Reiterative Synthesis by the Ribosome and Recognition of the N-Terminal Formyl Group by Biosynthetic Machinery Contribute to Evolutionary Conservation of the Length of Antibiotic Microcin C Peptide Precursor.

Escherichia coli microcin C (McC) is a representative member of peptide-nucleotide antibiotics produced by diverse microorganisms. The vast majority of biosynthetic gene clusters responsible for McC-like compound production encode 7-amino-acid-long precursor peptides, which are C-terminally modified by dedicated biosynthetic enzymes with a nucleotide moiety to produce a bioactive compound. In contrast, the sequences of McC-like compound precursor peptides are not conserved. Here, we studied the consequences of E. coli McC precursor peptide length increase on antibiotic production and activity. We show that increasing the precursor peptide length strongly decreases McC production by affecting multiple biosynthetic steps, suggesting that the McC biosynthesis system has evolved under significant functional constraints to maintain the precursor peptide length.

M icrocin C (McC) is a ribosomally synthesized and posttranslationally modified peptide (RiPP) produced by Escherichia coli cells bearing a plasmid-borne mcc gene cluster. McC is a heptapeptide adenylate that inhibits the growth of sensitive cells by a Trojan-horse mechanism. It is transported inside E. coli or other closely related bacterial cells by the YejABEF transporter, which specifically recognizes the peptide moiety of the prodrug (1). Inside the cell, the peptide is degraded by aminopeptidases, leading to the release of a toxic, nonhydrolyzable aspartamide adenylate (2). This compound structurally mimics aspartyl adenylate, an intermediate of a reaction catalyzed by aspartyl-tRNA synthetase (3). Inhibition of this essential enzyme by processed McC leads to an accumulation of uncharged tRNA Asp and cessation of protein synthesis and growth (4).
The E. coli mcc gene cluster contains a five-gene operon, mccABCDE, and a nearby independently expressed mccF gene (Fig. 1). The 7-amino-acid-long McC precursor (MccA; MRTGNAN), the product of the mccA gene, is adenylated by the MccB enzyme in an ATP-dependent process (5). The adenosine moiety is attached to the C terminus of the peptide via a nonhydrolyzable N-P bond. The peptide adenylate produced by MccB is additionally decorated with an aminopropyl group attached to the phosphoramidate by the joint activity of MccD and the N-terminal domain of a bifunctional MccE enzyme (6). The C-terminal domains of MccE and MccF contribute to the autoimmunity of the peptide-adenylate-producing cells by detoxifying the processed McC that accumulates in their cytoplasm (7,8). The mccC gene encodes the MccC pump protein required for the extrusion of mature McC as well as premature McC without the aminopropyl group.
Though the aminopropyl moiety increases the toxicity of McC by ca. 4-to 10-fold for some strains of E. coli, the peptide adenylate itself also inhibits bacterial growth (9). Thus, the minimal set of three genes in an mcc cluster, encoding the precursor peptide, the adenylating enzyme, and the export pump protein, should be sufficient for production of McC-like antibacterial compounds. Bioinformatics searches reveal that genes coding for MccB-like proteins can be found in various bacteria that are phylogenetically distant from E. coli (10).
Here, we report that extension of the gene encoding the E. coli MccA heptapeptide dramatically decreases peptide-nucleotide production. We show that the fitness gain of confining mccA to encode a heptapeptide depends on at least two parameters. First, in vitro ribosomal synthesis of the MccA heptapeptide proceeds via multiple rounds of mccA mRNA translation without the dissociation of the ribosome from the template, in line with earlier work on minigene expression (11). Ribosome recycling without mRNA dissociation is much less effective for open reading frames (ORFs) encoding longer MccA variants. We suggest that selective amplification of MccA heptapeptide synthesis compared to the synthesis of longer peptides results also in more efficient McC production inside the cell, where intracellular mRNAs compete with mccA mRNA for the ribosomes. Second, N-terminal formylation of the MccA heptapeptide promotes efficient adenylation by the MccB enzyme, and the formyl group on the heptapeptide adenylate specifically increases antibacterial activity. We speculate that these factors together contribute to the observed conservation of MccA homolog lengths in diverse bacteria.

