Purification and Biochemical Characterization of the Mycobacterium tuberculosis (cid:1) -Ketoacyl-acyl Carrier Protein Synthases KasA and KasB*

Mycolic acids are vital components of the Mycobacterium tuberculosis cell wall, and enzymes involved in their formation represent attractive targets for the discovery of novel anti-tuberculosis agents. Biosynthesis of the fatty acyl chains of mycolic acids involves two fatty acid synthetic systems, the multifunctional polypeptide fatty acid synthase I (FASI), which performs de novo fatty acid synthesis, and the dissociated FASII system, which consists of monofunctional enzymes, and acyl carrier protein (ACP) and elongates FASI products to long chain mycolic acid precursors. In this study, we present the initial characterization of purified KasA and KasB, two (cid:1) -ketoacyl-ACP synthase (KAS) enzymes of the M. tuberculosis FASII system. KasA and KasB were expressed in E. coli and purified by affinity chromatography. Both enzymes showed activity typical of bacterial KASs, condensing an acyl-ACP with malonyl-ACP. Consistent with the proposed role of FASII in mycolic acid synthesis, analysis of various acyl-ACP substrates indicated KasA and KasB had higher specificity for long chain acyl-ACPs containing at least 16 carbons. Activity of KasA and KasB increased with use of M. tuberculosis AcpM, suggesting that structural differences between AcpM and E.

The cell wall of Mycobacterium tuberculosis is a complex structure containing many components that contribute to the communication between the bacterial cell and its host (1) as well as its structural integrity and characteristic impermeability (2)(3)(4). In fact, the innate impermeability displayed by M. tuberculosis to many common broad spectrum antibiotics and other small hydrophilic molecules can be directly attributed to the hydrophobic nature of the cell envelope (4). The unique core structure of the cell wall consists of a covalently linked complex of peptidoglycan, arabinogalactan, and mycolic acids (5)(6)(7). The latter, which are high molecular weight, ␣-alkyl, ␤-hydroxy fatty acids, are the largest fatty acids in nature, ranging from 70 to 90 carbons (for reviews, see Refs. 5 and 8). In addition to their characteristically long acyl chains, M. tuberculosis mycolic acids also contain a variety of functionalities, including desaturations and cyclopropyl rings, ␣-methyl-branched methyl ethers, and ␣-methyl-branched ketones which define the ␣-, methoxy-, and ketomycolates, respectively (8,9). The low permeability of the cell can be directly attributed to the nature of functional groups in mycolic acids and their effect on the fluidity of the cell wall (2,5,10). Variations in functional group structure can also have effects on the ability of M. tuberculosis to grow in macrophages (11). In addition to the influence of mycolic acids on cell wall structure and function, cell wall components such as trehalose dimycolate (cord factor) have been shown to exhibit immunomodulatory activities that can enhance the pathogenicity of M. tuberculosis (12).
The importance of mycolic acids to bacterial survival and pathogenesis has generated much interest in the enzymes responsible for their biosynthesis, based on the hypothesis that inhibitors of these proteins will be potential antimycobacterial agents. Indeed, the cessation of mycolic acid synthesis is one of the primary effects of isoniazid (INH), 1 a front-line anti-tuberculosis drug (13). Various other compounds including ethionamide (14,15), isoxyl (15,16), and thiolactomycin (TLM) (17), have also been shown to inhibit mycolic acid synthesis.
