In plants and bacteria that use a Type II fatty acid synthase, isozymes of acyl-acyl carrier protein (ACP) thioesterase (TE) hydrolyze the thioester bond of acyl-ACPs, terminating the process of fatty acid biosynthesis. These TEs are therefore critical in determining the fatty acid profiles produced by these organisms. Past characterizations of a limited number of plant-sourced acyl-ACP TEs have suggested a thiol-based, papain-like catalytic mechanism, involving a triad of Cys, His, and Asn residues. In the present study, the sequence alignment of 1019 plant and bacterial acyl-ACP TEs revealed that the previously proposed Cys catalytic residue is not universally conserved and therefore may not be a catalytic residue. Systematic mutagenesis of this residue to either Ser or Ala in three plant acyl-ACP TEs, CvFatB1 and CvFatB2 from Cuphea viscosissima and CnFatB2 from Cocos nucifera, resulted in enzymatically active variants, demonstrating that this Cys residue (Cys348 in CvFatB2) is not catalytic. In contrast, the multiple sequence alignment, together with the structure modeling of CvFatB2, suggests that the highly conserved Asp309 and Glu347, in addition to previously proposed Asn311 and His313, may be involved in catalysis. The substantial loss of catalytic competence associated with site-directed mutants at these positions confirmed the involvement of these residues in catalysis. By comparing the structures of acyl-ACP TE and the Pseudomonas 4-hydroxybenzoyl-CoA TE, both of which fold in the same hotdog tertiary structure and catalyze the hydrolysis reaction of thioester bond, we have proposed a two-step catalytic mechanism for acyl-ACP TE that involves an enzyme-bound anhydride intermediate.

Introduction

De novo fatty acid biosynthesis is an iterative metabolic process, commonly primed with the acetyl moiety from acetyl-CoA and extended by two carbon atoms per cycle, via reactions with malonyl-ACP. Plant and bacterial organisms use a type II FAS (fatty acid synthase) system, which integrates the capabilities of discrete enzymes that each catalyzes distinct reactions in the fatty acid elongation cycle [1]. Acyl-acyl carrier protein (ACP) thioesterase (TE) plays an essential role in the termination of acyl chain elongation in these de novo FAS systems [2,3] by catalyzing the hydrolysis of the thioester bond of the acyl-ACP to release the free fatty acid product of the system. Acyl-ACP TEs can exhibit diverse substrate specificities, and terminate fatty acid biosynthesis to produce saturated and unsaturated fatty acids ranging from 4- to 18-carbons in chain length [4]. These FatB-type TEs can be categorized into three classes: Class I acyl-ACP TEs act primarily on 14- and 16-carbon substrates, Class II acyl-ACP TEs have broad substrate specificities with major activities on 8- and 14-carbon substrates and Class III acyl-ACP TEs act predominantly on 8-carbon substrate. For example, acyl-ACP TE CvFatB2 from Cuphea viscosissima and CnFatB2 from Cocos nucifera are Class I TEs, while CvFatB1 from C. viscosissima is a Class III TE [4]. The Class I acyl-ACP TEs are recognized as the more ubiquitous type, which are responsible for producing the fatty acids incorporated into cell membranes, while the Class II and III TEs have evolved to produce medium chain fatty acids that accumulate in certain plant seed oils [3]. Because of the important role in determining fatty acid chain length, acyl-ACP TEs have been widely studied and used to modify the fatty acid composition of oil crops for improved nutritional value or for industrial uses [58].

These acyl-ACP TEs have been the targets of microbial metabolic engineering for the production of biorenewable chemicals [9,10]. This application has increased the demand for specific fatty acids across a broad range of chemical functionalities. As an essential enzyme that controls the amount and composition of fatty acid output from one of the most important metabolic pathways, there is fundamental significance to investigate the catalytic mechanism of acyl-ACP TE as part of an effort to understand the regulation of the FAS pathway. Understanding the catalytic mechanism of acyl-ACP TEs would be beneficial for the targeted engineering of a more efficient acyl-ACP TE. Although substrate specificities of many acyl-ACP TEs have been functionally characterized [4], the catalytic residues and catalytic mechanisms for these enzymes are still not well understood [11,12].

Several efforts have been made to identify the catalytic residues and decipher the catalytic mechanism shared by acyl-ACP TEs, specifically from plants. The sensitivity of plant acyl-ACP TEs to thiol inhibitors suggests that these enzymes employ a cysteine in catalysis [13,14]. Conserved cysteine and histidine have been proposed as catalytic residues based on a site-directed mutagenesis study [15]. A third catalytic residue, asparagine was also identified based on structure modeling and mutational studies [11]. Based on these results, a papain-like catalytic triad consisting of the amino acid residues, Asn, His, and Cys, was proposed for plant acyl-ACP TEs. More recently, based on the crystallographic structure of a California Bay Laurel (Umbellularia californica) acyl-ACP TE, an alternative two-step catalytic mechanism involving a mixed anhydride intermediate has been proposed [12].

In the past decade, the number of acyl-ACP TE sequences in public databases has exponentially increased, with the majority being from bacterial sources. These data indicate that the bacterial and plant acyl-ACP TEs share little sequence similarity [16], yet higher-order structural data indicate that both bacterial (PDB Accession 2ESS, 2OWN, 4GAK) and plant acyl-ACP TE (PDB 5X04) exhibit the same ‘hotdog' tertiary structure. With the availability of these additional sequence and structural information, in the present study we reinvestigated the catalytic residues of acyl-ACP TE. Alignment of plant and bacterial acyl-ACP TE sequences revealed several universally conserved residues. Combining with structural modeling of these enzymes, we predicted the catalytic residues for acyl-ACP TEs. Site-directed mutagenesis demonstrated that Asp309 and Glu347 in C. viscosissima acyl-ACP TE (CvFatB2; GenBank Accession AEM72523.1) are involved in catalysis, instead of the previously proposed Cys348 [15]. Based on the identification of these additional catalytic residues, a two-step catalytic mechanism that uses a mixed anhydride intermediate is proposed for the acyl-ACP TE-catalyzed reaction.