RESULTS
Bioinformatics analysis. We performed a bioinformatics search for genes encoding MccB-like proteins in the NCBI protein database. Manual searches revealed the existence of short genes encoding putative peptide precursors in the vicinity of most such mccB homologs. All precursors contained a C-terminal asparagine residue required for installation of the N-P bond by MccB, and their open reading frames were preceded by a strong Shine-Dalgarno sequence. Some such MccA-MccB pairs were previously validated by the reconstitution of enzymatic activity on chemically synthesized MccA peptide precursors by cognate MccB homologs in vitro (10).
The phylogenetic tree of MccB homologs is presented in Fig. 2. The tree has a deeply rooted branch that separates MccB sequences into two major clades. An analysis of predicted and, in some cases, validated MccA precursors showed that they fall into two distinct groups. The vast majority of precursors are 7 (or more rarely 8) amino acids long and are associated with the major clade of MccB homologs (blue in Fig. 2). In the second much smaller group, the precursors are considerably longer, from 17 to 56 amino acids, and are associated with a different clade of MccB enzymes (red in Fig. 2). The unrooted tree was generated using RAxML (24) with 400 bootstrap replicates. The numbers at the nodes indicate the bootstrap values. Only bootstrap values greater than 70% are shown. Scale bar shows the number of inferred amino acid substitutions per site. Triangle marks the outgroup consisting of PaaA homologs not involved in the production of McC-like compounds. Each terminal node of the tree corresponds to the sequence representing a cluster of identical proteins and is labeled by the full systematic name of an organism (for the full data set see Table S1 in the supplemental material). The names of the organisms bearing the validated mcc clusters are shown in red. For other organisms, precursor peptides are putative and obtained by manual sequence analysis of mcc-like clusters. The lengths (in amino acids) of the verified or predicted microcin C precursor peptides are indicated both by labels and by the color strip. The color palette is given on the bottom left. White indicates that we were unable to predict the peptide. The blue and the red branches correspond to MccB subfamilies associated with 7-amino-acid and longer MccA precursors, respectively. Apart from universally conserved N-terminal methionine and C-terminal asparagine, the sequences of the MccA peptide precursors associated with the major MccB clade differ greatly from each other (see Table S1 in the supplemental material). Their length, however, appears to be well conserved, suggesting high selection pressure for heptapeptide precursors among the biosynthetic clusters of McC-like compounds.
Inefficient in vivo production of McC-like compounds with elongated peptide chains. To explain the preference for seven-residue-long MccA precursor peptides in naturally occurring mcc-like clusters, two sets of plasmids for the production of E. coli McC or McC-like molecules were created. In the first set, the pA7-ap vector contains the full mcc cluster expressed from its natural promoters (Fig. 3A, left) and was expected to produce fully matured McC. The pA11-ap and pA15-ap plasmids are pA7-ap derivatives encoding 11-and 15-amino-acid MccA-based peptides MGGGMRTGNAN and MASTA GGGMRTGNAN, respectively (the sequence of the natural MccA heptapeptide is underlined). The string of three glycines (highlighted in bold) was chosen to extend the length of the natural heptamer without providing any sequence-specific interactions. To prevent possible artifacts caused by an extensive glycine stretch, a tetrapeptide string (ASTA) composed of small amino acids was incorporated in the A15 peptide. The second set of plasmids (pA7, pA11, and pA15) are pA7-ap and its derivatives lacking the mccD and mccE genes responsible for installation of the aminopropyl moiety on the peptidyl adenylate (Fig. 3A, right). The E. coli pepABN (BW25113 ΔpepA ΔpepB ΔpepN) strain lacking peptidases responsible for intracellular McC processing (2) was used as a production host to avoid intracellular degradation of the peptide moiety of the McC products. Cells transformed with either one of the McC production plasmids grew normally. Antibiotic activity was measured after overnight growth by depositing aliquots of cultured medium on freshly seeded lawns of McC-sensitive E. coli tester cells (Fig. 3B). Clear growth inhibition zones were observed around the points where aliquots of cultured medium from cells carrying pA7-ap, pA7, and pA11-ap were Mechanism of Microcin C Biosynthesis ® deposited, with growth inhibition zones observed with medium from pA11-ap cells being the smallest. No antibiotic activity was detectable in cultured medium of cells carrying pA11, pA15, or pA15-ap. In what follows, we analyze the reason(s) for the observed dramatic difference in active peptide-nucleotide production caused by the elongation of mccA.