Biosynthesis of mycolic acid precursors requires the interaction of two fatty acid synthase (FAS) systems, the multifunctional polypeptide, FASI, and the dissociated FASII system, the latter composed of monofunctional enzymes and a discrete acyl carrier protein (ACP) 2 (8) (Fig. 1). Although the specific details of mycolic acid synthesis are not completely understood, the mycobacterial FASI system appears to be responsible for the de novo synthesis of C 16 -26 fatty acyl primers (18), which are then passed to the FASII system and elongated to produce intermediates of the long meromycolate chain (19). Such intermediates can be modified and condensed with ␣-branch fatty acids to form mature mycolic acids (8). The availability of the M. tuberculosis genome sequence has allowed the identification of putative genes encoding proteins homologous to other bac- 1 The abbreviations used are: INH, isoniazid; ACP, acyl carrier protein; KAS, ␤-ketoacyl-ACP synthase; TLM, thiolactomycin; FAS, fatty acid synthase; IPTG, isopropyl ␤-D-thiogalactopyranoside; DTT, dithiothreitol; MOPS, 3-[N-morpholino]propanesulfonic acid; MES, 2-[Nmorpholino]ethanesulfonic acid; CER, cerulenin; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid. 2 The abbreviation ACP is typically used to describe acyl carrier protein. When the source of ACP is relevant, i.e. from either Escherichia coli or M. tuberculosis, the abbreviations ACP Ec and AcpM are used, respectively, to differentiate the two.
Of particular interest in the FASII system are the roles played by three condensing enzymes, FabH, KasA, and KasB. Recent reports describe the isolation of the FASII initiation enzyme FabH (21,22) and show that this enzyme is inhibited by TLM but not by cerulenin (CER) (21). TLM also inhibits KasA and KasB, both putatively involved in fatty acid elongation, based on overexpression of the kasA and kasB genes in Mycobacterium bovis BCG, which resulted in increased resistance to TLM in vitro and in vivo (25). Distinct roles for KasA and KasB in mycolic acid synthesis were postulated based upon increases in the incorporation of radioactivity into fatty acids in various FAS assays after overexpression of either kasA or kasB (25). The antimycobacterial activity of TLM suggests the essentiality of at least one of the three condensing enzymes, as is the case for orthologs in other bacterial species such as E. coli FabB (26) and Streptococcus pneumoniae FabH (27) and FabF, 3 thus making them attractive targets for the development of screens to identify new lead compounds. Therefore, we purified and characterized KasA and KasB, and despite many similarities, these enzymes display differing enzymatic properties and may indeed play distinct roles in mycolic acid biosynthesis.

EXPERIMENTAL PROCEDURES
Materials and Instrumentation-Chemicals and reagents were purchased from Sigma unless otherwise noted. All cloning steps were performed in E. coli DH5␣ cells purchased from Life Technologies, Inc. E. coli BL21(DE3) cells and expression vector pET28a(ϩ) were purchased from Novagen. M. tuberculosis H37Rv genomic DNA was obtained from Dr. John Belisle at Colorado State University. Q-Sepharose and HiTrap chelating columns were purchased from Amersham Pharmacia Biotech. Materials for polyacrylamide gel electrophoresis were purchased from Novex. An AKTAprime liquid chromatography system (Amersham Pharmacia Biotech) was utilized for protein purification. Mass spectra were obtained using a SELDI protein biology system purchased from Ciphergen Biosystems, Inc. Enzyme assays were performed on a SpectraMax Plus 384 plate reader (Molecular Devices).
Phylogenetic Analysis of KasA and KasB-Representative sequences among selected classes of related fatty acid biosynthetic enzymes were aligned with the MACAW manual sequence alignment program (28 (30)). In each case, such positions were found to be in predicted loop regions that suggested structural inequivalence among the aligned sequences in these regions. Thus, these regions, including the non-homologous domains of the multi-domain fatty acid synthase proteins, were removed from the alignment before phylogenetic analyses. In the final edited alignment, 395 of the 414 amino acids of M. tuberculosis KasA were maintained.
Phylogenetic analyses were based on the total number of sites in edited multiple-sequence alignments. Trees were constructed by neighbor-joining and maximum-parsimony methods. Neighbor-joining trees were built from pairwise distances between amino acid sequences based on the Dayhoff PAM120 substitution matrix (31) using the programs NEIGHBOR and PROTDIST of the PHYLIP 3.572c package (at genetics.washington.edu/phylip.html (32)). The programs SEQBOOT and CONSENSE were used to estimate the confidence limits of branch points from 1000 bootstrap replicates. Maximum parsimony construction was accomplished with the software package PAUP*, version 4.0 (33). The number and length of minimal trees were estimated with 100 replicate random heuristic searches. Confidence limits for branch points were estimated from 1000 bootstrap replicates.