Experimental

Computational modeling and predictions

Multiple sequence alignment of 28 functionally characterized acyl-ACP TEs [4], including 16 plant and 12 bacterial sequences, was generated with PROMALS3D using the default parameters [17] and visualized with Jalview [18]. Multiple sequence alignment of a larger set of acyl-ACP TE sequences collected from the ThYme database (1019 sequences collected as of 15 October 2012) was performed with Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) using the default parameters.

The secondary structure of CvFatB2 was predicted using the PSIPRED Protein Structure Prediction Server (http://bioinf.cs.ucl.ac.uk/psipred/) [19]. The tertiary structure of CvFatB2 was predicted using I-TASSER (http://zhanglab.ccmb.med.umich.edu/I-TASSER/) with default parameters [20], and was viewed and analyzed with the PyMOL Molecular Graphics System, Version 1.5.0.4 (Schrödinger, LLC).

Site-directed mutagenesis

Three acyl-ACP TE coding cDNAs were previously cloned from C. nucifera (CnFatB2) and C. viscosissima (CvFatB1, and CvFatB2) [4]. The cDNA encoding the mature part of each protein (i.e. without the N-terminal chloroplast targeting peptide sequence) was codon-optimized for expression in Escherichia coli, chemically synthesized and cloned into a pUC57 vector such that expression was transcriptionally driven by the lacZ promoter [4]. Using these plasmids, mutations at specific conserved sites were generated with the QuikChange site-directed mutagenesis kit (Agilent Technologies, U.S.A.), following the manufacturer's instructions. Residues targeted for mutations were Asn321, His323, Glu357, and Cys358 in CnFatB2; Asn315, His317, Glu351, and Cys352 in CvFatB1; and Asp309, Asn311, His313, Glu347, and Cys348 in CvFatB2. For each acyl-ACP TE, Cys was mutated to Ser or Ala, Glu was mutated to Asp, Gln, or Ala, and both Asn and His were mutated to Ala. The Asp309 of CvFatB2 was mutated to Ala, Asn, and Glu. Authenticity of all mutants was confirmed by sequencing both strands of all constructs.

In vivo activities of acyl-ACP TE mutants

E. coli strain K27 (fadD88) contains a mutation in the fadD gene, which impairs β-oxidation of fatty acids and results in the accumulation of free fatty acids in the growth medium [21,22]. Each acyl-ACP TE was expressed in the K27 strain, and free fatty acids that accumulated in the medium were extracted and analyzed [4]. Fatty acid extracts from four independent colony isolates for each construct were independently analyzed via GC–MS. The total concentration of fatty acids produced by each acyl-ACP TE was obtained by subtracting the concentration of fatty acids produced by E. coli expressing a control plasmid (pUC57) that lacked an acyl-ACP TE sequence.

Determination of acyl-ACP TE protein expression level

Using purified CvFatB2 protein, polyclonal antibodies were produced in a mouse host at the Hybridoma Facility of Iowa State University (http://www.biotech.iastate.edu/facilities/hybridoma/). To assess the acyl-ACP TE protein expression level, soluble proteins were extracted from microbial cell pellets and analyzed by immuno-blotting. Cell pellets were resuspended and incubated for 30 min in Lysis Buffer (50 mM Tris–HCl pH 8.0, 150 mM NaCl, 10% glycerol, 0.6 mM phenylmethanesulfonyl fluoride, and 0.2 mg/ml lysozyme). Following sonication for 5 s, the suspension was subjected to centrifugation at 13 000×g for 5 min, and the soluble protein fraction was recovered in the supernatant. Protein concentration was quantified using the Bio-Rad Protein Assay Kit (Bio-Rad, U.S.A.). The expression level of the wild-type and point-mutant derivatives of CvFatB2 were quantified by first subjecting 35 µg of soluble crude protein extract to SDS–PAGE in a 12% polyacrylamide gel, and transferring the separated proteins to nitrocellulose membrane. Membranes were reacted with the primary CvFatB2 antibody at 1:2000 dilution, and the secondary antibody Goat-Anti-Mouse IgG (H + L)-HRP (Bio-Rad, U.S.A.) at 1:3000 dilution. Immuno-detection was performed using an enhanced chemifluorescence western-blotting detection kit according to the manufacturer's instructions (Thermo Scientific, U.S.A.) and visualized using a ChemiDoc™ XRS+ System (Bio-Rad, U.S.A.). Images were analyzed with the Image Lab™ Software (Bio-Rad, U.S.A.). The protein expression level for CvFatB2 in E. coli strain K27 was defined as 100% and the relative expression level of each mutant protein was calculated using the ratio of the band intensity of each mutant to the wild-type protein.

Purification of acyl-ACP TE mutant proteins

Each acyl-ACP TE was expressed in E. coli using a modified pET30b(+) vector, named pET30f. In this modified vector, expression was under the control of the T7 promoter, but the original His-tag, thrombin protease site, S-tag, and enterokinase protease sites present in pET30b(+) were replaced by a His-tag and a TEV protease recognition site. Each expression construct was confirmed by sequencing both strands of the vectors. Expression was conducted in the host, E. coli strain BL21 Star (DE3) (Life Technologies, U.S.A.). High-level expression of acyl-ACP TEs was initiated from a single colony inoculation of an overnight 5 ml of culture of LB medium supplemented with 50 mg/l kanamycin. This culture was used to inoculate a 1 l LB medium supplemented with 50 mg/l kanamycin and grown at 37°C. When the OD600 reached 0.5–0.7, isopropyl-β-d-galactopyranoside (IPTG) was added to a final concentration of 0.4 mM, and the incubation continued at 25°C for 16–20 h. The cells were harvested by centrifugation, and resuspended with Lysis Buffer containing 20 mM imidazole, followed by centrifugation at 10 000×g for 30 min. The supernatant was applied to a Ni-NTA column (5 Primer, U.S.A.), pre-equilibrated with buffer A (50 mM Tris–HCl pH 8.0, 150 mM NaCl, 10% glycerol) containing 20 mM imidazole. After successive washing of the column with buffer A supplemented with 20, 40, and 60 mM imidazole, the acyl-ACP TE protein was eluted with 250 mM imidazole in buffer A. An Amicon Ultra 10 kDa centrifugal filter (EMD Millipore, U.S.A.) was used for buffer exchange and concentration of each protein preparation.