McC variants with longer peptide parts do not accumulate inside producing cells. We wondered if longer peptides, once modified, are inefficiently exported from the producing cells. In this case, adenylated peptides would accumulate inside cells, since processing of their peptide parts would be prevented by the triple pepABN mutation in our production host. Antibiotic activity of extracts prepared from cells carrying mcc plasmids encoding MccA of different lengths was determined on McCsensitive E. coli BL21(DE3) cell lawns (Fig. 3C). Extracts from cells carrying pA7 and pA7-ap plasmids inhibited cell growth, while no growth inhibition activity was detected in extracts from cells carrying plasmids with elongated mccA genes. Thus, it appears that either accumulation or bioactivity of elongated peptide adenylates, rather than their export, is compromised.
The rate of synthesis of mccA RNA is not affected by increased length of the mccA reading frame. We previously showed that efficient synthesis of McC requires transcription termination induced by ribosome binding to the MccA heptapeptide ORF (12) (Fig. 4). This leads to accumulation of ribosome-bound monocistronic mccA mRNA from which the MccA peptide is produced in large excess over McC biosynthetic enzymes encoded by downstream genes (12). To establish whether extension of mccA disrupts the transcription-translation coupling necessary for mccA mRNA production, total RNA was purified from cultures transformed with pA7, pA11, and pA15, and the amount of mccA mRNA was determined by Northern blotting. We found similar amounts of mccA probe-hybridizing short RNA in each culture (Fig. 4); the transcripts detected in pA11-and pA15-carrying cells were longer than in pA7-carrying cells, as  expected. Thus, the observed reduction in the production of microcins with longer peptide parts cannot be explained by lower abundance of longer mccA mRNAs.
Templates encoding longer MccA peptides are translated less efficiently. Decreased production of longer MccA peptide adenylates can be caused by lowered synthesis of elongated precursor peptides. It was previously shown that short open reading frames (between 2 and 8 codons) can be reiteratively translated without ribosome dissociation from mRNA (11). Such a mechanism could increase the yield of peptide product per mRNA by bypassing the rate-limiting initiation step in which the mccA mRNA is competing with other mRNAs in the cell. The mccA gene of E. coli and most other validated McC-like precursor peptide genes are seven codons long and are therefore within the range where reiterative translation can be expected. To check if wild-type mccA mRNA can be reiteratively translated by the ribosome without dissociation and exchange with other mRNAs, we used a previously described in vitro translation competition assay (11,13). In vitro-synthesized mccA mRNA was preincubated with ribosomes, 3 H-formylmethionyl (fMet)-tRNA fMet (present in excess over ribosomes) (see Materials and Methods), and initiation factors to allow initiation complex formation. Next, components necessary for translation elongation (elongation and termination factors, tRNAs, amino acids and cofactors, and tRNA-recharging system) were added. A 10-fold molar excess (over mccA) of competitor mRNA was also added, as needed. The competitor mRNA has a strong ribosome-binding sequence and encodes a peptide whose synthesis is not supported by the mix due to the omission of an encoded amino acid. Thus, once the ribosome is dissociated from the mccA template, it is effectively and irreversibly sequestered at the competitor template and no longer able to initiate translation of free mccA mRNAs. We monitored the amounts of synthesized MccA peptides per active ribosome as a function of time in the presence (n) and in the absence (N) of competitor mRNA (13) (see also Materials and Methods for details) and plotted n versus N in each case, as shown in Fig. 5. If ribosomes do not dissociate from the mccA ORF in the course of the experiment, N equals n (Fig. 5, dotted line). If ribosomes redistribute to competitor template, the MccA peptide synthesis rate decreases with time and n becomes less than N. The data fitting section below the curves in Fig. 5 allows one to determine n rec , or ribosome recycle number, i.e., the number of peptide molecules synthesized per active ribosome before ribosome dissociation. An n rec value of zero means that the ribosome dissociates after completing the synthesis of the first MccA peptide. As can be seen from Fig. 5, data fitting for the wild-type mccA translation (A7) yielded an n rec value of 71. Thus, the mccA message is translated multiple times before dissociation. The n rec values for mcc templates encoding longer peptides, A11 and A15, were 30 and 6, respectively. Thus, while extension of the mccA ORF does not prevent reiterative translation in vitro, it significantly decreases it, which may contribute to the decreased production of longer peptides in vivo, although the extent of this effect inside the cells is hard to estimate.