Protein Expression and Purification-The pETkasA and pETkasB plasmids were electroporated into E. coli BL21(DE3) cells, and single colony transformants were grown at 37°C to mid-log phase (A 600 ϭ 0.7) in LB broth containing kanamycin (50 g/ml) and glucose (1%). For KasA, the cells were induced with 0.3 mM IPTG at 37°C for 2.5 h. KasB cultures were equilibrated to 18°C and induced with 0.3 mM IPTG for ϳ24 h at 18°C. Cells were harvested by centrifugation, resuspended in Buffer A (50 mM sodium phosphate, pH 7.5, 500 mM NaCl, 10 mM imidazole, 5 mM ␤-mercaptoethanol (5 ml/g)), and lysed by the addition of 2 mg of lysozyme followed by 3 cycles of freezing and thawing.
The KasA lysate was centrifuged at 20,000 ϫ g for 20 min, the supernatant was discarded, and the pellet was dissolved in 20 ml of Buffer A containing 8 M urea (Buffer AЈ). The urea solution was centrifuged at 27,000 ϫ g, and the supernatant containing denatured KasA was loaded onto a 5-ml nickel-chelating column. The column was washed with 5 column volumes of Buffer AЈ and eluted using a 100-ml linear gradient of 10 -500 mM imidazole in Buffer AЈ. Fractions containing KasA were identified by SDS-PAGE, pooled, and diluted to an approximate concentration of 0.2 mg/ml in Buffer AЈ. KasA was refolded during stepwise dialyses (6 -8-h/step) against 2 liters of 50 mM sodium phosphate, pH 7.5, 0.5 M NaCl, 2 mM DTT containing 4, 2, and 1 M urea and then dialyzed against 4 liters of 50 mM sodium phosphate, pH 7.5, 0.5 M NaCl, 2 mM DTT. Finally, the sample was dialyzed overnight against 2 liters of 50 mM Tris-HCl, pH 9.5, 0.3 M NaCl, 2 mM DTT, and 50% glycerol and filtered through 0.45 M filter. Aliquots of KasA were snap-frozen in liquid nitrogen and stored at Ϫ80°C.
The KasB lysate was centrifuged at 27,000 ϫ g, and the pellet was discarded. The soluble lysate was loaded onto a 5-ml nickel-chelating column, washed with 5 column volumes of Buffer A, and eluted with a 100-ml linear gradient of 10 -500 mM imidazole in Buffer A. Fractions containing KasB were identified by SDS-PAGE, pooled, and dialyzed against 4 liters of 50 mM sodium phosphate, 0.3 M NaCl, 2 mM DTT, 5% glycerol, after which KasB was divided into 100-l aliquots, snap-frozen in liquid nitrogen, and stored at Ϫ80°C.
Enzyme Assays-A continuous assay format was used to monitor KasA and KasB activity by coupling the condensing activity of the KAS enzymes to a ␤-ketoacyl-ACP reductase, either M. tuberculosis MabA or S. pneumoniae FabG (obtained from Joshua West, GlaxoSmithKline). Both MabA and FabG reduce ␤-ketoacyl-ACP intermediates to the corresponding ␤-hydroxyacyl-ACPs, enabling the reaction course to be monitored spectrophotometrically by following the oxidation of NADPH to NADP ϩ at 340 nm. MabA was purified by expressing the mabA gene as an N-terminal hexahistidine-tagged protein in E. coli followed by affinity chromatography of the soluble fraction of the whole-cell lysate. 4 Although various concentrations of enzyme and ACP substrates were utilized in the experiments described, the details of which are included in the figure legends and text, typical KasA assays (100-l total volume) contained 96 -240 nM KasA, 3 M MabA or FabG, 26 -54 M malonyl-ACP Ec (AcpM), 6 -10 M acyl-ACP Ec (AcpM), 50 M NADPH, and 0.01% CHAPS in 50 mM HEPES buffer, pH 6.8. KasB reactions (100 l total volume) contained 460 -920 nM KasB, 3 M MabA or FabG, 26 -54 M malonyl-ACP Ec , 6-10 M acyl-ACP Ec , 50 M NADPH, and 0.01% CHAPS in 50 mM sodium phosphate buffer, pH 7.0. For KasB assays using AcpM, typical conditions were as described above with 140 -185 nM KasB, 6 -10 M acyl-AcpM, and 18 -26 M malonyl-AcpM. Reactions were preincubated for 5 min at 37°C before initiation and allowed to proceed for 20 min at 37°C unless specified in half-area 96-well microtiter plates (Costar). Initial rates were determined by measuring the decrease in A 340 at either 10-or 30-s intervals throughout the course of the reaction and expressed as pmol/min.

