AccD6, a Member of the Fas II Locus, Is a Functional Carboxyltransferase Subunit of the Acyl-Coenzyme A Carboxylase in Mycobacterium tuberculosis

ABSTRACT The Mycobacterium tuberculosis acyl-coenzyme A (CoA) carboxylases provide the building blocks for de novo fatty acid biosynthesis by fatty acid synthase I (FAS I) and for the elongation of FAS I end products by the FAS II complex to produce meromycolic acids. The M. tuberculosis genome contains three biotin carboxylase subunits (AccA1 to -3) and six carboxyltransferase subunits (AccD1 to -6), with accD6 located in a genetic locus that contains members of the FAS II complex. We found by quantitative real-time PCR analysis that the transcripts of accA3, accD4, accD5, and accD6 are expressed at high levels during the exponential growth phases of M. tuberculosis in vitro. Microarray analysis of M. tuberculosis transcripts indicated that the transcripts for accA3, accD4, accD5, accD6, and accE were repressed during later growth stages. AccD4 and AccD5 have been previously studied, but there are no reports on the function of AccD6. We expressed AccA3 (α3) and AccD6 (β6) in E. coli and purified them by affinity chromatography. We report here that reconstitution of the α3-β6 complex yielded an active acyl-CoA carboxylase. Kinetic characterization of this carboxylase showed that it preferentially carboxylated acetyl-CoA (1.1 nmol/mg/min) over propionyl-CoA (0.36 nmol/mg/min). The activity of the α3-β6 complex was inhibited by the ε subunit. The α3-β6 carboxylase was inhibited significantly by dimethyl itaconate, C75, haloxyfop, cerulenin, and 1,2-cyclohexanedione. Our results suggest that the β6 subunit could play an important role in mycolic acid biosynthesis by providing malonyl-CoA to the FAS II complex.

Tuberculosis causes 2 million deaths each year, according to the World Health Organization. Mycobacterium tuberculosis, the pathogen that causes the disease, infects 8 million people each year and is one of the world's deadliest pathogens (9). The ongoing AIDS pandemic has developed a deadly synergy with tuberculosis, which is the leading cause of death among AIDS patients (2). Multidrug-resistant M. tuberculosis strains have been emerging rapidly (9), and the need for identifying novel drug targets in this pathogen has become urgent. The cell wall of M. tuberculosis is lipid enriched and acts as an impermeable barrier to many common broad-spectrum antibiotics (14).
The first committed step of fatty-acid biosynthesis, which is the biotin-dependent carboxylation of acyl-coenzyme A (CoA) to produce malonyl-CoA and methylmalonyl-CoA, is catalyzed by the acyl-CoA carboxylase. The reaction consists of two catalytic steps, which involve the biotin carboxylase and the carboxyltransferase (8). In M. tuberculosis, the biotin carboxylation step is catalyzed by the ␣ subunit; there are three open reading frames (ORFs) that can encode the ␣ subunit (accA1 to -A3) in the genome. Carboxyl transfer is catalyzed by the ␤ subunit, and there are six ␤ subunits (accD1 to -D6) in the genome of the pathogen (6).
