Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Jan 6.
Published in final edited form as: Structure. 2014 Dec 4;23(1):21–33. doi: 10.1016/j.str.2014.10.010

FadA5 a thiolase from Mycobacterium tuberculosis - a unique steroid-binding pocket reveals the potential for drug development against tuberculosis

Christin M Schaefer 1, Rui Lu 2, Natasha M Nesbitt 2, Johannes Schiebel 3, Nicole S Sampson 2,*, Caroline Kisker 1,*
PMCID: PMC4286435  NIHMSID: NIHMS641784  PMID: 25482540

Summary

With the exception of HIV, tuberculosis (TB) is the leading cause of mortality among infectious diseases. The urgent need to develop new anti-tubercular drugs is apparent due to the increasing number of drug resistant Mycobacterium tuberculosis (Mtb) strains. Proteins involved in cholesterol import and metabolism have recently been discovered as potent targets against TB. FadA5, a thiolase from Mtb, is catalyzing the last step of the β-oxidation reaction of the cholesterol side-chain degradation under release of critical metabolites and was shown to be of importance during the chronic stage of TB infections. To gain structural and mechanistic insight on FadA5 we characterized the enzyme in different stages of the cleavage reaction and with a steroid bound to the binding pocket. Structural comparisons to human thiolases revealed that it should be possible to target FadA5 specifically and the steroid-bound structure provides a solid basis for the development of inhibitors against FadA5.

Introduction

The World Health Organization (WHO) is increasing its surveillance of resistant Mycobacterium tuberculosis (Mtb) strains, the causative agent of tuberculosis (TB). Out of the 8.6 million reported tuberculosis cases in 2012, 5.2% were caused by multidrug- and extensively drug-resistant strains leading to a total of 1.3 million deaths worldwide (WHO, 2013), and the number infected with drug-resistant strains is increasing.

The WHO estimates that one third of the world's population is currently infected with latent TB and at least 5% of these people develop an active form of the disease within their life time. An important focus towards drug development is therefore the capability of the drug to target not only active, but also latent TB (WHO, 2013) for which the Mtb cholesterol metabolism pathway constitutes a potential drug target (Ouellet et al., 2011). It has been shown that cholesterol catabolism plays an important role in tubercular survival in host macrophages and in the mouse model of infection (Chang et al., 2009; McLean et al., 2009; Nesbitt et al., 2010; Pandey and Sassetti, 2008; Yam et al., 2009).

A cluster of genes responsible for cholesterol catabolism and import has been recently identified (Nesbitt et al., 2010; Van der Geize et al., 2007). The mycobacterial cell entry transport system 4 (Mce4), a multi-subunit ATP-binding-cassette-like (ABC-like) transport system, for example, is used for cholesterol import and is required for the chronic phase of TB infections in the mouse model (Miner et al., 2009; Pandey and Sassetti, 2008). The igr (intracellular growth) operon is required for growth of Mtb in vitro using cholesterol as a carbon source, for intracellular growth in macrophages, and for growth in the mouse model of infection (Chang et al., 2007; Chang et al., 2009). In this pathway, acetyl-Coenzyme A (acetyl-CoA) and propionyl-CoA, as well as more complex metabolites (Wipperman et al., 2014), are generated.

Dubnau et al. investigated which genes are preferentially expressed during infection of human macrophages with Mtb. fadA5 was one of the genes they found to be up-regulated (Dubnau et al., 2002). The fadA5 gene is located in the Mtb cholesterol catabolism cluster, and was annotated as encoding a thiolase (Nesbitt et al., 2010; Van der Geize et al., 2007). Recently, a phylogenetic study of thiolases in Mtb and Mycobacterium smegmatis categorized FadA5 as a member of the TFEL (trifunctional enzyme-like thiolases, type-1) class. This class includes the Mtb trifunctional enzyme (mtTFE), which harbors a predicted binding site for a bulky fatty acid tail (Anbazhagan et al., 2014).

Based on the up-regulation studies, the catalytic role of FadA5 and its influence on the virulence of the Mtb pathogen were investigated. In a mouse model of infection, a ΔfadA5 mutant Mtb strain displayed an attenuated disease phenotype with reduced colony-forming units in comparison to the wild-type strain during the chronic phase of infection. Thus fadA5 is important for Mtb survival in vivo, and may be a potential drug target (Nesbitt et al., 2010).

Biochemical studies with FadA5 established that this enzyme is a thiolase, and the side chain degradation of cholesterol was identified as the point of action (Nesbitt et al., 2010) (Figure 1). Thiolases catalyze the degradative cleavage of a β-ketoacyl-CoA to acetyl-CoA and a two-carbon shortened acyl-CoA (Figure 1A) (Gilbert, 1981; Gilbert et al., 1981). The reaction proceeds through nucleophilic addition of an active site cysteine at the β-keto moiety of the β-ketoacyl-CoA and elimination of acetyl-CoA to form an acyl-cysteine intermediate (Figure 1A, state I → II). Upon release of the acetyl-CoA product from the enzyme, CoA is bound (Figure 1A, state II → III). The acyl-cysteine intermediate undergoes a nucleophilic attack by the CoA to regenerate the free cysteine, and to form the two-carbon shortened acyl-CoA (Figure 1A, state III → IV). If the β-ketoacyl-CoA bears a methyl group at the α-carbon, propionyl-CoA is formed instead of acetyl-CoA (Figure 1B). Since the first product acetyl-CoA has to be released prior to binding of the second substrate CoA, catalysis can be described as a ping-pong mechanism.

Figure 1. Proposed thiolase reaction mechanism and role of FadA5 in the side chain degradation of cholesterol.

Figure 1

Open in a new tab

A: The large blue arrows indicate the two reaction directions of the thiolase, the degradation is described from left to right and the biosynthesis from right to left. The roman numerals stand for the enzyme states during thiolysis and the small blue arrows depict the electron transfer during the degradative thiolase reaction; Adapted from (Modis and Wierenga, 2000). B: FadA5 catalyzes the C-C cleavage step of the β-oxidation and therefore the cleavage of keto-thioesters into acetyl- (red) or propionyl- (blue) CoA and the steroid product. Adapted from (Nesbitt et al., 2010).

Although degradation is the thermodynamically favored direction, the carbon-carbon cleavage reaction is reversible. Thiolases are thus also found in biosynthetic pathways, and the most intensive structural analysis to date has been performed on the biosynthetic thiolase from Zoogloea ramigera (Kursula et al., 2002; Modis and Wierenga, 1999, 2000). The conserved active sites of thiolases include a nucleophilic cysteine, a general acid/base cysteine and a histidine (Haapalainen et al., 2006).

Towards further deciphering the role of FadA5 in cholesterol metabolism, we solved the structure of FadA5 and characterized its kinetics with a steroid-CoA substrate. We present the first structures of this enzyme in the apo form as well as an active site variant C93S in complex with its CoA ligand and with a non-covalently bound steroid. Our structural characterization of a bound steroid and Coenzyme A is the first example of a thiolase (like) enzyme crystallized in the presence of a steroid and reveals first insights into steroid-enzyme-interactions, as well as regions of protein rigidity and flexibility that might serve as a starting point for future inhibitor design.

