The primary step of biotin synthesis in mycobacteria

Significance

Biotin is an enzyme cofactor required for growth of mycobacteria, and the Mycobacterium tuberculosis biotin synthetic pathway has become a validated target for antitubercular compounds. Current pathway lacks information on synthesis of the pimelate moiety that provides 7 of the 10 biotin carbon atoms. We report the first step in mycobacterial pimelate synthesis parallels the pathway first reported in Escherichia coli, in which BioC catalyzed O-methylation of a malonyl-acyl carrier protein primer allows entry into fatty acid synthesis. The mycobacterial Tam proteins functionally replace the E. coli BioC protein, although these proteins are annotated as performing a different function Moreover, tam gene function is required for biotin synthesis by Mycobacterium smegmatis and its transcription is repressed by exogenous biotin.

Abstract

Biotin plays an essential role in growth of mycobacteria. Synthesis of the cofactor is essential for Mycobacterium tuberculosis to establish and maintain chronic infections in a murine model of tuberculosis. Although the late steps of mycobacterial biotin synthesis, assembly of the heterocyclic rings, are thought to follow the canonical pathway, the mechanism of synthesis of the pimelic acid moiety that contributes most of the biotin carbon atoms is unknown. We report that the Mycobacterium smegmatis gene annotated as encoding Tam, an O-methyltransferase that monomethylates and detoxifies trans-aconitate, instead encodes a protein having the activity of BioC, an O-methyltransferase that methylates the free carboxyl of malonyl-ACP. The M. smegmatis Tam functionally replaced Escherichia coli BioC both in vivo and in vitro. Moreover, deletion of the M. smegmatis tam gene resulted in biotin auxotrophy, and addition of biotin to M. smegmatis cultures repressed tam gene transcription. Although its pathogenicity precluded in vivo studies, the M. tuberculosis Tam also replaced E. coli BioC both in vivo and in vitro and complemented biotin-independent growth of the M. smegmatis tam deletion mutant strain. Based on these data, we propose that the highly conserved mycobacterial tam genes be renamed bioC. M. tuberculosis BioC presents a target for antituberculosis drugs which thus far have been directed at late reactions in the pathway with some success.

Biotin (vitamin B7 or vitamin H), an essential enzyme cofactor throughout biology, is a covalently attached cofactor required for key carboxylation and decarboxylation reactions in fatty acid synthesis, gluconeogenesis, and amino acid metabolism. Biotin consists of an ureido ring fused to a thiophene ring and a valeric acid side chain. The enzymes that assemble the fused heterocyclic rings of biotin are conserved in all biotin-producing bacteria, plants, and fungi (1). The appearance of Mycobacterium tuberculosis strains resistant to the historically effective drugs against tuberculosis has led to a striking global increase in deaths due to this pathogen, and thus the development of new drug targets is a priority.

Biotin biosynthesis is an excellent target, because this pathway is required for the growth of mycobacteria in the laboratory (2, 3) and also for the establishment and persistence of chronic infections in a murine model of tuberculosis (4). Moreover, mammals cannot synthesize biotin, and thus inhibitors of this pathway should not interact with host metabolism. Therefore, biotin synthesis is considered a validated target for the development of M. tuberculosis antimicrobials (4–8), and numerous lead compounds have been reported (5, 7–10). Our knowledge of the mycobacterial biotin synthetic pathway is incomplete, however. In this report, we address the mechanism of synthesis of pimelic acid moiety, the source of 7 of the 10 biotin carbon atoms. Diverse pathways exist for the synthesis of this precursor; the known pathways, those of Escherichia coli and Bacillus subtilis, involve enzymes of fatty acid synthesis (Fig. 1). In E. coli, ¹³C-NMR analyses of the incorporation patterns of various biotin precursors labeled with ¹³C demonstrated that pimelic acid moiety is formed from three acetate units incorporated in a head-to-tail manner, as seen in fatty acid and polyketide synthesis (11, 12). A similar head-to-tail ¹³C-labeling pattern is seen in B. subtilis, although the labeling pattern differs from that of E. coli, because unlike in E. coli, free pimelic acid is a pathway intermediate (13). In both cases, the seventh pimelate carbon originates as CO₂.

Fig. 1.

