|Home | About | Journals | Submit | Contact Us | Français|
Mycobacteria have a unique cell wall consisting of mycolic acids, very long chain lipids that provide protection and allow the bacteria to persist within human macrophages. Inhibition of cell wall biosynthesis is fatal for the organism and a starting point for the discovery and development of novel antibiotics. We determined the first crystal-structures of KasA, a key enzyme involved in the biosynthesis of long-chain fatty acids, in its apo-form and bound to the natural product inhibitor thiolactomycin (TLM). Detailed insights into the interaction of the inhibitor with KasA and the identification of a polyethylene glycol molecule which mimics a fatty acid substrate of approximately 40 carbon atoms length, represent the first atomic view of a mycobacterial enzyme involved in the synthesis of long chain fatty acids and provide a robust platform for the development of novel TLM analogs with high affinity for KasA.
Tuberculosis (TB), an infectious disease caused by Mycobacterium tuberculosis, remains one of the leading causes of death in the world. It is estimated that tuberculosis causes 8.8 million new infections and 1.6 million deaths each year (WHO, 2007). The rapid emergence of new strains, which are resistant against most of the known antibiotics, has increased the complexity of TB treatment. Patients infected with multi-drug resistant strains of M. tuberculosis (MDR-TB) require a longer, more costly therapeutic regime (Dye et al., 2002). In addition, the recent appearance of strains that are resistant to both first and second line antibiotics (extensively drug resistant TB, XDR-TB), represents a severe threat since these strains are virtually untreatable (Jain and Mondal, 2008). Consequently, it is important to identify new drug targets and develop new chemotherapeutics that circumvent existing drug resistance mechanisms.
The mycobacterial cell wall is essential for the pathogen's survival. It is lipid-rich, highly impermeable and thereby provides protection from many antibiotics and allows the bacteria to persist and to proliferate in macrophages (Daffe and Draper, 1998; Ying Yuan, 1998). Mycolic acids which are long chain α-alkyl-β-hydroxy fatty acids constitute up to 60 % of the cell wall and are mainly responsible for the low permeability of the waxy cell envelope (Asselineau and Lederer, 1950; Barry et al., 1998). Unlike other bacteria, mycobacteria require two distinct fatty acid synthesis pathways to generate these long chain fatty acids, the mammalian-like FAS-I and the bacteria-like FAS-II pathway (Kremer et al., 2000). The large multifunctional polypeptide complex within the FAS-I pathway is capable of de novo fatty acid synthesis and generates fatty acids with a chain length of C14-16 (Smith et al., 2003). These short fatty acids are transferred to the FAS-II system, where they are elongated to fatty acids up to 56 carbons in length and serve as precursors for mycolic acids (Kremer et al., 2002; Lu et al., 2004). In this pathway, an acyl carrier protein (AcpM) shuttles the growing acyl chain between discrete monofunctional enzymes that catalyze the individual steps (Figure 1, (Campbell and Cronan, 2001)). Growing evidence points towards a direct interaction of FAS-II enzymes with each other in interconnected specialized complexes that are essential for mycobacterial survival (Kremer et al., 2003; Veyron-Churlet et al., 2005; Veyron-Churlet et al., 2004). The molecular basis for these interactions, however, remain sketchy and, moreover the capability of mycobacterial enzymes to interact with and efficiently transfer the extremely long hydrophobic fatty acids from one protein to the next within a cytosolic environment is not understood.
KasA, the mycobacterial β-ketoacyl ACP synthase I is an important key enzyme within the FAS-II system. It catalyses the condensation between malonyl-AcpM and the growing acyl chain via a ping pong mechanism, the first of four steps in the fatty acid elongation cycle (Figure 1). In a first step, the acyl chain is transferred to the active site cysteine resulting in an acylated KasA intermediate. Subsequently, the acyl chain is elongated by two carbon atoms derived from the second substrate malonyl-AcpM in a condensation reaction with the KasA intermediate (Kremer et al., 2002). KasA has been shown to be essential in mycobacteria: conditional depletion of KasA induces cell lysis (Bhatt et al., 2005) and transposon site hybridization has demonstrated that KasA is essential for cell growth (Sassetti et al., 2003).