Results

FadA5 cleaves 3,22-dioxo-chol-4-ene-24-oyl-CoA to yield 3-OPC-CoA and AcCoA

In a previous report we explored the steady-state kinetics of FadA5 with acetoacetyl-CoA (AcAc-CoA) and CoA as substrates (Nesbitt et al., 2010). Although FadA5 cleaved AcAc-CoA to yield acetyl-CoA (Ac-CoA), the low catalytic activity (KmA = 464 ± 207 μM, kcatapp = 0.076 ± 0.002 s−1, kcatapp/KmA = 1.64 ± 0.45 ×102 M−1s−1, at 50 μM CoA) strongly suggested that AcAc-CoA is not the physiologically relevant substrate for this enzyme. Metabolite analysis upon disruption of fadA5 in Mtb identified the loss of androstenedione and androstadienedione accumulation in the mutant strain (Nesbitt et al., 2010). The altered metabolic profile therefore led to the hypothesis that FadA5 catalyzes the thiolysis of a keto CoA-ester formed during the β-oxidation of the cholesterol side chain. Based on these results we synthesized the proposed steroid substrate 3,22-dioxo-chol-4-ene-24-oyl-CoA (Figure 1B, compound 2) to probe FadA5's catalytic activity. FadA5 was assayed in the thiolytic direction with 3,22-dioxo-chol-4-ene-24-oyl-CoA and CoA as substrates and the enzyme reaction products were analyzed by MALDI-TOF mass spectrometry. Both 3-oxo-pregn-4-ene-20-carboxyl-CoA (3-OPC-CoA) and acetyl-CoA were formed as predicted (Figure S1). Negative controls without the enzyme or substrates were performed and no cleavage activity was observed.

FadA5 preferentially cleaves steroid CoA substrates

Upon determination that FadA5 can utilize a steroid-CoA ketoester as a substrate, we undertook steady-state kinetic analyses to determine the extent of substrate specificity in the thiolytic direction. As reported previously with a cytoplasmic thiolase from rat liver (Middleton, 1974), we observed substrate inhibition by CoA. Therefore, the highest concentration of CoA used was 34 μM. The steady-state kinetics of FadA5 with 3,22-dioxochol-4-ene-24-oyl-CoA and CoA followed a bi-bi (ping-pong) mechanism as determined by the best global fit of the initial velocities to the steady-state bi-bi kinetic model assessed by chi-square values (Martin, 1997). The KmA for 3,22-dioxo-chol-4-ene-24-oyl-CoA is 11.8 ± 1.4 μM; the KmB for CoA is 3.10 ± 0.39 μM, and kcat is 0.725 ± 0.038 s−1 resulting in a kcat/KmA = 6.14 ± 0.13 ×104 M−1s−1 (Table 1). FadA5 is thus nearly 400-fold more specific for 3,22-dioxo-chol-4-ene-24-oyl-CoA as a substrate compared to AcAc-CoA, which is due to a 40-fold reduction in KmA and a 10-fold increase in kcat with the steroid substrate. Therefore, we now classify FadA5 as a degradative thiolase or more precisely a 3-ketoacyl-CoA thiolase.

Table 1.

FadA5 steady-state kinetics parametersa

Substrate Km (μM) kcat (s−1) kcat/Km (M−1S−1)
graphic file with name nihms-641784-t0007.jpg 11.8 ± 1.4 0.725 ± 0.038 (6.14 ± 0.13) ×104
graphic file with name nihms-641784-t0008.jpg b 464 ± 207 0.076 ± 0.002 (1.64 ± 0.45) ×102
Open in a new tab
a

The steady-state kinetics of FadA5 were measured in multiple independent replicates. The errors are the standard deviations of the fit.

b

The values were obtained as the apparent Km and kcat in the presence of 50 μM CoA (Nesbitt et al., 2010).

Overall structure of FadA5

To provide insight into the structural prerequisites required to accommodate a bulky steroid as a substrate, we solved the structures of FadA5 in its apo–form as well as complexes containing the wild type (WT) or C93S variant protein with CoA, acetyl-CoA or the steroid OPC (3-oxo-pregn-4-ene-20-carboxylic acid) and CoA. The C93S variant was generated to prevent substrate turnover. The first structure containing WT FadA5, CoA or acetyl-CoA (acetylated WT(-acetyl)-CoA complex) was solved by molecular replacement using the structure of the putative acetyl-CoA acetyltransferase from Thermus thermophilus (pdb entry 1ulq), which displayed the highest (43%) sequence identity to FadA5, as a search model. All subsequent structures were solved using the initial FadA5 structure as the search model.

Commonly, thiolases are described with a three sub-domain architecture (Sundaramoorthy et al., 2006). In this three sub-domain model (Figure 2A), residues 1 to 125 and 246 to 266 of FadA5 form sub-domain I (N-terminal sub-domain), residues 126 to 245 form sub-domain II (lid sub-domain) and residues 267 to 391 form sub-domain III (C-terminal sub-domain). Subdomains I and III are structurally and functionally related. These two sub-domains are mainly characterized by a central β-sheet (β1, β3, β4, β5 and β8 in sub-domain I and β9 to β12 in sub-domain III) that is surrounded by three larger α-helices (α1 to α3 as well as α8, α9 and α11). The two sub-domains harbor the active site residues of FadA5, cysteine 93 within η3 (η = 310helix) in sub-domain I, histidine 347, between helices α10 and α11, and cysteine 377 at the end of β11 in sub-domain III (Figure 2A). The positioning of C93within helix η3 and right before α3 is likely to lower the pKa of this important nucleophile through its orientation with respect to the dipole moment of the helix (Hol, 1985; Lodi and Knowles, 1993). Deprotonation of C93 increases its nucleophilicity, which is important for the initial thiolysis step (Figure 1A, step I) (Price et al., 2003). In this step, C93 is deprotonated by H347 to form a thiolate, which subsequently attacks the β-ketoester of the substrate. The sub-domain I/III architecture provides a stable and rather inflexible platform for the active site, whereas subdomain II sits on top of the two other domains and occludes the active site residues from the solvent thus being called the lid sub-domain (Sundaramoorthy et al., 2006). The lid subdomain comprises an extended structure consisting of three helices, α4 to α6, and two shorter β strands, β6 and β7, interrupted by larger disordered regions. Analysis of the ligand-bound structures revealed the importance of this sub-domain for ligand binding, as at least half of the protein-ligand interactions are located within this domain (see below).

Figure 2. Overall structure of FadA5.

Figure 2

Open in a new tab

A: The sub-domain structure of the FadA5 apo monomer is shown as a cartoon, consisting of an N-terminal sub-domain (I, orange), a lid sub-domain (II, blue) and a C-terminal subdomain (III, green). The thiolase active site residues C93, H347 and C377, which are buried between sub-domains I and III and are ‘occluded’ from the solvent by sub-domain II, are shown in stick presentation. (η = 310helix) B: FadA5 forms the typical thiolase dimer (green monomer chain A, purple monomer chain B)

The functional entity of FadA5 is a dimer (Figure 2B), both in the crystal and in solution as observed by multi angle light scattering (MALS) (Supplemental methods, Table S1 and Figures S2 A/B). The degradative thiolase class can form dimers or tetramers, the latter are built by dimers of dimers, but biosynthetic thiolases are reported to solely form tetramers (Kursula et al., 2002; Modis and Wierenga, 1999). Interestingly, the thiolase fold and the general dimerization motif seems to be highly conserved, despite the fact that the sequence identity between FadA5 and thiolases from other organisms does not exceed 40% (Figure S3).

Structural rigidity of FadA5

We solved FadA5 structures representing four different states of the enzyme: the above described apo structure of WT FadA5, the acetylated WT(-acetyl)-CoA complex, the acetylated C93S-CoA complex, and the OPC-complex structure. These states represent different stages of the thiolytic cleavage reaction (Figure 1A). In the acetylated WT structure FadA5 is found to be in complex with CoA and acetyl-CoA in chain A, whereas in chain B only CoA is bound in two alternate conformations. Based on the electron density, the active site cysteine C93 was modeled to be either in a non-acetylated or in the acetylated C93 (SCY) form. All serine 93 residues of the acetylated C93S-CoA complex are acetylated (OAS) and the enzyme is in complex with a partially occupied CoA (Figure S4A). The last structure is a high-resolution structure of the C93S variant in complex with the hydrolytically cleaved, product fragments CoA and OPC (OPC-complex). The structures described herein were refined to resolutions between 1.7 and 3.0 Å with R-factors ranging between 14.5 and 24.7% and Rfree-factors ranging between 17.4 and 28.7% respectively (Table 2). All 391 residues of the FadA5 protein could be built for the four structures with the exception of residues 138-143 in chain A of the apo structure (Figure 3A, arrow 1) and residues 133-147 of chain D in the OPC-complex (not shown).

