The Creatininase Homolog MftE from Mycobacterium smegmatis Catalyzes a Peptide Cleavage Reaction in the Biosynthesis of a Novel Ribosomally Synthesized Post-translationally Modified Peptide (RiPP)*

Most ribosomally synthesized and post-translationally modified peptide (RiPP) natural products are processed by tailoring enzymes to create complex natural products that are still recognizably peptide-based. However, some tailoring enzymes dismantle the peptide en route to synthesis of small molecules. A small molecule natural product of as yet unknown structure, mycofactocin, is thought to be synthesized in this way via the mft gene cluster found in many strains of mycobacteria. This cluster harbors at least six genes, which appear to be conserved across species. We have previously shown that one enzyme from this cluster, MftC, catalyzes the oxidative decarboxylation of the C-terminal Tyr of the substrate peptide MftA in a reaction that requires the MftB protein. Herein we show that mftE encodes a creatininase homolog that catalyzes cleavage of the oxidatively decarboxylated MftA peptide to liberate its final two residues, including the C-terminal decarboxylated Tyr (VY*). Unlike MftC, which requires MftB for function, MftE catalyzes the cleavage reaction in the absence of MftB. The identification of this novel metabolite, VY*, supports the notion that the mft cluster is involved in generating a small molecule from the MftA peptide. The ability to produce VY* from MftA by in vitro reconstitution of the activities of MftB, MftC, and MftE sets the stage for identification of the novel metabolite that results from the proteins encoded by the mft cluster.

Peptide-derived natural products are implicated in myriad biological functions and hold great potential as novel therapeutic agents. Despite the enormous sequence diversity and wide range of modifications that are encountered, these metabolites are either synthesized by non-ribosomal peptide synthetases (1,2) and matured by tailoring enzymes or ribosomally synthesized and post-translationally modified pep-tides (RiPPs) 3 (3). RiPPs have emerged as an important class of peptide-derived natural products. The gene encoding the RiPP precursor peptide is often co-localized with genes that encode the maturase enzymes, making discovery of novel RiPP biosynthetic clusters by bioinformatics possible. Given the diverse structures of mature RiPPs, a wide variety of novel RiPP tailoring reactions are suspected to exist in nature, although the mechanisms underlying many of these reactions remain unknown.
In 2011, Haft (19) proposed the presence of a novel peptidederived metabolite in the Mycobacterium genus based solely on the clustering of the mftA gene that encodes for a short peptide with the mftB and mftC genes that encode a short protein and a member of the radical SAM superfamily, respectively (Fig. 1A). It was noted that this clustering of genes in the Mycobacterium genus is analogous to the clustering of the pqqA, pqqD, and pqqE genes, which are required for the biosynthesis of pyrroloquinoline quinone (PQQ) cofactor as well as bacteriocin biosynthetic gene clusters (19). PQQ is a small molecular weight redox mediator derived from the 23-amino acid peptide encoded by the pqqA gene (20 -22). By analogy, it was proposed that the newly described cluster produces a hypothetical metabolite, mycofactocin, which may be a putative small molecule redox cofactor derived from the peptide MftA (19).
We recently reported that the radical SAM enzyme MftC catalyzes the oxidative decarboxylation of the C-terminal Tyr residue of the MftA peptide in a reaction that also requires the MftB accessory protein (Fig. 1B) (13). Subsequently, Latham and co-workers (14) demonstrated that MftB binds MftA with a submicromolar K D (ϳ100 nM) and MftC with a low micromolar K D (2 M). Although the activity of MftC is not analogous to that of the radical SAM enzyme PqqE in PQQ biosynthesis (12), it is still possible that the formation of the oxidatively decarboxylated MftA peptide represents the first step in a biosynthetic pathway to a novel metabolite.
In addition to the mftB and mftC genes, the mftA gene is nearly universally observed clustered with genes that encode proteins that display amino acid sequence similarity to a heme/ flavin dehydrogenase (mftD), creatininase (mftE), and glycosyltransferase (mftF) (Fig. 1A). We hypothesized that if MftC does indeed catalyze the first transformation in the biosynthesis of a novel MftA-derived metabolite then one of these gene products may modify the oxidatively decarboxylated MftA product. One would predict that if the putative metabolite derived from MftA is a small molecule redox cofactor analogous to PQQ then there must be a step subsequent to MftC that further processes the modified MftA.