Inhibitors of FAS-II enzymes, with the first line antibiotic isoniazid that targets InhA as the most prominent example, impair the integrity of the cell wall and thereby act as bactericidal agents (Slayden et al., 1996). The natural product inhibitor thiolactomycin (TLM) is a promising lead compound for the development of potent FAS-II inhibitors. TLM has favorable physicochemical properties, is effective in mouse infection models and it has been shown to inhibit the mycobacterial β-ketoacyl synthases KasA and KasB, with KasA being the most sensitive (Kremer et al., 2000; Schaeffer et al., 2001). Recent kinetic studies revealed that TLM binds to both, the free enzyme and the acylated form of KasA (Machutta et al., unpublished data). Furthermore, it preferentially binds to the acyl-enzyme intermediate and shows a slow binding step during the inhibition reaction, which plays a crucial role for the in vivo activity of the compound.
Here, we describe the first structures of Mycobacterium tuberculosis KasA. Active KasA was obtained through overexpression in Mycobacterium smegmatis as an expression host. Since the inhibitor can bind to the apo form as well as the acylated form of the enzyme we solved binary complexes of TLM bound to KasA using the acyl-enzyme mimic C171Q KasA and the wild-type enzyme. The structures provide detailed insight into the interaction of TLM with KasA and facilitate a molecular explanation for the preferential binding to the acylated form of KasA. The presence of bound polyethylene glycol (PEG) molecules delineate for the first time how a long chain fatty acid is accommodated in the acyl binding channel and suggest how the substrate is efficiently transferred from one protein to the next within the FAS-II system.
In order to evaluate the interactions of TLM with KasA as a platform for rational inhibitor design and to explore the molecular basis for the preferential binding of TLM to acyl-KasA, we determined the X-ray structures of wild-type KasA and the acyl enzyme mimic C171Q, both unliganded (apo) and with bound TLM at resolutions ranging between 1.8 and 2.2 Å (Table 1). Wild-type KasA crystallized in space group P3121 with one monomer in the asymmetric unit while the C171Q variant crystallized in space group P31 with eight monomers in the asymmetric unit. Superposition of the respective monomers within the functional dimers resulted in an rmsd value of 0.30 Å for the KasA C171Q TLM bound structure and 0.26 Å for the KasA C171Q apo structure, indicating that there are no significant differences between the monomers within the functional dimer. The overall structure of KasA is shown in Figure 2. The homodimeric assembly of the protein in the crystal structures is consistent with the knowledge that KasA forms a homodimer in solution. Each monomer is comprised of a core domain and a cap. The core domain can be divided into two halves with similar topology, a mixed five-stranded β-sheet covered on each face by α-helices. Residues 2-259 form the N-terminal half and residues 260-416 the C-terminal half of the protein (Figure 2A). The two core domains form a five-layered αβαβα structure, characteristic of the thiolase superfamily and can be compared to the structures of other β-ketoacyl ACP synthases (Heath and Rock, 2002; Mathieu et al., 1994; Olsen et al., 1999; Price et al., 2001). Residues of the catalytic triad (Cys171, His311 and His345) are located in the core domain. While the catalytically essential cysteine that becomes covalently modified during the reaction is located in the N-terminal domain and lies at the N-terminus of an α-helix, all other catalytic residues are contained within the C-terminal core domain. The more flexible cap consists of helices α2, α5 and α9 which together with α5′ of the second monomer form the acyl-binding channel (vide infra).