CD spectra of acyl-ACP TE proteins

To assess the extent of folding of each purified protein, 0.2 mg/ml of protein, dissolved in 2 mM sodium phosphate buffer, pH 8.0, was subjected to far-UV CD spectrum analysis using a Jasco J-710 spectropolarimeter (Jasco, U.S.A.) at 25°C. The cell path length was 0.1 cm, and the spectra were an average of two scans collected at a speed of 50 nm/min and a step resolution of 0.2 nm. Protein secondary structure was predicted from CD spectra using the JFIT software [23] (B. Rupp; http://www.ruppweb.org/cd/cdtutorial.htm). Statistical analyses via χ2 tests were conducted to compare the secondary structures of mutant and wild-type proteins.

Results

Sequence alignment analysis of acyl-ACP TEs

Twenty-eight acyl-ACP TEs, including 16 plant TEs and 12 bacterial TEs, have previously been functionally characterized [4]. Comparison of the sequences of these acyl-ACP TEs indicates that overall they share 1.2% identity and 44% similarity. However, the region between residue positions 301 to 350 of CvFatB2, defined as the catalytic domain, shows higher conservation among these 28 proteins (86% similarity and 10% identity; Figure 1). This region is even more highly conserved among the 16 plant TEs (∼36% identity). These sequence alignments revealed six residues that are absolutely conserved among all 28 TE sequences, irrespective of whether they are sourced from plants or bacteria; these residues are Trp192, Asp309, Asn311, His313, Tyr319, and Glu347 (numbered relative to the sequence of CvFatB2, GenBank Accession AEM72523.1; Figure 1). Among these six conserved residues, five are located within the putative catalytic domain (Figure 1), which also contains the previously proposed active-site motifs of N311Q(K)HVN(S)N (Motif I) and Y343RR(K)EC(Q/T) (Motif II) of acyl-ACP TEs [15]. Motif I is conserved among all plant and bacterial sequences, whereas Motif II is conserved only among plant TEs.

A 50-amino acid portion of a multiple sequence alignment of 28 acyl-ACP TEs.

Figure 1.
A 50-amino acid portion of a multiple sequence alignment of 28 acyl-ACP TEs.

Two active-site motifs (Motif I and Motif II) are indicated by lines at the top of the alignment. Arrows identify the previously proposed catalytic residues, N311, H313, and C348 (numbered relative to the sequence of CvFatB2). Asterisks identify other residues that are conserved (D309, Y319, and E347) in this region. Another conserved residue W192 is not shown here.

Figure 1.
A 50-amino acid portion of a multiple sequence alignment of 28 acyl-ACP TEs.

Two active-site motifs (Motif I and Motif II) are indicated by lines at the top of the alignment. Arrows identify the previously proposed catalytic residues, N311, H313, and C348 (numbered relative to the sequence of CvFatB2). Asterisks identify other residues that are conserved (D309, Y319, and E347) in this region. Another conserved residue W192 is not shown here.

The catalytic residues of acyl-ACP TEs were previously proposed to be Asn311, His313, and Cys348 (numbered relative to the sequence of CvFatB2) [11]. However, when one compares the sequences of all 28 TE sequences in Figure 1, the Cys348 residue that is located in the proposed active-site Motif II was conserved only in plant TE sequences, whereas the Asn311 and His313 residues that are located in the proposed active-site Motif I are conserved among all 28 TEs aligned in Figure 1. Furthermore, no corresponding cysteine residue in this region could be identified in any of the bacterial acyl-ACP TE sequences analyzed herein. Interestingly, residue Glu347, which is located in the proposed active-site Motif II immediately adjacent to, and upstream of the proposed Cys catalytic residue, is conserved among all 28 TE sequences.

To further investigate the level of conservation of these potential catalytic residues (i.e. Trp192, Asp309, Asn311, His313, Tyr319, Glu347, and Cys348) among a wider collection of TEs, a total of 1019 sequences were collected from the ThYme database [16] and these were also aligned (Supplementary Table S1). Residues Trp192, Asp309, Asn311, His313, Tyr319, Glu347, and Cys348 (positions relative to CvFatB2) are conserved among 95.8, 99.8, 93.1, 99.9, 98.7, 85.9, and 21.8% of these sequences, respectively (Supplementary Table S1). We conclude therefore that the highly conserved residues, Trp192, Asp309, Asn311, His313, Tyr319, and Glu347, may play important structural or catalytic roles in TE functionality. The previously proposed catalytic residue, Cys348 is only conserved among 22% of the examined 1019 sequences, and this conservation is primarily restricted to plant-sourced enzymes, suggesting that its role in catalysis may not be direct.

Predicted structure of a plant acyl-ACP TE

Using the I-Tasser algorithm [20], we modeled the three-dimensional structure of the C. viscosissima acyl-ACP TE CvFatB2 using the crystallographically determined TE structure as a template, 2OWN from Lactobacillus plantarum. The resulting predicted structure is similar to a previous structural prediction of plant acyl-ACP TEs modeled using the more distant E. coli TEII (PDB IC8U) [11] and similar to the recently experimentally determined structure of a California Bay Laurel TE (PDB 5X04). These structures comprise two helix/4-stranded sheet domains in the canonical ‘hotdog' fold [11,16]. The two hotdog domains are linked by a long coil structure, and the two proposed active-site motifs, Motif I and Motif II, are located in the C-terminal hotdog structure on the edge of the cleft between the two hotdog domains.