Table 2.

Data collection and refinement statistics

acetylated FadA5 WT-(acetyl-)CoA complex FadA5 WT apo acetylated FadA5 C93S-CoA complex FadA5 C93S OPC-complex
Description acetylated thiolase (SCY) + acetyl-CoA + CoA apo acetylated thiolase (OAS) + CoA thiolase in complex with a cleaved product (CoA + OPC)
Relevant buffers
Protein buffer 20 mM bicine pH 8.5, 250 mM NaCl
Crystallization condition 0.1 M MES pH 6.5, 4% 1,4-dioxan, 1.8 M (NH4)2SO4 0.1 M MES pH 6.1, 5% DMSO, 1.8 M (NH4)2SO4 0.1 M Bis-tris pH 6.5, 0.2 M NaCl, 25% PEG 3350 0.1 M MES pH 6.7, 23% PEG 4000
Cryoprotectant solution 0.1 M MES pH 6.5, 1.6 M (NH4)2SO4, 10% 1,4-dioxan, 25% glycerol + 10x ACO 0.1 M MES pH 6.1, 2.0 M (NH4)2SO4, 5% DMSO, 27% glycerol 0.1 M Bis-tris pH 6.5, 0.3 M NaCl, 27% PEG 3350, 25% glycerol + 10x ACO 0.1 M MES pH 6.7, 27% PEG 4000, 25% PEG 400 + 10x 4-OPC-CoA
Data collection
Wavelength (Å) 0.9184 0.9184 0.9537 0.9184
Temperature (K) 100 100 100 100
Space group P43212 P41212 P1211 P1211
Unit cell parameters
a/b/c(Å) 124.68/124.68/124.71 128.31/128.31/114.07 127.61/105.16/147.66 76.82/100.39/107.98
α/β/γ 90°/90°/90° 90°/90°/90° 90°/107°/90° 90°/100°/90°
Resolution (Å) 36.0–1.95 52.0-2.7 58.8-3.0 47.5–1.7
Number of subunits 2 2 8 4
Total reflections 1,051,278 (64,269) 190,844 (20,941) 144,763 (21,071) 1,080,779 (154,493)
Unique reflections 72,009 (4,358) 25,129 (3,358) 70,941 (10,279) 175,728 (25,549)
Completeness (%) 100.0 (100.0) 94.4 (99.5) 95.0 (95.0) 99.6 (99.7)
ø Redundancy 14.6 (14.7) 7.6 (7.7) 2.0 (2.0) 6.0 (6.2)
<I / σ (I)> 13.8 (2.5) 11.8 (2.2) 4.6 (1.3) 12.2 (2.7)
Rmergea (%) 13.3 (108.0) 14.3 (85.3) 17.1 (63.4) 7.5 (62.3)
Wilson B-factor (Å2)b 23.7 37.4 26.5 19.3
Refinement
Total number of atoms 6,504 6,172 23,784 13,496
Rwork / Rfree(%) 19.3 / 23.1 17.1 / 24.8 24.7 / 28.7 14.5 / 17.4
RMSD
    bond angle (°) 1.3581 1.6953 1.4556 1.096
    bond length (Å) 0.0114 0.0122 0.0123 0.007
Estimated coordinate error (Å) (Refmac / Phenix) 0.09 0.24 0.38 0.15
Average B factor (Å2)c detailed list Table S2 29.9 42.6 43.9 31.7
Ramachandran plot (MolProbity)d
Most favored (%) 97.9 93.2 96.3 97.3
Allowed (%) 2.1 6.0 3.5 2.5
Disallowed (%) 0 0.8 0.2 0.2
pdb code 4ubv 4ubw 4ubu 4ubt
Open in a new tab

Values in parentheses refer to the highest-resolution shell

a

Rmerge = ΣhklΣi|Ii - <I>| / ΣhklΣiIi

c

baverage (Winn et al., 2011)

Figure 3. Rigidity of the FadA5 active site and ligand position.

Figure 3

Open in a new tab

A: Zoom into flexible parts of the superimposed sub-domain II in FadA5 apo (green) and the acetylated C93S-CoA complex (brown). Arrow 1 indicates a region of disordered residues in the apo structure, arrow 2 shows the additional helix α3‘ that is formed in the acetylated FadA5 structures and the OPC-complex and arrow 3 points to a quite flexible region between L201 and Q206 with high B-factors in all four structures. B: Detailed stereo view of a superposition of the active site residues C93/S93/SCY93/OAS93, H347 and C377 of all obtained structures; indicating that mainly residue 93 adopts different conformations. (FadA5 apo (green); acetylated WT(-acetyl)-CoA complex (pink and magenta); acetylated C93SCoA complex (brown); OPC-complex (blue)) C: Superposition of residue 93 in FadA5 apo and one of the other structures. Arrow 4 indicates movement of S93 in comparison to C93 (left). Arrows 5 point to the position of SCY93 and C93 in chain A of the acetylated WT structure (middle). Arrow 6 indicates the position of OAS93 in the acetylated C93S-CoA structure (right). Color code as above. D: Superposition of the (acetyl-)CoA traces of the three complex structures shown in stereo. The pink and rosé acetyl-CoA and CoA traces belong to the acetylated WT(-acetyl)-CoA complex structure, where both ligands are binding to chain A. Arrow 7 shows the additional helix α3‘. Color code as above. E: Two of the four FadA5 structures (acetylated WT(-acetyl)-CoA and OPC-complex structure) contain a central water molecule in the active site that is also present in the biosynthetic thiolase (black) structures at the same position. Color code as above.

A similarity analysis performed with PDBeFold revealed a high overall structural similarity of the different FadA5 structures, indicated also by low RMSD values ranging from 0.30 to 0.52 Å. This observation is reinforced by comparing the FadA5 root mean square deviation (RMSD) values to the FadA5 apo structure and the thiolase from T. thermophilus (pdb entry 1ulq) (Tables S3 and S4 for cross structure statistics). Differences among the structures are mainly restricted to the lid sub-domain and are located in the loop region between β5 and α4, an additional helix α3’ (R136 to I139), which is only present in the complex structures, and a flexible region close to hairpins β6 and β7. The described differences are illustrated in Figure 3A where the strongest diverging FadA5 structures, the apo and acetylated C93S-CoA complex, are superimposed.

Interestingly, this rigidity among thiolases seems to be well conserved and was also observed for other thiolases (Modis and Wierenga, 2000). The rigidity can even be extended to the position of the active-site residues, which remain in the same conformation in the apo and in the ligand bound structures (Figure 3B/C).

The side chain of residue 93 adopts different orientations, and as described above, alternative conformations can be observed as well (S93, Figure 3C, arrow 4). Those movements were elucidated as significant, as they are significantly higher than the coordinate errors (Table 2). In addition, the acetylated and non-acetylated states of this residue are present, even within the same monomer. C93 is partially acetylated in the WT(-acetyl)-CoA complex, which means that one and the same position is occupied by alternative conformations of a C93 or an acetylated C93 (SCY) (see Figure 3C, arrows 5). In the acetylated C93S-CoA complex we observe the acetylated S93 (OAS) without alternative conformations (see Figure 3C, arrow 6).

