The Mechanisms of Human Hotdog-fold Thioesterase 2 (hTHEM2) Substrate Recognition and Catalysis Illuminated by a Structure and Function Based Analysis^,

Abstract

The focus of this paper is the hotdog-fold thioesterase THEM2 from human (hTHEM2; Swiss-Prot entry Q9NPJ3). In an earlier communication (Cheng, Z., Song, F., Shan, X., Wei, Z., Wang, Y., Dunaway-Mariano, D., and Gong, W. (2006) Crystal structure of human thioesterase superfamily member 2, Biochem Biophys Res Commun 349, 172-177.) we reported the apo crystal structure of hTHEM2. Herein, we report the results of an extensive hTHEM2 substrate screen, the structure determination of hTHEM2 complexed with the inert substrate analog undecan-2-one-CoA (in which O=C-CH₂-S substitutes for O=C-S) and the kinetic analysis of active site mutants. The work described in this paper represents the first reported structure-function based analysis of a human hotdog-fold thioesterase. The catalytic mechanism proposed involves the Asp65/Ser83 assisted attack of a water molecule at the Gly57/Asn50 polarized thioester C=O and the Asn50 assisted departure of the thiolate leaving group. Thioesterase activity was observed with acyl-CoAs but not with the human acyl-ACP or with an acyl-Cys peptide. The medium-to-long chain fatty acyl-CoAs displayed the smallest K_m values. The substrate specificity profile was analyzed within the context of the liganded enzyme to define the structural determinants of substrate recognition. Based on the results of this structure-function analysis we hypothesize that the physiological role of hHTEM2 involves catalysis of the hydrolysis of cytosolic medium-to-long chain acyl-CoA thioesters.

The focus of this paper is the single domain acyl-CoA hotdog fold thioesterase THEM2 from human (hTHEM2; Swiss-Prot entry Q9NPJ3). hTHEM2 occurrence is most predominant in the kidney (1, 2) and although its physiological function is not known, studies have shown that hTHEM2 is regulated by the hepatocyte nuclear factor 4alpha (3), which in turn might play an important role in renal cell carcinoma (4). In this investigation we use a structure-function based approach to characterize hTHEM2 as a catalyst for thioester hydrolysis.

The hTHEM2 three-dimensional structure and thioesterase activity are traits of the hotdog-fold thioesterase family. The structural unit of this family is a monomer that consists of a five or six-stranded β-sheet that wraps around a long α-helix, reminiscent of a hotdog. The association of two subunits generates an elongated β-sheet, along which the two α-helices run anti-parallel to one another. Each dimer contains two substrate-binding sites, located at the respective ends of the subunit interface. Two clades are distinguished by “back-to-back” (sheet-to-sheet) (prototype 4-hydroxybenzoyl-CoA (4-HBA-CoA) thioesterase from Arthrobacter sp strain SU (5); Figure 1A) vs. “face-to-face” (helix-to-helix) (prototype 4-hydroxybenzoyl-CoA thioesterase from Pseudomonas sp strain CBS3(6); Figure 1B) dimer-dimer association.

Figure 1.

(A) The Pymol generated stereoview of the back-to-back tetramer of Arthrobacter 4-HBA-CoA thioesterase (5) (left) and the dissected catalytic scaffold (right). (B) The Pymol generated stereoview of the front-to-front tetramer of Pseudomonas 4-HBA-CoA thioesterase (6) (left) and the dissected catalytic scaffold (right). In both figures the subunits are colored individually to show the dimer units (blue and salmon; cyan and magenta). The 4-hydroxy-phenyacyl ligands are shown in black. The catalytic scaffold of the blue (protomer I) and salmon (protomer II) colored dimer is shown in red (the loop leading to the N-terminus of the polarizing α-helix) and green (the midsection of the polarizing α-helix). The catalytic scaffolds are shown in dissection to left in order to highlight the alignment of the catalytic carboxylate residue (yellow carbon atoms, red oxygen atoms) with the C=O carbon atom of the trunctated ligand (black) and the backbone amide NH of the helix N-terminal residue (white carbon atoms; blue nitrogen and red oxygen atoms) with the ligand C=O oxygen atom.

The catalytic scaffold of the hotdog-fold thioesterase is formed by the N-terminus of the elongated α-helix and its connecting loop from one subunit (protomer I), and the center segment of the elongated α-helix on the opposing subunit (protomer II). In Figure 1, these two segments are colored red and green, respectively. The backbone amide NH of the N-terminal residue of the α-helix in protomer I engages in hydrogen-bond interaction with the substrate thioester C=O, thereby orienting and polarizing the C=O for nucleophilic attack (see Figure 1). In the Pseudomonas 4-HBA-CoA thioesterase clade the catalytic carboxylate residue is positioned on this same segment (Figure 1B) whereas in the Arthrobacter 4-HBA-CoA thioesterase clade it is positioned on the α-helix of protomer II (Figure 1A).

In an earlier communication (7) we reported the apo crystal structure of hTHEM2, which revealed that it is a member of the Arthrobacter 4-HBA-CoA thioesterase clade. Limited activity screening showed that hTHEM2 possesses acyl-CoA thioesterase activity (6), however the identity of its physiological substrate and the mechanism by which it catalyzes the hydrolysis of this substrate remained to be defined. Herein, we report the findings from a structure-function analysis which integrates the results obtained from an extensive hTHEM2 substrate screen, structure determination of hTHEM2 complexed with the inert substrate analog undecan-2-one-CoA (in which O=C-CH₂-S substitutes for O=C-S) and kinetic analysis of hTHEM2 site-directed mutants.

Experimental

Commercial Materials

All restriction enzymes and T4 DNA ligase were purchased from Invitrogen (Carlsbad, CA). Pfu Turbo DNA polymerase was purchased from Strategene (Cedar Creek, TX). Oligonucleotide primers were custom-synthesized by Invitrogen. Competent Escherichia coli cells (BL21-DE3) were obtained from Strategene and the pET22b vector was purchased from QIAGEN (Valencio, CA). DNA sequence determination was performed by the DNA Sequencing Facility of the University of New Mexico. Acetyl-CoA, 2-butenoyl-CoA, n-propionyl-CoA, hexanoyl-CoA, methylmalnoyl-CoA, ß-methylcrotonyl-CoA, tiglyl-CoA, linoleoyl-CoA, n-butyryl-CoA, ß-hydroxybutyryl-CoA, n-octanoyl-CoA, n-palmitoyl-CoA, glutaryl-CoA, malonyl-CoA, crotonyl-CoA, DL-3-hydroxy-3-methylglutaryl-CoA, palmitoleoyl-CoA, myristoyl-CoA, decanoyl-CoA, oleoyl-CoA, lauroyl-CoA, stearoyl-CoA, phenylacetyl-CoA and arachidonoyl-CoA were purchased from Sigma-Aldrich (St. Louis, MO). The tripeptide NAc-Gly-Ala-Cys-S-palmitoyl was custom synthesized by EZBiolab (Westfield, IN). All other chemicals were purchased from Sigma-Aldrich except where noted.