Identification and Phylogenetic Analysis of KasA and KasB-
The presence of a bacterial FASII system in M. tuberculosis is well established, and sequencing of the M. tuberculosis genome confirmed the presence of numerous genes postulated to comprise a single FASII system (8,18,20). Located within a cluster of FASII-related genes including fabD and acpM are two genes, kasA and kasB, that have been predicted to encode KAS enzymes (20,25). Also, sequence alignments of KasA and KasB along with E. coli FabF and other related structures suggest that KasA and KasB are related to FabF (25). In the present context, phylogenetic analysis was applied to examine the relationship of KasA and KasB to selected classes of FAS enzymes (Fig. 2). M. tuberculosis and M. leprae KasA sequences share 93% identity and 99% similarity, and their KasB sequences share 92% identity and 98% similarity (Table I). M. tuberculosis KasA and KasB sequences are also closely related to one another, sharing 67% identity and 86% similarity (Table  I). Neighbor-joining and maximum parsimony methods (Fig. 2) showed strong statistical support in terms of bootstrap values and minimal-length trees for placing these enzymes together in a clade distinct from the other classes of fatty acid biosynthetic enzymes. Additionally, these phylogenetic and sequence homology analyses indicate that KasA and KasB are more closely related to FabF and FabB than to any of the multi-subunit fatty acid synthases.
Expression The tree shown was constructed using the neighbor-joining method as implemented by the program NEIGHBOR of the PHYLIP 3.572c package (30). The scale bar represents 0.1 expected amino acid replacements per site as estimated by the program PROTDIST using the Dayhoff PAM120 substitution matrix (29). The same branch order was obtained by maximum parsimony construction with the software package PAUP* (31). Branch points supported in more than 90% of 1000 bootstrap replications for both methods are indicated with asterisks (*). Although additional sequences were used to generate the alignment and trees (see "Experimental Procedures"), for clarity only relationships among sequences from Mycobacterium, human, Saccharomyces, and select bacterial pathogens are shown.
that KasA and KasB are more identical to each other than to homologous bacterial enzymes suggests that either kasA or kasB arose by gene duplication as opposed to horizontal transfer from another species. It was therefore of considerable interest to attempt to biochemically differentiate the two enzymes by comparing their enzymatic characteristics.
Over the course of these studies, numerous approaches to express KasA as a soluble protein in E. coli were attempted, but none were successful. Varying the level and rate of expression by lowering the temperature of induction, changing the amount of glucose in the medium, and titrating IPTG concentrations failed to produce any soluble KasA (data not shown). Several E. coli strains including BL21(DE3), NovaBlue(DE3), C41(DE3), and C43(DE3) (37) were utilized to express KasA under varying conditions, but the product was always insoluble (data not shown). Additionally, expression of KasA in its native form or fused to maltose-binding protein did not improve solubility.
