The Mechanism of Microcin C Resistance Provided by the MccF Peptidase*

The heptapeptide-nucleotide microcin C (McC) is a potent inhibitor of enteric bacteria growth. Inside a sensitive cell, McC is processed by aminopeptidases, which release a nonhydrolyzable aspartyl-adenylate, a strong inhibitor of aspartyl-tRNA synthetase. The mccABCDE operon is sufficient for McC production and resistance of the producing cell to McC. An additional gene, mccF, which is adjacent to but not part of the mccABCDE operon, also provides resistance to exogenous McC. MccF is similar to Escherichia coli LdcA, an l,d-carboxypeptidase whose substrate is monomeric murotetrapeptide l-Ala-d-Glu-meso-A2pm-d-Ala or its UDP-activated murein precursor. The mechanism by which MccF provides McC resistance remained unknown. Here, we show that MccF detoxifies both intact and processed McC by cleaving an amide bond between the C-terminal aspartate and the nucleotide moiety. MccF also cleaves the same bond in nonhydrolyzable aminoacyl sulfamoyl adenosines containing aspartyl, glutamyl, and, to a lesser extent, seryl aminoacyl moieties but is ineffective against other aminoacyl adenylates.

Microcins are a class of small (less than 10 kDa) antibacterial peptides produced by Escherichia coli and its close relatives (1,2). Microcins are produced from ribosome-synthesized, geneencoded precursors. Precursors of post-translationally modified microcins are heavily modified by dedicated maturation enzymes (3,4). One post-translationally modified microcin, microcin C (McC), 4 is a heptapeptide with covalently attached C-terminal modified adenosine monophosphate (Fig. 1A) (5,6) Once inside a sensitive cell, McC is processed, and the product of processing (Fig. 1A), a nonhydrolyzable analog of aspartyl-adenylate (Fig. 1A) that is an intermediate of reaction catalyzed by aspartyl-tRNA synthetase (AspRS), inhibits translation by preventing the synthesis of Asp-tRNA Asp (7). McC processing involves deformylation of the N-terminal Met residue by peptide deformylase, followed by degradation of the peptide moiety by any one of the three broad specificity aminopeptidases, peptidases A, B, and N (8). Whereas unprocessed McC has no effect on the aminoacylation reaction, processed McC has no effect on the growth of sensitive cells at concentrations at which intact McC efficiently inhibits growth (7). Thus, McC is a Trojan-horse inhibitor (7,9): the peptide moiety is required for the entry of unprocessed McC into sensitive cells where it must be processed by peptidases to release the aminoacyl-nucleotide part of the drug, which is required for the inhibition. Other known Trojan horse inhibitors targeting aminoacyl-tRNA synthetases include albomycin, a nonhyrolyzable seryl pyrimidyl attached to ferritin transport moiety (10), and agrocin 84, a nonhydrolyzable leucyl adenylate modified by an opine necessary for transport inside agrobacterial cells (11)(12)(13). Albomycin targets SerRS, whereas agrocin 84 targets LeuRS.
Although some McC processing should occur inside the cytoplasm of the producing cells, no poisoning of cells in the producing culture is observed, and cells remain viable. Resistance to internally produced McC is provided by the action of the MccC pump and through acetyltransferase activity of the C-terminal domain of the MccE protein. MccE CTD acetylates the amino group of processed McC, thus preventing its interaction with (and inhibition of) AspRS (14).
The natural plasmid responsible for McC production, pMcc7, is a large low copy number plasmid (15). The mccABCDE McC biosynthesis operon was subcloned from pMcc7 during early studies aimed at elucidating the minimal determinants sufficient for McC production and immunity. However, it was observed that both the growth rate and levels of McC resistance of cells harboring pMcc7 were higher than the corresponding rates and levels in cells harboring multicopy plasmids carrying the mccABCDE operon (16). It was subsequently shown that a fragment of pMcc7 containing an additional gene, located immediately downstream of the mccABCDE operon and transcribed in an opposite direction, leads to McC resistance when placed on a multicopy plasmid (16). This gene was named mccF (Fig. 1B). The mechanism by which MccF contributes to McC resistance was not defined, however. Here, we show that MccF is a peptidase that is able to specifically cleave both intact and processed McC at the carboxamide bond connecting, respectively, the peptidyl or aminoacyl moieties and the nucleotide moiety. The results thus describe a novel mechanism of detoxification of antibacterial compounds based on aminoacyl adenylates. Given the abundance of MccF homologs, the mechanism may be widespread in the bacterial world.