Previously, the catalytic activities of the ␣ 3 , ␤ 4 , and ␤ 5 subunits were studied (10,11,22,24). However, the levels of expression of the various subunits have not been examined. Our analysis of transcripts from M. tuberculosis cells indicate that the ␣ 3 , ␤ 4 , ␤ 5 , ␤ 6 , and ε ORFs are the main subunits regulated during cell growth. To determine whether the highly expressed ␤ 6 subunit possessed enzymatic activity and to assess its substrate specificity, we expressed and purified the ␤ 6 subunit and reconstituted it with the purified ␣ 3 subunit. We report that an active acyl-CoA carboxylase was reconstituted with the purified ␤ 6 and ␣ 3 subunits and that it preferentially carboxylated acetyl-CoA over propionyl-CoA. This is the first report showing that ␤ 6 , which is a member of a fatty acid synthase II (FAS II) gene locus, is a functional carboxyltransferase of the acyl-CoA carboxylase in M. tuberculosis, and these results suggest that ␤ 6 might make a significant contribution to mycolate biosynthesis. purchased from Sigma and Fisher Scientific. All media were purchased from Difco. Nucleotide primers were synthesized by Integrated DNA Technologies, Inc. qPCR analysis. Total RNA (5 g) was treated with 2.5 units of DNase I (Ambion) according to the manufacturer's instructions. First-strand cDNA was synthesized using SuperScript III and 6 g of random primers (Invitrogen). The cDNA was purified with a QIAquick PCR purification kit (QIAGEN). The primers used are listed in Table 1. The primers corresponding to each ORF were designed to amplify less than 100 bp of product. Quantitative real-time PCR (qPCR) was performed in triplicate with three different concentrations of the diluted cDNAs (0.5 ng, 1 ng, and 5 ng) and 25 l of iQ SyBr Green Supermix (2ϫ Premixture; Bio-Rad) in a total volume of 50 l. The qPCR was carried out at 95°C for 20 s and 60°C for 20 s (40 cycles). Calculated thresholds were determined using the maximum-curvature approach after determination of per-well baseline cycles (iCycler; Bio-Rad). The amplified DNA samples were further subjected to melting-curve analysis and 1.2% agarose gel electrophoresis to verify the amplification product. Relative quantities of target ORF transcripts were calculated using the cycle threshold (C T ) values of the housekeeping gene, sigA (Rv2703), as an internal standard as described previously (12,23).
Microarray hybridization. The Institute for Genomic Research (Pathogen Functional Genomics Resource Center [http://www.tigr.org]) provided the Mycobacterium tuberculosis genome microarray for this study. The genome microarray consisted of 70-mer oligonucleotides representing 4,127 ORFs from the M. tuberculosis reference strain H37Rv and 623 unique ORFs from strain CDC1551 that are not present in the H37Rv strain's annotated gene complement (98% of H37Rv ORFs). The full 70-mer complement was printed four times on the surface of the microarray. Total M. tuberculosis RNA was isolated using a TRIzol (Invitrogen) extraction and RNeasy (QIAGEN) purification as described previously (32). Two-color (Cy3 and Cy5) hybridization was used for the microarray analysis. Generally, RNA extracted from cells growing exponentially at an optical density (A 600 ) of approximately 0.5 in Middlebrook 7H9 medium (pH 7.0) was used to create fluorescent Cy3-labeled reference cDNA for each experiment (mid-log phase). The reference Cy3-labeled cDNA was hybridized together with the Cy5-labeled cDNA synthesized from RNA extracted from cells when the optical density was approximately doubled (late-log phase). Labeled cDNA was prepared as follows. Ten micrograms of total RNA and 2 g of random oligonucleotide hexamers (Invitrogen, CA) were incubated for 15 min at 65°C, cooled on ice, combined with Stratascript reverse transcriptase buffer, 0.5 mM dATP, 0.5 mM dGTP, 0.5 mM dCTP, 0.02 mM dUTP, 1.5 nmol aminoallyl-dUTP (Amersham, NJ), and 2 l Stratascript reverse transcriptase II (Invitrogen, CA) in a total volume of 30 l. This mixture was incubated for 10 min at room temperature and at 42°C overnight. Synthesized cDNA was purified with a QIAquick PCR purification kit (QIAGEN, CA) according to the manufacturer's recommendations. The aminoallyl-dUTP-conjugated cDNA probes were spin dried (SpeedVac; Savant) and resuspended in 50 mM sodium carbonate buffer (pH 9.3) and cyanine monofunctional dyes (Amersham) for 1 h at room temperature, followed by quenching with 4 M hydroxylamine. The cyanine-labeled probes were purified again with a QIAquick PCR purification kit. All hybridizations were performed with dye reversal replicates. QuantArray (version 3.0; Perkin-Elmer) was used for 16-bit TIFF image quantification and initial data visualization. The hybridization signal was subjected to normalization and clustering by using an open-source R (version 2.1.1) package (http://www .bioconductor.org/).