Despite the fact that the molar protein to TLM ratio for wild-type KasA complex formation was 1:200, KasA is not fully occupied by TLM, consistent with the biochemical observation that TLM binds only weakly to the free enzyme. Nevertheless the conformation of the TLM molecule is clearly defined and allows an unambiguous assignment of its protein interactions (see Supplementary Figures). TLM binds to the malonyl binding pocket of KasA. The methyl groups C-9 and C-10 are positioned in two hydrophobic pockets formed by Pro280, Gly318 and Phe402, Phe237, respectively, and the intercalation of the isoprenoid tail into the space between two peptide bonds, namely Ala279 - Pro280 and Gly403 - Phe404, further stabilizes the interaction. The isoprenoid moiety of TLM points towards an extended lipophilic pocket where two important water molecules are present in all structures and stabilize the loop from Asp273 to Pro280 (Figure 3A). Two strong hydrogen bonds are formed between the O-1 oxygen and the nitrogens of the active site histidines, His311 and His345 (Figure 3B). Upon binding of TLM to the active site, Phe404 and the associated loop comprising residues Phe402-Gly406 are shifted by 0.9 Å to avoid clashes with the inhibitor. Additionally, the main chain oxygens of Phe402 and Val278 are rotated away from the TLM molecule. Binding of TLM to the active site in M. tuberculosis KasA can be compared to the related E. coli β-ketoacyl ACP synthase I (FabB, 33 % sequence identity to KasA) in the presence of TLM (PDB code 1fj4 (Price et al., 2001); Figure 3B). Unexpectedly, the isoprenoid tail of TLM and, most strikingly, the active site residue Phe404 adopt different conformations. Phe404 acts as a gatekeeper to the adjacent acyl channel and assumes a closed conformation in both the free and TLM-bound wild-type KasA structures. This feature has not been observed in the TLM bound and apo structures of ecFabB, where the corresponding residue Phe392 is always observed in an open conformation leading to an almost perpendicular orientation of its aromatic ring relative to the TLM thiolactone ring. In contrast, the aromatic ring of Phe404 in KasA is oriented parallel to the thiolactone ring and TLM binding does not appear to alter the position of Phe404 in KasA which remains in the ‘closed’ conformation in both wild-type structures.
Based on the report that the C163Q mutant of E. coli FabF (β-ketoacyl ACP synthase II, 40 % sequence identity to KasA) mimicked the structural change accompanying acylation (Wang et al., 2006), we used the C171Q mutant for structural studies of the intermediate state of KasA. For the C171Q KasA TLM complex, co-crystallization with a protein to inhibitor ratio of 1:10 led to a structure that was fully occupied with TLM. TLM forms the same hydrophobic interactions to those described for the wild-type KasA TLM complex and, likewise, the hydrogen bonds with the active site histidines are unchanged (Figure 3C). One fundamental difference is, however, the orientation of the Phe404 side chain in the C171Q KasA structure. Phe404 adopts an open conformation which is also observed in the apo C171Q KasA structure. The side chain oxygen of the mutated Gln171 generates a hydrogen bond to the amide nitrogen of Phe404, thereby shifting the phenylalanine side chain by 60° and by 3.3 Å out of the active site relative to the position of this residue in the KasA wild-type structure. In this open conformation the Phe404 aromatic ring forms an energetically favorable edge-to-face interaction with the thiolactone ring of TLM (Hunter et al., 1991). The very flexible loop containing residues Phe402, Gly403, Gly405 and Gly406 which immediately precede and follow Phe404 is shifted out of the binding pocket by about 1.5 Å, while Phe210 and helix α9 are moved away from the cavity by ~2 Å. Thereby th e solvent accessible volume of the active site cavity is increased from 97 Å3 in the wild-type to 150 Å3 in the mutant structure (Dundas et al., 2006). Possibly due to additional space in the cavity of the KasA C171Q mutant, the isoprenoid tail of TLM in the KasA C171Q structure adopts a different conformation to that observed in the binary wild-type KasA TLM structure. Calculations of the pathways leading from the inside of the pocket to the outside solvent with CAVER (Petrek et al., 2006) showed that the entrance to the active site is more restricted in the wild-type enzyme (2 Å radius) than in the mutant (2.5 Å radius), suggesting that the binding site is more accessible for TLM, and presumably also the malonyl group, when the enzyme is acylated (Figure 4). These results support the contention that replacement of the active site cysteine with a