The six highly conserved residues (Trp192, Asp309, Asn311, His313, Tyr319, and Glu347) identified in the multiple sequence alignment are similarly oriented in the L. plantarum TE structure and in the homology model of CvFatB2 (Figure 2). One of these residues, Trp192 in CvFatB2, that resides in the N-terminal hotdog domain and is far removed from the active site, has been recently shown to determine the cavity size that binds the acyl-moiety of the acyl-ACP substrate [24]. The other conserved residues (Asp309, Asn311, His313, Tyr319, and Glu347 in CvFatB2) are in the C-terminal hotdog domain. Among these residues, Tyr319 is located in the central α-helix and is distant from the proposed active site, indicating that it may be of structural but not of catalytic importance. In the predicted CvFatB2 structure (Figure 2), the previously proposed Cys348 catalytic residue [15] is oriented distantly from the other proposed catalytic residues, His313 and Asn311. Moreover, no analogous Cys residue can be identified that resides nearby His177 and Asn175 in the 2OWN tertiary structure from the L. plantarum TE (equivalent to His313 and Asn311 in the CvFatB2 sequence). Together, these data suggest that Cys348 may not be involved in TE catalysis. In contrast, residues Asp309 and Glu347, together with the previously proposed catalytic residues Asn311 and His313, are on a coil structure and cluster in close proximity to each other. We therefore suggest that these latter four residues may be directly involved in TE catalysis.

Comparison between the experimentally determined structure of L. plantarum acyl-ACP TE and the computationally modeled CvFatB2 structure.

Figure 2.
Comparison between the experimentally determined structure of L. plantarum acyl-ACP TE and the computationally modeled CvFatB2 structure.

Structural overlay of CvFatB2 model with crystal structure of L. plantarum acyl-ACP TE (2OWN) (A), and a magnified overlay view of the active sites (B). The CvFatB2 predicted structure and 2OWN structure are shown in magenta and blue, respectively. The conserved residues identified from the multiple sequence alignment of 1019 acyl-ACP TEs are shown as stick models (i.e. W192, D309, N311, H313, Y319, and E347 in CvFatB2). The imperfect alignment of the side chain conformations of these residues between the CvFatB2 model and the 2OWN structure may be attributable to the low sequence identity (26%) between the sequences. Residues N311, H313, and C348 in CvFatB2 were previously proposed as the catalytic residues. C348 orients distantly from the other two catalytic residues in the CvFatB2 model and no cysteine can be identified nearby in the active site structure of 2OWN.

Figure 2.
Comparison between the experimentally determined structure of L. plantarum acyl-ACP TE and the computationally modeled CvFatB2 structure.

Structural overlay of CvFatB2 model with crystal structure of L. plantarum acyl-ACP TE (2OWN) (A), and a magnified overlay view of the active sites (B). The CvFatB2 predicted structure and 2OWN structure are shown in magenta and blue, respectively. The conserved residues identified from the multiple sequence alignment of 1019 acyl-ACP TEs are shown as stick models (i.e. W192, D309, N311, H313, Y319, and E347 in CvFatB2). The imperfect alignment of the side chain conformations of these residues between the CvFatB2 model and the 2OWN structure may be attributable to the low sequence identity (26%) between the sequences. Residues N311, H313, and C348 in CvFatB2 were previously proposed as the catalytic residues. C348 orients distantly from the other two catalytic residues in the CvFatB2 model and no cysteine can be identified nearby in the active site structure of 2OWN.

Mutagenesis-based evaluation of potential catalytic residues of acyl-ACP TE

To verify the structure modeling and homology-based prediction that Asp309, Glu347, Asn311, and His313 are catalytic residues, and that Cys348 is not involved in catalysis, site-directed mutants at these positions were generated for the C. viscosissima TE CvFatB2. The in vivoactivity of each mutant was determined by expressing each protein in E. coli strain K27, and the free fatty acids that accumulated in the media were measured as a quantification of the catalytic activity of each TE. We initially evaluated mutants of Cys348, which, prior studies have suggested, may be part of the catalytic triad that catalyzes the hydrolysis of the thioester bond in acyl-ACP [15]. Replacing this Cys residue with either Ser or Ala (i.e. CvFatB2-C348S and -C348A) did not inactivate the enzyme. The wild-type enzyme CvFatB2 produced 108 µM of free fatty acids, while CvFatB2-C348A and C348S exhibited ∼60 and ∼130% of the wild-type activity, respectively (Figure 3A). In contrast, all mutants of Asp309, Asn311, His313, and Glu347 exhibited dramatically decreased activity. The activity of mutants E347D, E347Q, E347A, N311A, H313A, D309A, D309E, and D309N ranged from 0 to 7.6% of wild-type activity (Figure 3A). Statistical analysis (i.e. Student's t-test) indicates that the total free fatty acid produced by these mutants was not significantly different from the free fatty acid production of the control strain that did not express any acyl-ACP TE gene, suggesting these mutants completely lost activity.

In vivo activities of the wild-type and mutant CvFatB2 enzymes.

Figure 3.
In vivo activities of the wild-type and mutant CvFatB2 enzymes.

Total free fatty acid production (A) and normalized relative activities (B) of wild-type and mutant forms of CvFatB2. TE activity was determined as described in the Experimental section, by expressing each protein in E coli strain K27 and analyzing the free fatty acid accumulation in the culture. To compare the enzymatic activities of CvFatB2 mutants and wild type, the total fatty acid accumulation of each mutant was first normalized to the protein expression level (red line) and then normalized relative to the wild-type TE activity. The protein expression levels of wild-type and mutant TEs were assessed via western blot analyses (original western blot image is shown in Supplementary Figure S1). The protein expression level of CvFatB2 in E. coli strain K27 was defined as 100%, and the relative expression level of each mutant protein was calculated using the ratio of band intensity of mutant and wild-type proteins. All data were gathered from four replicates for each TE protein. Asterisks indicate statistically significant difference (Student's t-test P < 0.05) compared with wild-type activity, and arrow heads indicate no significant difference between the fatty acid production of TE mutants and control strain without TE.

Figure 3.
In vivo activities of the wild-type and mutant CvFatB2 enzymes.

Total free fatty acid production (A) and normalized relative activities (B) of wild-type and mutant forms of CvFatB2. TE activity was determined as described in the Experimental section, by expressing each protein in E coli strain K27 and analyzing the free fatty acid accumulation in the culture. To compare the enzymatic activities of CvFatB2 mutants and wild type, the total fatty acid accumulation of each mutant was first normalized to the protein expression level (red line) and then normalized relative to the wild-type TE activity. The protein expression levels of wild-type and mutant TEs were assessed via western blot analyses (original western blot image is shown in Supplementary Figure S1). The protein expression level of CvFatB2 in E. coli strain K27 was defined as 100%, and the relative expression level of each mutant protein was calculated using the ratio of band intensity of mutant and wild-type proteins. All data were gathered from four replicates for each TE protein. Asterisks indicate statistically significant difference (Student's t-test P < 0.05) compared with wild-type activity, and arrow heads indicate no significant difference between the fatty acid production of TE mutants and control strain without TE.