Formylated heptapeptide MccA is a preferred substrate for MccB. In previous work, it was shown that E. coli MccB is able to modify a wide range of substrates in vivo and in vitro, including MBP protein with attached A7 peptide (14). To test MccB activity on elongated substrates encoded by pA11 and pA15, we incubated purified MccB with ATP and chemically synthesized A11 and A15 peptides and measured adenylated peptide yields by high-pressure liquid chromatography (HPLC) (Fig. 6). As can be seen, comparable yields of the adenylated MccA A7, A11, and A15 variants were observed. We next repeated the experiment with N-terminally formylated versions of each peptide that correspond to the products of ribosomal synthesis. Formylation of the A7, A11, and A15 peptides resulted in an approximately 8-fold increase, 2-fold increase, and 3-fold decrease in adenylated product yields compared to the yields obtained with A7, A11, and A15, respectively. We conclude that under our conditions, the presence of the formyl group on the N-terminal methionine residue selectively amplified the yields of MccA heptapeptide adenylation compared to adenylation yields of longer peptides A11 and A15. Reaction products were purified as a single HPLC peak, and their identity was confirmed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) (Fig. 7A). When 10 M solutions of each compound were tested for antibiotic activity, clear growth inhibition zones were observed in all cases (Fig. 7B). MIC mea-  surements performed using serial dilutions of each compound revealed that the desformylated adenylate of the 11-amino-acid peptide was 2-fold more active than the corresponding adenylated heptapeptide A7-McC, while activity of the adenylated A15 peptide was similar to that of A7-McC (Fig. 7B, top left panel). Aminopropylation increased the toxicity of each of the peptidyl adenylates by ca. 2-fold (Fig. 7B, top versus bottom panels). The formylation of the N terminus had a more complex effect on bioactivity. It slightly reduced the MICs of both elongated peptide adenylates (with or without aminopropyl) but stimulated, by ca. 4-fold, the activity of heptapeptide adenylates (Fig. 7B, left versus right panels). Overall, it appears that while either formylation of the N terminus or limited elongation of the 7-amino-acid MccA peptide leads to increased bioactivity of the resulting peptide adenylate, there is no synergy between these two effects and formylation makes a larger contribution to bioactivity than peptide length.

DISCUSSION
The primary impetus for this work came from our desire to understand why many mcc operons from diverse bacteria encode peptide precursors that are 7 amino acids in length (Fig. 2). A phylogenetic analysis revealed a well-supported large clade of MccB family proteins associated with either validated or predicted precursor peptides. Most precursors were 7 amino acids long, suggesting that peptide length is a conserved and potentially functionally important feature of microcin C-like compounds. Previously, we showed that in the case of E. coli McC, a decrease of the peptide moiety length by just 1 amino acid prevented facilitated transport by the YejABEF transporter into sensitive cells, thus abolishing activity (15). Conversely, increasing the peptide moiety length appeared to have a marginal effect on the bioactivity of McC derivatives prepared by enzymatic synthesis in vitro (10), indicating that there may not be a stringent upper limit of precursor peptide length. Surprisingly, we here observed that increasing the length of the E. coli mccA peptide precursor gene has a dramatic negative effect on the antibiotic activity of producing cells. Given that McC biosynthesis and the mechanism of action are well understood, this activity decrease could a priori have been due to (i) lower rates of synthesis of longer MccA peptides, (ii) less efficient adenylation of longer peptides, (iii) impaired export of mature compounds, (iv) decreased antibacterial activity of products with longer peptides, or any combination of these factors. We experimentally tested each one of the possibilities listed above and found that the length of the MccA product contributes to McC production and activity on multiple levels. Somewhat surprisingly, increasing the length of the mccA open reading frame had no effect on the transcription-translation coupling in the mcc operon. This follows Mechanism of Microcin C Biosynthesis ® from the observation that the termination mechanism that ensures increased accumulation of mccA transcript over the transcripts of McC biosynthesis genes is not affected by increasing the mccA gene length from 7 to 15 codons. At the same time, translation of mccA mRNA does indeed depend on the length of the ORF. Specifically, we showed that mRNA encoding a 7-amino-acid MccA peptide very efficiently underwent reiterative translation in the presence of competitor RNA. This ability to withstand the competition and reinitiate translation without ribosome dissociation was decreased (but not abolished) by increasing the mccA reading frame up to 15 codons. While at present it is not possible to assess the significance of in vitro-determined numbers of synthesis rounds without dissociation for in vivo MccA synthesis, it is reasonable to assume that such reiterative translation contributes to higher yields of heptameric MccA and, hence, to the increased yield of the final product.