Ligand and water-binding at the active site

In general, the occupancy for acetyl-CoA and CoA varies among the structures and is the highest for the CoA in the OPC-complex. The quality of the electron density for the CoA ligand is a result of the higher resolution of the OPC-complex, but may also be due to the presence of the additional ligand, a cleaved product, in this structure. Nevertheless, the overall orientation of the acetyl-CoA and CoA ligand is similar in all complex structures (Figure 3D). Although the orientation of the 3’-phosphate-ADP part of the ligand is quite conserved, the pantothenic acid of the lower occupied ligand orientations and the (acetylated) cysteamine can adopt different orientations in the structures, and therefore, interact differently with the active site and its surroundings.

A thiolase-specific water molecule was observed in the active sites of the acetylated WT(-acetyl)-CoA (Z253) and the OPC-complex structures (Z25) (Figures 3E and 4). This water was also observed in other thiolase structures (Haapalainen et al., 2006; Modis and Wierenga, 2000), and thus seems to be highly conserved. It is part of the oxyanion hole 1 that is especially important for biosynthetic thiolase reactions, where the oxyanion hole stabilizes the transition state intermediates (Kursula et al., 2002; Kursula et al., 2005).

Figure 4. Interaction network of bound acetyl-CoA and CoA.

Figure 4

Open in a new tab

A: Stereo view of the acetyl-CoA (gray) and FadA5 (acetylated WT(-acetyl)-CoA structure) interaction via water (red spheres) mediated hydrogen bonds (dashed lines) as well as direct contacts and a π-cation-stacking interaction (orange ellipse) of the adenosine moiety of CoA with R221. The active site residues interact with each other via direct and water mediated contacts, a central position is assumed by water molecule Z253. B: CoA (gray) and FadA5 (OPC-complex) interact similarly as described in A, a central position is assumed by water molecule Z25. The residues, are colored according to the sub-domain color code.

Even though most thiolase structures share significant structural similarity, biosynthetic and degradative thiolases differ with respect to the location of water molecules near their active sites. FadA5 is comparable to the degradative thiolase from Saccharomyces cerevisiae (Mathieu et al., 1997), but differs from the above-mentioned biosynthetic thiolase from Zoogloea ramigera and an investigated human thiolase (Kursula et al., 2005). For the latter thiolases, a water channel to the bulk solvent water and a water reservoir located near the active site are described, whereas only the reservoir can be observed in the FadA5 structures, but the channel where the water chain would be located is blocked by two hydrophobic residues, I314 and V322.

The acetylated WT(-acetyl)-CoA complex

The acetylated WT(-acetyl)-CoA complex was obtained after the protein was incubated with a ten-fold molar excess of acetyl-CoA. Both monomers in this structure contain a partially occupied native (C93) and an acetylated active site cysteine, C93 (SCY93). In chain A, a mixture of acetyl-CoA (50%) and CoA (40%) is present according to the electron density, whereas chain B only contains CoA as a ligand, but also in a higher (40%) and a lower (30%) occupied conformation. The ligand densities are not very well defined, and in both chains, a third even less occupied conformation might be present but was not modeled. In addition to the ligand densities we observe diffuse difference density in the steroid-binding pocket that can not be explained as ligand or solvent molecules, and therefore remains uninterpreted.

In the degradative thiolase reaction, an acylated cysteine is formed upon nucleophilic attack of the cysteine on the β-carbonyl (Figure 1A, state I -> II). The intermediate state (Figure 1A, state II) is mimicked in our structure by an acetylation of C93 where the larger steroid (R) is substituted by a methyl group that does not reach into the void of the proposed steroid-binding pocket. Even though the electron density for SCY93 as well as for the acetyl-CoA and CoA was only partially present (Figure S4B) in the acetylated WT(-acetyl)-CoA complex structure we were able to observe different conformations of the active site cysteine 93 compared to the apo structure. The sulfur is shifted by 25° in the direction of the active site H347 and also the native C93 of the complex structure is shifted to a bigger extent by 95° (see Figure 3C, arrow 5).

The shift of the sulfur in both SCY93s positions the acetyl-group directly into the open space of the steroid-binding pocket and a substitution of the methyl group of the acetyl by the steroid would be readily accommodated. This movement of the active site C93 or its covalent modification respectively is also described for other thiolases (Modis and Wierenga, 2000), and the alternative conformations obtained is also found for the Z. ramigera thiolase (Kursula et al., 2002).

In both protein chains, we observe a stronger binding for one of the two described ligands, indicated by a higher number of direct and indirect contacts in comparison to the other ligand. Acetyl-CoA in chain A assumes three direct contacts with residues Q151, R221 and H347 and twelve additional water-mediated interactions. Residues of sub-domain II provide half of those contacts, but the acetyl-CoA is also connected to the catalytic residues C93 and C377 by the oxyanion hole water Z253 (Figure 4A). The plane of the ligand's nucleotide heterocycle is positioned parallel to the guanidine function of R221 and forms a π-cation stacking interaction at a distance of 3.7 Å (Figure 4A orange ellipse). The CoA ligand of chain A forms a total of only five water-mediated interactions to the protein, and in addition, a weaker π-cation stacking interaction with R221, as the parallel geometry seems to be imperfect (Figure S4C).

The catalytic triad residues are also connected to each other via a network of water-mediated and direct contacts. SCY93 does not form any interaction with the catalytic triad or the ligand, but C93 is in direct contact to H347. C377 is located farther away, and therefore, only interacts via the oxyanion hole water with C93. In chain B, the higher occupied CoA adopts 14 direct or water-mediated contacts, including the catalytic triad, whereas the lower occupied CoA only forms six contacts with the protein and is not interacting with the active site.

FadA5 in complex with a hydrolyzed steroid-CoA (OPC)

The C93S variant of FadA5 was co-crystallized with a 10-fold molar excess of the enzyme reaction product 3-OPC-CoA, which led to a complex containing FadA5, free CoA and the 3-oxo-pregn-4-ene-20-carboxylic acid (OPC), a steroid derivative, in our structure. The OPC-complex structure contains four protein chains (A-D) in the asymmetric unit. An interesting aspect of the OPC-complex structure is the structural diversity of these different subunits. Residues 133 – 147 of sub-domain II were not resolved in chain D, but could be modeled in the other chains. This disorder also extends to the active site. The CoA ligand was modeled for all four chains, whereas OPC could not be modeled for chain D.

The following analysis is based on chain A in the OPC-complex structure. Serine 93 within the catalytic triad adopts two different conformations pointing into opposite directions, but neither of these conformations is identical to the C93 conformation of the apo protein (Figure 3B/C). One of these conformations, S93-B, is shifted by −80° in comparison to apo C93, whereas the second denoted as S93-A differs by an angle of 79°.

As discussed above, neither the product nor CoA binding induce a large conformational change within the overall structure, and thus CoA assumes similar contacts with the active site residues as described for the acetyl-CoA of chain A or CoA of chain B in the acetylated WT(-acetyl)-CoA complex structure (Figure 4A). Beyond the active site, three direct interactions of CoA with residues Q151, T223 and S246 and one additional water-mediated interaction to Q247 are observed. The first three are partially present in the acetylated WT(-acetyl)-CoA complex structure, depending on the ligand orientation, but the last interaction could only be observed in the OPC structure (Figure 4B).

OPC (Figure 5A) is not the native product of the thiolase reaction; rather it was generated during crystallization by hydrolysis of the thioester bond of compound 3 (Figure 1B) to release free CoA. Given that OPC is a steroid, it is not surprising that it mainly forms hydrophobic interactions with the enzyme and is located in a pocket that is mostly built by the lid sub-domain, engaging residues L128 to I139 and N150 to Q151. An interesting contact is formed by N68 of the neighboring monomer that also contributes its side chain to the steroid-binding pocket. Although polar amino acids are also part of the pocket, the steroid-packing interactions are provided through main chain, Cβ or Cγ atoms (Figure 5A). Interestingly, the C3 ketone does not form any interaction with surrounding residues or water molecules. The carboxyl function of OPC forms one direct hydrogen bond to the amino group of residue G379 and it interacts with the B conformation of the active site S93. In addition to these interactions, the distance between the formerly covalently-connected atoms CAQ and S1P has increased to 3.9 Å.