Many RiPP biosynthetic pathways encode peptidases that engage in processing the propeptide into the mature product. Although a peptidase is not encoded by the Mft cluster, we were intrigued by the presence of a gene that encodes a protein with similarity to the creatininase subfamily of amidohydrolases. The prototypical creatininase hydrolyzes an amide bond in the cyclic metabolite creatinine to produce creatine (23)(24)(25). We reasoned that it might be possible that the creatininase homolog encoded in this cluster potentially processes the decarboxylated MftA to a smaller metabolite en route to the mature natural product.
Herein we report the biochemical characterization of the creatininase homolog encoded by the gene LI98_07105, here-after referred to as mftE, clustered with the mftA gene in the genome of Mycobacterium smegmatis. Indeed, MftE catalyzes the hydrolysis of the amide bond between Gly 28 and Val 29 of MftA peptide only after it has undergone the C-terminal oxidative decarboxylation. Chemical derivatization of the product of MftE is consistent with the valine containing a primary amine. Moreover, we observed that the oxidatively decarboxylated tyrosine displays an unexpected reactivity toward chemical modifiers, which may foreshadow the function of the mature metabolite.

Results
Expression and Purification of MftE-To determine the function of MftE, the homolog from M. smegmatis ATCC 7000840 (LI98_07105) was cloned, the His 6 -tagged recombinant protein was expressed in Escherichia coli, and the recombinant protein was purified to Ն95% purity (as judged by SDS-PAGE) by nickel affinity chromatography. Amino acid analysis of the protein was carried out to obtain a correction factor of 1.9 Ϯ 0.3 for the Bradford assays with BSA as standard. ICP-MS analysis of the protein revealed the presence of 0.39 Ϯ 0.06 mol of zinc, 0.21 Ϯ 0.03 mol of iron, and 0.042 Ϯ 0.002 mol of manganese per mol of MftE.
MftE Modifies the Decarboxylated MftA-The purified enzyme was incubated with MftA in the presence of MftB and MftC under various conditions and analyzed for MftA modification by UHPLC-MS (Fig. 2). Although unmodified and decarboxylated MftA co-elute at approximately 11 min, the two are readily distinguished by mass spectrometry monitoring at m/z 1661 and 1638, which correspond to the ϩ2 charge state mass envelopes of the unmodified and decarboxylated peptide, respectively ( Fig. 2A). When MftA is incubated with MftB, MftC, and reductant in the absence of SAM, the peptide remains unmodified in the presence of MftE. The deconvoluted mass spectrum (supplemental Fig. S1) 28 and Val 29 to liberate a shortened MftA peptide containing residues Glu 1 -Gly 28 (MftA ). The theoretical m/z for [M ϩ H] ϩ of an MftA peptide containing residues 1-28 (MftA(1-28)) is 3058.3867, which is within 2.6 ppm error of that observed in the deconvoluted mass spectrum of the peptide experimentally (see Fig. 2B).
To determine whether MftE is indeed catalyzing the hydrolysis of the amide bond, the chromatograms were analyzed for a peak corresponding to the putative valine-oxidatively decarboxylated tyrosine (VY*) product. In the UV traces of chromatograms corresponding to the reactions shown in Fig. 2, a new peak that elutes at approximately 7.3 min is observed that is only present when all of the reaction components are present (Fig. 3A). This species has an m/z of 235.1438, which is within 3.8 ppm of the theoretical mass (m/z ϭ 235.1447 for [M ϩ H] ϩ ) of the VY* resulting from the hydrolysis of the oxidatively decarboxylated MftA peptide (Fig. 3B). The VY* is not present in the UV or extracted ion chromatogram of the reactions where either SAM or MftE is omitted (Fig. 3C). Collectively, these results are consistent with MftE catalyzing the hydrolysis of the amide bond between Gly 28 and Val 29 to generate the VY* dipeptide.