Unexpectedly, the largest contributor to preferential synthesis of wild-type McC in vitro, compared to that of elongated peptide derivatives, is the formyl group present at the N-terminal methionine. This group leads to an almost 10-fold increase in the production yield of adenylated heptapeptide but has a much smaller or even inhibitory effect on the adenylation of 11-and 15-amino-acid-long derivatives. The observed  (16). The presence of a formyl group in longer peptides is not expected to have a stimulatory effect, since the positioning of the C-terminal aspartate in the MccB catalytic site allows only the N-terminal backbone residue positioned 7 amino acids away to make favorable interactions with the enzyme.
The MccD/E enzymes required for the secondary modification of peptide nucleotide phosphate with aminopropyl are encoded by both long and short MccA mcc operons (17). It is therefore an expected result that this secondary modification had no specific effect on bioactivity of McC-like compounds with different peptide lengths and generally increased the level of bioactivity. In contrast, formylation acts as a specific contributor to the activity of adenylated heptameric peptides. The presence of the formyl group makes natural E. coli McC a better antibacterial agent than either 11-or 15-amino-acid peptide adenylate with or without aminopropyl. We speculate that this effect may be due to better transport of formylated peptides of appropriate length by the YejABEF transporter, which was previously shown to recognize formylated peptides (15). In the absence of the N-terminal formyl group, increasing the peptide length up to a certain point also increases the bioactivity (14). Interestingly, in the case of the Yersinia pseudotuberculosis mcc operon, which encodes a 42-amino-acid-long MccA precursor, the final peptide-nucleotide product is processed to an 11-amino-acid peptide with attached C-terminal nucleotide, and this processing is required for bioactivity (17). We speculate that such processing may be a common feature of mcc operons with longer MccA precursors.

MATERIALS AND METHODS
Phylogenetic analysis. The initial data set consisted of 12 MccB homologs forming a branch on MccB/PaaA family tree that contained validated microcin C-like clusters (10). Protein sequences were retrieved from the NCBI protein database (https://www.ncbi.nlm.nih.gov/protein/) and aligned using MUSCLE v3.8.31 with defaults (18). Since the N-terminal RiPP precursor peptide recognition element (RRE) domain of MccB proved to be critical for microcin C processing (19), it was employed to distinguish the MccB subfamily from other members of the ThiF superfamily (20). The N-terminal 86-amino-acid portion of the alignment containing the RRE domain was used as a query for a PSI-BLAST search against protein reference sequences downloaded on 22 December 2018 (21). PSI-BLAST was iterated until convergence with an inclusion E value threshold of 0.005 and maximum of 500 hits. To remove redundancy, the resulting 202 MccB homologs were clustered using MMseqs2 with identity and coverage cutoffs of 0.95 and 0.9, respectively (22). One random representative sequence for each of 82 clusters was chosen for further analysis. Genomic regions of MccB homologs were analyzed with RODEO to check the presence of ThiF domain and a transport-related gene within corresponding gene cluster (23).The final set of MccB homologs was aligned with MUSCLE. Partial sequences were removed. The alignment was trimmed and extremely variable columns were removed. A maximum likelihood phylogenetic tree of the 71 remaining sequences was estimated using RAxML with an LG amino acid substitution model and gamma-distributed evolutionary rates (24). The LG substitution matrix was chosen for phylogenetic analysis according to the Akaike information criterion implemented in ProtTest 3.0 (25). Rapid bootstrap analysis and a search for the best tree were performed in a single run. Bootstrapping converged after 400 replications according to the autoMRE criterion. The tree was rooted using an outgroup consisting of PaaA proteins that have the same domain architecture but different function (20). The tree was then visualized using iTOL (26). The precursor peptides for each putative microcin C cluster were manually extracted from the corresponding genomic records on the basis of the following criteria: ORF with a strong ribosome binding site and C-terminal asparagine in the close proximity (as far as 300 bp from gene cluster) of MccB preferentially upstream of MccB and on the same strand.