Synthetic Substrates and Inhibitors

3-Hydroxyphenylacetyl-CoA, 3,4-dihydroxyphenylacetyl-CoA, 3,5-dihroxyphenacetyl-CoA, 4-hydroxyphenylacetyl-CoA, 3-hydroxybenzoyl-CoA, 4-chlorobenzoyl-CoA, 4-hydroxybenzoyl-CoA, 4-hydroxybenzyl-CoA, 4-hydroxyphenylacyl-CoA, phenylacetylmethyl-CoA were synthesized as previously reported. (6, 8-11)

Undecan-2-one-CoA was synthesized from bromoundecan-2-one (12) and CoA as follows. Bromoundecan-2-one (108 mg) was dissolved in ethanol (3 ml) and benzene (400 μl) and then added dropwise to a stirred solution of 68 mg of coenzyme A lithium salt in 3.5 ml deionized H₂O. The mixture was stirred for 16 h while maintaining the pH between 7 and 8 by the addition of 1.0 M LiOH in deionized water. The mixture was extracted three times with ethyl acetate. The aqueous fraction was concentrated by lyophilization and then chromatographed on a Sephadex-G15 (Amersham Pharmacia) column with deionized water serving as the eluant. The column factions were analyzed by reversed phase HPLC, and the desired fractions were combined to give pure undecan-2-one in 13 % yield. MW calculated: 935.2670; observed:[M^-] 934.2667. ¹H-NMR (D₂O solvent) ppm: 0.86 (s, 3H), 0.89 (t, 3H), 0.99 (s, 3H), 1.27 (m, 14H), 1.58 (m, 2H), 2.68 (m, 3H), 3.24 (q, 1H), 3.38 (t, 2H), 3.57 (m, 3H), 3.64 (m, 1H), 3.91 (m, 1H), 4.09 (s, 1H), 4.31 (s, 1H), 4.66 (s, 1H), 6.28 (d, 1H), 8.47 (s, 1H), 8.72 (s, 1H).

Holo-acyl carrier proteins (ACP) were synthesized by reacting the apo human cytosolic ACP with acyl-CoAs catalyzed by the human phosphopantetheinyl transferase (PPTase). Human cytosolic ACP fragment and PPTase plasmids were a kind gift from Dr. Stuart Smith, Children's Hospital Oakland Research Institute, Oakland, CA. The two recombinant proteins were purified as described earlier (13, 14). The reaction mixture contained 60 μM apo-ACP, 150 μM acyl-CoA and 2 μM PPTase in 20 ml of 20 mM Tris/1mM MgCl₂ (pH 7.0; 25 °C) for 3 h. The concentrated acyl-ACP was prepared using a PALL 10K centrifuge A device and the structure was verified by ES-MS. Oleoyl-ACP calculated: 13673, observed [M⁺] 13673, lauroyl-ACP calculated: 13597, observed [M⁺] 13598, octanoyl-ACP calculated: 13534, observed [M⁺] 13534, palmitoyl-ACP calculated: 13646, observed: [M⁺] 13647, palmitoeyl-ACP calculated: 13644, observed: [M⁺] 13645, myristoyl-ACP calculated: 13618, observed: [M⁺] 13619, steroyl-ACP calculated: 13678, observed: [M⁺] 13679.

Preparation of recombinant wild-type and mutant hTHEM

The C-terminal His₆-tagged wild-type hTHEM2 (Swiss-Prot entry Q9NPJ3) was prepared as previously described (7). Site directed mutagenesis was carried out using a PCR-based strategy with the WT-hTHEM2/pET-22b(+) plasmid as template, commercial primers (Invitrogen), the PCR kit supplied by Stratagene, and the Techgene thermal cycler manufactured by TECHNE (Princeton, NJ). The PCR products were used to transform competent E. coli BL21(DE3) cells and the plasmid was prepared using a QIAprep Spin Miniprep Kit (Qiagen). The sequence of the mutated gene was confirmed by DNA sequencing carried out by the Health Sciences Center of the University of New Mexico. The hTHEM2 mutants were purified to homogeneity (as shown by SDS–PAGE analysis) in a yield of ∼ 4 mg protein/gm wet cells by the same procedure (7) used to purify the wild-type enzyme.

Enzymatic assays

Steady-state kinetic methods were used to determine the k_cat and K_m for wild-type and mutant hTHEM2. The thioesterase activity was monitored at 412 nm (Δε = 13.6 mM^-1 cm^-1) by coupling the reaction of 5,5′-dithio-bis(2-nitrobenzoic acid) (DTNB) with the CoA liberated from the acyl-CoA substrate. All reactions were carried out in 50 mM K⁺HEPES containing 1 mM DTNB at pH 7.5 and 25 °C.

The initial velocity data, measured as a function of substrate concentration, were analyzed using Eq. (1) and the computer program KinetAsyst (IntelliKinetics, PA).

$$\text{V}={\mathrm{V}}_{\text{max}}[\text{S}]/([\text{S}]+{\text{K}}_{\mathrm{m}})$$ (Eq.1)

V, initial velocity; V_max, maximum velocity; [S], substrate concentration; and K_m, Michaelis constant. The k_cat was calculated from V_max/[E] where [E] is the total enzyme concentration determined using the Bradford method (15).

The inhibition constant K_i for competitive inhibition was obtained by fitting the initial velocities of reactions of 3-hydroxyphenyacetyl-CoA (at varied concentration) in the absence or presence of inhibitor (at 1 × K_i and 2 × K_i concentration) to equation 2,

$$\text{V}={\mathrm{V}}_{\text{max}}[\text{S}]/({\text{K}}_{\mathrm{m}}(1+[\text{I}]/{\text{K}}_{\mathrm{i}}+[\text{S}])$$ (Eq.2)

Where [I] is the concentration of the inhibitor and K_i is the inhibition constant.

The hTHEM2 catalyzed reactions of the acyl-ACPs were monitored by ES-MS (palmitoeyl-ACP and myristoyl-ACP) or by the DTNB based spectrophotometric assay. Reaction solutions monitored at 412 nm initially contained 2 μM hHTEM2, 5-100 μM acyl-ACP, 1 mM DTNB in 50 mM K⁺HEPES at pH 7.5 and 25 °C. Reaction solutions analyzed by ES-MS following incubation for 60 min initially contained 10 μM hHTEM2 and 80 μM palmitoeyl-ACP or myristoyl-ACP in 50 mM K⁺HEPES at pH 7.5 and 25 °C. The palmitoeyl-ACP ([M⁺] 13645 Da) or myristoyl-ACP ([M⁺] 13619 Da) signal observed in the control solution (lacking hHTEM2) were also present the spectra measured for reaction solutions. The signal for the product ACP on the other hand, was not observed.