Figure 5. Binding and steroid- binding pocket of the hydrolyzed steroid.

Figure 5

Open in a new tab

A: SIGMAA-weighted 2Fo-Fc electron density omit map (σ = 1.0; gray) of OPC (yellow; chemical structure on the left). OPC is shown with respect to the active site residues and residues of the lid sub-domain (blue) that are forming a hydrophobic pocket to bind OPC; the protein (OPC-complex structure, chain A) is depicted in cartoon representation in the according sub-domain color code. B: Surface presentation of the enzyme dimer (OPC-complex structure; chain B, gray) illustrating the positioning of CoA and OPC (both yellow) with respect to chain A (blue) where the more flexible lid sub-domain is highlighted by the deep blue coloring. OPC is buried in a hydrophobic pocket built mainly by residues of sub-domain II but also a small part of chain B. Main parts of CoA are solvent exposed. C/D: Movements of residues R136 (arrows 1/2) and Q151 (arrow 3) caused by OPC binding in the OPC-complex structure (blue) in comparison to the other structures are shown in stereo. The atoms used for the angle determination are labeled and indicated by yellow dots (residue color code: acetylated WT(-acetyl)-CoA complex, pink; apo, green; acetylated C93S-CoA complex, brown).

The position of the OPC binding pocket reveals intriguing insights with respect to the thiolase dimer. Whereas the lid sub-domain of chain A as well as residues from the N-terminal sub-domain of the other monomer occlude the hydrophobic OPC molecule from the solvent (Figure 5B, gray surface), CoA, on the other hand, is far more exposed to the solvent and does not reach beneath the protein surface.

In chains B and D, an alternative conformation of the CoA ligand could be modeled. The alternative conformation in chain B affects the thiol and the neighboring C2P only, whereas in the CoA of chain D a long stretch of the CoA chain, ranging from the thiol back to the C9P, adopts an alternative conformation. The two thiols of CoA in chain B are twisted by 96°, one thiol is pointing towards the OPC (SH-B) and superimposes with the thiol of chain A (Figure S5A), while the other thiol (SH-A) points away from the OPC (Figure S5B). Even though both conformations are well defined in the electron density maps, the one pointing towards the OPC assumes the higher occupancy. In this chain, in contrast to chain A, OPC and CoA assume hardly any direct contact, the distance is 3.4 Å, but the angle is not optimal for a hydrogen bond interaction and the most intriguing aspect is the positive difference density (Figure S5B, green) located around S93-B and between this residue and the OPC molecule. This difference density is also partially present in chain A, but the size and shape is most prominent for chain B and its ligands. We hypothesize that this density is a relic from the uncleaved 3-OPC-CoA that was originally used for crystallization; however modeling of 3-OPC-CoA into this difference density was not possible due to its very low occupancy.

As already mentioned, chains A to C contain the OPC molecule, but it was not possible to model OPC in chain D, although some partial density was observed in the binding pocket. In addition, the comparably high B-factors of the CoA ligand bound to chain D indicate a higher flexibility or disorder. The flexibility of the CoA ligand may be directly linked to the absence of the OPC, as described above. This strengthens our hypothesis that only upon occupation of the steroid-binding pocket the (acetyl-)CoA ligand is tightly bound or correctly positioned. We therefore propose that chain D represents an open enzyme state after catalysis when the product is released.

The steroid-binding pocket

Although the steroid-binding pocket is present in all described FadA5 structures (exceptions are: FadA5 apo chain A and OPC-complex chain D) the question arises, whether the rigidity of the overall structure extends to this area as well. A superposition of selected chains of the four structures shows that residues R136 and Q151 assume different conformations due to ligand binding (Figure 5C/D). The guanidine function of R136 in the OPC-complex structure (blue, bound OPC) is shifted by 84° - 90° in comparison to the other complex structures. Binding of the hydrophobic OPC thus led to an extended conformation of this arginine residue (Figure 5C, arrow 1). The apo structure (green) differs with respect to both polypeptides in the dimer. R136 in chain A of the apo structure (Figure 5C, arrow 2) is located approx. 6 Å farther away from the OPC binding site than in the OPC structure, which is most likely linked to the disorder after amino acid 137. Glutamine 151 assumes a bent conformation in the OPC-complex structure (see Figure 5D, arrow 3), but an extended conformation in the other structures. The bend angle, ∢(CDOPC, CBOPC, CDothers), of Q151 is between 90° and 110° (Figure 5D, yellow dots) in the different chains and is required to accommodate the OPC, which would not fit into the pocket if the glutamine would remain in its extended conformation. In chain D of the same structure, however, where no OPC could be modeled, this residue adopts two alternative conformations, one in the bent and one in an extended conformation.

Discussion

We pursued the structural and functional characterization of the mycobacterial FadA5 protein to provide insight into its mechanism and to utilize this knowledge for inhibitor design. Our different structures represent three of the five states in the thiolase reaction cycle (Figure 1A) and thus permit detailed mechanistic insights. For the first step of the cleavage reaction, the β-keto thioester substrate (Figure 1B compound 2) forms a complex with the apo enzyme. We solved the structure of the apo enzyme and showed that it is structurally highly similar to the complex structures with bound ligands. We can therefore hypothesize that in state I (Figure 1A), the only visible movement upon binding of the substrate would be a shift of C93 to permit entry to the steroid-binding pocket as observed in all of our complex structures.

State II of this cycle is represented by the acetylated WT(-acetyl)-CoA complex structure where the active site cysteine C93 of the acylated enzyme is in close proximity to an acetyl-CoA molecule that theoretically was just cleaved off the substrate. In our structure, the acylation with a bulky steroid is mimicked by the much shorter acetyl, but even in the presence of this shortened compound the orientation of the “real” substrate is unambiguous and is confirmed by the OPC-complex structure. The OPC-complex structure mimics state III (Figure 1A) and shows for the first time the occupation of the steroid-binding pocket with the required residue movements, i.e. Q151 and R136, for proper binding. These movements (Figure 5C/D) hold the shortened steroid in place until it is transferred to CoA. The mimic is not perfect, as we do not observe a covalent bond between the active site residue S93 and OPC. In the case of a covalent complex, one would expect a single conformation of S93 in the B orientation of the two observed conformations and an OPC molecule that is in slightly closer proximity to the CoA than observed in our structure. In addition to OPC, the complex contains a bound CoA molecule that can be activated by deprotonation with the help of a C377-thiolate. The activated CoA reacts with the covalently bound product that is thereby released from the enzyme. Finally, the enzyme is in complex with a more loose, noncovalently, bound intermediate (Figure 1A, state IV), for which we do not have structural information so far.

The general question that arises and cannot completely be solved here is the binding and release mechanism of the steroids. They are located in a hydrophobic binding pocket during catalysis, but due to the rigidity of the enzyme in the apo or ligand-bound state, it is not clear whether the compound is transferred along the CoA-binding site and the catalytic triad or if a larger reorientation of the steroid-binding loop occurs. The fact that a hydrophobic compound would have to pass along hydrophilic binding regions makes a reorientation or opening of the binding loop more plausible. This hypothesis is also supported by chain D of the OPC-complex structure where the OPC is absent from the binding pocket and a highly disordered binding loop (residues G133 – D147) and a more flexible bound CoA ligand are present. In this chain, we thus observe a snapshot with an opened binding loop that may just have released its product and displays a solvent exposed steroid-binding site. The lower occupied CoA (Figure S4B) or acetyl-CoA ligands observed in the other structures together with the more flexible bound CoA in chain D of the OPC complex suggest that high occupancy CoA binding in FadA5 only occurs when the steroid-binding pocket is occupied. However, we do not observe a structural difference in the CoA-binding pocket of the steroid-bound and steroid-free structures and can thus not explain this phenomenon. Based on our different enzyme snapshots we therefore hypothesize, despite the fact that the overall fold of the protein is rigid, that FadA5 undergoes larger reorientations to allow substrate entry and product release and that an occupation of both binding sites is essential for tight ligand binding, and for catalysis.