The biotin synthesis pathway of E. coli and alignments of BioC and Tam proteins. (A) The E. coli BioC-BioH pathway for the synthesis of pimelate comprises three main steps. The initiation step is catalyzed by BioC, an O-methyltransferase that transfers a methyl group from SAM to the O-carboxyl group of malonyl-ACP to give malonyl-ACP methyl ester, a primer dedicated to biotin biosynthesis. The second step is the chain elongation cycle of fatty acid synthesis using ACP as the acyl chain carrier. The methyl ester moiety allows elongation of a malonyl chain to a pimeloyl chain after two cycles of synthesis. Once the seven-carbon chain length is achieved, BioH terminates chain elongation by cleaving the methyl ester moiety to produce pimeloyl-ACP, a substrate for BioF in the third stage of biotin biosynthesis. Assembly of the bicyclic rings in the third stage requires four reactions catalyzed by BioF, BioA, BioD, and BioB. Pimeloyl-ACP is a dedicated precursor of the second stage of the pathway and provides the majority of the biotin carbon atoms. Following synthesis, biotin is covalently attached to a specific lysine residue of biotin-dependent proteins by biotin protein ligase. SAH, S-adenosylhomocysteine; SAM, S-adenosyl-l-methionine; AMTOD, S-adenosyl-2-oxo-4-thiomethylbutyrate; 5^’-DOA, 5′-deoxyadenosine. The IUPAC names for KAPA and DAPA are 8-amino-7-oxononanoate and 7,8-diaminononanoate, respectively. Conversion of DTB to biotin requires two SAM molecules per biotin molecule. (B) Clustal W alignments of M. smegmatis mc²155 Tam, M. tuberculosis H37Rv Tam together with E. coli MG1655 BioC, B. cereus ATCC10987 BioC, and E. coli MG1655 Tam. The accession numbers of the proteins are as follows: E. coli MG1655 BioC, NP_415298; B. cereus ATCC10987 BioC, NP_415298; M. smegmatis mc²155 Tam, WP_011727078; M. tuberculosis H37Rv Tam, NP_214808; and E. coli MG1655 Tam, NP_416036. The asterisks denote the residues shown to be essential for B. cereus BioC activity (19).

In 1963, Lezius et al. (14) suggested that pimelate moiety could be formed by condensation of three molecules of a malonyl thioester in two decarboxylative Claisen-like condensations to obtain pimeloyl-acyl carrier protein (ACP). However, the hydrophobicity of the active sites evident in fatty acid synthetic enzyme crystal structures (15) argued that the postulated free carboxyl group would not be tolerated. This dilemma is avoided in E. coli and many other bacteria by a pathway in which the free carboxyl group of malonyl-ACP is methylated by BioC, an S-adenosyl-l-methionine–dependent methyltransferase (16). Methylation eliminates the charge of the free carboxyl group, introduces a methyl group that mimics that of the canonical primer, and deceives the fatty acid biosynthetic enzymes into utilizing malonyl-ACP methyl ester as a substrate (16) (Fig. 1). Following two cycles of canonical fatty acid synthesis reactions, the methyl group of the seven-carbon pimeloyl-ACP methyl ester is removed by BioH, a short-chain fatty acid esterase, to give pimeloyl-ACP, which is condensed with l-alanine by BioF to start the assembly of the fused rings of biotin (17) (Fig. 1).

The E. coli protein of known function most similar to BioC (23% identity with several gaps), Tam is an O-methyltransferase that monomethylates and thereby inactivates trans-aconitate, a nonenzymatically formed toxic derivative of the citric acid cycle intermediate cis-aconitate (18). The identity value of Tam with BioC is similar to those values among the four proteins with demonstrated BioC activity, whereas only low levels of sequence conservation are seen in pairwise alignments. For example, although the Bacillus cereus BioC functionally replaces that of E. coli, the proteins show only 26.5% identity, and the alignment requires numerous gaps (19). A similar picture emerges for BioC proteins from other bacteria. Low-level but slightly higher pairwise sequence identities are seen for the annotated mycobacterial Tams with the Tams of known activity (E. coli Tam) or structure (Agrobacterium tumefaciens Tam; Protein Data Bank [PDB] ID code 2P35).

Given the low sequence conservation among and between BioCs and Tams, the idea that the annotated mycobacterial Tams might have BioC functions seemed worth exploring. Note that gene encoding proteins with pairwise sequence identities to known Tams and known BioCs are found in most bacterial genomes as well as in the yeast Saccharomyces cerevisiae, whose Tam1 is 23% identical to the E. coli Tam. Despite its low sequence identity, yeast Tam1 activity has been demonstrated in vitro, and disruption of the gene abolishes activity. The M. smegmatis Tam also can be aligned with the yeast enzyme, albeit with many large gaps.

We report that expression of M. smegmatis Tam restored growth to an E. coli ∆bioC strain. In addition, the purified Tam replaced BioC in the synthesis of dethiobiotin (DTB) in vitro, and disruption of the M. smegmatis tam gene resulted in biotin auxotrophy. Moreover, biotin supplementation repressed tam transcription in wild-type M. smegmatis cultures. The M. tuberculosis Tam replaced E. coli BioC in vivo and in vitro and complemented the M. smegmatis tam mutant. These data suggest that the E. coli pathway may extend to all mycobacteria.