The hTHEM2 catalyzed reaction of the tripeptide NAc-Gly-Ala-Cys-S-palmitoyl was monitored at 412 nm for 60 min using a reaction solution initially containing 10 μM hTHEM2 and 100 μM NAc-Gly-Ala-Cys-S-palmitoyl in 5 mM K⁺HEPES at pH 7.5 and 25 °C. In addition, ES-MS mass spec was used to analyze the reaction mixture initially containing 10 μM hHTEM2 and 100 μM NAc-Gly-Ala-Cys-S-palmitoyl in 5 mM K⁺HEPES at pH 7.5 and 25 °C following a 60 min incubation period. The ([M+Na⁺] = 551 and [M+K⁺] = 567 Da) signals observed for the acylated peptide in the control reaction (lacking hHTEM2) were also observed in the spectrum of the reaction solution, however the signals ([M+Na⁺] = 312 and [M+K⁺] = 328 Da) for the deacylated peptide were not observed.

hTHEM2 Crystallization and Structure Determination

The hTHEM2(undecan-2-one-CoA) complex was crystallized by using the hanging drop vapor diffusion method. The drops were set up by mixing 1 μl of solution containing 10 mg/ml hTHEM2, 2 mM undecan-2-one-CoA in 20 mM Tris (pH 7.5), 50 mM NaCl and 10 mM DTT and 1 μl of solution containing 15% PEG3350, 0.1 M ammonium citrate, dibasic, and 5-10% glycerol. A cloudy mixture initially formed, however after 1 wk crystals having the typical dimensions of 0.3 × 0.3 × 0.3 mm³ were observed. The crystals were removed using nylon cryo-loops and immediately flash-frozen at 100K in a stream of nitrogen gas (the glycerol served as the cryoprotectant). A 1.7 Å data set was collected on a FR-E+ Cu rotating-anode generator (Rigaku, Japan), equipped with a R-AXIS IV⁺⁺ imaging-plate detector. The data set was processed with HKL2000 (16). In order to determine the structure of hTHEM2(undecan-2-one-CoA) complex, molecular replacement was performed by MOLREP (17), using the tetrameric hTHEM2 structure (PDB accession ID 2F0X) as the search model. After manual adjustment in Coot (18), the structure was refined using isotropic B-factor refinement in the REFMAC5 (19) to 1.7 Å with R_work of 19.8% and R_free of 23.6%. The statistics of data reduction and structure refinement are summarized in Table 1. The stereochemical quality of the final model was examined by PROCHECK (20). The existence of 7 chloride ions was confirmed by an anomalous difference Fourier map constructed using the anomalous difference signal and phases calculated from protein only. The densities corresponding to the chloride ions and undecan-2-one-CoA sulfur atoms were observed in the map, as these two atoms have similar anomalous signals at the wavelength of Cu K-alpha radiation (f″-Cl = 0.7e, f″-S = 0.6e). The fo-fc omit map was constructed using phases calculated based on the hTHEM2(undecan-2-one-CoA) complex structure, in the absence of undecan-2-one ligand M, chloride ion, water ligand, using FFT (21). The decanoly-CoA ligand was modeled in place of the undecan-2-one ligand M and chloride ion by using Coot (18).

Table 1.

Statistics of data reduction and structure refinement.

Data collection statistics
Space group	P212121
Unit cell (Å)	a=91.0, b=110.6, c=119.8
Resolution range (Å)	50.00-1.7 (1.76-1.7)a
No. of unique reflections	134034
Average redundancy	4.5 (4.2)
Completeness	99.1(99.2)
Rmerge(%)b	5.0(50.2)
I/σ (I)	23.1(2.7)
Refinement statistics
Rcryst/ Rfree (%)c	19.8/23.6
R.m.s.d Bond length (Å)d	0.015
R.m.s.d Bond angle (°)	2.0
Number of protein residues	1098
Number of H2O molecules	863
Number of PEG fragments	4
Number of Cl-	7
Average temperature factor (Å2)
Protein main-chain atoms	25.0
Protein side-chains atoms	26.9
Inhibitor atoms	43.0
PEG stoms	66.0
Cl ions	31.0
Water molecules	40.6

^aData for the highest resolution bin is in parentheses.

^bRmerge = Σ|I_i - I_m|/ΣI_i, where I_i is the intensity of the measured reflection and I_m is the mean intensity of all symmetry-related reflections.

^cR_cryst = Σ‖F_obs| - |F_calc‖/Σ|F_obs|, where Fobs and F_calc are observed and calculated structure factors.

R_free = Σ_T‖F_obs| - |F_calc‖/Σ_T|F_obs|, where T is a test data set of about 10% of the total reflections randomly chosen and set aside prior to refinement.

^dR.m.s.d = Root-mean-square deviations

The atomic coordinates of hTHEM2(undecan-2-one-CoA) complex have been deposited into the Protein Data Bank (PDB accession code 35FO).

Results and Discussion

Overall structure of the hTHEM2(undecan-2-one-CoA) complex

Global

The structure of this complex was solved and refined to 1.7 Å resolution (Table 1). The C-terminus (residue 140 and the His₆ tag) is disordered in each subunit. Within one asymmetric unit, there are 2 tetramers of hTHEM2 (Figure 2A, subunits are labeled A-H). Eight undecan-2-one-CoA ligands (colored blue and labeled I-P in Figure 2A) are bound to the respective substrate-binding sites. Seven chloride ions (colored green in Figure 2A) are bound to seven of the eight subunits. For the eighth subunit, a water molecule is observed in place of the chloride ion (colored magenta in Figure 2A). The chloride ions were identified by the anomalous difference Fourier map (see Figure SI1; Supporting Information). Two subunits of each tetramer are bound with a PEG molecule (colored red in Figure 2A). Approximately five oxyethylene units of the PEG ligand are observed in a surface depression near the rear exit of the undecan-2-one-CoA binding site. The PEG and chloride ligands were acquired from the crystallization solution that contained 15% PEG3350 in the reservoir solution and 50 mM NaCl in the protein solution. The effect of NaCl on the k_cat and K_m of hHTEM2 catalyzed hydrolysis of 3-hydroxyphenylacetyl-CoA was measured over a NaCl concentration range of 0 to 100 mM. The results (Table SI1; Supporting information) showed that neither kinetic constant is altered. Likewise, there is no difference in the undecan-2-one-CoA inhibition constant K_i measured in the absence of NaCl (K_i = 3.0 ± 0.2 μM) and in the presence of 100 mM NaCl (K_i = 3.3 ± 0.3 μM) (Table 2). We therefore conclude that the presence of the chloride ion in the active site is not pertinent to the catalytic function of the enzyme or to the binding affinity of the undecan-2-one-CoA ligand. On the other hand, as will be described in the following section, the incidental chloride ion assumes the position in the active site that we anticipate would otherwise be occupied by the oxygen atom of the undecan-2-one O=C-CH₂-S unit. The presence of the chloride ion might be, at least in part, responsible for the orientation of the O=C-CH₂-S unit observed for undecan-2-one-CoA ligands J-P.