Bacterial strains and vectors. E. coli strain BW25113 ⌬pepA ⌬pepB ⌬pepN was used as McC producer. An E. coli BL21(DE3) strain was used as the McC-sensitive strain for growth inhibition tests. To create pA11-ap and pA15-ap, encoding extended mccA, double-stranded DNA (dsDNA) containing the extended mccA gene was prepared by two oligonucleotides annealing and inserted between SalI and NsiI of the pA7-ap vector (pp70 plasmid [12]). The pA7, pA11, and pA15 vectors were constructed from the corresponding pA7-ap, pA11-ap, and pA15-ap plasmids by the deletion of mccD, mccE, and mccF genes using Gibson Assembly Mix (NEB, USA).
Bioactivity assays. McC activity assay was performed as described previously (14), with a freshly seeded lawn of sensitive cells prepared from E. coli BL21(DE3) cells on M9 medium.
To determine intracellular McC content, the cell pellet obtained from 100 ml of overnight cell culture was resuspended in 5 ml of 1ϫ phosphate-buffered saline and disrupted by ultrasonication. The lysate was cleared by centrifugation, mixed with an equal volume of 0.1% trifluoroacetic acid (TFA) in acetonitrile, and incubated on ice for 30 min. The insoluble fraction was removed by centrifugation. The supernatant was vacuum dried, and the resulting pellet was dissolved in 100 l of water and cleared by Mechanism of Microcin C Biosynthesis ® centrifugation. Bioactivity was checked by applying 10-l drops of the cleared supernatant on the freshly prepared lawn of sensitive cells.
In vitro synthesis of Mc -like compounds. Synthetic peptides were purchased from GenScript Biotech Corp., USA. E. coli MccB was purified as described previously (10). Briefly, the BL21(DE3) cells transformed with pET-MccB vector (10) were grown at 37°C in 200 ml of LB medium supplemented with 50 g/ml ampicillin until the optical density at 600 nm reached 0.6. Protein production was induced with 0.1 mM isopropyl ␤-D-1-thiogalactopyranoside (IPTG) followed by growth at 18°C for 16 h. Cells were harvested and resuspended in 8 ml of BW buffer (20 mM Tris-HCl [pH 8.0], 500 mM NaCl, and 1 mM ␤-mercaptoethanol) supplemented with 0.2 mM phenylmethanesulfonyl fluoride (PMSF) and 1 mg/ml lysozyme. Cells were disrupted by sonication, and the insoluble fraction was precipitated by centrifugation at 30,000 ϫ g for 30 min at 4°C. The clarified cell lysate was loaded onto a 1-ml HiTrap Co 2ϩ -chelating HP Sepharose column (GE Healthcare) preequilibrated with BW buffer. The column was washed with 10 ml of BW buffer, and the MccB protein was eluted with EB buffer (50 mM Tris-HCl [pH 8.0], 0.5 M imidazole, 300 mM NaCl, 10% glycerol). Protein fractions were analyzed using Laemmli 10% SDS-PAGE. Protein concentration was measured according to the Bradford method.
For elongated McC-like compound synthesis, 100 M peptides were adenylated by 12 M MccB in the mixture with 100 mM Tris (pH 7.5), 50 mM NaCl, 10 mM MgCl 2 , 5 mM ATP, and 10 mM dithiothreitol (DTT) at 32°C for 8 h. Reaction mixtures were checked by MALDI-TOF MS for reaction completion, and compounds were purified by HPLC using an Eclipse Plus C 18 column (4.6 mm by 250 mm; particle size, 5 m; Agilent Technologies) and 0% to 15% linear elution gradient of acetonitrile in 0.1% TFA. The fractions were vacuum dried and dissolved in deionized water for concentration measurements. The published extinction coefficient for AMP (E mM ϭ 15.4 at 259 nm in 0.01 M PO 4 [pH 7.0]) was used to calculate peptidyl nucleotide concentrations (27).