Results

We studied M. smegmatis, a fast-growing nonpathogenic soil bacterium both in vivo and in vitro, and extended the in vitro studies to the pathogen M. tuberculosis. Although the in vitro studies were often performed in parallel, we first present the M. smegmatis results to provide context for the M. tuberculosis studies.

Complementation of E. coli ∆bioC Biotin Auxotrophy by Expression of the M. smegmatis Tam Protein.

We used the E. coli ΔbioC strain STL23 (19) to test the activity of the M. smegmatis Tam protein, with B. cereus BioC serving as a control. The M. smegmatis tam gene was inserted into the arabinose promoter vector pBAD322, and the resulting plasmid transformed into the ∆bioC biotin auxotroph STL23. All strains grew well in the presence of biotin, indicating that the expressed protein was not toxic under these growth conditions. Growth of the E. coli ΔbioC strain transformants was tested in the absence of biotin on minimal medium plates. Expression of M. smegmatis tam gave only weak (albeit reproducible) growth of the E. coli ΔbioC strain in the absence of biotin. In contrast, B. cereus bioC expression gave robust growth without arabinose induction (Fig. 2A).

Fig. 2.

Complementation of E. coli ∆bioC strain STL23 by expression of M. smegmatis and M. tuberculosis Tam proteins. (A) Growth of strain STL23 containing plasmids that express M. smegmatis tam, M. tuberculosis tam (E. coli codons), or B. cereus bioC in M9 minimal salts plus 0.4% glycerol and either arabinose or biotin. To prevent cross-feeding, plates divided into sectors by plastic walls were used. The M. smegmatis Tam, M. tuberculosis Tam, and B. cereus BioC proteins were expressed from plasmids pZH101, pZH102, and pZH103, respectively, which were derived from vector pBAD322. The plate below the cartoon contained biotin. (B) Growth of strain STL23 carrying plasmids expressing M. smegmatis tam or M. tuberculosis tam in the presence of M. smegmatis acpP, M. tuberculosis acpA, or M. tuberculosis acpM. E. coli strain STL23 contained a plasmid expressing a tam gene as in A. The M. smegmatis acpP, M. tuberculosis acpA, and M. tuberculosis acpM proteins were expressed from plasmids pZH104, pZH105, and pZH106, respectively, which were derived from vector pHSG575. The upper row of plates contained arabinose, whereas the lower plates lacked arabinose. In addition, IPTG denotes IPTG induction. (C) Growth of strain STL23 expressing B. subtilis Sfp, a mycobacterial Tam and a cognate ACP. The plasmids expressing tam genes were the same as in B. B. subtilis Sfp was expressed from pNRD136 (20). All plates contained IPTG, whereas the upper and lower rows of plates were plus or minus arabinose as in B. The plus biotin was as in B. Glycerol was the primary carbon source and provided a basal level of transcription from the arabinose promoter.

The weak complementation seen for M. smegmatis tam could be due to poor interactions between the Tam proteins and E. coli malonyl-ACP. Indeed, several bacterial ACPs fail to replace function of the E. coli AcpP in vivo (20). The M. smegmatis acpP encodes a protein of 100 amino acid residues, compared with the 79-residue E. coli ACP (the proteins are 38% identical). To test the possibility of incompatible ACPs, the M. smegmatis acpP was obtained through PCR and ligated to vector pHSG575, and this plasmid was cotransformed into corresponding strains. No detectable improvement in the growth of these transformants was seen (Fig. 2B).

A straightforward and previously reported (20) explanation for these results is that the E. coli holo-ACP synthase (AcpS) failed to transfer the 4′-phosphopantetheinyl moiety from CoA to the mycobacterial apo-ACPs to form holo-ACP, the active form in fatty acid synthesis. Therefore, we expressed M. smegmatis AcpS in the foregoing strains to counter this possibility. In another set of strains, we expressed B. subtilis Sfp, a 4′-phosphopantetheinyl transferase of very broad acceptor protein specificity (21). The M. smegmatis acpS or B. subtilis sfp expression plasmids were transformed into the E. coli bioC mutant STL23 expressing the Tam and ACP proteins, and the resulting strains were streaked on minimal M9 medium in the absence of biotin. In the strain expressing Sfp and the cognate AcpP, M. smegmatis Tam allowed strong growth of the E. coli ΔbioC strain with arabinose even in the absence of induction of Sfp by isopropyl β-d-1-thiogalactopyranoside (IPTG) (Fig. 2C). For unknown reasons, the M. smegmatis AcpS protein failed to aid growth of strain STL23 expressing a mycobacterial Tam and the cognate AcpP.

Fatty Acid Synthesis-Dependent DTB Production In Vitro.