Figure 2.

(A). The asymmetric unit of hTHEM2 bound with undecan-2-one-CoA defining two tetramers colored gray (subunits A-D) and brown (subunits H-K) bound with undecan-2-one-CoA ligands (I-L and M-P, respectively) colored blue. The seven chloride ions are represented as green colored spheres and the water molecule that replaces the chloride ion in one subunit is represented as a red colored sphere. The PEG ligands are colored red. (B) The fo-fc electron density map of undecan-2-one-CoA ligand M. The densities of the map (shown in blue) are contoured at the level of 2 sigma. The undecan-2-one-CoA carbon atoms are colored cyan, the hTHEM2 carbon atoms are colored gray, and all oxygen, nitrogen and sulfur atoms are colored red, blue and yellow respectively. (C). The superposition of the eight undecan-2-one-CoA ligands. The atoms of these ligands are colored according to isotropic B values in order to illustrate the change in ordered structure as one proceeds from the nucleotide unit (red to orange to yellow: B factors 60 to 50 to 40) to the pantetheine arm to the O=C-CH₂S unit (green to blue: B factors 30 to 20) and progressively through the hydrocarbon tail. The average B value for an amino acid side chain is 27 (Table 1).

Table 2.

Steady-state constants for hTHEM2 wild-type catalyzed hydrolysis of acyl-CoA thioesters in 50 mM K⁺HEPES (pH 7.5 and 25 °C) and 1 mM DTNB. The error limits shown reflect the fit to a data set and do not represent the standard deviation calculated from repetitive determination.

Substrate	kcat (s-1)	Km(μM)	kcat/Km (M-1 s-1)
Short Chain Fatty Acyl-CoAs
n-Decanoyl-CoA	(1.6± 0.1) × 10-2	4.9 ± 0.6	3.3 × 103
n-Octanoyl-CoA	(4.7 ± 0.1) × 10-3	(2.6 ± 0.3) × 101	1.8× 102
n-Hexanoyl-CoA	(2.2 ± 0.1) × 10-2	(7.0 ± 0.6) × 101	3.1 × 102
n-Butyryl–CoA	(4.7± 0.5) × 10-2	(6.0 ± 0.1) × 102	7.8 × 101
2-Butenoyl-CoA	(1.4 ± 0.5) × 10-2	(2.2 ± 0.3) × 102	6.4 × 101
β-Methylcrotonyl-CoA	(2.3± 0.1) × 10-2	(1.2 ± 0.1) × 102	1.9 × 102
Crotonyl-CoA	(7.6 ± 0.2) × 10-3	(1.8 ± 0.1) × 102	4.7 × 102
Tiglyl-CoA	(2.0± 0.2) × 10-2	(2.5 ± 0.5) × 102	8.2 × 101
n-Propionyl-CoA	(3.6 ± 0.3) × 10-2	(1.9 ± 0.4) × 102	1.9 × 102
Acetyl-CoA	(5.4 ± 0.2) × 10-2	(2.9 ± 0.8) × 102	1.8 × 102
Medium-Long Chain Fatty Acyl-CoAs
Lauroyl-CoA	(1.9 ± 0.1) × 10-2	9.9 ± 0.8	1.9 × 103
Myristoyl-CoA (25 °C)	(2.0± 0.1) × 10-2	5 ± 1	4.0 × 103
(37 °C)	(7.0 ± 0.1) × 10-2	9 ± 2	7.7 × 103
n-Palmitoyl-CoA	(4.4 ± 0.1) × 10-3	(1.6 ± 0.1) × 101	2.8 × 102
Palmitoleoyl-CoA	(1.7± 0.1) × 10-2	9 ± 2	1.9 × 103
Stearoyl-CoA	(1.1 ± 0.1) × 10-2	(2.0 ± 0.4) × 101	5.5 × 102
Oleoyl-CoA	(1.1 ± 0.1) × 10-2	9 ± 1	1.3 × 103
Linoleoyl-CoA	(1.7± 0.1) × 10-2	(3.1 ± 0.4) × 102	5.4 × 101
Arachidonoyl-CoA	(2.2 ± 0.1) × 10-2	(2.0 ± 0.2) × 101	1.1 × 103
Aromatic Acyl-CoAs
3-Hydroxyphenylacetyl-CoA	1.4 ± 0.1	(4.1 ± 0.4) × 101	3.5 × 104
3,4-Dihydroxyphenylacetyl-CoA	(1.9 ± 0.1) × 10-1	(1.0 ± 0.1) × 101	1.8 × 104
3,5-Dihydroxyphenylacetyl-CoA	(1.9 ± 0.1) × 10-1	(2.5 ± 0.3) × 101	7.9 × 103
4-Hydroxyphenylacetyl-CoA	(1.4 ± 0.1) × 10-1	(4.9 ± 0.4) × 101	2.9 × 103
Phenylacetyl-CoA	(8.4 ± 0.6) × 10-2	(2.4 ± 0.3) × 102	3.5 × 102
3-Hydroxybenzoyl-CoA	(1.1 ± 0.1) × 10-2	(3.2 ± 0.2) × 101	3.5 × 102
4-Hydroxybenzoyl-CoA	No activity	-	-
4-Chlorobenzoyl-CoA	(3.2 ± 0.2) × 10-3	(1.3 ± 0.2) × 102	2.5 × 101
Polar Aliphatic Acyl-CoAs
β-Hydroxybutyryl-CoA	(1.8 ± 0.1) × 10-1	(1.1 ± 0.2) × 102	1.6 × 103
Glutaryl-CoA	(5.0 ± 0.2) × 10-2	(6.7 ± 0.5) × 102	7.5 × 102
3-Hydroxy-3-methylglutaryl-CoA	(7.9 ± 0.2) × 10-2	(2.9 ± 0.2) × 103	2.7 × 101
Malonyl-CoA	(6.6 ± 0.6) × 10-2	(2.8 ± 0.8) × 102	2.4× 102
Methylmalonyl-CoA	(8.3± 0.1) × 10-2	(2.1 ± 0.1) × 102	3.9 × 102
Fatty acyl-ACPS
Lauroyl-ACP	No activity
Myristoyl-ACP	No activity
Palmitoleoyl-ACP	No activity
Octanoyl-ACP	No activity
Oleoyl-ACP	No activity
Acylated Cys Peptide
AC-GAC(Palmitoylated)-NH2	No activity
Inhibitors	Ki(μM)
Undecan-2-one-CoA 0 mM NaCl	3.0 ± 0.2
100 mM Nacl	3.3 ± 0.3
Phenylacetylmethyl-CoA	500 ±100
DesulfoCoA	9 ± 1