For adenylation efficiency measurements, 500 M peptide was incubated with 5 M MccB at 28°C for 20 min, and then the reaction was quenched with 0,1% TFA and analyzed using C 18 reversed-phase chromatography (Symmetry C 18 , 5 m, 4.6 mm by 150 mm; Waters). A 0% to 15% linear gradient of acetonitrile in 0.1% TFA was applied for elution. The identities of the compounds in the chromatographic fractions were confirmed by MALDI-TOF MS.
Purification of MccD and MccE enzymes and in vitro aminopropylation of the compounds were performed as described in reference 6.
MALDI-TOF MS analysis. Sample aliquots (1 to 2 l) were mixed on a steel target with 0.5 l of 20 mg/ml 2,5-dihydroxybenzoic acid in 0.5% TFA and 30% acetonitrile (ACN) water solution (Aldrich). Mass spectra were recorded on an UltrafleXtreme MALDI-TOF/TOF mass spectrometer (Bruker Daltonics) equipped with a neodymium laser. The [MH] ϩ molecular ions were measured in reflector mode; the accuracy of mass peak measurement was within 0.1 Da.
The mccA transcript detection. An aliquot of an overnight culture was diluted 100-fold with LB containing 100 g/ml carbenicillin and grown for 4 h at 37°C (OD 600 reached approximately 2.0). Cells from 0.5-to 3-ml culture aliquots were collected by centrifugation, the supernatant was discarded, and cell pellets were frozen in liquid nitrogen and stored at Ϫ70°C until further use. Cell pellets were resuspended in 30 to 50 l of 20 mM Tris-HCl (pH 8.0) with 1 mM EDTA, treated with 10 mg/ml lysozyme, mixed with 500 l of TRIzol reagent (Invitrogen), and incubated for 5 min at room temperature. One hundred microliters of chloroform was added, and the samples were incubated for 15 min on ice until phase separation occurred. After a 15-min centrifugation at 13,000 ϫ g, the aqueous phase (ϳ300 l) was transferred to a fresh tube. RNA was ethanol precipitated and then redissolved in 20 to 50 l of diethyl pyrocarbonate (DEPC)-treated water. RNA concentration was determined using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA).
For Northern blot hybridization, the RNA samples were diluted to a final concentration of 0.25 g/l with formamide loading buffer (95% formamide, 0.025% xylene cyanol, 0.025% bromophenol blue), and 1 g of RNA was loaded on an 8% (19:1) denaturing polyacrylamide gel. After electrophoretic separation, RNA was electroblotted to a Hybond XL membrane (Amersham) using a wet transfer system (Bio-Rad). The transfer was performed for 1 h at 250 mA in 0.5ϫ Tris-borate-EDTA (TBE) buffer. The membrane was air dried and baked using "optimal cross-link" settings of the UV cross-linker (UVP). The membrane was blocked with 3 ml of ExpressHyb buffer (Clontech) for 30 min at 37°C; 20 pmol of 5= 32 P-labeled oligonucleotide probe was added, and hybridization was performed overnight at 37°C. The membrane was washed twice for 10 min with 2ϫ SSC (1ϫ SSC is 0.15 M NaCl plus 0.015 M sodium citrate) with 0.1% SDS and once for 10 min with 0.1ϫ SSC with 0.1% SDS. The results were visualized using Storm Phosphorimager (GE Healthcare).
In vitro translation. DNA templates were prepared by PCR with a universal forward primer containing the T7 promoter sequence. RNA templates were prepared by transcription with T7 polymerase (40 mM Tris [pH 8], 20 mM MgCl 2 , 5 mM spermidine, 25 mM DTT, 0.01% Triton X-100, 5 mM ATP, 4 mM each GTP, UTP, and CTP, 0.02 M T7 polymerase, 0.005 U/l inorganic pyrophosphatase, 1 U/l inhibitor of RNases, and 1 ng/l DNA). After RNA synthesis, the reaction mixture was treated with DNase I (1 U per 250 l reaction mixture) for 15 min, phenol extracted twice, and applied to a NAP 10-gel filtration column (GE Healthcare) to remove unincorporated nucleotides. The eluate was precipitated, redissolved in water, and used as a template for in vitro translation.