In vitro synthesis of DTB, the last intermediate of the biotin synthesis pathway (Fig. 1) provides a highly sensitive and efficient means of studying biotin pathway function in vitro (16). In the present experiment, the in vitro system used cell-free extracts of Δbio strains expressing M. smegmatis Tam that had been subjected to ammonium sulfate precipitation and dialysis, manipulations designed to remove small molecules and deplete the extracts of ACP and acyl-ACP intermediates (pure malonyl-ACP plus the small molecules required for fatty acid and DTB synthesis are added to the assay) (19). For a bioassay of DTB synthesis, biotin-starved cells of E. coli strain ER90 (ΔbioF, bioC, and bioD) were added to biotin-free minimal medium. Strain ER90 is defective in synthesis of KAPA, DAPA, and DTB but converts DTB to biotin (22). In this bioassay, the test samples diffuse from filter disks placed on the agar surface into agar seeded with strain ER90. If growth occurs, a redox indicator in the agar becomes reduced to form a bright-red, insoluble formazan deposit covering an area proportional to the concentration of the biotin pathway intermediate (23).

Here, a plasmid encoding M. smegmatis tam (pZH101) was transformed into strain STL96, a ∆bioC strain that carries multicopy plasmid pCY123, which encodes the E. coli biotin operon altered by an inactivating, in-frame deletion within bioC (16, 24). Thus, bioA, bioB, bioD , and bioF are overexpressed from the plasmid, whereas bioH is robustly expressed from the chromosome. Cell-free extracts were prepared as described in Experimental Procedures. Extracts of cells expressing M. smegmatis tam produced DTB only when the cultures had been grown with arabinose induction, thereby demonstrating a dependence on Tam induction (Fig. 3A). A similar result was seen in extracts of cells expressing the M. smegmatis Tam and AcpP plus Sfp (Fig. 4). Omission of either Tam or malonyl-ACP blocked DTB synthesis, indicating that Tam activity requires malonyl-ACP.

Fig. 3.

M. smegmatis Tam restores DTB synthesis in the BioC-dependent bioassay. (A) DTB assay for cell-free extract of the ∆bioC E. coli strain STL96, carrying plasmids expressing either M. smegmatis Tam or M. tuberculosis Tam (E. coli codons). The no Tam control was the extract of a strain carrying the empty vector. The host strain also carried the compatible multicopy plasmid pCY123, which expresses the E. coli biotin operon lacking BioC activity due to an in-frame deletion. The M. smegmatis Tam and M. tuberculosis Tam were expressed from plasmids pZH101 and pZH102, as shown in Fig. 1A. The 20 nM biotin control was the empty vector strain plus biotin. Arabinose addition was done as in Fig. 2A. (B) In vitro DTB assay for testing activities of the M. smegmatis and M. tuberculosis Tams. N-His-tag MsTam and C-His-tag MsTam are N-terminal and C-terminal hexahistidine-tagged forms of M. smegmatis Tam. The native Ms Tam and Mtb Tam are the untagged native forms of M. smegmatis and M. tuberculosis Tam. Bc BioC is the native form of B. cereus BioC. Negative controls: no Mal-ACP, no malonyl-ACP was added to the reactions; no Tam, no Tam was added. (C) SDS- PAGE of native form of M. smegmatis Tam. L, protein ladder. Lanes 1 to 6, gradient elution of anion-exchange chromatography using a HiTrap SP FF column from 0.3 M LiCl to 0.5 M LiCl. The native M. smegmatis Tam is a protein of 25.8 kDa as calculated from the gene sequence.

Fig. 4.

In vitro DTB synthesis activities of mycobacterial Tams when coexpressed in the ∆bioC E. coli strain STL96 with Sfp and a cognate ACP before extract preparation. (A) The M. tuberculosis (Mtb) Tam expressed with its cognate AcpA. (B) The M. tuberculosis (Mtb) Tam expressed with its cognate AcpM (left quadrants) and M. smegmatis Tam expressed with its cognate AcpP. The ring of the 20 nM biotin control in A is larger than that of Fig. 3, because the plates were intentionally dried less than those in Fig. 3, to allow increased diffusion. The no Tam control is the strain carrying the empty vector.

DTB Synthesis In Vitro with Purified M. smegmatis Tam.

E. coli BioC is an O-methyltransferase that methylates the free carboxyl of malonyl-ACP (19). However, on overexpression, E. coli BioC protein invariably forms insoluble inclusion bodies, and the refolded protein have only traces of activity (16). BioC homologs from a series of diverse bacteria also form insoluble inclusion bodies, the exception being B. cereus BioC (16, 25). Note that high-level expression of B. cereus BioC is toxic to E. coli. Toxicity almost certainly is related to depletion of the malonyl-ACP building block of fatty acid synthesis by conversion to its O-methyl ester (19, 25). To avoid these possible problems, we expressed the Tam protein from a moderately induced arabinose promoter on a medium copy number plasmid.