To ensure that the differences in fatty acid yields generated by the wild-type and mutant variants of CvFatB2 TE can be attributed to the individual mutations that disrupt active site function and not to differences in the levels of expression of the heterologous TEs, we evaluated the soluble protein expression level of all CvFatB2 TEs in E. coli strain K27. Specifically, total soluble protein extracts from each TE-expressing strain were assessed via western blot analysis using the anti-CvFatB2 antibody (Supplementary Figure S1). The expression of mutant CvFatB2-C348S was ∼65% of the wild-type level, while the expression of all other mutants was higher than the wild-type TE, in the range of 130 to 700% of the wild-type levels (Figure 3B and Supplementary Figure S1). Therefore, the total fatty acid production of CvFatB2 and its mutants was normalized to allow for this difference in the observed protein expression level. The normalized activity of C348A and C348S was therefore 50 and 200% of the wild-type enzyme, respectively.

Similarly, mutations of Asp309, Asn311, His313, and Glu347 resulted in the reduction in relative TE activity to less than 1% of wild-type activity (Figure 3B). Together, these data demonstrate that reduction in in vivo activities of the mutant enzymes is not attributable to different expression levels of mutant TE proteins, but these are due to the specific alterations in the ability of the protein to support catalysis, caused by mutations in individual residues.

To assess whether these mutations in CvFatB2 resulted in altered protein folding properties, far-UV CD spectra were obtained from the purified wild-type and all mutant CvFatB2 proteins. Comparing the spectrum of each mutant protein to the wild-type indicates no significant differences in the secondary structure between these proteins (Figure 4). The secondary structures of CvFatB2 and its mutants were calculated using the program, Jfit [23]. The wild-type CvFatB2 protein exhibited 11.9% α-helix, 24.0% β-sheet, and 64.1% coil structures, and each mutant protein shared a similar secondary structure profile that did not differ significantly from wild-type (χ2 test P > 0.1). These results indicate therefore that the loss of catalytic activity associated with the Asp309, Asn311, His313, and Glu347 mutants was not caused by incorrect folding of mutant proteins, but due to the specific changes of these active site residues and thus inactivation of the enzyme.

Far-UV CD spectra of CvFatB2 and its mutants.

Figure 4.
Far-UV CD spectra of CvFatB2 and its mutants.

The spectra of the site-directed mutants are nearly identical with the wild-type CvFatB2 protein, indicating that the mutations did not affect the overall fold of the proteins.

Figure 4.
Far-UV CD spectra of CvFatB2 and its mutants.

The spectra of the site-directed mutants are nearly identical with the wild-type CvFatB2 protein, indicating that the mutations did not affect the overall fold of the proteins.

The role of these specific residues in TE catalysis was further evaluated by mutating the equivalent residues in an additional two acyl-ACP TEs, C. nucifera TE CnFatB2, and C. viscosissima TE CvFatB1. Prior analyses of these TEs indicated that they each show different substrate specificities, with CnFatB2 hydrolyzing acyl-ACPs of 14- and 16-carbon fatty acids, and CvFatB1 hydrolyzing acyl-ACPs of 8- and 10-carbon fatty acids [4]. The wild-type enzymes CnFatB2 and CvFatB1, when expressed in E. coli K27, produced 211 and 715 µM of free fatty acids, respectively (Figure 5). Replacing the Cys residue previously proposed as having a catalytic role [15] (i.e. Cys358 in CnFatB2 and Cys352 in CvFatB1) with either Ser or Ala in both acyl-ACP TEs resulted in less than 30% reduction in activity, as compared with the wild-type enzymes (Figure 5). In contrast, mutations of other residues that we propose to be involved in catalysis (i.e. Asn321, His323, and Glu357 in CnFatB2 and Asn315, His317, and Glu351 in CvFatB1) caused dramatic decreases of the activity of these enzymes (Figure 5). These mutagenesis results with acyl-ACP TEs that display different substrate specificities are consistent with our conclusions reached with CvFatB2, supporting the inference that a Cys residue is not directly involved in catalysis of acyl-ACP TE, but the adjacent Glu may be one of the catalytic residues.

In vivo activities of the wild-type and mutant CnFatB2 and CvFatB1 enzymes.

Figure 5.
In vivo activities of the wild-type and mutant CnFatB2 and CvFatB1 enzymes.

Total free fatty acid production of site-directed mutants of CnFatB2 (A) and CvFatB1 (B). Free fatty acid production of these enzymes was determined as described in the Experimental section. Data were gathered from four replicates for each TE enzyme. Asterisks indicate statistically significant difference (Student's t-test P < 0.05) compared with wild-type activity, and arrow heads indicate no significant difference between the fatty acid production of TE mutants and control strain without TE.

Figure 5.
In vivo activities of the wild-type and mutant CnFatB2 and CvFatB1 enzymes.

Total free fatty acid production of site-directed mutants of CnFatB2 (A) and CvFatB1 (B). Free fatty acid production of these enzymes was determined as described in the Experimental section. Data were gathered from four replicates for each TE enzyme. Asterisks indicate statistically significant difference (Student's t-test P < 0.05) compared with wild-type activity, and arrow heads indicate no significant difference between the fatty acid production of TE mutants and control strain without TE.