glutamine mimics structural changes caused by acyl-enzyme formation. Similar observations are reported for the E. coli FabF enzyme where Phe400 adopts a closed conformation in the apo enzyme but an open conformation in the acyl-enzyme and the C163Q mutant structures (PDB codes 2gfw, 2gfv, 2gfy and 2gfx (Wang et al., 2006)). In FabF as well as in KasA this structural rearrangement has great impact on the binding efficiency of inhibitors. Binding of TLM to C171Q KasA perturbs the protein structure only very slightly and just the main chain oxygen of Val278 undergoes a significant movement upon TLM binding to avoid steric clashes with the inhibitor. These structural differences suggest that the binding pocket is already preformed to accept the TLM molecule, providing a molecular explanation for the differential binding behavior of TLM to KasA as observed by Machutta et al. (unpublished data). In this work, direct binding experiments also revealed that addition of TLM to the acyl-enzyme form of KasA resulted in a much slower decrease in the fluorescence signal than the instantaneous change observed for the free enzyme, which suggested that there is a slow-onset component to the interaction of TLM with acyl-KasA. However, no structural evidence could be found to explain this phenomenon.
The acyl channel in KasA is clearly defined in two of our structures due to the presence of a bound PEG molecule with a distance of approximately 5.7 Å from the active site glutamine 171 and mimics an acyl chain of about 40 carbon atoms in length (Figure (Figure2B2B and and5A).5A). These structures thus provide for the first time a view into the hydrophobic acyl binding cavity, which is lined with numerous hydrophobic amino acids and perfectly accommodates the growing fatty acid chain (Figure 5B). A superposition with the E. coli FabB structure in complex with a short fatty acid (PDB code 1f91) clearly displays the difference between proteins capable of binding short versus long fatty acid chains. Initially the two fatty acids superimpose but then the PEG molecule extends through the channel to the surface of the protein, bends back into the channel and terminates in a hydrophobic pocket (Figure 5). In the E. coli structure, side-chain and backbone atoms from a helix comprising residues Trp195-Gly205 significantly truncate the acyl binding channel. This helix is reoriented in KasA and thereby permits binding of the longer acyl chain substrates. Additionally, long polar side chains point into the channel in E. coli FabB, namely Glu200 and Gln113, as well as Pro110, thereby blocking further access. In KasA short hydrophobic side chains Ala209, Ile122 and Ala119 facilitate fatty acid binding (Figure 5C).
The acyl channel of KasA is directly accessible through two openings, the malonyl binding pocket and the opening of the acyl channel at the surface of the protein. However, migration of such a long fatty acid chain through the malonyl binding channel past the hydrophilic and charged active site residues appears to be energetically and sterically unfavorable. Equally unfavorable is the migration of the pantetheine group through the hydrophobic environment of the long acyl channel to the active site cysteine. Both entries would afford a threading mechanism of a long fatty acid which would be too time consuming for efficient transfer between the different enzymes involved in the individual steps of fatty acid synthesis. A separate mechanism of substrate binding and product release therefore seems plausible. Comparisons of the wild-type KasA structures with the C171Q KasA structures indicate that a polypeptide segment in KasA containing helices α5 and α6 (residues 115-147) shows increased temperature factors of about 60 Å2 relative to the average of the whole protein (40 Å2) in the wild-type apo KasA structure. The temperature factors in the respective residues of the mutant structures remain at the same level as the average. These findings suggest that this loop remains very flexible in the absence of a bound acyl substrate and becomes ordered upon binding of the fatty acid to KasA. It is conceivable that residues 115-147 move in a concerted way in a scissor like motion in the dimer, thereby providing direct access to the long acyl channel (Figure 6). Most likely the opening of the channel is further guided by the interaction with the acyl carrier protein. Our findings are further supported by Sachdeva et al., who characterized the binding mode of fatty acids to M. tuberculosis FabH (β-ketoacyl ACP synthase III), and proposed an open conformational state of FabH where the acyl channel is accessible for acyl substrates or inhibitors (Sachdeva et al., 2008).
