60 min 20 μL aliquots were quenched by adjusting the pH to 1-2 by adding formic acid. The aliquots were then lyophilized to remove the solvents and labelled by NEM following the protocol described previously ⁸. The labelled peptides were then desalted by ziptip, lyophilized and digested by trypsin. Peptide fragments were analyzed by MALDI-TOF MS and MS/MS was performed on a Thermo Q Exactive HF-X Mass Spectrometer.

In vitro Analysis of Aminopeptidases

To characterize the proteolysis activity of aminopeptidases, 100 μM of modified or unmodified SflA was incubated with 10 μM aminopeptidase in 20 mM Tris•HCl buffer at pH 7.5. The reaction was run at room temperature and 1 μL aliquots were removed at multiple time points, diluted by 20 μL water before desalted by ziptip and peptide fragments were analyzed on MALDI-TOF MS.