Plasmids expressing M. smegmatis Tam were constructed with either N-terminal or C-terminal hexahistidine tags. Although both tagged proteins were soluble and readily purified by standard metal chelation chromatography, they were unable to replace E. coli BioC in the in vitro DTB synthesis assay. This was not unprecedented; a tagged version of B. cereus BioC also lacked function (26) in vitro. The tagged Tam proteins also failed to allow growth of the ∆bioC strain STL23.

Native B. cereus BioC was previously purified via three column chromatography steps (19), and essentially the same three chromatographic steps were used to purify the native M. smegmatis Tam (Fig. 3C). DTB was produced on addition of purified native M. smegmatis Tam to cell extracts of the ∆bioC strain STL96 (Fig. 3B).

Regulation and Deletion of the M. smegmatis tam Gene.

Biotin is required for bacterial growth, although only a few hundred molecules/cell are needed (13, 27, 28). The foregoing data argue very strongly that M. smegmatis Tam acts as a biotin synthesis O-methyltransferase in vitro and in E. coli. If so, then Tam expression might be regulated by biotin availability, and disruption of the tam gene should result in biotin auxotrophy.

Expression of M. smegmatis tam in cultures of different growth phases grown with or without biotin supplementation was assayed by qRT-PCR. The addition of biotin resulted in decreased expression of M. smegmatis tam (Fig. 5A). The tam transcriptional level was at least fivefold higher in cells of all growth phases when the cells were grown on minimal medium lacking biotin (Fig. 5A). In early stationary stage cells (36 h), the tam transcriptional level was almost twofold higher than in late-exponential phase cells (24 h). Expression of tam was decreased in the late stationary phase (48 h), although it remained higher than seen in the mid-exponential phase (12 h). These data provide further evidence of the involvement of M. smegmatis tam in biotin synthesis.

Fig. 5.

(A) Effects of biotin supplementation on expression, biotin auxotrophy, and complementation of the M. smegmatis ∆bioC (∆tam) mutant strain. Shown are qRT-PCR analyses of the relative transcription levels of M. smegmatis tam in the wild-type strain grown with and without biotin. Cultures were grown for the time periods shown on the abscissa. The levels of tam transcripts are given relative to the level of cells grown for 12 h with biotin. Open columns, with biotin (20 nM) supplementation; solid columns, without biotin supplementation. (B) Growth of the M. smegmatis ∆bioC (∆tam) mutant strain on minimal medium with and without biotin. (C) Growth on minimal medium of the ∆bioC (∆tam) mutant strain expressing different tam or bioC genes. M. smegmatis bioC (tam), M. tuberculosis tam (E. coli codons), E. coli bioC, and E. coli tam were expressed from plasmids pZH112, pZH113, pZH114, and pZH115, respectively, all derived from pMind. Vector denotes the empty pMind vector. Strains were grown on M. smegmatis minimal medium with induction by tetracycline added at 20 ng/mL (D) Growth of M. smegmatis ∆bioC (∆tam) mutant strain carrying the native M. tuberculosis tam or driven by the vector promoter plus the promoter region of M. smegmatis bioC (tam) (denoted pMs+) expressed from plasmids pZH116 and pZH117, respectively. Vector denotes the empty pMind vector as a negative control. One plate sector contained 20 nM biotin as a positive control.

To test this hypothesis further, a M. smegmatis ∆tam mutant was constructed (Experimental Procedures). The ∆tam mutant failed to grow on minimal medium unless biotin (10 nM) was added (Fig. 5B). For complementation testing, M. smegmatis tam, E. coli tam, and E. coli bioC gene fragments were amplified from 20 bp upstream of the translation initiation site to the transcription terminator and then ligated to the shuttle plasmid, pMind, behind a tetracycline-inducible promoter. The pMind-derived plasmids were selected in E. coli and transformed into the M. smegmatis tam mutant strain by electroporation. Colonies were selected by resistance to hygromycin and kanamycin and then tested on minimal medium with or without biotin supplementation. M. smegmatis tam and E. coli bioC supported weak but significant growth in the absence of biotin supplementation when 20 ng/mL tetracycline was added as an inducer (Fig. 5C). As expected from the biotin auxotrophy of ∆bioC strains, E. coli tam failed to complement growth of the mutant strain.

Activity of the M. tuberculosis Tam.