The β-ketoacyl synthase KasA in the Mycobacterium tuberculosis fatty acid biosynthesis pathway is a validated but unexploited target for the development of novel TB chemotherapeutics. In the present work we have characterized the interaction of KasA with TLM. Our studies reveal an intricate binding mode of the inhibitor and direct evidence why the inhibitor is more readily bound by the acylated rather than the apo enzyme. The presence of bound PEG molecules provide for the first time insights how a long-chain fatty acid is accommodated in a FAS-II enzyme and rationalizes how it could be efficiently transferred from one protein to the next in this pathway to yield the precursors for the essential waxy mycobacterial cell wall.
Comparison of the four crystal structures shows that the main structural change upon KasA acylation involves a rotation of the gatekeeper residue Phe404 from a closed to an open conformation. This observation is based on the presumption that replacement of the catalytic cysteine with a glutamine causes similar structural changes to KasA that occur upon acylation. These structural changes appear to be driven by the formation of hydrogen bonds between the oxyanion hole and the carbonyl groups of either the glutamine side chain or the acyl-enzyme thioester. Formation of these hydrogen bonds causes Phe404 to rotate, resulting in an increase in the size of the malonyl binding pocket and a widening of the entrance to the pocket which facilitates TLM binding. These rearrangements in KasA combined with the energetically favorable edge-to-face interaction between Phe404 and TLM provide an explanation for the increase in affinity of the inhibitor for the acylated enzyme.
Careful comparison of the C171Q KasA structures in the apo and inhibitor-bound state does not reveal any major change in structure that could account for the slow-onset component to the inhibition of the acylated enzyme by TLM. While slow-onset inhibition has been observed in a number of systems, structural explanations for this kinetic phenomenon have only been elucidated in a few cases and are generally presumed to involve the formation of a final enzyme-inhibitor complex (EI*) via a conformational change following formation of the initial EI complex. In the FAS-II pathway, slow-onset inhibition of the enoyl-ACP reductase enzyme (FabI, InhA in M. tuberculosis) by inhibitors such as triclosan leads to ordering of a loop of amino acids close to the active site (Sivaraman et al., 2003; Stewart et al., 1999). However, no such structural change is observed for KasA. While conversion of EI to EI* in KasA could involve numerous subtle changes in structure, it is also possible that the structure of the TLM bound C171Q KasA corresponds to the initial EI complex. Further studies are in progress to elucidate the structural basis for slow-onset inhibition, and to develop additional TLM analogs that retain this mode of inhibition while also retaining the ‘drug-like’ properties of the parent molecule.
Previous studies (Kim et al., 2006) aimed towards the modification of TLM at the 5 position due to the observation that the isoprene side chain of TLM is pointing towards a lipophilic pocket, which could be filled and thereby increase the affinity of the lead compound. Although this pocket is slightly larger in KasA compared to homologous enzymes of E. coli, our structures clearly reveal, that this pocket can not accommodate a longer hydrophobic chain, since two important water molecules are present in all structures and stabilize the loop from Asp273 to Pro280. It was also attempted to remove the double bonds in the isoprene side chain, which again yielded no improvement of the lead compound. In the KasA structures the isoprene side chain is sandwiched between two peptide bonds. This arrangement only allows the position of a planar group whereas non-planar groups would sterically interfere with either of the two walls formed by residues 279-280 and 403-404.