To confirm the identity of the MftE catalyzed reaction, we probed the reaction with two MftA peptides that were isotopically enriched at the C-terminal Tyr only ([ 13 C 9 , 15 30 MftA peptide with all of the reaction components (Fig. 5A). The UHPLC-MS data show that when MftA is uniformly labeled with 13 C and 15 N at position 30 the peak for the putative VY* product shifts 9.0238 atomic mass units to m/z 244.1676, which is consistent with the isotopically enriched product containing eight 13 C and one 15  HCD MS/MS Fragmentation of VY*-To further probe the structure of the VY*, we subjected the unlabeled and isotopically enriched VY* dipeptide to HCD fragmentation (35% collision energy) in the mass spectrometer (Fig. 6A). Only two peaks were observed in the resulting MS/MS fragmentation spectrum. The species with m/z values of 217.1327 and 136.0750 in the spectrum obtained with unlabeled VY* are consistent with the loss of water (theoretical m/z ϭ 217.1341, 6.5 ppm error) and decarboxylated tyrosine (theoretical m/z ϭ 136.0762, 8.8 ppm error) (Fig. 6A, a). The MS/MS fragmentation spectrum of each of the isotopically enriched VY* dipep-   30 MftA, respectively (Fig. 6A, b and c). These fragmentation results establish the identity of VY*. Chemical Modification of the VY*-The data presented thus far are consistent with MftE catalyzing the hydrolysis of the amide bond between Gly 28 and Val 29 to generate the VY* product. One can draw at least three distinct structures of the VY* dipeptide product, each of which has a theoretical m/z of 235.1447 corresponding to [M ϩ H] ϩ (Fig. 6B). Although the ability to fragment to Val and Tyr* strongly support the notion that it retains a free N terminus at Val, we carried out modification studies to unambiguously establish this and rule out forms where the amino group on valine cyclizes with Tyr*.
PITC and 1-fluoro-2,4-dinitrophenyl-5-L-alanine amide (Marfey's reagent) are commonly used to chemically modify primary amines on amino acids. In the presence of PITC, a new SAM-dependent peak is observed at 14 (Fig. 7B). The fragmentation of the molecular ions by HCD leads to five related fragments (Fig. 7C), which are consistent with the structures shown in supplemental Fig. S4. Taken together, the fragmentation data allow us to assign the peak at m/z 370.1572 and the corresponding peaks with labeled peptides to PITC-VY* (Fig. 7D).
We next examined the modification of VY* by Marfey's reagent and the corresponding fragmentation products when unlabeled or isotopically enriched peptides were used. Fig. 8A Fig. S4. D, the proposed structure and corresponding theoretical masses of the PITC-derivatized VY* dipeptide based on the HCD fragmentation results with the natural abundance and isotopically enriched MftA peptides.
across the peak indicated by • in the extracted ion chromatograms shows that the observed mass in each spectrum is consistent with the absence or presence of the appropriate 13 C and 15 N isotopes, and each is within 3.2 ppm error relative to the theoretical mass for [M ϩ H] ϩ (Fig. 8B). The HCD fragmentation of the molecular ion peak reveals seven related fragments (Fig. 8C). Surprisingly, the fragments in the unlabeled and the isotopically enriched Marfey's reagent-modified VY* are most consistent with the Marfey's reagent modifying the oxidatively decarboxylated tyrosine. We note that Tyr* is an enamide, which although not as reactive as an enamine is still nucleophilic and can react with Marfey's reagent to alkylate the ␣-carbon. It is also possible that alkylation takes place at the ␤-carbon. Because the MS data do not distinguish between these in the foregoing discussion, Fig. 8 shows both the ␤-carbon (D) and ␣-carbon (E) adducts, and supplemental Fig. S5 shows the assignment of the fragments.
To further probe this unexpected reactivity between Marfey's reagent and the VY* product, we first incubated the VY* dipeptide with PITC and then treated the PITC-modified VY* with Marfey's reagent. The extracted ion chromatograms in Fig. 9A show a peak with a retention time of approximately 16 min in each of the spectra that is consistent with the theoretical m/z corresponding to [M ϩ H] ϩ of the unlabeled or the isotopically enriched VY* containing both PITC and Marfey's reagents. Inspection of the mass spectra reveals that the observed m/z of each species is within 3.5 ppm error of the theoretical m/z of each PITC-Marfey's reagent-modified VY* dipeptide (Fig. 9B). The molecular ion peaks were subjected to HCD fragmentation (Fig. 9C). Supplemental Fig. S6 shows the structures that are most consistent with the ions. As for the VY* dipeptide modified by Marfey's reagent alone, the peaks with m/z values of 343.1032 and 388.1246 in the unlabeled VY* PITC-Marfey's reagent modified product and the corresponding peaks in the isotopically enriched product are consistent with Marfey's reagent reacting with the oxidatively decarboxylated tyrosine instead of the valine. The data clearly show that the peptide is modified by PITC at the N terminus and Marfey's reagent at the decarboxylated tyrosine. As noted above, we cannot distinguish between modification at the ␣or ␤ positions, and so both possibilities are shown in the figure (Fig. 9, D and E).