The two hTHEM2 tetramers of the asymmetric unit are identical, and are formed by the back-to-back (i.e., sheet-to-sheet) association of dimer 1 (formed by subunits A and B or E and F) and dimer 2 (formed by subunits C and D or G and H) (Figure 2A). The subunit organization observed for the apo hTHEM2 (7) and undecan-2-one-CoA bound hTHEM2 are identical as is the subunit fold (RMSD of 0.603 for the alpha carbon atoms of the superimposed apo and liganded tetramers) (Figure SI2; Supporting Information).

The undecan-2-one-CoA ligands

The electron density map of undecan-2-one-CoA ligand M bound at the interface of subunits D and E is presented in Figure 2B. Of the eight undecan-2-one-CoA ligands, only ligand I (bound at the interface of subunits A and B) shows poor electron density. An overlay of the eight undecan-2-one-CoA ligands is shown in Figure 2C. The atoms of each ligand are colored according to their isotropic B factors to illustrate that the atoms of the pantetheine arm and the O=C-CH₂S unit are highly ordered, whereas the atoms of nucleotide and decanyl units are significantly less ordered than the atoms of the surrounding amino acid side chains.

The substrate binding site

The undecan-2-one-CoA (K_i = 3 μM; Table 2) ligands mark the locations of the four substrate-binding sites of each tetramer (Figure 2A). A comparison of the liganded hTHEM2 tetramer to that of the Arthrobacter 4-HBA-CoA thioesterase tetramer bound with the inert substrate analog 4-hydroxyphenacyl-CoA (5) reveals the conservation of the substrate binding mode between these two thioesterases (see Figures 3A and 1A). Although of the same clade, hTHEM2 does not share significant sequence identity with the Arthrobacter 4-HBA-CoA thioesterase.

Figure 3.

(A) The Pymol generated stereoview of the back-to-back tetramer of hTHEM2 bound with undecan-2-one-CoA for illustration of subunits as functional protomers and the location of the catalytic scaffold (as described in the Introduction section and illustrated for the prototypes of the two hotdog-fold thioesterase clades in Figure 1). The subunits are colored individually to show the dimer units (blue and salmon; cyan and magenta). The undecan-2-one-CoA ligands are shown in black. The catalytic scaffold of the blue (protomer I) and salmon (protomer II) colored dimer is shown in red (the loop leading to the N-terminus of the polarizing helix) and green (the midsection of the polarizing helix). The location of the protomer III loop (residues 108-113) that interacts with the CoA pyrophosphate moiety of the undecan-2-one-CoA ligand bound of the opposing dimer is identified by * (B) The Pymol generated stereoview of the hHTEM2 active site showing the amino acid residues that immediately surround the undecan-2-one-CoA ligand M. The ligand carbon atoms are colored cyan, the carbon atoms of the amino acid residues of the catalytic site are colored green, and all other carbon atoms are colored gray. The oxygen atoms are colored red, nitrogen atoms blue and sulfur atoms yellow. (C) The Pymol generated stereoview of the zoom-in of hHTEM2 active site showing the amino acid residues that immediately surround the undecan-2-one-CoA ligand M (left) and the modeled decanoyl-CoA ligand (right). The coloring scheme used is the same as that described in (B). (D) The Pymol generated stereoview of the superimposed active site residues of hHTEM2 (cyan), Arthrobacter 4-HBA-CoA thioesterase (16) (green), and the E. coli PaaI (23) (magenta). The figure was created by first super-positioning the three structures using the N-terminus of the polarizing helix to set the structural alignment.

The undecan-2-one-CoA ligand is bound at the interface formed by three adjacent hTHEM2 subunits, two (protomers I and II) of which belong to the same dimer and the third (protomer III) to the opposing dimer (Figure 3A and Figure SI3; Supporting information). The catalytic scaffold consists of the connecting loop and N-terminal residue of the α-helix of protomer I (colored red in Figure 3A) and the midsection of the α-helix of protomer II (colored green in Figure 3A). The scaffold is very similar to that observed in the Arthrobacter 4-HBA-CoA thioesterase (5) (see Figure 1A).

The adenine 3′-monophosphate-5′-pyrophosphate unit binds to the protein surface at the entrance of the active site, whereas the pantetheine arm and acyl tail snake through a channel formed at the interface of protomers I and II. The residues that surround the undecan-2-one-CoA ligand M are illustrated in Figure 3B. The zoom-in snap shot of the catalytic site (Figure 3C) reveals that O=CCH₂S unit is “flipped” away from the N-terminal residue (Gly47) of the protomer I α-helix N-terminus, and consequently the formation of a hydrogen bond between the O=C and the Gly57 backbone amide NH is not possible. The chloride ion, which occupies the site that we would anticipate to be occupied by the O=C for hydrogen bond formation with the Gly57 backbone amide NH, might promote this particular binding mode. The undecan-2-one-CoA ligand I is bound to the one active site that does not contain the chloride ion (Figure 2A). The O=C-CH₂S unit of this ligand appears to be rotated in such a way as to bring the oxygen atom in closer contact with the Gly57 backbone amide NH (Figure SI4).

The substrate model

Whereas the ligand O=C-CH₂S is misaligned with the N-terminal residue (G57) of the protomer I polarizing α-helix, it is aligned with the Lys136 of protomer II (Figure 3C). Because Ala replacement of Lys136 resulted in a relatively small perturbation in catalytic efficiency (k_cat reduced 4-fold and K_m increased 3-fold; Table 3) we contend that the orientation of the O=C-CH₂S is an artifact of crystallization. In order to obtain a “virtual” picture of the catalytic site in the enzyme-substrate complex, the decanoyl-CoA ligand was modeled in place of the undeac-2-one-CoA ligand M and the incidental chloride ion. The resulting model (see Figure 3B and 3C) shows the substrate thioester C=O to be within hydrogen bond distance of the Gly57 backbone amide and the Asn50 side chain, and the thioester sulfur atom to be within hydrogen bond distance of the Asn50 side chain. Otherwise, the substrate ligand conformation is identical to that of undecan-2-one-CoA ligand M (Figure 3B). An overlay of the undecan-2-one-CoA ligand M and the modeled decanoyl-CoA ligand is shown in Figure SI5 of Supporting Information.