Molecular Cloning and Recombinant Protein Purification-
The mccF gene was amplified by PCR with primers containing engineered XhoI and BplI restriction sites using pBM43 plasmid as a template (18). PCR product was digested with XhoI and BplI, and the mccF fragment was cloned into appropriately treated pET19 expression vector. The S118A substitution was introduced into cloned mccF by site-directed mutagenesis using appropriate oligonucleotides. Recombinant MccF and MccF S118A were overproduced in E. coli BL21(DE3) cells. The cells were grown at 37°C in 400 ml of LB supplemented with ampicillin (100 g/ml) until A 600 of culture reached 0.6, and expression of plasmid-borne mccF was induced by the addition of 0.2 mM isopropyl ␤-D-thiogalactopyranoside. After induction, the cells were grown at 30°C until A 600 reached 1. The cells were harvested by centrifugation and resuspended in 6 ml of buffer A (20 mM Tris-HCl, pH 8.0, 50 mM NaCl). After a 1-h incubation on ice, the cells were disrupted by sonication, and cell debris was removed by centrifugation. The supernatant was loaded on a 1-ml chelating HiTrap column (GE Healthcare) charged with Ni 2ϩ according to the manufacturer's instructions. The column was washed with buffer A containing 50 mM imidazole, and the bound protein was eluted with the same buffer containing 300 mM imidazole. The resulting MccF proteins were 95% pure as judged by Coomassie staining of SDS gels. Glycerol was added to a final concentration 50%, and the samples were stored at Ϫ20°C until further use.
In Vivo Sensitivity Test-E. coli BL21(DE3) cells carrying pET19, pET19mccF, and pET19mccF S118A plasmids were grown in 5 ml of LB supplemented with ampicillin (100 g/ml) at 37°C until A 600 reached ϳ1. Cells carrying pet28mccE (14) and pet11aspRS (7) plasmid were grown in 5 ml of LB supple-mented with kanamycin (50 g/ml) and isopropyl ␤-D-thiogalactopyranoside (0.1 mM) at 30°C overnight (ϳ18 h). 100 -200 l of culture was mixed with 5 ml of melted top (0.75%) LB agar, and the mixture was poured on a surface of LB agar plates supplemented with kanamycin (in the case of pet28mccE and pet11aspRS plasmids) or LB agar plates supplemented with ampicillin (in the case of pET19-based plasmids). The sensitivity of cells to different compounds was measured by placing Effect of serine-peptidase inhibitor PMSF on MccF activity was studied by including 4 mM PMSF in the starting reaction mixture. The reactions were terminated by flash-freezing in liquid nitrogen and lyophilization. Dried residue was dissolved in 5 l of ultrapure water. Aliquots of dissolved reaction products were placed on the surface of LB agar plates overlaid with 5 ml of soft LB agar seeded with 0.1 ml of E. coli BL21(DE3) overnight culture. The plates were incubated at room temperature overnight, and the sizes of growth inhibition zones were determined as described above.
Preparation of S30 Extracts-E. coli cells were grown in 50 ml of LB medium containing 50 g/ml ampicillin until mid log phase. After centrifuging at 3,000 ϫ g for 10 min, the supernatant was discarded, and the cell pellet was resuspended in 40 ml of buffer containing 20 mM Tris-HCl or HEPES-KOH, pH 8.0, 10 mM MgCl 2 , 100 mM KCl. The cell suspension was centrifuged again as above. This procedure was repeated two times. The pellet was resuspended in 1 ml of the same buffer containing 1 mM DTT and kept at 0°C. The cells were disrupted by several successive rounds of sonication. The lysate was centrifuged at 15,000 ϫ g for 30 min at 4°C. The supernatant was aliquoted and stored at Ϫ80°C until further use.
The tRNA Aminoacylation Reaction-The tRNA aminoacylation reactions were performed as described in Ref.