Microscopic analysis of E. coli BL21(DE3)(pETkasA) cells induced with IPTG suggested that inclusion bodies were produced upon induction of KasA (data not shown). Therefore, KasA was purified under denaturing conditions (Fig. 3A) and refolded during the slow removal of the denaturant (8 M urea) by dialysis. Complete removal of urea in the penultimate dialysis step resulted in the formation of a white colloidal precipitate that was determined to contain ϳ50% of the KasA in the sample (data not shown). The choice of buffer at the final dialysis step was critical for maximal solubility of KasA, the best being 50 mM Tris, pH 9.5 buffer. Dialysis into this buffer after removal of urea resulted in the resolubilization of ϳ90% of the KasA (Fig. 3B). Any residual precipitate was removed by filtration. Refolding KasA in pH 9.5 buffer resulted in no colloidal precipitate, but under those conditions, no active enzyme was obtained.
The authenticity of KasA was confirmed by sequencing of the N-terminal 25 amino acids, revealing the sequence GSSHHH-HHHSSGLVPRGSHMASVSQ, which matches that predicted for hexahistidine-tagged KasA with the initiation methionine cleaved off. Total amino acid analysis of refolded KasA indicated no other detectable protein (data not shown). Approximately 20 mg of Ͼ95% pure refolded KasA was obtained from 0.5 liters of culture.  Unlike KasA, ϳ10% of the KasB expressed in E. coli BL21(DE3) cells at 18°C was soluble in Buffer A. However, attempts to improve solubility using many of the techniques described above had no effect (data not shown). KasB was purified under native conditions from the soluble lysate of IPTG-induced cells (Fig. 3C). Combined fractions of KasB were Ͼ95% pure as judged by SDS-PAGE analysis, and the N-terminal sequence, determined to be GSSHHHHHHSSGLVPRG-SHMVGVPPLAGAS, was consistent with hexahistidine-tagged KasB lacking the initiation methionine. As with KasA, total amino acid analysis showed the presence only of KasB (data not shown). 15 mg of KasB was purified from 2 liters of culture.
Effects of Acyl-ACP and Acyl-CoA Substrates on KasA and KasB Specific Activity-The KAS activities of KasA and KasB were confirmed using a continuous assay format coupled to a ␤-ketoacyl-ACP reductase. The mycobacterial FASII system is incapable of de novo fatty acid synthesis but instead elongates acyl primers (C 16 -26 ) to larger mycolic acid intermediates (38). For this reason, purified KasA and KasB activities were initially tested using C 16:0 acyl substrates. Specific activities of KasA and KasB were determined using a variety of buffers ranging from pH 6.5 to 8, and KasA was found to be most active in HEPES pH 6.8 buffer, whereas KasB had highest activity in phosphate pH 7 buffer. Both enzymes displayed ϳ2-3-fold higher specific activities at 37°C compared with 30 or 42°C. Therefore, KasA and KasB were characterized enzymatically using these optimized conditions.
The linearity of the assays was determined by varying the amounts of KasA and KasB in the reaction. The activity of KasA increased linearly with increasing enzyme concentration, indicating that KasA was the limiting reagent in the assay (Fig. 4C). KasB activity also increased in a linear fashion with enzyme concentration (Fig. 4D).
Kinetic Parameters for KasA and KasB Substrates-Kinetic parameters for two long chain acyl-ACP Ec substrates (C 16:0and C 20:0 -ACP Ec ) and malonyl-ACP Ec were determined for KasA and KasB (Table II). The apparent K m values for C 16:0and C 20:0 -ACP Ec were similar for KasA and KasB, ranging from 1.4 to 3.2 M. V max values were also comparable with C 16:0 -ACP Ec , whereas the V max for KasB with C 20:0 -ACP Ec was nearly 2-fold lower than that for KasA, suggesting that C 20:0 -ACP Ec was a relatively inefficient substrate for KasB. KasB assays generally required higher enzyme concentrations to obtain observable rates as compared with KasA, which is reflected in the ϳ3-fold higher k cat value for KasA as compared with that of KasB. The apparent K m and V max values for malonyl-ACP Ec were nearly identical for both; however, the catalytic efficiency (k cat /K m ) for malonyl-ACP Ec was ϳ1 log lower than that for C 16:0 -ACP Ec , presumably a consequence of the higher K m for malonyl-ACP Ec . These results suggest that KasA and KasB exhibit similar binding affinities for each of the substrates, with the exception of KasB and C 20:0 -ACP Ec , but that the rate of turnover (k cat ) is higher for KasA.