Table 3.

Steady-state constants for hTHEM2 wild-type and mutants catalyzed hydrolysis of 3-hydroxyphenylacetyl-CoA in 50 mM K⁺HEPES (pH 7.5 and 25 °C) and 1 mM DTNB. The error limits shown reflect the fit to a data set and do not represent the standard deviation calculated from repetitive determination.

Enzyme	kcat (s-1)	Km (μM)	kcat/Km (M-1s-1)
WT	1.4 ± 0.1	(4.1 ± 0.4) × 101	3.5 × 104
D65A	<1.0× 10-5	-	-
D65N	(3.6 ± 0.1) × 10-1	(3.6 ± 0.3) × 102	1.0 × 103
D65E	(1.6 ± 0.1) × 10-1	(2.3 ± 0.3) × 102	6.7 × 102
H56A	(6.2 ± 2) × 10-1	(2.6 ± 0.2) × 101	2.4 × 104
S83A	(1.5 ± 0.1) × 10-1	(3.1 ± 0.5) × 102	5.8 × 102
T69A	(6.6 ± 0.4) × 10-1	(1.6 ± 0.2) × 102	4.1 × 103
H134A	(1.6 ± 0.1) × 10-1	(6.2 ± 0.4) × 102	1.7 × 102
N50A	(7.0 ± 1) × 10-3	(3.0± 0.5) × 102	2.3 × 101
K136A	(3.2 ± 0.1) × 10-1	(1.2 ± 0.1) × 102	2.6 × 103
H137A	(8.3 ± 0.8) × 10-1	(9.0 ± 1.5) × 102	9.2 × 102

The hHTEM2 Catalytic Mechanism

The active site residues of special interest include Asp65, Gly57, Asn50, Ser83, His56 and Thr69. Each residue is conserved among the hTHEM2 eukaryotic sequence homologs having at least 42 % sequence identity with hTHEM2 (the sequence alignment is shown Figure SI6; Supporting Information). The counter parts to these residues in the like-clade members Arthrobacter 4-HBA-CoA thioesterase (5) and E. coli phenylacetyl-CoA thioesterase (PaaI) (22) are shown in Figure 3D. Gly57 is conserved as Gly65 in Arthrobacter 4-HBA-CoA thioesterase and as Gly53 in E. coli PaaI. The hTHEM2 Asp65 and Asn50 are conserved as Asp61 and Asn46 in the E. coli PaaI and are conservatively substituted with Glu73 and Gln48 in the Arthrobacter 4-HBA-CoA thioesterase. His56 is conserved as His64 in Arthrobacter 4-HBA-CoA thioesterase and as His52 in E. coli PaaI. Residues that are not conserved are Ser83, which is replaced by Gly93 in the Arthrobacter 4-HBA-CoA thioesterase and by Ala77 in E. coli PaaI and Thr69, which is conserved in Arthrobacter 4-HBA-CoA thioesterase as Thr71 but replaced by Ala65 in E. coli PaaI.

The Gly53 of Arthrobacter 4-HBA-CoA thioesterase orients and polarizes the substrate thioester C=O, whereas the Glu73 functions in nucleophilic catalysis, the Thr71 helps orient the water molecule that hydrolyzes the mixed anhydride intermediate, the Gln48 activates the CoA leaving group and the His52 participates in a hydrogen bond network which includes the substrate ring C(4)OH substituent (5, 23-24). Ala replacement of Arthrobacter 4-HBA-CoA thioesterase Thr71 resulted in a 3-fold reduction in k_cat and no change in K_m. The T69A hTHEM2 mutant k_cat is reduced 2-fold and the K_m is increased 4-fold (Table 3). The Thr69 of hTHEM2 forms a “long” hydrogen bond with the Asp65 carboxylate as does the His56. Ala replacement of the hTHEM2 His56 reduces the k_cat value 2-fold and increases the K_m value 6-fold (Table 3). In comparison, Ala replacement of His64 has a dramatic effect on the Arthrobacter 4-HBA-CoA thioesterase catalytic efficiency as indicated by a 4000-fold reduction in the k_cat value and a 500-fold increase in the K_m value. Ala replacement of the Arthrobacter 4-HBA-CoA thioesterase Gln58 reduces the k_cat value 53-fold without having a significant impact on the K_m value. Ala replacement of the hHTEM2 Asn50 has a more pronounced effect: the k_cat value is reduced 200-fold and the K_m value is increased 10-fold (Table 3).

The greatest deviation in the proposed mechanisms of hHTEM2 and Arthrobacter 4-HBA-CoA thioesterase catalysis is the roles of the respective catalytic carboxylate residues Asp65 and Glu73. Conservative replacement of the Glu73 with Gln results in a 2,000-fold reduction in the k_cat value, consistent with its suggested role in nucleophilic catalysis (5, 23-24). Whereas the substitution of the hTHEM2 Asp65 with Ala is not tolerated, conservative replacement with Asn or Glu results in relatively small reductions in the hTHEM2 turnover rate (the k_cat values of the D65N² and D65E mutants are reduced only 3-fold and 10-fold, respectively) (Table 3).³ This finding is consistent with a less demanding role for Asp65 wherein it collaborates with Ser83 to orient and activate the water nucleophile. The replacement of the hHTEM2 Ser83 with Ala reduces the k_cat value 10-fold and increases the K_m value 10-fold.

The catalytic mechanism that we propose for hTHEM2 catalysis is depicted in Figure 4. The reaction pathway is shown as a two-step sequence, however we do not dismiss the possibility that the acyl transfer is concerted. The key components of this mechanism are the orientation and polarization of the thioester C=O and the water nucleophile. The N-terminus residue of the polarizing α-helix of protomer I Gly57 is suggested collaborate with Asn50, also of protomer I, in binding and polarizing the substrate thioester C=O. The Asp65 and Ser83, both of protomer II, are believed to collaborate in the orientation and activation of the water nucleophile. Asn50 might facilitate the departure of the thiolate leaving group. The proposed mechanism is different from, and much less elaborate than, the general base catalysis mechanism proposed for the T. thermophilus PaaI (25). This enzyme is reported to employ an asymmetric substrate-induced fit mechanism. We found no structural evidence for substrate-induced fit operative in hTHEM2 (vide supra).

Figure 4.

A cartoon created in Chemdraw that depicts the proposed hTHEM2 catalytic mechanism. Although the reaction is shown to involve two chemical steps, we do not rule out the possibility of a concerted mechanism.