An expression plasmid carrying a synthetic gene with E. coli codons that encoded the M. tuberculosis Tam (62% identical to M. smegmatis Tam) showed no detectable ability to support biotin-independent growth of the E. coli bioC strain STL23 under conditions in which the M. smegmatis Tam gave weak complementation (Fig. 2A). Moreover, little or no complementation was seen when the host strain also expressed either of the M. tuberculosis ACPs (AcpA or AcpM) (Fig. 2B). However, when the host strain expressed Sfp and M. tuberculosis Tam plus either AcpA or AcpM, growth proceeded in the absence of biotin (Fig. 2C). Therefore, M. tuberculosis Tam had BioC complementation activity but only in the presence of a cognate holo-ACP. To confirm these in vivo results, we performed in vitro tests with extracts of derivatives of E. coli ∆bioC strain STL96 expressing Sfp and M. tuberculosis Tam plus either AcpA or AcpM and observed DTB synthesis (Fig. 4). In the absence of the cognate ACPs, little or no DTB synthesis was observed in ∆bioC cell extracts (Fig. 3A) either when M. tuberculosis Tam had been expressed in the cells used for extract preparation or when purified M. tuberculosis Tam had been added to ∆bioC cell extracts (Fig. 3B). As was seen in the M. smegmatis experiments, the M. tuberculosis AcpS was unable to replace Sfp.

Expression of the synthetic gene encoding the M. tuberculosis Tam failed to complement the M. smegmatis ∆tam strain (Fig. 5C). However, that gene was constructed with E. coli codons, which differ markedly from those used by mycobacteria (29). Therefore, we synthesized a tam gene with the native M. tuberculosis sequence and ligated the gene into pMind. The pMind-derived plasmid was selected in E. coli and transformed into the M. smegmatis ∆tam mutant strain by electroporation as described above. The resulting strain grew in the minimal medium without biotin addition (Fig. 5D), and thus the M. tuberculosis Tam functionally replaced the M. smegmatis Tam.

Discussion

We have demonstrated that the physiological function of the annotated M. smegmatis Tam is synthesis of the biotin pimelate moiety. Several lines of evidence indicate that M. smegmatis Tam methylates the free carboxyl of malonyl-ACP. The protein replaces E. coli BioC in DTB synthesis in vivo and in vitro (Figs. 2 and 3). Moreover, deletion of the M. smegmatis tam gene results in biotin auxotrophy, thereby establishing the role of the protein in biotin synthesis (Fig. 5). Expression of E. coli BioC allows growth of the M. smegmatis biotin auxotroph, providing further confirmation of the physiological activity of M. smegmatis Tam. Finally, transcription of the M. smegmatis tam gene is repressed by biotin supplementation (Fig. 5). Thus, we have renamed the M. smegmatis tam gene bioC.

Note that although M. smegmatis bioC expression is repressed by biotin, the 250-bp sequence upstream of the coding sequence lacks a recognizable sequence for binding of BioQ, the repressor of biotin gene transcription in M. smegmatis (30–32). This argues for a complex mode of transcriptional control of the biotin synthetic genes of this bacterium. Recently, mutants lacking activity of pyruvate carboxylase, a biotin-requiring enzyme that converts pyruvate to oxaloacetate, were found to be defective in biosynthesis due to decreased transcription of the biotin ring synthesis genes (3). The working model is that the carboxylase is required for synthesis of the ligand that induces BioQ. The M. tuberculosis pyruvate carboxylase could replace that of M. smegmatis, but that of B. subtilis could not. Another M. smegmatis biotin synthetic gene, bioA, also lacks a BioQ-binding site (30, 31), although whether bioA transcription responds to biotin is unclear. It should also be noted that M. tuberculosis lacks a recognizable bioQ gene.

It seems most likely that the M. tuberculosis Tam is essential for biotin synthesis in this pathogen. Validating this hypothesis would require in vivo and genetic experiments paralleling those that we have performed with M. smegmatis. However, since BSL3 biosafety containment is required to grow M. tuberculosis, in vivo analyses must be left to the small number of laboratories authorized to culture this highly pathogenic bacterium. Given such validation, M. tuberculosis Tam could become a good target for development of antitubercular compounds.

The cognate partner of E. coli BioC is BioH, the esterase that removes the methyl group from pimeloyl-ACP methyl ester (16, 25) (Fig. 1). This cleavage has two essential functions: to prevent further elongation of the acyl chain and to free the carboxyl group for ligation to the biotin-dependent enzymes (17). Therefore, we expect that M. smegmatis encodes a protein that cleaves the methyl group from pimeloyl-ACP methyl ester. It will be challenging to identify the M. smegmatis esterase, owing to the known diversity of enzymes that catalyze cleavage of the methyl group from pimeloyl-ACP methyl ester together with the large number of candidate proteins encoded in the M. smegmatis genome. Seven distinct pimeloyl-ACP methyl esterases from diverse bacteria have been identified, and all but BioH and BioG are restricted to specific bacterial species (33–37). These esterases are encoded by genes that are sometimes located within biotin gene clusters and in other cases are free-standing genes that lack a genome context (as does bioH in E. coli). To date, all known esterases that cleave pimeloyl-ACP methyl ester are members of the α,β-hydrolase family and contain the signature Ser-His Asp catalytic triad (33–36), and M. smegmatis has more than 30 genes annotated as encoding such proteins. M. tuberculosis genomes show a similarly large number of α,β-hydrolase encoding genes, although annotations of the sequenced genomes report variable numbers of such genes.