Discussion

Confirmation that Cys348 is not the catalytic residue

The Asn-His-Cys catalytic triad model of the acyl-ACP TE catalytic mechanism was proposed at a time when only a few functionally characterized acyl-ACP TE sequences, primarily from plants, were available for a sequence-homology-based prediction [15]. More recently, a larger number of acyl-ACP TEs have been isolated and characterized from diverse plants and bacteria [4] and high-throughput sequencing projects have generated numerous putative acyl-ACP TE sequences from both plants and bacteria. In addition, several crystal structures of acyl-ACP TEs are now available (PDB Accession 2ESS, 2OWN, 4GAK, and 5X04). With this abundance of new sequence and structural data comes an increased ability to computationally predict the catalytic residues in acyl-ACP TEs, based upon both primary and tertiary structural modeling. In addition, the widely used approach of analyzing free fatty acid concentrations and distributions produced by the heterologous expression of TEs in E. coli K27 provides an efficient and fast method to study the activities of acyl-ACP TEs. Using multiple sequence alignments, we first identified that previously proposed catalytic Cys residue (Cys348 in CvFatB2) is not conserved. Subsequent site-directed mutagenesis experiments and activity analyses in E. coli K27 verified that this residue is not directly involved in the catalytic site of acyl-ACP TEs.

Replacing Cys348 with Ser resulted in a mutant that was still active as compared with the wild-type enzyme [15]. It may be possible that the Ser replacement may still support catalysis due to the ability of the hydroxyl group of the Ser residue to act as a nucleophile, as it does in serine proteases [25]. However, when Cys348 was mutated to Ala, which would be predicted to have no catalytic function, the mutant still retained more than 50% of activity, demonstrating that the Cys residue is not a member of the catalytic triad. This result, however, is in contrast with a previous report, where mutating Cys to Ala in the UcFatB1 TE totally inactivated the TE enzyme [15]. In this study, therefore, we analyzed Cys-to-Ala mutants with two additional enzymes, CnFatB2 and CvFatB1 and consistently observed that these mutants also retained high levels of catalytic activity. Collectively, these data therefore establish that acyl-ACP TEs do not use a papain-like protease mechanism.

The prior identification of a catalytic Cys residue was, in part, based on the observation that plant acyl-ACP TEs displayed sensitivity to thiol inhibitors, and the Cys to Ser replacement mutation converted the enzyme to being sensitive to serine hydroxyl-reactive reagents [13,14]. Specifically, 5,5′-dithiobis(2-nitrobenzoic acid) inhibited 92% of the UcFatB1 activity, iodoacetamide inhibited 72% of this activity, and phenylmethylsulfonyl fluoride showed 90% inhibition when the supposed catalytic Cys residue was replaced with a serine residue [15]. The inhibitory mechanisms of these reagents, however, involve the covalent modification of the residue side chains; 5,5′-dithiobis(2-nitrobenzoic acid) and iodoacetamide covalently react with the thiol of cysteine, and phenylmethylsulfonyl fluoride reacts with the hydroxyl group of serine. Our finding that the adjacent Glu residue (Glu347 in CvFatB2) may be the catalytic residue, instead of Cys348 can therefore provide an explanation for these prior chemical modification results. Namely, these bulky chemical modification reagents may force a conformational change of the adjacent Glu347 catalytic residue, or the modification blocks access of substrates into the active site, and thus inhibits the TE activity.

Identification of Asp309 and Glu347 as potential catalytic residues

Mutation of the residue Glu347 to either the similar acidic residue Asp, or a non-active residue, Ala resulted in almost complete loss of enzymatic activity, suggesting that Glu347 is a catalytic residue. Although Asp shares similar chemical properties with Glu, the smaller side chain of the former may thereby not support catalytic activity. This suggests that the proximity of Glu347 to other catalytic residues or substrates is necessary for catalysis. Moreover, mutation of the Glu to Gln in CvFatB2 resulted in substantial loss of activity, suggesting that the negative charge of the carboxylic group of Glu347 is also important.

Sequence and structural analyses also suggested a catalytic role for residue Asp309. Similar to the rationale in considering Glu347, mutating residue Asp309 to Ala nearly inactivated the enzyme. Moreover, mutation to the larger Glu-residue or the uncharged Asn-residue also reduced enzymatic activity. Collectively, these data indicate that residue Asp309 is a catalytic residue and the charge and position of the carboxylic group on this residue appears to play an important role in catalysis. Residues Asn311 and His313 were also mutated to Ala in our study, and as expected, only negligible activities were retained by these mutants. Mutation of the corresponding residues in another two acyl-ACP TEs (CnFatB2 and CvFatB1) yielded the same results, further supporting the potential catalytic function of these latter two residues.

A proposed catalytic mechanism for acyl-ACP TEs

The mutational analyses suggest that Asp309, Asn311, His313, and Glu347 are catalytic residues and therefore TE enzymes may function via a different catalytic mechanism than the previously proposed thiol-dependent papain-like protease mechanism [15]. TE families, including acyl-CoA TE and acyl-ACP TE, primarily have two different structural folds: the hotdog fold and α/β-hydrolase fold [16]. Whereas the Ser-His-Asp catalytic triad is highly conserved among TEs having the α/β-hydrolase-fold, the TEs that share a hotdog-fold exploit a variety of different catalytic mechanisms, utilizing different sets of catalytic residues [16].

Upon reviewing the catalytic mechanisms reported for several hotdog-fold TEs [16,26], two TEs that catalyze the analogous reaction to acyl-ACP TEs but belong to two distinct sequence-based phylogenetic clades of TE families were of interest: a 4-hydroxybenzoyl-CoA TE (4HBT) from Pseudomonas sp. strain CBS3, and 4HBT from Arthrobacter sp. strain SU [27,28]. The Pseudomonas 4HBT exploits a two-residue catalytic mechanism, utilizing Tyr24 and Asp17 residues. The catalytic residue Tyr24 is located at the N-terminus of an α-helix and is responsible for the polarization of the thioester carbonyl group of the substrate via hydrogen bonding with its backbone amide. The catalytic residue Asp17 acts as a nucleophile to attack the thioester carbonyl carbon to form an anhydride intermediate, which then undergoes hydrolytic cleavage at the hydroxybenzoyl carbonyl carbon atom [29]. In comparison, the Arthrobacter 4HBT TE exploits a four-residue catalytic mechanism, utilizing the Glu73, Gly65, Gln58, and Tyr77 residues. In this mechanism, the Glu73 functions as a nucleophilic catalyst, Gly65 and Gln58 stabilize the thioester moiety of the substrate via hydrogen bonding, and Tyr77 orients the water nucleophile for attack at the carbonyl carbon of the enzyme-anhydride intermediate [30]. Although Pseudomonas 4HBT and Arthrobacter 4HBT have different catalytic residues and different active site structures, they share commonalities in the catalytic mechanisms: (1) both utilize an acidic residue as the nucleophilic catalyst; (2) the backbone amide group of a residue at the N-terminus of an α-helix (Tyr24 in Pseudomonas 4HBT and Gly65 in Arthrobacter 4HBT) hydrogen bond with the carbonyl group of the substrate thioester to make the carbonyl carbon more susceptible to nucleophilic attack; and (3) both active sites form an enzyme-anhydride intermediate.