Recombinant-protein expression and preparation of cell extracts. The ORF for ␤ 6 (accD6; Rv2247) was cloned by PCR using Pfu Turbo HotStart DNA polymerase (Stratagene) from the genomic DNA of M. tuberculosis H37Rv. The expression construct was prepared in pET 200 directional-TOPO expression vector (Invitrogen), and sequence integrity was confirmed by DNA sequencing. BL21 Star (DE3) host cells were transformed with the expression construct, and the overnight culture was diluted 1:50 in fresh medium. IPTG (isopropyl-␤-Dthiogalactopyranoside) was added to a final concentration of 1 mM when the culture reached an optical density at 600 nm (OD 600 ) of 0.7, and the induction was carried out for 4 h in a 37°C shaker. The ␣ 3 subunit was expressed and biotinylated with the E. coli biotin ligase as described previously (22). Following induction, the cells were washed and resuspended in lysis buffer (50 mM sodium phosphate, 300 mM NaCl, 1 mM phenylmethylsulfonyl fluoride). The cells were disrupted by sonication using a Branson Sonifier 450 (Branson Ultrasonics Corp.). The cell lysates were clarified by centrifugation at 12,000 ϫ g, and the supernatants were used for purification of the expressed proteins. The ␤ 5 (Rv3280) and ε (Rv3281) subunits were expressed in E. coli as described previously (22).
Purification, SDS-PAGE analysis, and dialysis of expressed proteins. The clear supernatants obtained from the above-mentioned lysates were applied to TALON cobalt affinity resin (BD Biosciences), and the His-tagged ␤ 6 and ␣ 3 (after biotinylation) were purified by the batch/gravity procedure of the manufacturer with the following modifications: the bound proteins were washed with 10 mM imidazole and eluted from the affinity resin with 100 mM imidazole and 500 imidazole elution steps. The eluted fractions were analyzed by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and fractions containing pure protein were pooled and dialyzed against 100 mM potassium phosphate (pH 8.0) containing 10% glycerol at 4°C. The ␤ 5 and ε subunits were purified as described previously (22).
Reconstitution of the carboxylase complex. Following dialysis, the purified ␣ 3 subunit (100 g/ml) was mixed with the purified ␤ 6 subunit (100 g/ml), and the purified ␣ 3 , ␤ 6 , and ε subunits were mixed together in a 1:1:2 molar ratio and incubated on ice for 6 h. Following this, the solutions were concentrated four-to Acyl-CoA carboxylase assay. The carboxylase activity of the reconstituted ␣ 3 -␤ 6 complex was measured by following the incorporation of the radiolabel from NaH 14 CO 3 into acid-stable reaction product using a modified procedure of Hunaiti and Kolattukudy (15). The reaction was carried out in a 100-l volume containing 100 mM potassium phosphate, pH 8.0, 300 g bovine serum albumin, 3 mM ATP, 5 mM acetyl-CoA or propionyl-CoA, 5 mM NaH 14 CO 3 , 5 mM MgCl 2 , and 20 to 30 g of reconstituted enzyme. Following incubation at 30°C for 2 h, the reaction was stopped with 150 l of 6 M HCl, and the entire solution was evaporated to dryness in a heating block at 100°C. The residue was resuspended in 100 l water, and the radioactivity was measured by liquid scintillation counting. The effects of inhibitors were determined by preincubating the reconstituted ␣ 3 -␤ 6 complex with the inhibitor for 10 min at 24°C, after which the percent inhibition above the respective solvent control was determined. The kinetic parameters for the ␣ 3 -␤ 6 carboxylase were calculated using nonlinear regression analysis with the Michaelis-Menten equation, and the 50% inhibitory concentrations (IC 50 s) were determined from the sigmoidal dose-response equation (GraphPad Prism version 3.02; GraphPad Software).
Microarray accession number. The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (GEO [http://www.ncbi .nlm.nih.gov/geo/]) and are accessible through GEO Series accession number GSE5977.

Analysis of carboxylase subunit transcript levels.