MftB Is Not Required for the Reaction Catalyzed by MftE-It has been established that MftC catalyzes the oxidative decarboxylation of MftA in an MftB-dependent manner (13,14). Because MftB interacts with MftA, it is possible that MftE functions in an MftB-dependent manner similar to the activity observed with MftC. To probe whether MftE requires the presence of MftB to catalyze the hydrolysis of the oxidatively decarboxylated MftA, reactions were set up as described below in the absence of the MftE. In these experiments, after an initial 6-h incubation of MftA with MftB and MftC, MftB and MftC were removed. The reaction was split into two equal aliquots, and MftE was added to one. Both were incubated for an additional 3 h and analyzed for the formation of the VY* dipeptide by UHPLC-MS. Fig. 10A shows that the oxidatively decarboxylated MftA was generated by MftC in the presence of MftB, SAM, and reductant and in the absence of MftE, and there is no species in the extracted ion chromatogram corresponding to MftA(1-28) (Fig. 10A) or to the VY* dipeptide (Fig. 10B) after the 3-h incubation postfiltration. However, UHPLC-MS analysis of reactions that contained MftE shows that MftE catalyzes the hydrolysis of the oxidatively decarboxylated MftA peptide forming the both MftA(1-28) (Fig. 10A) and the VY* dipeptide (Fig. 10B). This lack of MftB requirement contrasts with activity of MftC, which is dependent on MftB (13, 14). Therefore, MftE catalyzes the cleavage of the peptide only after the MftB-dependent decarboxylation of MftA has taken place. Summary-The biochemical and mass spectral data shown here collectively show that MftE catalyzes the cleavage of MftA* to form VY* and MftA(1-28).

Discussion
Members of the amidohydrolase superfamily catalyze hydrolysis of a diverse array of substrates (26). Within the vast superfamily, there are examples of enzymes, such as renal dipeptidase, that catalyze the hydrolysis of peptide amide bonds (27,28). However, to our knowledge, MftE is unique in catalyzing the hydrolysis of a peptide bond in a 30-amino acid modified protein. Importantly, the enzyme does not catalyze the hydrolysis of the amide bond on the unmodified MftA peptide precursor, strongly supporting its role as downstream of MftB and MftC in the biosynthetic pathway involving MftA.
A search of the Protein Data Bank with the sequence of MftE reveals similarity to creatininase homologs whose structures are known. Sequence alignments (Fig. 11A)    that MftE retains the active site residues that bind the divalent cations that activate a water molecule for hydrolysis and to stabilize the oxyanion intermediate generated during catalysis (Fig.  11B) (25,29). On the basis of sequence conservation, we postulate that MftE utilizes a similar catalytic mechanism (Fig. 11C). The bioinformatics investigation by Haft (19) first identified a subset of microorganisms that cluster mftA, mftB, mftC, mftD, mftE, and mftF genes in their chromosomes. It was speculated that the MftA peptide that is encoded by the mftA gene in these organisms is the precursor to a novel redox metabolite analogous to PQQ (19). We have recently shown that MftB and MftC are required for the oxidative decarboxylation of the Tyr residue at the C terminus of MftA (13). Here we showed that the creatininase homolog MftE, which is encoded by the mftE gene, catalyzes the second step in the maturation of MftA. Although at this point the structure and biological function of the metabolite that is ultimately derived from MftA are not known, the reactivity of the modified Tyr residue may foreshadow both. Studies are now underway to characterize the remaining two proteins encoded by the genes in the mft cluster to complete the in vitro reconstitution of the biosynthetic pathway and identify the novel natural product that builds on VY*.