Inactivation of Microcin C by the MccF Peptidase
where indicated, 1 l of 100 nM solution of purified MccF were added. The reactions that contained inhibitors requiring processing by peptidases (i.e. McC) were preincubated 15 min prior to the addition of MccF and carrying out the first aminoacylation reaction (the time of the first aminoacylation reaction was taken as 0 in this case). The reactions that contained inhibitors that do not require processing (i.e. DSA) were assembled by simultaneous inhibitor and MccF to cell extract (a zero time point), followed by performing aminoacylation reactions at the times indicated. Individual reactions were incubated for addition time intervals indicated in appropriate figures, and 15 l of the following aminoacylation mixture was added to individual reactions: 30 mM Tris-HCl, pH 8.0, 1 mM DTT, 5 g/liter bulk E. coli tRNA, 3 mM ATP, 30 mM KCl, 8 mM MgCl 2 , and 40 M of specified radiolabeled amino acid. After a 5-min incubation at room temperature, the reaction products were precipitated in cold 10% TCA on Whatman 3MM papers. After thorough washing with cold 10% TCA, the papers were washed twice with acetone and dried on a heating plate. Following the addition of scintillation liquid, the amount of radioactivity was determined in a scintillation counter. After a 2-h incubation at 37°C, the reactions were frozen and lyophilized. Antimicrobial activity of reaction aliquots was monitored as described above. For mass spectrometric analysis, the samples were dissolved in acetonitrile. Aliquots (0.5 l) of the sample were mixed on a steel target with 1 l of 2,5-dihydroxybenzoic acid (Aldrich) solution (20 mg/ml in 20% acetonitrile, 0.5% trifluoroacetic acid), and the droplet was left to dry at room temperature. The mass spectra were recorded on Ultraflex II MALDI-ToF-ToF mass spectrometer (Bruker Daltonik, Germany) equipped with a neodymium laser. The [MH] ϩ molecular ions were measured in reflector and tandem (Lift) mode; the accuracy of mass peak measurement was 0.1 Da for parent ions and 0.5 Da for daughter ions.

MALDI-MS Characterization of MccF Reaction Products-
For peptide mass fingerprint analysis, the Biotools 3.0 software combined with MASCOT search engine (Matrix Science, London, UK) was used. Searches were performed using N-formyl as the variable modification, and with no limitations on proteolysis type using the database of the United States National Center for Biotechnological Information in the E. coli database subset.
HPLC Analysis-The reverse phase C18 column was equilibrated with buffer A (25 mM thriethylamine acetate buffer, pH 7.5, in H 2 O). Probes were diluted in 1 ml of buffer A and injected to the column. Then a column was ramped to buffer B (25 mM TEAB in CH 3 CN). The flow rate was 1 ml/min throughout. The retention times are relative to the flow start. We also tested three synthetic McC analogs: XDSA, XESA, and XLSA (17), for their ability to inhibit the growth of cells overproducing MccF. The synthetic compounds contain a hexapeptide corresponding to the first six amino acids of McC attached to, respectively, aspartyl sulfamoyl adenosine DSA (nonhydrolyzable analog of aspartyl adenylate, inhibits AspRS), glutamyl sulfamoyl adenosine ESA (inhibits GluRS), and leucyl sulfamoyl adenosine LSA (inhibits LeuRS). As shown elsewhere (17), synthetic compounds retain the Trojan horse mechanism of McC and target AspRS (XDSA), GluRS (XESA), and LeuRS (XLSA). Cells overproducing MccF were resistant to XDSA and XESA but were as sensitive to XLSA as control cells harboring pET vector ( Fig. 2A). In contrast, cells overproducing MccE were resistant to all three synthetic McC analogs, as expected ( Fig. 2A). MccF-overproducing cells were fully sensitive to albomycin ( Fig. 2A), whereas cells overproducing MccE were resistant, in agreement with earlier data (14). The effect of MccF overproduction on sensitivity to DSA and LSA was also investigated (ESA was excluded from the analysis because it does not permeate E. coli cells). The results showed that cells overproducing MccF and MccE were resistant to DSA, whereas only MccE-overproducing cells were resistant to LSA (Fig. 2B). Based on these results, we conclude that when overexpressed, MccF alone can provide resistance to McC and its synthetic derivatives targeting AspRS or GluRS, as well as to processed  (Fig. 3B). In contrast, this reaction was efficiently rescued by the addition of MccE (data not shown, see also Ref. 14). Additional experiments demonstrated that the addition of MccF had no effect on aminoacylation reactions inhibited by several other aminoacyl adenylates (isoleucyl-sulfamoyl-adenosine, lysyl-sulfamoyladenosine, phenylalanyl-sulfamoyl-adenosine, glycyl-sulfamo-  A pairwise BLAST comparison reveals that the MccF sequence is highly similar, over its entire length, to E. coli LdcA (an e value of Ͻ10 Ϫ15 ). LdcA is an L,D-carboxypeptidase