KasA and KasB Are Specific for Long Chain Acyl-ACP Substrates-Although the use of C 16:0 -and C 20:0 -ACP Ec for determination of kinetic parameters confirmed that KasA and KasB utilize long chain acyl-ACP derivatives, these experiments did not rule out their ability to utilize short chain acyl-ACPs as primers for elongation. Therefore, several substrates of varying acyl chain lengths were tested under saturating conditions, and specific activities were determined. As shown in Fig. 5A, KasA specific activity increased with acyl chain length. The specific activity with C 12:0 -ACP Ec was ϳ2.4-fold higher than that with C 4:0 -ACP Ec , whereas C 16:0 -and C 20:0 -ACP Ec represented a ϳ2.5-fold increase over that with C 12:0 -ACP Ec . It is clear from these data that KasA has higher activity with long chain acyl-ACP substrates than those with short or medium length acyl groups. In contrast to KasA, KasB activity peaked with C 16:0 -ACP Ec , which had a specific activity ϳ5.7-fold higher than that for C 12:0 -or C 20:0 -ACP Ec (Fig. 5B).
KasA and KasB are more active with acyl-ACP substrates in which the acyl substituent is longer than those readily utilized by other bacterial proteins in the FabF family (36,39), and these experiments give some insight into the substrate specificities of KasA and KasB. In vitro synthesis of defined acyl-ACPs greater than C 20:0 represents a challenge in further determining the substrate specificity of these enzymes due to the inability of E. coli Aas to efficiently utilize fatty acids longer than C 20:0 .
KasA and KasB Have Higher Specific Activities with AcpM than ACP Ec Substrates-Although the initial characterization of KasA and KasB was done using the more readily available acyl-ACP Ec and malonyl-ACP Ec substrates, it is important to analyze the effect of using the physiologically relevant M. tuberculosis AcpM. To address this, C 16:0 -AcpM and malonyl-AcpM were synthesized and evaluated under saturating conditions.
As shown in Fig. 6, the specific activities of both KasA and KasB were higher with AcpM substrates as compared with ACP Ec . The activity of KasA increased nearly 2-fold using AcpM (161 Ϯ 11 pmol/min/g) over that observed for ACP Ec (93 Ϯ 8 pmol/min/g). KasB activity was ϳ3.5-fold greater with AcpM (82 Ϯ 3 pmol/min/g) than with ACP Ec (23 Ϯ 5 pmol/ min/g). Whether there are changes in substrate binding affinity with respect to a ACP source cannot be inferred from these data, but it is clear that the use of AcpM enhances the rate at which substrate is turned over.
Effects of TLM and CER on KasA and KasB-Inhibitors of bacterial FASII-condensing enzymes include TLM (40,41), a natural product inhibitor of FabH, FabF, and FabB, and CER, a fungal epoxide that irreversibly inhibits members of the FabF-FabB class (42,43). To evaluate the sensitivity of KasA and KasB to these inhibitors, the effects of single concentrations (100 M) of TLM and CER were tested with both ACP Ecand AcpM-based substrates. Both TLM and CER inhibited KasA (Fig. 7A) and KasB (Fig. 7B) activity. Inactivation of KasA and KasB by CER was dependent on the length of preincubation time before initiation of the reaction (Fig. 7C). The effects of TLM on KasA and KasB differed; greater than 90% of KasA activity was inhibited compared with only ϳ65% inhibition of KasB. There was no statistically significant difference in the sensitivity of KasA and KasB to TLM and CER with regard to ACP Ec or AcpM, suggesting similar mechanisms of inhibition for each enzyme irrespective of the ACP source.