Comparison of the active site structures of CvFatB2 and Pseudomonas 4HBT (1BVQ) reveals that the catalytic residue Asp309 in CvFatB2 has the same position and orientation as the catalytic residue Asp17 in 1BVQ and they are located in well-aligned loop structures (Figure 6A,B). Moreover, the backbone amide group of Asp316 in CvFatB2 perfectly aligns with the backbone amide group of the catalytic residue Tyr24 in 1BVQ, though the side chains of these two residues lie in opposite directions. In addition, residue Asp316 is highly conserved (88% conservation) among the 1019 TEs sequences that were compared. Based on our experimental results and the comparison of the CvFatB2 acyl-ACP TE-modeled structure with Pseudomonas 4HBT structure (PDB 1BVQ), we propose a two-step catalytic mechanism for acyl-ACP TEs, as exemplified by the CvFatB2 catalytic residues. In the first step, similar to the mechanism of 4HBT, the carbonyl group in the thioester bond of the acyl-ACP substrate is stabilized via hydrogen bonding with backbone amide group of residue Asp316, and residue Asp309 initiates the nucleophilic attack on the carbonyl carbon of the thioester to form an anhydride intermediate. In the second step, catalytic residue His313 co-ordinated by Asn311 and Glu347 orients and activates a water molecule to hydrolyze the enzyme-anhydride intermediate and release the free fatty acid product (Figure 6C). A similar two-step catalytic mechanism has recently been proposed based on the experimentally determined Bay Laurel acyl-ACP TE crystal structure [12]. Although both mechanisms indicate an anhydride intermediate between the acyl-substrate and the carboxylate nucleophile, the detailed mechanisms are proposed differently, including how the substrate is stabilized, how the nucleophilic attack is initiated, how the anhydride intermediate is hydrolyzed, and how His313 and Glu347 function in the catalysis.

Comparison of the structural model of CvFatB2 with the crystal structure of Pseudomonas 4HBT (1BVQ) and the proposed catalytic mechanism for TE catalysis.

Figure 6.
Comparison of the structural model of CvFatB2 with the crystal structure of Pseudomonas 4HBT (1BVQ) and the proposed catalytic mechanism for TE catalysis.

(A) Alignment of the active site structures of CvFatB2 (magenta) and 1BVQ (green); highlighted is the loop structure that contains the catalytic residues, which appear to align. (B) Magnified view of ‘stick-model' of the catalytic residues of CvFatB2 (magenta) and 1BVQ (green); (C) Proposed catalytic mechanism for acyl-ACP TE indicating the involvement of residues Asp309, Asn311, His313, and Glu347 of CvFatB2.

Figure 6.
Comparison of the structural model of CvFatB2 with the crystal structure of Pseudomonas 4HBT (1BVQ) and the proposed catalytic mechanism for TE catalysis.

(A) Alignment of the active site structures of CvFatB2 (magenta) and 1BVQ (green); highlighted is the loop structure that contains the catalytic residues, which appear to align. (B) Magnified view of ‘stick-model' of the catalytic residues of CvFatB2 (magenta) and 1BVQ (green); (C) Proposed catalytic mechanism for acyl-ACP TE indicating the involvement of residues Asp309, Asn311, His313, and Glu347 of CvFatB2.

Abbreviations

     
  • 4HBT

    4-hydroxybenzoyl-CoA TE

  •  
  • ACP

    acyl carrier protein

  •  
  • FAS

    fatty acid synthase

  •  
  • IPTG

    isopropyl-β-d-galactopyranoside

  •  
  • PDB

    Protein Data Bank

  •  
  • TE

    thioesterase

Author Contribution

F.J. conducted all experiments. F.J., M.D.Y.-N., and B.J.N. conceived the study, analyzed the data, and wrote the manuscript. All authors read and approved the final manuscript.

Funding

This work was partially supported by the U.S. National Science Foundation through its Engineering Research Center Program (Award No. EEC-0813570), which supports the Center for Biorenewable Chemicals (CBiRC), headquartered at Iowa State University and including Rice University, the University of California, Irvine, the University of New Mexico, the University of Virginia, and the University of Wisconsin-Madison.