As cholesterol metabolism became the focus of being an anti-TB drug target, the question of FadA5‘s potential and the feasibility of drug development arises. Despite their sequence differences, thiolases display a highly conserved fold throughout the entire enzyme class which leads to the question whether FadA5 can be specifically targeted without inhibiting human thiolases as well?

Humans contain six different thiolases and three of them, the cytosolic thiolase (CT), the mitochondrial T2-thiolase (T2) and the peroxisomal AB-thiolase (AB) have been structurally characterized; for the remaining three thiolases (mitochondrial T1-thiolase (T1), peroxisomal SCP2-thiolase (SCP2) and mitochondrial trifunctional enzyme thiolase (TFE)) models were generated. As shown in Figure 6A the CT-, the T2- and the T1-thiolase are structurally diverging the most compared to FadA5. These thiolases form a tetramerization loop instead of a steroid-binding pocket. Residues emerging from this loop cut directly through the position of the OPC. Human AB- and SCP2-thiolase are structurally more closely related to FadA5 and we cannot exclude the possibility that they would be targeted by FadA5 inhibitors. However, superposition of the AB-thiolase and FadA5 revealed that some residues within the steroid-binding loop of the AB-thiolase are disordered and others replaced by shorter ones so that the observed shielding of the steroid may not be possible (Figure 6B). Concerning the SCP2 thiolase, only 25% of the generated models adopted a similar steroid-binding pocket as observed in FadA5 (not shown).

Figure 6. Superposition of FadA5 with different human thiolases.

Figure 6

Open in a new tab

Superposition of the FadA5 OPC-complex monomer (blue) with structures or models of different human thiolases. OPC is depicted in surface representation (yellow, red). Regions in the human thiolases which are located in close proximity to the bound OPC are also shown in transparent surface representation. The steroid-binding loop (arrow 1) and the tetramerization loop (arrow 2) are also shown. A: Superposition with structures of CT- (1wl4, violet) and T2-thiolases (2ib7, pink) as well as a model of the T1-thiolase (green). B: Superposition with the structure of the AB-thiolase (2iik, green). C: Superposition with the model of the TFE-thiolase (cyan). The CoA molecule is shown in all-sticks presentation.

Lastly, the mitochondrial trifunctional enzyme thiolase seems to be the closest relative to FadA5 and is therefore probably the most critical to be targeted by FadA5 drug candidates. However, a structure prediction using the I-TASSER server (Roy et al., 2010; Zhang, 2008) suggested that the TFEs are structurally more related to the T2 and CT enzymes, by forming a tetramerization loop instead of a steroid-binding pocket, and therefore targeted FadA5 inhibition may be feasible (Figure 6C).

To provide initial insights about how FadA5 inhibitors should look like, a hotspot analysis (Supplemental experimental procedures) of the steroid- and CoA-binding pocket, using HotspotsX (Neudert and Klebe, 2011), was performed. The analysis indicated that inhibitory potency might be gained for example by involving the carbonyl oxygen of the catalytic residue H347 in a hydrogen or halogen bond (Figure S6).

FadA5's preference for a steroid β-keto-CoA thioester substrate in combination with previous metabolic profiling data (Nesbitt et al., 2010) strongly supports that its in vivo catalytic activity is the catalysis of thiolytic degradation of the cholesterol side chain. Our structural analysis of four FadA5 structures thus provides first insights into steroid complex formation and ligand binding for a degradative thiolase. In particular, the OPC-complex structure permitted the observation of how a large hydrophobic compound is accommodated in the substrate-binding pocket and oriented with respect to the active site.

Based on our structural data, combined with our hotspot analysis, and the preliminary structural homology analysis demonstrating the low similarity of FadA5 compared to the human thiolases, we are optimistic that specific targeting of FadA5 is feasible. As a starting point steroid-like compounds could be used for inhibitor design with specific and targeted functionalization of the ring-system extending towards the CoA-binding pocket.

Experimental Procedures

FadA5 expression and purification

See Experimental procedures in the supplement for the protein expression and purification.

FadA5 Kinetics

FadA5 was assayed for thiolytic activity with 3,22-dioxo-chol-4-ene-24-oyl-CoA and CoA as substrates at 30 °C. The thiolytic activity was followed at 303 nm using a UV/vis spectrophotometer with a Peltier temperature controller (Shimadzu Scientific Instruments) by monitoring the disappearance of the Mg2+-3,22-dioxo-chol-4-ene-24-oyl-CoA complex (Middleton, 1974; Nesbitt et al., 2010). The assay was carried out with 28 nM FadA5 in 100 mM Tris-HCl, pH 8.1 containing 25 mM MgCl2, 1.25 μM to 60 μM 3,22-dioxo-chol-4-ene-24-oyl-CoA and 1.07 μM to 34 μM CoA. The initial velocity was determined from the linear portion of the reaction progress curve, and the initial burst of activity before reaching steady-state was disregarded. Steady-state kinetic parameters were obtained from global hyperbolic fits of the initial velocity data using Grafit4. The substituted-enzyme (bi-bi, ping pong) mechanistic equation (1) provided the best fit to the data, where A stands for 3,22-dioxo-chol-4-ene-24-oyl-CoA; B stands for CoA; KB is the Km for CoA; KA is the Km for 3,22-dioxo-chol-4-ene-24-oyl-CoA; Vmax is the maximal velocity.

v=Vmax[A][B]KB[A]+KA[B]+[A][B] (1)

See Experimental procedures in the supplement for the substrate synthesis.

Crystallization and data collection

The detailed crystallization and cryoprotectant conditions are listed in Table 2. FadA5 WT apo (12 mg/ml) was mixed in a one to one ratio with the crystallization buffer and set up for crystallization at 20 °C. For co-crystallization with acetyl-Coenzyme A (dissolved in water) the protein (FadA5 WT and FadA5 C93S) samples (12 mg/ml) were incubated with a 10-fold molar excess of ACO for 30 min at 4 °C, prior to crystallization. The samples were mixed in a one to one ratio with the crystallization buffer (Table 2) and were set up for crystallization at 20 °C. For co-crystallization trials 3-oxo-pregn-4-ene-20-carboxyl-CoA (3-OPC-CoA, Figure 1, compound 3) was dissolved in the protein buffer and added to the FadA5 C93S variant (12 mg/ml) in a 10-fold molar excess; the mixture was immediately used for crystallization in a one to one ratio with its crystallization solution (Table 2). Prior to flash-freezing, the crystals were briefly transferred into a cryoprotectant solution (Table 2).

All FadA5 wild type data sets were collected at BESSY II, beamline MX 14.1 with a MAR mosaic 225 detector. The CoA containing C93S variant data set was collected with a MAR mosaic 225 detector at the ESRF beamline BM 14U. The C93S containing 3-OPC-CoA data set was collected at BESSY II, beamline MX 14.1 with a PILATUS 6M detector.