Experimental Procedures
Cloning of the mftE-The LI98_07105 gene that encodes for the creatininase homolog, referred to as mftE, was PCR-amplified from the M. smegmatis ATCC 700084 genome using the forward primer 5Ј-AAAAAAGCTAGCATGAATTCGGCT-TACCATCGGCACG-3Ј and the reverse primer 5Ј-AAAAA-ACTCGAGTCATGTCAGCAATCCGTTCCGG-3Ј. The itali-   FIGURE 11. A, amino acid sequence alignments among MftE and creatininase homologs with known structures. Single, fully conserved residues are indicated with asterisk. Conservation between groups with strongly similar properties is indicated with a colon, and conservation between groups with weakly similar properties is indicated by a period. The red boxes highlight the His, Glu, and Asp residues that are conserved in MftE and map to residues that are structurally observed to coordinate the two divalent metals in the structures of the creatininase homologs (Protein Data Bank codes 1J2U, 1O3K, 3LUB, and 3NO4). B, active site of creatininase from Pseudomonas putida (Protein Data Bank code 1J2U) showing the six conserved residues highlighted in the sequence alignment (25). cized sequences show engineered NheI and XhoI restriction sites in the forward and reverse primers, respectively, and the stop codon in the reverse primer is underlined. The purified PCR product was subsequently digested with NheI and XhoI and ligated into pET28JT vector (30) that was similarly digested with NheI and XhoI for recombinant expression of protein containing a tobacco etch virus protease cleavable N-terminal His 6 tag. Standard Sanger sequencing at the University of Michigan DNA Sequencing Core confirmed the sequence of the mftE-pET28JT plasmid.
Expression of Recombinant MftE-E. coli Rosetta 2 (DE3) cells were transformed with the mftE-pET28JT plasmid. Large scale cultures containing Lennox broth growth medium (12 ϫ 1 liter in 2-liter baffle flasks) supplemented with 34 g/ml kanamycin and 30 g/ml chloramphenicol were inoculated with 0.1 liter of mftE-pET28JT Rosetta 2 (DE3) overnight culture/liter of fresh medium and grown at 37°C. The mftE gene was induced upon the addition of isopropyl ␤-D-1-thiogalactopyranoside to a final concentration of 0.1 mM when the A 600 nm reached ϳ0.8. Cells were allowed to express the MftE at 37°C. After 5 h, the cells were harvested by centrifugation at 5000 ϫ g, flash frozen in liquid nitrogen, and stored at Ϫ80°C.
Purification of the MftE-The purification of MftE was performed at 4°C. Cells (ϳ15 g) were suspended in ϳ0.5 liter of 0.05 M potassium phosphate (pH 7.4), 0.5 M KCl, 0.05 M imidazole, and 1 mM PMSF and lysed using a Branson digital sonifier operated at 50% amplitude. The lysate was clarified by centrifugation at 18,000 ϫ g for 35 min at 4°C prior to loading on a 5-ml HisTrap HP column (GE Healthcare) that had been charged with nickel sulfate and equilibrated in 0.05 M potassium phosphate (pH 7.4) containing 0.5 M KCl and 0.05 M imidazole. After loading the clarified lysate, the column was washed with 15 column volumes of equilibration buffer, and MftE was eluted with a linear gradient from 0.05 to 0.5 M imidazole over 20 column volumes. Fractions containing the MftE were identified by SDS-PAGE, pooled, and dialyzed overnight against 0.02 M Tris⅐HCl (pH 8.0), 0.15 M NaCl, and 10 mM DTT at 4°C with two changes of the dialysis buffer. After dialysis, the MftE was concentrated to a minimal volume (ϳ1.5 ml), flash frozen in small aliquots, and stored at Ϫ80°C. The concentration of the MftE was determined by the Bradford method with BSA as the standard but corrected for the actual amino acid concentration as described below.
Purification of MftB and MftC-The MftB and MftC proteins were expressed and purified as described previously (13).