To determine which of the multiple carboxylase subunit genes are actually expressed in M. tuberculosis, we analyzed the expression levels of all of the ORFs encoding the carboxylase subunits by qPCR and microarray analyses. RNA isolated from cells at exponential growth phase was subjected to qPCR, and each C T value was normalized to the level of the internal control sigA qPCR C T value as described previously (12,23). By comparison with the normalized C T values of sigA, we determined the relative levels of expression of the ACC subunits during exponential growth of M. tuberculosis. The results showed that the levels of expression of the ␤ 4 , ␤ 5 , ␤ 6 , and ε subunits were highest among the carboxylase subunit ORFs when the mycobacterial cells were in exponential growth phase (Fig. 1). ␣ 3 was the most highly expressed gene among the acyl-CoA carboxylase ␣-subunit ORFs. To monitor gene expression changes during cell growth, independent batches of cells were subjected to microarray analysis. Comparison of transcription profiles obtained at various growth stages indicated that the ␣ 3 , ␤ 4 , ␤ 5 , ␤ 6 , and ε ORFs were the most significantly repressed subunits among all the acyl-CoA carboxylase ORFs as the mycobacterial cells transitioned from mid-log to late-log and stationary phases (Fig. 2), while the levels of transcripts for other ACC subunits did not change much.
Purification of ␣ 3 and ␤ 6 subunits. Although the ␤ 6 (accD6) ORF is located within a cluster of genes encoding FAS II components (Fig. 3) which have been shown to be involved in mycolic acid biosynthesis (28), the enzymatic activity of ␤ 6 has never been investigated. Since ␤ 6 was reported to be essential for the viability of M. tuberculosis, along with ␣ 3 , ␤ 4 , and ε subunits (26), and there is no ␣ subunit in the genetic neighborhood of ␤ 6 , we deduced that the ␣ 3 -␤ 6 complex might constitute a functional acyl-CoA carboxylase. To investigate the carboxyltransferase activity of ␤ 6 , we expressed ␤ 6 and ␣ 3 as His-tagged fusion proteins in E. coli and purified them by cobalt affinity chromatography. The ␣ 3 subunit was biotinylated using the E. coli biotin ligase prior to purification as described previously (22). The fractions eluted from the TALON resin were analyzed on SDS-PAGE (Fig. 4), and the fractions containing pure protein were used for reconstitution of the acyl-CoA carboxylase. The ε and ␤ 5 subunits were purified as described previously (22). Reconstitution and characterization of the enzymatic activity of ␣ 3 -␤ 6 carboxylase. Following dialysis, the ␣ 3 and ␤ 6 subunits were mixed together in a 1:1 molar ratio, and the reconstituted ␣ 3 -␤ 6 complex was assayed for acyl-CoA carboxylase activity. The ␣ 3 -␤ 6 complex carboxylated acetyl-CoA with high specific activity (1.1 nmol/mg/min) and propionyl-CoA at lower levels (0.36 nmol/mg/min). The activity increased linearly with time and protein concentration ( Fig. 5A and B) and displayed typical Michaelis-Menten kinetics for bicarbonate (Fig. 5C), acetyl-CoA and propionyl-CoA (Fig. 5D), and ATP (Fig. 5E). Under the assay conditions, saturation was reached with 2 mM bicarbonate, 5 mM acetyl-CoA or 2.5 mM propionyl-CoA, and 0.5 mM ATP. The optimal MgCl 2 concentration was 10 mM for acetyl-CoA carboxylase (ACC) activity and 5 mM for propionyl-CoA carboxylase (PCC) activity. The kinetic parameters for the ␣ 3 -␤ 6 complex are tabulated in Table 2.
Effects of inhibitors on ␣ 3 -␤ 6 . The carboxylase activities of ␣ 3 -␤ 6 were inhibited by avidin and several other inhibitors. As shown in Table 3, at 100 M, dimethyl itaconate and C75 inhibited the ACC activity of ␣ 3 -␤ 6 by 96% and 87%, respectively, and the PCC activity by 53% and 54%, respectively. Haloxyfop, cerulenin, and 1,2-cyclohexanedione inhibited the enzymatic activities significantly. The dose dependence of inhibition by dimethyl itaconate and C75 of the ACC and PCC activities of ␣ 3 -␤ 6 was investigated, and the results are shown in Fig. 6. From this sigmoidal dose-response analysis, the IC 50 s for dimethyl itaconate were calculated to be 8.2 M for ACC and 10.8 M for PCC. For C75, the IC 50 s were 50.2 M for ACC and 101.5 M for PCC.