Structure determination

All data sets were integrated with iMosflm (Leslie, 1992) and scaled with Scala (Evans, 2006) or Aimless (Evans, 2011). For phasing of the first data set (acetylated WT(-acetyl)-CoA) molecular replacement was performed with Phaser (McCoy et al., 2007) using the putative acetyl-CoA acetyltransferase from Thermus thermophiles (pdb entry 1ulq) with 43% sequence identity to FadA5 as a search model. Automatic model building was conducted with the help of ARP/wARP (Langer et al., 2008) and buccaneer (Cowtan, 2006). Manual model building was subsequently performed with Coot (Emsley and Cowtan, 2004) and the structure was further refined with REFMAC (Murshudov et al., 1997) (Winn et al., 2011).

For all following data sets molecular replacement was performed using chain A of the acetylated WT(-acetyl)-CoA structure (excluding ligands and solvent molecules) as the search model using Phaser. Model building for the FadA5 WT apo structure was performed as described above. For refinement and model building of the acetylated C93S-CoA complex thin shell and non-crystallographic symmetry (NCS) refinement, including the CoA ligand, was used. Refinement cycles for the OPC-complex were conducted with Phenix (Adams et al., 2010). The PRODRG server (Schuttelkopf and van Aalten, 2004) was used to create library files for geometry restraints as well as structure files of acetyl-Coenzyme A (ACO = abbreviation in the coordinate file), acetylated cysteine (SCY), acetylated serine (OAS), the free acid of the product (OPC) and Coenzyme A (CoA). MolProbity (Chen et al., 2010) analysis was performed to validate the structures. Figures of all structures were generated using PyMOL (DeLano, 2002) after secondary structure assignment by DSSP (Joosten et al., 2011; Kabsch and Sander, 1983) whereas sequence alignments and comparisons were performed with CLUSTAL W (Thompson et al., 1994) and ESPript (Gouet et al., 2003).

Supplementary Material

Highlights.

  • Structural characterization of the mycobacterial FadA5 thiolase

  • First structure of a thiolase in complex with a bound steroid

  • Elucidation of preference for specific steroids

  • Similarity analysis with human thiolases showed feasibility of targeting FadA5

Acknowledgements

We thank Prof. Courtney Aldrich, U Minnesota for providing a fadD17 clone and Liqun Wang for preparing the C93S mutant clone. We also thank Prof. Christoph Sotriffer (University of Wuerzburg) and Prof. Gerhard Klebe (University of Marburg) for their support concerning the hotspot analysis. This work was supported in part by the National Institutes of Health and the National Science Foundation (NIH AI092455 (N.S.S.), NIH AI085349 (N.S.S.), NIH AI065251 (N.S.S.), NIH HL53306 (N.S.S.), NIH RR021008 (N.S.S.), NSF BIO1039771 (NMR)) and the Deutsche Forschungsgemeinschaft (grants SFB630 to C.K. and Forschungszentrum FZ82 to C.K.). We thank the staff of beamlines MX 14.1 at BESSY II, Berlin and BM14U ate the ESRF, Grenoble for technical support.

Footnotes

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.

Accession numbers

The coordinates and structure factors of FadA5 in its apo form as well as the different complexes have been deposited in the PDB with the accession codes 4ubt, 4ubu, 4ubv, and 4ubw, respectively.