On the basis of our crystal structures, we suggest a novel approach towards improvement of the lead compound TLM. The bound PEG molecules in our crystal structures directly point towards the possibility to combine the TLM molecule with a polyethylene glycol molecule of defined length. This so called PEGylation is a common process to modify drugs or therapeutic molecules, mostly, to improve solubility and decreases immunogenicity. Additionally, PEGylation can increase the stability of the drug and the retention time of the conjugates in blood, thereby allowing a reduced dosing frequency and reduced cost (Veronese and Pasut, 2005). In case of drug development against tuberculosis, where current treatment can take longer than 12 months, this improvement is very desirable. A PEG chain with a length of 40-50 atoms attached to TLM could also have the advantage of achieving increased specificity, targeting only the mycobacterial FAS-II enzymes without affecting the human system that is not able to accommodate such long substrates. Further studies are in progress to prove if a combined TLM-PEG molecule is able to bind to KasA and trigger the conformational change of the gatekeeper residue Phe404 to induce the favourable interaction between the TLM thiolactone ring and the phenylalanine side chain.
The binding of long PEG chains in the acyl channels of the KasA structures allowed us to unambiguously assign the mode of substrate binding and the residues important for fatty acid binding. The FAS-II pathway in M. tuberculosis synthesizes fatty acids with a length of up to C56, in contrast to the much shorter lipids produced by the FAS-II systems in organisms such as E. coli. The visualization of a PEG molecule in the acyl channel, which is truncated in the homologous protein from E. coli, provides clear insight into how KasA in M. tuberculosis accommodates long acyl chains, which can contain up to 56 carbon atoms and thereby efficiently build the mycolic acids required for its cell wall. Previous studies (Sachdeva et al., 2008) on the mycobacterial enzyme FabH presented the first clues on how a long acyl chain enters the channel. Our structural analyses reveal that KasA could adopt a related mechanism. Two helices, α5 and α6, comprising residues 115-147 are very flexible in the free enzyme and could move in a scissor-like fashion in the two monomers to accommodate the long fatty acid substrates in the substrate binding pocket. This movement would ensure a facilitated uptake and subsequent release of the substrate, thus allowing an efficient transfer between the different enzymes in the pathway.
Wild-type KasA and the C171Q mutant were expressed in Mycobacterium smegmatis strain mc2155 using the mycobacterial expression vector pFPCA (Changsen et al., 2003). The plasmid was transformed into M. smegmatis competent cells by electroporation. Colonies selected on 7H10 solid media containing 50 μg/mL kanamycin, 200 μg/mL ampicillin and 10 μg/ml cyclohexamide were cultivated in 7H9 liquid media supplemented with 0.5 % glycerol and antibiotics, grown to an optical density (OD600) of 0.6-0.8 and protein expression was induced with 0.2 % acetamide. After 36 h cells were harvested by centrifugation and frozen at −80 °C. For purification cell pellets were thawed in 20 mM CHES (N-cyclohexyl-2-aminoethanesulfonic acid) pH 9.5, 500 mM NaCl and 5 mM imidazole and lysed by sonication. The lysate was centrifuged and the supernatant applied to a HisTrap HP column (GE Healthcare). A gradient of 40 mM – 1 M imidazole was used for elution. The peak fractions were pooled and diluted to reduce the NaCl concentration to 50 mM. The protein solution was applied to a MonoQ 10/100 column (GE Healthcare) and eluted with a gradient of 50 mM – 1 M NaCl. The fractions containing the pure protein were applied to a Superdex 26/60 200 size exclusion column (GE Healthcare) that was equilibrated with 20 mM CHES pH 9.5 and 500 mM NaCl. Wild-type KasA and the C171Q KasA mutant were concentrated to final concentrations of 4.3 mg/mL and 8.5 mg/mL, respectively, prior to crystallization.