Our finding that biotin synthesis by the mycobacterial BioC/Tam proteins either requires (M. tuberculosis) or was markedly stimulated (M, smegmatis) by their cognate ACPs when coexpressed with the Sfp modification enzyme strongly supports the idea that pimelate synthesis in these bacteria proceeds via fatty acid synthesis. A puzzle is the inability of the cognate AcpS proteins to replace Sfp. A possible explanation for these results is that the only annotated mycobacterial AcpS proteins—those that we tested—specifically modify the polyfunctional fatty acid synthases encoded immediately upstream of the acpS genes. Such modification specificity is seen in the yeast polyfunctional fatty acid synthases, proteins that share marked structural and mechanistic similarities with the mycobacterial synthases (38, 39).

Experimental Procedures

Bacterial Strains, Media, and Plasmids.

The plasmids and strains used in this study are listed in SI Appendix, Tables S1 and S2. M. smegmatis mc²155 (now Mycolicibacterium smegmatis mc²155) was grown at 37 °C in Middlebrook 7H9 liquid medium or on 7H10 agar plates. The minimal medium for E. coli was M9, and the minimal medium for M. smegmatis was modified from 7H9 (40): 4 ml glycerol, 1.5 g Na₂HPO₄, 1.5 g KH₂PO₄, 0.4 g Na citrate, 0.025 g MgSO₄, 0.5 g (NH₄)₂SO₄, 0.5 g l-glutamic acid, 0.04 g ferric ammonium citrate, 1 mM CaCl₂, 1 mg ZnSO₄, 1 mg CuSO₄, 1 mg pyridoxine, and 7 g ADC-41 per liter, at a final pH of 6.6 ± 0.2. Antibiotics were used at final concentrations of 30 μg/mL kanamycin, 50 μg/mL carbenicillin, 50 μg/mL hygromycin, and 10 μg/mL cycloheximide. Tetracycline was added at 20 ng/mL as an inducer. Sucrose, l-arabinose, and glycerol were added at 10%, 0.02%, and 0.4%, respectively. Biotin was added at 20 nM, and IPTG was added at 0.5 μM for preparation of cells for native form protein purification and at 1 mM otherwise. E. coli strains were grown as described previously (16, 19).

Plasmid Constructions.

The plasmids and PCR amplification primers used are listed in SI Appendix, Tables S1 and S2, respectively. Strain STL23 was complemented with either plasmid-encoded tams or bioC genes under control of an arabinose-dependent promoter. Plasmid pZH101 was constructed by inserting the PCR-amplified M. smegmatis tam gene using primers Ms tam NcoI and Ms tam HindIII. The PCR amplified M. smegmatis MC²155 tam gene (MSMEG_0629; GenBank accession no. WP_011727078) was inserted into pBAD322 cut with NcoI and HindIII to give pZH101. Plasmid pZH102 was obtained by digestion of B. cereus bioC -pET28(b) with NcoI and HindIII, and the fragment was ligated with pBAD322 (41). Two M. tuberculosis tam genes were synthesized by Integrated DNA Technologies (IDT) as gBlocks gene fragments encoding the Tam of M. tuberculosis strain H37Rv (Rv0294; GenBank accession no. NP_214808). One of these genes had E. coli codons, and the other had the native M. tuberculosis codons. The E. coli codon version was digested with NcoI and HindIII, then ligated with pBAD322 to obtain plasmid pZH103. The M. smegmatis acpP gene was obtained by PCR using primers Ms acpP SalI F and Ms acpP HindIII R), digested with SalI and HindIII, and then ligated to vector pHSG575 digested with the same enzymes (pZH104). M. tuberculosis acpA (pZH105) and acpM (pZH106) were synthesized by IDT gBlocks from the M. tuberculosis H37Rv genome sequence, digested with BamHI and PstI or SalI and HindIII, respectively, and then ligated to pHSG575 cut with the same enzymes. The M. smegmatis acpS gene with E. coli codons was synthesized by IDT gBlocks, digested with XmaI and HindIII, and then ligated to pI1030, digested with same enzymes to obtain pZH118. When an E. coli strain carried more than one plasmid, the plasmids had compatible replication origins and selective markers.