References

  1. Adams PD, Afonine PV, Bunkoczi G, Chen VB, Davis IW, Echols N, Headd JJ, Hung LW, Kapral GJ, Grosse-Kunstleve RW, et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr. 2010;66:213–221. doi: 10.1107/S0907444909052925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Anbazhagan P, Harijan RK, Kiema TR, Janardan N, Murthy MR, Michels PA, Juffer AH, Wierenga RK. Phylogenetic relationships and classification of thiolases and thiolase-like proteins of Mycobacterium tuberculosis and Mycobacterium smegmatis. Tuberculosis (Edinb) 2014 doi: 10.1016/j.tube.2014.03.003. [DOI] [PubMed] [Google Scholar]
  3. Chang JC, Harik NS, Liao RP, Sherman DR. Identification of Mycobacterial genes that alter growth and pathology in macrophages and in mice. J Infect Dis. 2007;196:788–795. doi: 10.1086/520089. [DOI] [PubMed] [Google Scholar]
  4. Chang JC, Miner MD, Pandey AK, Gill WP, Harik NS, Sassetti CM, Sherman DR. igr Genes and Mycobacterium tuberculosis cholesterol metabolism. J Bacteriol. 2009;191:5232–5239. doi: 10.1128/JB.00452-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chen VB, Arendall WB, 3rd, Headd JJ, Keedy DA, Immormino RM, Kapral GJ, Murray LW, Richardson JS, Richardson DC. MolProbity: all-atom structure validation for macromolecular crystallography. Acta crystallographica. Section D, Biological crystallography. 2010;66:12–21. doi: 10.1107/S0907444909042073. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cowtan K. The Buccaneer software for automated model building. 1. Tracing protein chains. Acta crystallographica. Section D, Biological crystallography. 2006;62:1002–1011. doi: 10.1107/S0907444906022116. [DOI] [PubMed] [Google Scholar]
  7. DeLano WL. The PyMOL Molecular Graphics System on World Wide Web. 2002 [Google Scholar]
  8. Dubnau E, Fontan P, Manganelli R, Soares-Appel S, Smith I. Mycobacterium tuberculosis genes induced during infection of human macrophages. Infect Immun. 2002;70:2787–2795. doi: 10.1128/IAI.70.6.2787-2795.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallogr D Biol Crystallogr. 2004;60:2126–2132. doi: 10.1107/S0907444904019158. [DOI] [PubMed] [Google Scholar]
  10. Evans P. Scaling and assessment of data quality. Acta Crystallogr D Biol Crystallogr. 2006;62:72–82. doi: 10.1107/S0907444905036693. [DOI] [PubMed] [Google Scholar]
  11. Evans PR. An introduction to data reduction: space-group determination, scaling and intensity statistics. Acta Crystallogr D Biol Crystallogr. 2011;67:282–292. doi: 10.1107/S090744491003982X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. French S, Wilson K. On the treatment of negative intensity observations. Acta Crystallographica Section A. 1978;34:517–525. [Google Scholar]
  13. Gilbert HF. Proton transfer from acetyl-coenzyme A catalyzed by thiolase I from porcine heart. Biochemistry. 1981;20:5643–5649. doi: 10.1021/bi00523a001. [DOI] [PubMed] [Google Scholar]
  14. Gilbert HF, Lennox BJ, Mossman CD, Carle WC. The relation of acyl transfer to the overall reaction of thiolase I from porcine heart. The Journal of biological chemistry. 1981;256:7371–7377. [PubMed] [Google Scholar]
  15. Gouet P, Robert X, Courcelle E. ESPript/ENDscript: Extracting and rendering sequence and 3D information from atomic structures of proteins. Nucleic acids research. 2003;31:3320–3323. doi: 10.1093/nar/gkg556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Haapalainen AM, Merilainen G, Wierenga RK. The thiolase superfamily: condensing enzymes with diverse reaction specificities. Trends Biochem Sci. 2006;31:64–71. doi: 10.1016/j.tibs.2005.11.011. [DOI] [PubMed] [Google Scholar]
  17. Hol WG. The role of the alpha-helix dipole in protein function and structure. Progress in biophysics and molecular biology. 1985;45:149–195. doi: 10.1016/0079-6107(85)90001-x. [DOI] [PubMed] [Google Scholar]
  18. Joosten RP, te Beek TA, Krieger E, Hekkelman ML, Hooft RW, Schneider R, Sander C, Vriend G. A series of PDB related databases for everyday needs. Nucleic Acids Res. 2011;39:D411–419. doi: 10.1093/nar/gkq1105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kabsch W, Sander C. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers. 1983;22:2577–2637. doi: 10.1002/bip.360221211. [DOI] [PubMed] [Google Scholar]
  20. Kursula P, Ojala J, Lambeir AM, Wierenga RK. The catalytic cycle of biosynthetic thiolase: a conformational journey of an acetyl group through four binding modes and two oxyanion holes. Biochemistry. 2002;41:15543–15556. doi: 10.1021/bi0266232. [DOI] [PubMed] [Google Scholar]
  21. Kursula P, Sikkila H, Fukao T, Kondo N, Wierenga RK. High resolution crystal structures of human cytosolic thiolase (CT): a comparison of the active sites of human CT, bacterial thiolase, and bacterial KAS I. J Mol Biol. 2005;347:189–201. doi: 10.1016/j.jmb.2005.01.018. [DOI] [PubMed] [Google Scholar]
  22. Langer G, Cohen SX, Lamzin VS, Perrakis A. Automated macromolecular model building for X-ray crystallography using ARP/wARP version 7. Nature protocols. 2008;3:1171–1179. doi: 10.1038/nprot.2008.91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Leslie AGW. Recent changes to the MOSFLM package for processing film and image plate data. Joint CCP4 + ESF-EAMCB Newsletter on Protein Crystallography, No. 26. 1992 [Google Scholar]
  24. Lodi PJ, Knowles JR. Direct evidence for the exploitation of an alpha-helix in the catalytic mechanism of triosephosphate isomerase. Biochemistry. 1993;32:4338–4343. doi: 10.1021/bi00067a024. [DOI] [PubMed] [Google Scholar]
  25. Martin RB. Disadvantages of double reciprocal plots. J. Chem. Educ. 1997;74:1238–1240. [Google Scholar]
  26. Mathieu M, Modis Y, Zeelen JP, Engel CK, Abagyan RA, Ahlberg A, Rasmussen B, Lamzin VS, Kunau WH, Wierenga RK. The 1.8 A crystal structure of the dimeric peroxisomal 3-ketoacyl-CoA thiolase of Saccharomyces cerevisiae: implications for substrate binding and reaction mechanism. J Mol Biol. 1997;273:714–728. doi: 10.1006/jmbi.1997.1331. [DOI] [PubMed] [Google Scholar]
  27. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC, Read RJ. Phaser crystallographic software. J Appl Crystallogr. 2007;40:658–674. doi: 10.1107/S0021889807021206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. McLean KJ, Lafite P, Levy C, Cheesman MR, Mast N, Pikuleva IA, Leys D, Munro AW. The Structure of Mycobacterium tuberculosis CYP125: molecular basis for cholesterol binding in a P450 needed for host infection. J Biol Chem. 2009;284:35524–35533. doi: 10.1074/jbc.M109.032706. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Middleton B. The kinetic mechanism and properties of the cytoplasmic acetoacetyl-coenzyme A thiolase from rat liver. Biochem J. 1974;139:109–121. doi: 10.1042/bj1390109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Miner MD, Chang JC, Pandey AK, Sassetti CM, Sherman DR. Role of cholesterol in Mycobacterium tuberculosis infection. Indian J Exp Biol. 2009;47:407–411. [PubMed] [Google Scholar]
  31. Modis Y, Wierenga RK. A biosynthetic thiolase in complex with a reaction intermediate: the crystal structure provides new insights into the catalytic mechanism. Structure. 1999;7:1279–1290. doi: 10.1016/s0969-2126(00)80061-1. [DOI] [PubMed] [Google Scholar]
  32. Modis Y, Wierenga RK. Crystallographic analysis of the reaction pathway of Zoogloea ramigera biosynthetic thiolase. Journal of Molecular Biology. 2000;297:1171–1182. doi: 10.1006/jmbi.2000.3638. [DOI] [PubMed] [Google Scholar]
  33. Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr. 1997;53:240–255. doi: 10.1107/S0907444996012255. [DOI] [PubMed] [Google Scholar]
  34. Nesbitt NM, Yang X, Fontan P, Kolesnikova I, Smith I, Sampson NS, Dubnau E. A thiolase of Mycobacterium tuberculosis is required for virulence and production of androstenedione and androstadienedione from cholesterol. Infect Immun. 2010;78:275–282. doi: 10.1128/IAI.00893-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Neudert G, Klebe G. DSX: a knowledge-based scoring function for the assessment of protein-ligand complexes. J Chem Inf Model. 2011;51:2731–2745. doi: 10.1021/ci200274q. [DOI] [PubMed] [Google Scholar]
  36. Ouellet H, Johnston JB, de Montellano PR. Cholesterol catabolism as a therapeutic target in Mycobacterium tuberculosis. Trends Microbiol. 2011;19:530–539. doi: 10.1016/j.tim.2011.07.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Pandey AK, Sassetti CM. Mycobacterial persistence requires the utilization of host cholesterol. Proceedings of the National Academy of Sciences of the United States of America. 2008;105:4376–4380. doi: 10.1073/pnas.0711159105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Price AC, Rock CO, White SW. The 1.3-Angstrom-resolution crystal structure of beta-ketoacyl-acyl carrier protein synthase II from Streptococcus pneumoniae. J Bacteriol. 2003;185:4136–4143. doi: 10.1128/JB.185.14.4136-4143.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Roy A, Kucukural A, Zhang Y. I-TASSER: a unified platform for automated protein structure and function prediction. Nat Protoc. 2010;5:725–738. doi: 10.1038/nprot.2010.5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Schuttelkopf AW, van Aalten DM. PRODRG: a tool for high-throughput crystallography of protein-ligand complexes. Acta crystallographica. Section D, Biological crystallography. 2004;60:1355–1363. doi: 10.1107/S0907444904011679. [DOI] [PubMed] [Google Scholar]
  41. Sundaramoorthy R, Micossi E, Alphey MS, Germain V, Bryce JH, Smith SM, Leonard GA, Hunter WN. The crystal structure of a plant 3-ketoacyl-CoA thiolase reveals the potential for redox control of peroxisomal fatty acid beta-oxidation. Journal of Molecular Biology. 2006;359:347–357. doi: 10.1016/j.jmb.2006.03.032. [DOI] [PubMed] [Google Scholar]
  42. Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic acids research. 1994;22:4673–4680. doi: 10.1093/nar/22.22.4673. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Van der Geize R, Yam K, Heuser T, Wilbrink MH, Hara H, Anderton MC, Sim E, Dijkhuizen L, Davies JE, Mohn WW, et al. A gene cluster encoding cholesterol catabolism in a soil actinomycete provides insight into Mycobacterium tuberculosis survival in macrophages. Proceedings of the National Academy of Sciences of the United States of America (PNAS) 2007;104:6. doi: 10.1073/pnas.0605728104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. WHO . Global tuberculosis report 2013. World Health Organisation; 2013. [Google Scholar]
  45. Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Emsley P, Evans PR, Keegan RM, Krissinel EB, Leslie AG, McCoy A, et al. Overview of the CCP4 suite and current developments. Acta crystallographica. Section D, Biological crystallography. 2011;67:235–242. doi: 10.1107/S0907444910045749. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Wipperman MF, Sampson NS, Thomas ST. Pathogen roid rage: Cholesterol utilization by Mycobacterium tuberculosis. Critical reviews in biochemistry and molecular biology. 2014 doi: 10.3109/10409238.2014.895700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Yam KC, D'Angelo I, Kalscheuer R, Zhu H, Wang JX, Snieckus V, Ly LH, Converse PJ, Jacobs WR, Jr., Strynadka N, et al. Studies of a ring-cleaving dioxygenase illuminate the role of cholesterol metabolism in the pathogenesis of Mycobacterium tuberculosis. PLoS Pathog. 2009;5:e1000344. doi: 10.1371/journal.ppat.1000344. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Zhang Y. I-TASSER server for protein 3D structure prediction. BMC Bioinformatics. 2008;9:40. doi: 10.1186/1471-2105-9-40. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS641784-supplement.docx (19.6MB, docx)

RESOURCES