Effect of the epsilon subunit on the ␣ 3 -␤ 6 complex and stoichiometry of the ␣ 3 -␤ 6 carboxylase. We examined the ef-fect of the epsilon (ε) subunit on the activity of the ␣ 3 -␤ 6 complex. In our previous report on the effect of the ε subunit on the carboxylase activities of the ␣ 3 -␤ 5 complex, we determined that the optimal molar ratio of ␣ 3 to ␤ 5 to ε was 1:1:2 (22). Therefore, we incubated the ␣ 3 , ␤ 6 , and ε subunits together at the same molar ratio of 1:1:2 prior to assay. As shown in Fig. 7, the ε subunit inhibited both the ACC and PCC activities of the ␣ 3 -␤ 6 complex. However, the ε subunit stimulated the ACC and PCC activities of the ␣ 3 -␤ 5 complex, as reported previously (reference 22 and data not shown). We analyzed the stoichiometry of the reconstituted ␣ 3 -␤ 6 carboxylase by gel filtration chromatography on Sepharose CL-6B (a 1,100-mm by 5-mm column equilibrated with 50 mM potassium phosphate buffer, pH 7.0, and 50 mM NaCl). The reconstituted ␣ 3 -␤ 6 complex eluted in a peak with the apparent molecular mass of a dodecamer comprising six molecules each of the ␣ 3 and ␤ 6 subunits (data not shown).

DISCUSSION
Mycolic acid biosynthesis in M. tuberculosis involves FAS I, which carries out the de novo synthesis of C 16 -C 26 fatty acids (16), and the FAS II multisubunit complex, which is incapable of de novo fatty-acid biosynthesis but elongates C 14 and C 16 primers to produce meromycolates (28). The FAS II complex in M. tuberculosis is located in a genetic locus that also contains a lone, uncharacterized acyl-CoA carboxylase carboxyltransferase subunit, accD6 (␤ 6 ) (6), which could provide malonyl-CoA to the ␤-ketoacyl-ACP synthases and to FAS I. Malonyl-CoA is converted to malonyl-ACP by the malonyl-CoA-ACP transacylase (FabD) (19) and is utilized by the ␤-ketoacyl-ACP synthases (KasA/KasB) (5,18,27) in successive reactions with the other enzymes of the FAS II complex to produce the meromycolic acids. Protein-protein interaction analyses have shown that FabD is a structural component of the FAS II complex (30) and that KasA, KasB, InhA, MabA (FabG1), and FabH (␤-ketoacyl-ACP synthase III) interact with each other (31).
The reconstituted ␣ 3 -␤ 6 acyl-CoA carboxylase, reported here, preferred acetyl-CoA over propionyl-CoA, in contrast to the ␣ 3 -␤ 5 complex, which preferred propionyl-CoA over acetyl-CoA (10,22). This observation leads us to suggest that M. tuberculosis may utilize the ␤ 6 subunit with ␣ 3 to provide malonyl-CoA to FAS I and to the FAS II complex for de novo fatty-acid biosynthesis and mycolic acid biosynthesis, respectively, and the ␤ 5 subunit with ␣ 3 for providing methylmalonyl-CoA for branched-fatty-acid biosynthesis. The rate of acetyl-CoA carboxylation by the ␣ 3 -␤ 5 complex increased when the ε subunit was bound to it (10,22), and thus, the ␣ 3 -␤ 5 -ε complex could also provide malonyl-CoA for fatty-acid biosynthesis in mycobacteria. The ␣ 3 subunit also interacts with the ␤ 4 (AccD4) subunit, which is possibly involved in the carboxylation of fatty acids that may then be incorporated into mycolic acids (11,22,24).