N-terminal or C-terminal hexahistidine-tagged forms of M. smegmatis Tam were generated by PCR using primers Ms tam NdeI and Ms tam HindIII (N-terminal) and Ms tam NcoI and Ms tam XhoI (C-terminal) primers. The resulting products were digested with the same enzymes, then ligated to digested pET28(b) to construct pZH107 (N-terminal) and pZH108 (C-terminal). For expression and purification of the native M. smegmatis Tam and M. tuberculosis Tam, the tam sequences were obtained by NcoI and HindIII digestion and ligated to vector pET28(b) cut with the same enzymes to yield pZH109 (Ms Tam) and pZH110 (Mtb Tam) for expression of the native forms of the proteins.

Plasmids p2NIL and pGAOL19 (42), obtained from Addgene, were used to construct the M. smegmatis tam knockout plasmid. The 500 bp upstream of M. smegmatis tam were amplified with Q5 DNA polymerase (New England BioLabs) using primers Ms tam knock P1 BamHI and P2 SacI with M. smegmatis genomic DNA as the template. The downstream were amplified in the same way using primers Ms tam knock P3 SacI and P4 PstI. The two fragments were purified and joined by overlapping PCR using the outside primers. The fused product was then digested with BamHI and HindIII and inserted between the same sites of suicide vector p2NI (42). The transformants were selected by kanamycin resistance and screened by PCR. The marker hygromycin plus the SacB counterselection cassette of pGAOL19 obtained by PacI digestion was ligated to the PacI-linearized p2NIL-derived plasmid to generate pZH111. Hygromycin-resistant transformants were selected and confirmed by PCR.

For expression of genes in M. smegmatis, the PCR-amplified fragments containing M. smegmatis tam, M. tuberculosis tam (native codons), E. coli tam, and E. coli bioC were digested with BamHI and SpeI and ligated with the mycobacterial tetracycline regulatory plasmid pMind (43) (Addgene), digested with the same enzymes. Transformants were selected by kanamycin resistance and screened by PCR.

Construction of the M. smegmatis tam Deletion Mutant Strain.

The strain with a deleted tam gene was generated by a two-step process to generate an unmarked mutant strain (42). Single cross-over transformants were initially obtained by electroporation and selected on plates containing hygromycin and kanamycin. Selection for the second cross-over was done as follows. A single crossover colony was cultured in 7H9 liquid medium lacking antibiotics for 36 h until the OD reached 0.2 to 0.4, then diluted to several different concentrations in 7H10 liquid medium lacking antibiotics and containing 10% sucrose grown for 36 h and then plated on 7H9 plates. Kanamycin-sensitive colonies were isolated and analyzed by PCR. Owing to the lack of a selection marker, the second cross-over can result in either the wild-type allele or the mutant allele necessitating screening by PCR.

RNA Isolation and qRT-PCR.

The cultures were harvested at the time points corresponding to early-log, exponential, and stationary phases with or without biotin. The cell pellets were suspended in 5 mL of fresh protoplast buffer (15 mM Tris⋅HCl pH 8, 0.45 M sucrose, 8 mM EDTA, and 4 mg/mL lysozyme) and incubated at 37 °C for 1 h. The total RNA was isolated from the whole-cell lysates using the RNeasy Miniprep Kit (Qiagen) following the manufacturer’s instructions. The contaminating DNA was digested with the RNase-free DNase Set (Qiagen), and the concentration and purity of the RNA were determined using a NanoDrop 2000C spectrophotometer (Thermo Fisher Scientific). The Omniscript Reverse Transcription Kit (Qiagen) was used for real-time qRT-PCR. The quantification of gene expression and the melting curve analyses were performed with the Mastercycler ep realplex 4S PCR system (Eppendorf) and iQ SYBR Green Supermix (Bio-Rad). The constitutively expressed 16S rRNA gene served as the reference against which to standardize all samples and replicates.

Protein Purification and Preparations of Cell-Free Extracts and in Vitro DTB Synthesis.

The native forms of M. smegmatis Tam and M. tuberculosis tam were expressed in E. coli BL21 Turner as described previously for B. cereus BioC (19). In brief, the E. coli ∆bioC strain STL96 was transformed with either plasmid pZH101 or plasmid pZH102. Cultures were grown in LB-M9 medium at 37 °C and then inoculated into 1 L of 2XYT-M9 medium containing 50 μg/mL kanamycin and 0.5 μM IPTG. The cultures were shaken at room temperature for 17 h. The cells were harvested by centrifugation, and the cell pellets were stored at −80 °C. The M. tuberculosis Tam was expressed and purified by the same procedure, although the yields were much lower, and concentration of the protein by ultrafiltration was required.

For cell-free extract preparations plasmids pZH101 and pZH102 were transformed into the E. coli strain STL96 to generate strains ZH104 and ZH105, respectively. Strains ZH104 and ZH105 were grown at 37 °C to OD 0.6 to 0.8 in 250 mL of M9 minimal medium containing 2 nM biotin. The cell-free extracts were prepared and DTB assays conducted as described previously (19).

Data Availability Statement.

All study data are included in the main text and SI Appendix.