The divergent effects of TLM on KasA and KasB in the single concentration experiment were verified by obtaining IC 50 values for TLM using C 16:0 -AcpM and malonyl-AcpM (Fig. 8). For KasA, TLM had an IC 50 of 20 M (Fig. 8A) whereas that for KasB was 90 M (Fig. 8B). The 20 M IC 50 for TLM against KasA was slightly less than that reported for M. tuberculosis FabH (24 M) (21). DISCUSSION Bioinformatics provides powerful tools to analyze genes and their encoded proteins to glean clues about function and physiological role. Phylogenetic analyses are particularly useful in visualizing the relatedness of an uncharacterized target sequence to those from other organisms, especially if there are numerous homologous proteins spanning several enzymes classes. For example, the FASII KAS family includes FabB, FabF, and FabH, comprising the KASI, KASII, and KASIII classes of condensing enzymes, respectively. Members of these classes share varying degrees of homology both to each other and to domains of multi-functional FAS and polyketide synthases. Our analyses confirm that KasA and KasB are indeed more closely related to FabF and FabB, thereby placing them into the KASI/II class of FASII proteins. The biochemical activity of KasA and KasB corroborates these data, exemplified by their specificity for acyl-ACP substrates as opposed to acyl-CoAs.  However, KasA and KasB can be clearly differentiated from most KASI/II enzymes by their increased activity with long chain acyl-ACP substrates than with shorter acyl primers. Edwards et al. (36) report that recombinant E. coli FabF and FabB displayed the highest activity using C 6:0 to C 12:0 acyl-ACP Ec substrates with minimal utilization of C 16:0 -ACP Ec . This is in direct contrast to the specificity of KasA and KasB, both of which demonstrated higher activity for acyl-ACPs greater than C 16:0 and little activity with C 4:0 -and C 12:0 -ACP. These findings are consistent with the putative role of KasA and KasB in mycolic acid biosynthesis. Both enzymes are proposed to elongate long chain acyl primers generated by FASI with subsequent conversion to acyl-ACP by FabH. The specific roles of KasA and KasB in fatty acid elongation are still unclear; however, the data suggest that both enzymes do indeed have distinctive characteristics and, therefore, may play unique roles in mycolic acid synthesis. KasA was suggested to be the condensing enzyme responsible for elongation of the initial acyl primers passed to the FASII system from FASI via FabH, and KasB was proposed to be involved later in the pathway (25). Data obtained with the purified enzymes do not contradict this scenario and support the hypothesis that these enzymes make separate contributions to fatty acid elongation. KasA and KasB share similar characteristics with respect to use of long chain acyl-ACPs, effects of temperature on activity, and sensitivity to CER, but there are also significant differences including contrasting enzymatic properties, sensitivity to TLM, activity and stability in various buffers, and solubility when expressed in E. coli.
Since the K m values of C 16:0 -and C 20:0 -ACP Ec suggest similar binding affinities for KasA and KasB, the higher relative specific activity of KasA reflects higher catalytic constants (k cat ) for KasA. Given the prediction that KasB may be important in the later stages of fatty acid elongation (25), it is plausible that the C 16:0 and C 20:0 acyl primers are significantly shorter than those encountered by KasB in vivo, accounting for the lower specific activity exhibited by KasB with those substrates. It is unclear why KasB was able to utilize C 16:0 -ACP Ec much more efficiently than C 20:0 -ACP Ec , and further studies with long chain (ϾC 20 ) acyl-ACP substrates may help to delineate the full range of chain-length specificity for KasA and KasB. Both enzymes demonstrated higher specific activities with AcpMbased substrates than those incorporating ACP Ec , and this increase was more evident for KasB than KasA. Intuitively, it might be expected that the physiologically relevant AcpM is a better substrate, but the molecular reason for this is unclear. There are many differences between AcpM and ACP Ec including size and the substitution of various conserved residues (in AcpM), which have been suggested to play key roles in KAS recognition of ACP Ec (24). It is plausible that some of these differences may affect the binding of ACP Ec to KasA and KasB; however, kinetic studies with AcpM-based substrates will be necessary to determine whether AcpM binds more tightly to the enzyme or if the increase in rate with AcpM is due to greater catalytic efficiency.