whose substrate is monomeric murotetrapeptide l-Ala-D-Glu-meso-A 2 pm-D-Ala or its UDP-activated murein precursor (20). The enzyme removes the C-terminal L-Ala residue from its substrates. The amino acid sequence alignment of MccF, LdcA, and a homologous protein encoded by the biosynthetic cluster of agrocyn 84, a Trojan horse antibiotic targeting LeuRS (9), is shown in Fig. 1C. We hypothesized that MccF is also a peptidase and that it contributes to McC resistance through its peptidase activity. Two experiments support this view. First, a mutation substituting an evolutionary conserved serine residue Ser 118 that likely forms the catalytic triad of the MccF peptidase catalytic center for alanine was created (Fig. 1C). An expression plasmid carrying the mutant gene overproduced MccF at high level, and all overproduced protein was soluble (data not shown), yet the cells overproducing the mutant protein were fully sensitive to McC, indicating that the point substitution annihilated MccF protective function (Fig. 4A). The mutant protein was purified and found to be unable to detoxify McC in vitro (Fig. 4B). Second, the in vitro test described above was repeated in the presence of PMSF, an inhibitor of serine proteases. As can be seen (Fig. 4B) McC peptide lacking C-terminal modification (i.e. with C-terminal aspartic acid residue). The tandem mass spectrometric analysis was performed for the 792-Da product and for synthetic N-formylated peptide identical to the peptide part of McC. The spectra of fragmentation were found to be very similar (data not shown) and were identified using a Mascot search through the E. coli subset of the NCBI database as N-formylated MRTGNAD peptide (the product of the MccA gene containing a terminal aspartate instead of asparagine) with a score of 80.

Overproduction of MccF Renders Cells Resistant to McC-If
The 404-Da mass peak matches the mass of a modified nucleoside phosphoramide that should be formed upon hydrolysis of the carboxamide bond of McC. The identification of this nucleotide fragment is supported by the results of MALDI-MS analysis of reaction products of MccF incubation with McC derivative lacking the propylamine group (21). In this case, the heavier (peptide) compound peak remained the same as in the case of mature McC. In contrast, the 404-Da peak was absent; instead, a 347-Da peak corresponding to the expected nucleotide fragment lacking the propylamine group was detected. MALDI-MS analysis of MccF-treated XDSA and XESA was also performed. In this case, the reaction products contained, in addition to mass peak corresponding to MRTGNAD/E peptide, a 347-Da mass peak, corresponding to sulfamoyl adenosine (data not shown; note that the masses of sulfamoyl adenosine  1A). Based on the results of MS analysis and mutagenesis data (above), we conclude that MccF is a highly unusual serine peptidase that cleaves McC, its maturation derivatives, and synthetic analogs at the amide bond connecting the peptidyl part with the nucleotide part of these compounds.
In Vitro MccF Treatment of DSA and ESA Leads to Production of Antibacterially Active Sulfamoyl Adenosine-Mass spectrometric identification of the products of incubation of pure MccF with DSA, ESA, and LSA was also performed. A mass peak corresponding to LSA remained unaltered upon MccF treatment, and no new products were observed, in agreement with in vivo and in vitro data presented above (Fig. 5D). In contrast, treatment of DSA and ESA with MccF led to a complete disappearance of mass peaks corresponding to the original compounds. Instead, a mass peak corresponding to sulfamoyl adenosine (SA) generated by the cleavage of carboxamide bond between the aminoacyl and nucleotidyl parts of initial aminoacyladenylates was observed in both cases (Fig. 5, C and  E). The complementing aminoacyl moieties were not detected in the MALDI experiment because of interference from matrix peaks. Both DSA and LSA remained intact upon incubation with MccF mutant described above (data not shown). Several additional aminoacyl sulfamoyl adenylates (listed above) were also tested. Of these, only seryl-sulfamoyl-adenosine was found to be a substrate for MccF-catalyzed cleavage (Fig. 5F).
An unexpected result was obtained when MccF-treated ESA or DSA reactions were applied onto lawns of McC-sensitive E. coli cells. In control reactions (incubation of ESA or DSA without MccF), no growth inhibition was observed in the case of ESA, whereas DSA produced growth inhibition, as expected.
Surprisingly, upon incubation with MccF, reactions containing either compound led to the appearance of exceptionally large growth inhibition zones (Fig. 6A). The zones of identical size were observed when MccF-treated DSA was tested on lawns of cells overproducing MccF, MccE, or AspRS from expression plasmids (Fig. 6B). Cells overproducing either of these proteins were resistant to DSA and McC, as expected.