Isoniazid was shown to inhibit the biosynthesis of mycolic acids (29), and the reductases involved in fatty-acid elongation isolated from Mycobacterium avium were shown to be selectively inhibited by the drug (17). More recent work has shown that the reductases are the targets of isoniazid inhibition in M. tuberculosis (1,25). In isoniazid-treated M. tuberculosis cul-   The ␣ 3 -␤ 6 complex was preincubated with 1 unit avidin (binds 1 g D-biotin) or 100 M inhibitor and used in the assay. Percent inhibition above the respective solvent control is given as mean Ϯ standard deviation from three independent assays of the carboxylase with inhibitor.

VOL. 189, 2007
MYCOBACTERIAL ␣ 3 -␤ 6 IS A FUNCTIONAL ACYL-CoA CARBOXYLASE losis ␤ 6 subunit (6, 7, 13). Since ␤ 6 is predicted to be an essential gene (26), it may not be possible to disrupt the gene to show a direct biochemical consequence. Our current and previous observations showed that the ␣ 3 , ␤ 4 , ␤ 5 , and ␤ 6 ORFs are highly expressed in M. tuberculosis and that their products have catalytic activities (22). Interestingly, M. leprae, which has lost a significant portion of its genome, has retained only ␣ 3 , ␤ 4 , ␤ 5 , and ␤ 6 as functional ORFs among the ACC-encoding ORFs in its genome (7). Our microarray results indicated that the transcripts for the ␣ 3 , ␤ 4 , ␤ 5 , ␤ 6 , and ε subunits were repressed as mycobacterial cells entered late log phase, when they possibly begin to encounter unfavorable growth conditions, like nutrient starvation, suggesting that these carboxylase subunits may be less important for the mycobacterium during the late-log and stationary growth phases. The inhibition of the carboxylase activities of the ␣ 3 -␤ 6 complex when reconstituted with the ε subunit indicated that the ε subunit hindered interaction between the ␣ 3 and ␤ 6 subunits. Analysis of the stoichiometry of the ␣ 3 -␤ 6 complex indicated that it is reconstituted as a dodecamer containing six molecules each of the ␣ 3 and ␤ 6 subunits, similar to the ␣ 3 -␤ 5 complex, which was reported to exist as a dodecamer (10).
Inhibitors that were previously shown to target a particular enzyme have subsequently been demonstrated to inhibit novel targets. C75 is a synthetic inhibitor of fatty acid synthase and is a potential antiobesity and antitumor drug (20). C75-CoA was recently shown to inhibit carnitine palmitoyl transferase I in a novel mode of action (3). In another example, itaconate, which is a potent inhibitor of isocitrate lyase, was shown to inhibit propionyl-CoA carboxylase in cell extracts of Rhodospirillum rubrum (4). Isocitrate lyase is a glyoxylate shunt pathway enzyme that is used by M. tuberculosis to utilize fatty acids as a carbon source (21). Therefore, we investigated the effects of several inhibitors that have been shown to inhibit fatty-acid metabolism but had not been shown to inhibit ACC activity. Interestingly, we found that C75 and dimethyl itaconate severely inhibited the carboxylase activities of ␣ 3 -␤ 6 in our assays (Fig. 6). Further work on the mechanism of action of these inhibitors is needed, and the characterization of such inhibitors, which target multiple biochemical pathways, may lead to the identification of potent drug candidates against M. tuberculosis.
FIG. 6. Inhibition of acyl-CoA carboxylase activities of ␣ 3 -␤ 6 by dimethyl itaconate and C75. Inhibitions by dimethyl itaconate of ACC (■) and PCC (e) activities (A) and by C75 of ACC (OE) and PCC (‚) activities (B) were measured after preincubation of the respective inhibitor at the indicated concentrations with reconstituted ␣ 3 -␤ 6 for 10 min at 24°C prior to the carboxylase assay. Percent inhibition above the solvent (dimethyl sulfoxide) control was plotted as the mean with standard deviation .   FIG. 7. Effect of the epsilon subunit on the ␣ 3 -␤ 6 carboxylase. The purified ␣ 3 and ␤ 6 subunits were preincubated together in a 1:1 molar ratio, and the ␣ 3 , ␤ 6 , and ε subunits were preincubated together in a 1:1:2 molar ratio on ice for 6 h prior to a carboxylation assay with acetyl-CoA or propionyl-CoA. Values represented are means and standard errors.