The sensitivity of KasA and KasB to TLM was significantly different, the former being more sensitive. Overexpression of M. tuberculosis FabH in M. bovis BCG did not confer TLM resistance (21), suggesting FabH is not the primary TLM target in mycobacteria, and these data correlate with the in vitro IC 50 data. The 4.5-fold difference in TLM IC 50 between KasA and KasB implies that overexpression of KasB in M. tuberculosis should not have any effect on the TLM MIC. However, this evidence runs contrary to a report that overexpression of KasB resulted in TLM resistance (25). It is possible that the physiological activity of KasB with its native substrates, predicted to be much longer than C 16:0 , may be more sensitive to TLM. for 5 min at 37°C in the assay well before adding MabA or FabG. Reactions were initiated by adding a mixture of both C 16:0 -ACP Ec and malonyl-ACP Ec (black bars) or C 16:0 -AcpM and malonyl-AcpM (gray bars). Conditions for each assay were 120 nM KasA, 10 M C 16:0 -ACP Ec (or AcpM), 32 M malonyl-ACP Ec (or AcpM) (A) and 920 nM KasB, 10 M C 16:0 -ACP Ec or 6 M C 16:0 -AcpM, 54 M malonyl-ACP Ec or 18 M malonyl-AcpM (B). The percent inhibition values for TLM and CER are indicated. C, effects of preincubation time on CER inhibition of KasA (ࡗ) and KasB (‚). CER was incubated with enzyme in reaction buffer before initiation with a mixture of C 16:0 -ACP Ec , malonyl-ACP Ec , FabG, NADPH. Reaction conditions were as described for panels A and B.
CER acts as an irreversible inhibitor of KASI/II enzymes by forming a covalent complex with the active site cysteine (44). These enzymes are inactivated by CER in a time-dependent fashion that is governed by the rate of enzyme-inhibitor complex formation under varying assay conditions (28). Like their KASI/II counterparts, KasA and KasB are also sensitive to CER to varying extents depending on the time of preincubation. For example, KasA and KasB activity is reduced ϳ40% with a 5-min preincubation in 100 M CER, but when that time is doubled, the enzymes are effectively inhibited. In the absence of K I and k inact values for KASI/II enzymes, this time dependence makes it difficult to compare CER inhibition data among them, especially without standardized assay conditions. Despite these issues, however, CER remains a useful tool for characterization of KAS enzymes.
Other compounds, which inhibit the growth of mycobacteria, including isoxyl (16), ethionamide (14,15), and the enoyl-ACP reductase inhibitor INH (45), have been shown to disrupt the synthesis of mycolic acids. INH has also been reported to affect multiple components in mycolic acid synthesis including KasA (46,47); however, confirmation of this using purified KasA may be difficult because the active form of INH cannot be reliably generated in vitro. It is clear that inhibition of mycolic acid synthesis plays an important role in the anti-mycobacterial action of these compounds, and there is no doubt that the enzymes involved comprise an attractive pathway for the discovery of novel chemotherapeutic agents. The purification and characterization of KasA and KasB is an important step in the characterization of mycolic acid synthetic pathway components as drug discovery targets. However, the in vitro essentiality of the kasA and kasB genes in M. tuberculosis has not been determined, and accordingly, it is not yet appropriate to say that both of these enzymes are indeed valid targets for antimycobacterial drug discovery. Despite the differences in enzymatic properties of KasA and KasB, both are able to utilize C 16:0 -ACP as a substrate, and it is not known if deletion of one gene will result in compensation of the lost activity by the second gene product. Despite this, the in vitro effects of TLM on KasA and KasB activity in conjunction with the overexpression data of Kremer et al. (25) provide compelling evidence that inhibition of one or both of these enzymes may lead to the discovery of lead compounds that could form the basis of an optimization program designed to deliver novel antimycobacterial drugs.