Because in our MccF-treated DSA or ESA samples the initial compounds were completely converted to SA (and a corresponding amino acid) as judged by MALDI-MS analysis, we hypothesized that the modified nucleotide was by itself responsible for the strong antibacterial action observed. Indeed, published literature contains reports documenting potent antibacterial activity of SA (22). SA was prepared by chemical synthesis and was found to be identical to SA generated by MccF treatment of DSA by both MALDI MS and HPLC retention time (Fig. 6, compare C and E). Synthetic SA indeed strongly inhibited E. coli growth, including the growth of E. coli overproducing MccF, MccE, or AspRS (Fig. 6B). In contrast and as expected, aspartic or glutamic acid solutions had no effect on cell growth at these concentrations (data not shown). We therefore conclude that the appearance of antibacterial activity of ESA, and increased activity of DSA observed upon MccF treatment is due to the accumulation of SA that results from MccFmediated cleavage.
MccF Homologs Encoded by the E. coli Genome and Agrocin 84 Biosynthetic Operon Are Unable to Detoxify McC and Processed McC Analogs-As mentioned above, the E. coli genome encodes a homolog of MccF, the product of ldcA gene. To determine whether sequence similarity between the two proteins leads to functional similarity, an E. coli strain lacking ldcA and a strain harboring an ldcA expression plasmid were obtained, respectively, from the Kejo and ASKA collections, and the sensitivity of these strains to McC was determined. No difference from control levels (an ldcA ϩ strain alone or harboring a vector plasmid) was observed. Moreover, the addition of purified recombinant LdcA protein had no effect on tRNA Asp aminoacylation reactions inhibited by McC at conditions when MccF efficiently relieved the inhibition (data not shown). We therefore conclude that the LdcA protein plays no role in McC resistance.
A plasmid expressing the MccF homolog encoded by the agrocin 84 biosynthetic cluster was also constructed. Cells expressing this protein were as sensitive to McC or DSA as control cells carrying the pET vector alone. Although agrocin 84 does not affect the growth of E. coli, its addition leads to inhibition of tRNA Leu   (and detoxifies) every aminoacyl sulfamoyl adenylate tested with the exception of PSA (14). MccF is also unable to cleave processed albomycin, an inhibitor of SerRS that contains a seryl residue attached to thioxylofuranosyl pyrimidine. This compound is acetylated by MccE (14). Elucidation of the molecular determinants of MccF substrate specificity will have to await structural information on MccF and its complexes with substrates.
The mechanism of MccF function to provide resistance to McC and other aminoacyl adenylates is formally similar to the mechanism of Hint protein action. Hint is a cellular purine nucleoside phosphoramidase that cleaves, among other substrates, certain aminoacyl adenylates (23). It remains to be seen whether E. coli hinT contributes to basal levels of resistance to McC and processed McC analogs.
It is interesting to speculate why the system of McC production requires three different proteins (the MccC export pump, the MccE acetyltransferase, and the MccF protease) for immunity. One idea is that MccF is a backup immunity system that is expressed continuously throughout the growth of mcc gene cluster carrying cell culture. The basal level of immunity afforded by MccF may guarantee that cells that initiate transcription of the mccABCDE operon early (and therefore start producing McC before other cells in the culture) do not gain undue advantage. Indeed, preliminary data indicate that although the mccABCDE operon transcription is activated in the stationary phase, mccF transcription is detected in the logarithmic phase of growth. 6 In the course of this work, we made a serendipitous observation that MccF treatment strongly increases the antibacterial potency of DSA and ESA. This effect was shown to be due to the accumulation of SA, which was earlier shown to be a broad spectrum antibacterial agent. The mechanism of SA antibacterial action is unknown. Our observation that MccF overproducing cells are resistant to DSA suggests, rather unexpectedly, that internally generated SA does not affect cell growth. If so, external SA should affect the cell surface.
An MccF homolog is encoded in the agrocin 84 biosynthesis cluster. The role of this gene in agrocin 84 synthesis and/or immunity of the producing cell is unknown. In the agrocin 84 molecule, the ␣-carboxyl group of modified leucine is attached to a modified adenosine monophosphate through an N-acyl phosphoramidate linkage also found in McC. We tested the ability of E. coli MccF and its agrobacterial homolog to detoxify agrocin 84 in E. coli extracts. No inhibition relief was observed, a result that appears in line with an observation that E. coli MccF does not cleave leucyl sulfamoyl adenylate. The agrobacterial homolog also did not cleave E. coli MccF substrates McC or DSA. These data suggest that agrobacterial MccF may not be directly involved in agrocin 84 resistance. Creation of an in vitro system of agrocin 84 maturation using purified proteins shall resolve this issue.