Amino Acid Analysis-MftE was subjected to amino acid analysis to determine the accuracy of the concentration measured by the Bradford method. A 0.1-ml aliquot of concentrated protein was desalted to 10 mM NaOH using an Illustra NICK column (GE Healthcare). The concentration of the eluent was determined by the Bradford method with BSA as the standard before being flash frozen and analyzed at the Molecular Structure Facility at the University of California-Davis. The analysis was performed on two independent preparations of the enzyme in triplicate. The correction factor was applied to all Bradford method-determined protein concentrations used in this investigation.  30 MftA Peptides-All MftA peptides were synthesized on a 0.025-mmol scale by solid phase peptide synthesis methodology using a PS3 peptide synthesizer (Protein Technologies Inc.) following the manufacturer's coupling protocol using Fmoc chemistries. All protected Fmoc-amino acids were purchased from Protein Technologies Inc. The Fmoc-[ 13 C 9 , 15 N]Tyr O-tert-butyl ether (-OtBu) and Fmoc-[ 13 C 5 , 15 N]Val were purchased from Cambridge Isotope Laboratories Inc. In brief, 0.15 g of 2-chlorotrityl chloride resin 100 -200 mesh (Chem-Pep) was loaded with either Fmoc-Tyr-OtBu or Fmoc-[ 13 C 9 , 15 N]Tyr-OtBu (ϳ0.2 mmol/g of resin) to obtain either unlabeled or isotopically enriched MftA, respectively. Prior to loading the Fmoc-Tyr onto the resin, the resin was washed three times with 5 ml of DMF and three times with 5 ml of dichloromethane (DCM). Either the unlabeled or 13 C 9 , 15 N-enriched Fmoc-Tyr-OtBu (0.03 mmol) was dissolved in 1 ml of 1:1 DMF/DCM containing 0.15 mmol of diisopropylethylamine (DIPEA). The Fmoc-Tyr/DIPEA solution was added to the 2-chlorotrityl chloride resin and incubated at room temperature to load the resin with the C-terminal residue of MftA. The resin was gently agitated every 5-10 min. After 1 h, the Fmoc-Tyr/ DIPEA solution was removed, and the resin was washed three times with 5 ml of DCM. The remaining sites on the 2-chlorotrityl chloride resin were capped by slowly washing the resin with 20 ml of 17:2:1 DCM:methanol:DIPEA. After capping, the Fmoc-Tyrloaded resin was washed three times with 5 ml of DCM and three times with 5 ml of DMF and transferred to the reaction vessel.

Analysis of Metal Ion Content in
All Fmoc-amino acids, including the Fmoc-[ 13 C 5 , 15 N]Val (0.15 mmol, 5 eq) were coupled by in situ activation with N-[(dimethylamino)-1H-1,2,3-triazolo-[4,5-b]pyridin-1-ylmethylene]-N-methylmethanaminium hexafluorophosphate N-oxide (0.15 mmol, 5 eq; ChemPep) in 0.6 M N-methylmorpholine. To obtain the peptides with a free C-terminal carboxylic acid from the resin, peptides were deprotected and cleaved from the resin by adding 6 ml of cleavage solution (92.5% (v/v) TFA, 2.5% (v/v) water, 2.5% (v/v) ethane dithiol, and 2.5% (v/v) triisopropyl silane) at room temperature with constant mixing. After 2 h, the cleavage reaction was filtered, and the peptide was precipitated in ϳ30 ml of ice-cold diethyl ether. Centrifugation of the ether suspension at 5000 ϫ g for 30 min at 4°C pelleted the precipitated peptide, allowing the supernatant to be decanted. The peptide was washed three times with ice-cold ether and centrifuged at 5000 ϫ g for 30 min at 4°C, and the supernatant was discarded. After the final wash, the peptide was dried under vacuum for 2 h. The dried peptide was subsequently dissolved in ϳ10 -20 ml of 50 mM ammonium bicarbonate (pH 7.7), frozen in liquid nitrogen, and lyophilized to dryness. Once dried, each crude peptide was dissolved to a final concentration of 7 mg/ml in 0.05 M Tris⅐HCl (pH 8.0) and 10 mM DTT under anaerobic conditions (Coy Laboratories anaerobic chamber with 95% N 2 , 5% H 2 atmosphere), aliquoted, flash frozen in liquid nitrogen, and stored at Ϫ80°C. The three synthetic peptides lacked the N-terminal methionine residue. of 25%. The doubly modified VY* products were fragmented with a collision energy of 35%. The full mass spectra and fragments were detected using the FT analyzer, which was set up as described above.
MftB Requirement for MftE Activity-Reactions to synthesize the natural abundance or isotopically enriched oxidatively decarboxylated MftA peptides were set up as described above in the absence of MftE to a final volume of 0.4 ml. After incubating for 6 h, the reactions were passed through a 10-kDa molecular mass cutoff spin filter to remove both MftB and MftC. The filtrates were split into two equal fractions (0.1 ml), and the MftE was added to one aliquot of filtrate to a final concentration of 0.05 mM. Each aliquot was incubated at room temperature for an additional 3 h, quenched via filtration, and analyzed for the presence of the VY* dipeptide by UHPLC-MS as described above.