KasA was crystallized by vapor diffusion, mixing equal volumes of protein and precipitant solution. For wild-type KasA, the precipitant contained 10 % isopropanol, 0.2 M NaCl, 0.1 M Hepes pH 7.5 and 10 mM TCEP (tris(2-carboxyethyl) phosphine hydrochloride). The precipitant solution of the C171Q KasA crystals contained 20 % PEG (polyethylene glycol) 3350 and 0.2 M potassium formate. Crystals appeared after two days. For the formation of the KasA inhibitor complexes, thiolactomycin (TLM, Sigma) was dissolved in isopropanol and added to the concentrated protein solution in a molar ratio of 10:1 for the C171Q KasA variant and 200:1 for wild-type KasA, prior to crystallization. Before cryo-cooling in liquid nitrogen, the crystals were briefly transferred into a cryoprotectant solution that contained the respective precipitant and 30 % glycerol for the wild-type KasA crystals and 25 % ethylene glycol for the C171Q KasA crystals. Diffraction data of wild-type KasA and the C171Q mutant were collected on a Rigaku MicroMax™-007HF generator with an Raxis HTC Detector. Datasets were processed using D*Trek implemented in the CrystalClear software (Rigaku). The data set of the binary C171Q KasA TLM complex was collected at the protein structure factory beamline BL14.2 at BESSYII, Berlin, and the data set of the binary wild-type KasA TLM complex was collected at beam line ID14 of the European Synchrotron Radiation Facility, Grenoble, both were indexed and integrated using Mosflm (Leslie, 1992) and scaled with Scala (CCP4, 1994). All structures were solved by molecular replacement using Phaser (McCoy et al., 2007). As a search model for the wild-type KasA apo structure, one monomer of the M. tuberculosis KasB structure (PDB code 2GP6 (Sridharan et al., 2007)) was used. For all subsequent structures the coordinates of the refined wild-type KasA apo structure were used. Data collection statistics are given in Table 1.
Model building and refinement of the wild-type KasA structures was carried out using alternating rounds of COOT (Emsley and Cowtan, 2004) for manual model building and REFMAC (Murshudov et al., 1997) for maximum likelihood refinement. The TLM molecule and water molecules were built into the electron density using COOT. Ramachandran statistics calculated by MolProbity (Davis et al., 2007) showed that 97.4 % of all residues are in the favored region, 99.8 % in the allowed region and one residue is an outlier in the wild-type KasA structure. In the wild-type KasA TLM bound structure, 97.9 % of the residues are in the favored region and 100 % are in the allowed region. Due to problems during refinement of the C171Q KasA structures with REFMAC, a twinning test with Phenix.xtriage (Zwart et al., 2008) was performed that indicated that the C171Q KasA TLM dataset was merohedrally twinned at 29 % and the C171Q apo KasA dataset at 45 %. Further refinement was carried out using Phenix.refine (Adams et al., 2004), taking into account the twin law and refining the twin fraction. Upon twin refinement, the density improved and the Rfree decreased which permitted further model building in COOT and allowed the unambiguous modeling of TLM, water molecules and PEG molecules into the electron density maps. To avoid bias of the Rfree due to the presence of 8-fold NCS and twinning, five percent of all reflections that were omitted during refinement to calculate the Rfree were selected in 20 thin resolution shells. In the C171Q KasA structure 95.7 % of the residues are in the favored region, 99.6 % in the allowed region and 13 residues are outliers. The C171Q KasA TLM bound structures has 96.7 % of the residues in the favored region, 99.6 % in the allowed region and 12 outliers. Refinement statistics are given in Table 1. All Figures were prepared using PyMOL (DeLano, 2002).
This work was supported in part by grant AI44639 from the National Institutes of Health P.J.T. and through the Deutsche Forschungsgemeinschaft (SFB 630 and Forschungszentrum FZ82). We thank the staff of BL 14.2 at BESSYII, Berlin and ID14 of the ESRF, Grenoble for technical support.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
†This work was supported in part by NIH grants AI44639, AI70383 and through the Deutsche Forschungsgemeinschaft (SFB630 and Forschungszentrum FZ82).
Accession codes Coordinates for the crystal structures of Mycobacterium tuberculosis KasA, KasA-TLM, C171Q KasA and C171Q KasA-TLM have been deposited with accession codes 2wgd, 2wge, 2wgf, 2wgg, respectively.