Structure-function Analysis of the hHTEM2 Acyl Binding Site

Substrate Screen

The acyl-CoAs listed (by structural category) in Table 2 were tested as substrates in order to define the hTHEM2 substrate range. Using the substrate k_cat/K_m value as the measure of substrate activity the results show that the medium-to-long chain fatty acyl-CoAs are preferred over short chain and very long chain fatty acyl-CoAs. With the exception of β-hydroxybutyryl-CoA, the small polar aliphatic acyl-CoAs are inferior substrates. Among the aromatic acyl-CoAs tested hydroxyphenylacetyl-CoA (but not phenylacetyl-CoA) is a good substrate. Because the k_cat values measured for all substrates (excluding 3-hydroxyphenylacetyl-CoA) are low, it is reasonable to assume that the substrate binding step is at thermodynamic equilibrium and therefore that the K_m value approximates the K_d value. Indeed, the K_m values measured for decanoyl-CoA and phenylacetyl-CoA agree with the K_i values measured for undecan-2-one-CoA and phenylacetylmethyl-CoA, respectively (Table 2). The K_m values reported in Table 2 range from 5 μM to 3 mM. The very short chain (nonpolar as well as polar) acyl-CoAs bind less tightly than do the medium to long chain fatty acyl-CoAs, which suggests that desolvation of the acyl chain contributes to favorable binding energy. The tight binding observed for the medium-to-long chain fatty acyl-CoAs is suggestive of structure based targeting of a class of molecules rather than the targeting of a specific metabolite.

Structure of the Acyl Binding Site

The hTHEM2 acyl binding site, formed by Ile52, Asn50, Ala51, Thr54, Leu55 and His56 of protomer I and by Met15, Glu22, Asn66, Thr69, Met70, Leu73, Pro80, Gly81 and Val 82 of protomer II, is a long, largely hydrophobic, opened-ended channel (see Figure 3B and Figure 5A). This accounts for why the medium-to-long chain fatty acyl-CoAs bind tightly. On the other hand, we are left to rationalize the observation that the turnover rates of the medium-to-long chain fatty acyl-CoAs are ∼100-fold slower than that measured for 3-hydroxyphenylacetyl-CoA, which we note is not a known human metabolite. The 170-fold tighter binding of undecan-2-one-CoA vs phenylacetylmethyl-CoA (Table 2) indicates that the linear aliphatic chain offers more favorable binding energy than does the aromatic ring and the k_cat measured phenylacetyl-CoA is comparable to those of the medium-to-long chain fatty acyl-CoAs.

Figure 5.

The Grasp generated representations of the steric and electrostatic topological features of the acyl-CoA binding sites of (A) hHTEM2 bound with undecan-2-one-CoA, (B) Arthrobacter 4-HBA-CoA thioesterase bound with 4-hydroxyphenacyl-CoA (16), (C) Pseudomonas 4-HBA-CoA thioesterase bound with 4-hydroxyphenacyl-CoA (17), and (D) H. influenza YciA bound bound with CoA (19). The electron poor surface is represented in blue color and the electron rich surface is represented in red color.

In order to explore why 3-hydroxyphenylacetyl-CoA is a superior substrate, a model of the hTHEM2(3-hydroxyphenylacetyl-CoA) complex (not shown) was generated which shows the potential for hydrogen bond interaction between the channel residue Glu22 and the substrate ring C(3)OH. The 10-fold higher turnover rate observed for the β-hydroxybutyryl-CoA compared to butyryl-CoA (Table 2) might also be attributed to interaction of the hydroxyl group with Glu22. It is conceivable that the hydroxyl groups of these two substrates help orient the thioester C=O in the active site and in this manner enhances the turnover rate. Given this, a β-hydroxy medium-to-long chain fatty acyl-CoA might prove to be a more active in vitro substrate for hTHEM2 than is the corresponding medium-to-long chain fatty acyl-CoA, however this has not been tested.⁴

Structure-function Analysis of the hHTEM2 CoA Binding Site

In this section we use structure-function analysis to address the question of whether hTHEM2 is specialized to distinguish between an acyl-CoA thioester, an acylated acyl carrier protein (acyl-ACP) and a protein acylated at a Cys residue.

Substrate Screen

The ACP of interest is the ACP domain of the human cytosol fatty acid synthase (FAS). The ACP domain has been prepared as a separate protein unit and charged to form the acyl-ACP using the human pantetheinephosphate transferase in conjunction with the corresponding acyl-CoA (13, 14). We prepared the fatty acyl-ACPs listed in Table 2 and tested them as substrates for hTHEM2 (see Material and Methods for details). Despite the high enzyme concentration and the extended incubation times used, the expected hydrolysis products were not observed. This finding indicates that hTHEM2 does not recognize fatty acyl-ACPs derived from the engineered (FAS) ACP domain as substrates.

For the purpose of probing the possibility that hTHEM2 might target a palmitoylated protein (6) in vivo the peptide Ac-GAC(palmitoylated)-NH₂ 1 was tested as substrate. Significantly, no activity was found.

These results, together with the substrate activities of the acyl-CoAs reported in the previous section, suggest that hTHEM2 requires CoA as the substrate thiol unit, which in turn demands that hTHEM2 has a CoA binding site. We tested this prediction by measuring the inhibition constant of the CoA analog desulfoCoA, in which the CH₂SH is replaced with CH₃. Because the activity assay is based on measuring CoA formation using DTNB, CoA itself could not be tested. The desulfoCoA was shown to be tight binding competitive inhibitor of hTHEM2 catalysis (K_i = 9 μM; Table 2).

The CoA Binding Site

Next, we examined the binding interactions with the CoA unit as observed from the structure of the hTHEM2(undecan-2-one-CoA) complex. A stereoview of the hTHEM2 CoA binding site residues is shown in Figure 3B and the topological and electrostatic features of the binding site are shown in Figure 5A. The CoA 3′-phosphate is within hydrogen bond distance (3.0 Å) of the protomer II His137 side chain. Ala replacement of this residue has a minimal effect on the k_cat, however the K_m is increased 460-fold (Table 3). His137 is conserved among vertebrate homologs (mammal, fish, frog), although in more distant homologs (tree, insect, nematode) it is replaced with Tyr or Ala (Figure SI6; Supporting Information). Direct binding interactions involving the hairpin loop (residues 108-113) of protomer III (see Figure 2A) include Lys108 ion pair formation (3.2 Å) and Lys111 backbone amide NH hydrogen bond formation (2.6 Å) with the 5′-α-phosphate (both Lys residues are conservatively replaced with Arg in some homologs) and Leu113 (3.1 Å; not conserved) and Thr112 (3.0 Å; not conserved) backbone amide NH hydrogen bond formation with the 5′-β-phosphate.

The isotropic B values which are high for the atoms of the undecan-2-one-CoA nucleotide unit, low for the atoms of the pantetheine unit and high again for the atoms of the acyl tail (Figure 2C) indicate that the pantetheine unit of the acyl-CoA substrate plays a key role in hTHEM2 substrate recognition. The hTHEM2 channel residues that engage pantetheine unit in hydrogen bond interaction include the backbone amide C=Os of residues Ser83 (3.1 Å to N(9)H; protomer I; stringently conserved), Tyr90 (3.1 Å, N(5)H; protomer II; conservatively replaced with Phe), and Met91 (2.9 Å, C(3)OH; protomer II; conservatively replaced with Leu). In addition, the pantetheine unit C(8)=O interacts with several water molecules that are in turn bound by protomer II residues Ser92 (conservatively replaced with Thr in some sequences), Leu55 (stringently conserved) and Thr54 (not conserved). Packing of the hydrocarbon moieties of the pantetheine unit with the hydrocarbon moieties of channel residues Ile52, Le55, Pro93, Lys108, Val82, Tyr112, Tyr90, and Leu113 (Figure 3A) is also strongly suggestive of electrostatic complementation between the binding site and the ligand.

The pantetheine unit can bridge the distance between the entrance to the active site (where the nucleotide is positioned) and the short tunnel in which the thioester C=O is positioned for reaction. In contrast, the Cys unit of the acylated peptide is not long enough to bridge this distance and therefore the inactivity of this thioester is easily accounted for. The FAS acyl-ACPs on the other hand do fulfill the distance requirement, yet the acyl-ACPs derived from the engineered ACP domain tested did not show substrate activity. This suggests that the topological and/or electrostatic surface of the ACP is not compatible with that of the hTHEM2. However, we cannot rule out the possibility that the acyl-ACP of the holo FAS is substrate active.

hTHEM2 Divergence of Structure and Function From its Bacterial Counterparts

The liganded structure of hTHEM2 provides the opportunity to compare its structural determinants of substrate recognition with those of its characterized bacterial counterparts. In Figure 5 the topological and electrostatic features that define the landscape of the hTHEM2 substrate-binding site are represented along with those of three bacterial hotdog-fold thioesterases. The Arthrobacter 4-HBA-CoA thioesterase (5) and the high activity, broad substrate range, cytosolic thioesterase YciA (26, 27) from H. influenza belong to the same clade as hTHEM2, whereas the Pseudomonas 4-HBA-CoA thioesterase (6, 28) belongs to the other clade. All three bacterial thioesterases catalyze the hydrolysis of their physiological acyl-CoA substrates with high catalytic efficiency (k_cat/K_m > 1 × 10⁶ M^-1s^-1) (26, 29-30). Structure determinations (5-6, 27) have shown that for each of these liganded bacterial thioesterases, as well as the liganded hTHEM2, the pantetheine unit of the CoA snakes through a narrow channel that leads to a short tunnel (where the thioester C=O binds) that in turn opens to the acyl-binding site (Figure 5). The CoA nucleotide binds to the protein surface (at the door to the channel), which can be best described as a canyon that is highly electropositive in the vicinity of the CoA nucleotide phosphoryl groups. Like hTHEM2, all three bacterial thioesterases bind CoA tightly and this tight binding appears to derive from favorable interaction (desolvation of the hydrophobic units and hydrogen bond formation with the polar units) with the pantetheine unit and from the electrostatic forces at protein surface that act on the nucleotide phosphate groups (27, 28).

The enclosed substrate binding pockets of the Arthrobacter and Pseudomonas 4-HBA-CoA thioesterases (Figure 5B and 5C) allow these enzymes to restrict their substrate range and thereby avoid unwanted hydrolysis of acyl-CoA metabolites. Owing to their “open” acyl-binding sites, the bacterial thioesterase YciA (26, 27) (Figure 5A) and hTHEM2 (Figure 5D) are less discriminatory. YciA is subject to strong CoA feedback inhibition (K_d ≪ 1 μM) but hTHEM2 is not.

hTHEM2 Function

Human hotdog-fold thioesterases have been linked to diseases that range from cancer, to inflammatory disease, to obesity (31-32). Although most of these enzymes are known to catalyze thioester hydrolysis in fatty acyl-CoAs, the connection between this activity and the physiology of the healthy vs. diseased human is elusive. On the cellular level, the biochemical role that a particular fatty acyl-CoA thioesterase might perform will depend in part on its location. hTHEM2 is cytosolic and therefore the fatty acyl-CoAs that it might encounter originate from fatty acids acquired by import across the plasma membrane or via biosynthesis on the cytoplasmic FAS. What is the purpose of hTHEM2 catalysis: to decrease the level fatty acyl-CoA, to recover the fatty acid or to recover the CoA? We do not yet know the answer to this question.

The combination of slow catalytic turnover coupled with tight substrate binding observed for hTHEM2 might the mechanism by which this thioesterase selects its medium-to-long chain fatty acyl-CoA substrates from the cytoplasmic metabolite pool. Alternatively, hTHEM2 might be activated by a protein partner. A recent report suggests that the phosphatidylcholine transfer protein (StarD2) binds to hTHEM2 and increases its thioesterase activity (2), yet we note that the 4-fold increase is seemingly insignificant, that the physical basis for formation of a complex between these two proteins is not obvious from inspection of their individual structures, and that a possible connection between the biochemical activities of these two proteins has not been made.

Conclusion

The work reported in this paper represents the first structure-function based analysis of a human hotdog-fold thioesterase. The mechanism of hydrolysis proposed involves the Asp65/Ser83 assisted attack of a water molecule at the Gly57/Asn50 polarized thioester C=O and the Asn50 assisted departure of the thiolate leaving group. The substrate specificity profile was analyzed within the context of the liganded enzyme and the ultimate conclusion that we reach is that hHTEM2 is, by structural design and by cellular location, set to target cytoplasmic medium-to-long chain acyl-CoA thioesters for hydrolysis. The most challenging objective, determination of the physiological function of hTHEM2, remains to be achieved.

Abbreviations used are

hTHEM2

human thioesterase superfamily member 2

4-HBA-CoA

4-hydroxybenzoyl-CoA

DTNB

5,5′-dithio-bis(2-nitrobenzoic acid)

HPLC

high performance liquid chromatography

ES-MS

electro spray mass spectroscopy

K⁺HEPES

potassium salt of N-2-hydroxyethylpiperzine-N′-ethanesulfonic acid

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

ACP

acyl carrier protein

PaaI

phenylacetyl-CoA thioesterase