Acknowledgments

The authors thank Drs M. Ann D.N. Perera and Zhihong Song of the W.M. Keck Metabolomics Research Laboratory at Iowa State University for assisting with fatty acid analysis, Dr Paul Kapke and Amanda Brockman of the Hybridoma Facility at Iowa State University for making the CvFatB2 antiserum, and Dr Joel Nott of the Protein Facility at the Iowa State University for assisting with acquiring the CD spectra. We thank Ms Jarmila Tvaruzkova for generating CvFatB1 mutants. We also thank Mr Colin Hueser for assisting with the generation of CvFatB2 mutants and purification of CvFatB2 mutant proteins.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
White
,
S.W.
,
Zheng
,
J.
,
Zhang
,
Y.M.
and
Rock
,
C.O.
(
2005
)
The structural biology of type II fatty acid biosynthesis
.
Annu. Rev. Biochem.
74
,
791
831
2
Voelker
,
T.
(
1996
)
Plant acyl-ACP thioesterases: chain-length determining enzymes in plant fatty acid biosynthesis
.
Genet. Eng.
18
,
111
133
3
Jones
,
A.
,
Davies
,
H.M.
and
Voelker
,
T.A.
(
1995
)
Palmitoyl-acyl carrier protein (ACP) thioesterase and the evolutionary origin of plant acyl-ACP thioesterases
.
Plant Cell
7
,
359
371
4
Jing
,
F.
,
Cantu
,
D.C.
,
Tvaruzkova
,
J.
,
Chipman
,
J.P.
,
Nikolau
,
B.J.
,
Yandeau-Nelson
,
M.D.
et al. 
(
2011
)
Phylogenetic and experimental characterization of an acyl-ACP thioesterase family reveals significant diversity in enzymatic specificity and activity
.
BMC Biochem.
12
,
44
5
Hawkins
,
D.J.
and
Kridl
,
J.C.
(
1998
)
Characterization of acyl-ACP thioesterases of mangosteen (Garcinia mangostana) seed and high levels of stearate production in transgenic canola
.
Plant J.
13
,
743
752
6
Hills
,
M.J.
(
1999
)
Improving oil functionality by tuning catalysis of thioesterase
.
Trends Plant Sci.
4
,
421
422
7
Ohlrogge
,
J.B.
(
1994
)
Design of new plant products: engineering of fatty acid metabolism
.
Plant Physiol.
104
,
821
826
8
Voelker
,
T.A.
,
Hayes
,
T.R.
,
Cranmer
,
A.M.
,
Turner
,
J.C.
and
Davies
,
H.M.
(
1996
)
Genetic engineering of a quantitative trait: Metabolic and genetic parameters influencing the accumulation of laurate in rapeseed
.
Plant J.
9
,
229
241
9
Lennen
,
R.M.
and
Pfleger
,
B.F.
(
2012
)
Engineering Escherichia coli to synthesize free fatty acids
.
Trends Biotechnol.
30
,
659
667
10
Zhang
,
X.J.
,
Li
,
M.
,
Agrawal
,
A.
and
San
,
K.Y.
(
2011
)
Efficient free fatty acid production in Escherichia coli using plant acyl-ACP thioesterases
.
Metab. Eng.
13
,
713
722
11
Mayer
,
K.M.
and
Shanklin
,
J.
(
2005
)
A structural model of the plant acyl-acyl carrier protein thioesterase FatB comprises two helix/4-stranded sheet domains, the N-terminal domain containing residues that affect specificity and the C-terminal domain containing catalytic residues
.
J. Biol. Chem.
280
,
3621
3627
12
Feng
,
Y.B.
,
Wang
,
Y.Y.
,
Liu
,
J.
,
Liu
,
Y.H.
,
Cao
,
X.P.
and
Xue
,
S.
(
2017
)
Structural insight into acyl-ACP thioesterase toward substrate specificity design
.
ACS Chem. Biol.
12
,
2830
2836
13
Pollard
,
M.R.
,
Anderson
,
L.
,
Fan
,
C.
,
Hawkins
,
D.J.
and
Davies
,
H.M.
(
1991
)
A specific acyl-ACP thioesterase implicated in medium-chain fatty-acid production in immature cotyledons of Umbellularia californica
.
Arch. Biochem. Biophys.
284
,
306
312
14
Davies
,
H.M.
,
Anderson
,
L.
,
Fan
,
C.
and
Hawkins
,
D.J.
(
1991
)
Developmental induction, purification, and further characterization of 12-0-ACP thioesterase from immature cotyledons of Umbellularia californica
.
Arch. Biochem. Biophys.
290
,
37
45
15
Yuan
,
L.
,
Nelson
,
B.A.
and
Caryl
,
G.
(
1996
)
The catalytic cysteine and histidine in the plant acyl-acyl carrier protein thioesterases
.
J. Biol. Chem.
271
,
3417
3419
16
Cantu
,
D.C.
,
Chen
,
Y.
and
Reilly
,
P.J.
(
2010
)
Thioesterases: a new perspective based on their primary and tertiary structures
.
Protein Sci.
19
,
1281
1295
17
Pei
,
J.
,
Kim
,
B.H.
and
Grishin
,
N.V.
(
2008
)
PROMALS3D: a tool for multiple protein sequence and structure alignments
.
Nucleic Acids Res.
36
,
2295
2300
18
Waterhouse
,
A.M.
,
Procter
,
J.B.
,
Martin
,
D.M.
,
Clamp
,
M.
and
Barton
,
G.J.
(
2009
)
Jalview Version 2 — a multiple sequence alignment editor and analysis workbench
.
Bioinformatics
25
,
1189
1191
19
Jones
,
D.T.
(
1999
)
Protein secondary structure prediction based on position-specific scoring matrices
.
J. Mol. Biol.
292
,
195
202
20
Zhang
,
Y.
(
2008
)
I-TASSER server for protein 3D structure prediction
.
BMC Bioinformatics
9
,
40
21
Klein
,
K.
,
Steinberg
,
R.
,
Fiethen
,
B.
and
Overath
,
P.
(
1971
)
Fatty acid degradation in Escherichia coli. An inducible system for the uptake of fatty acids and further characterization of old mutants
.
Eur. J. Biochem.
19
,
442
450
22
Overath
,
P.
,
Pauli
,
G.
and
Schairer
,
H.U.
(
1969
)
Fatty acid degradation in Escherichia coli. An inducible acyl-CoA synthetase, the mapping of old-mutations, and the isolation of regulatory mutants
.
Eur. J. Biochem.
7
,
559
574
23
Altschuler
,
E.L.
,
Hud
,
N.V.
,
Mazrimas
,
J.A.
and
Rupp
,
B.
(
1997
)
Random coil conformation for extended polyglutamine stretches in aqueous soluble monomeric peptides
.
J. Pept. Res.
50
,
73
75
24
Jing
,
F.
,
Zhao
,
L.
,
Yandeau-Nelson
,
M.D.
and
Nikolau
,
B.J.
(
2018
)
Two distinct domains contribute to the substrate acyl chain length selectivity of plant acyl-ACP thioesterase
.
Nat. Commun.
9
,
860
25
Hedstrom
,
L.
(
2002
)
Serine protease mechanism and specificity
.
Chem. Rev.
102
,
4501
4524
26
Pidugu
,
L.S.
,
Maity
,
K.
,
Ramaswamy
,
K.
,
Surolia
,
N.
and
Suguna
,
K.
(
2009
)
Analysis of proteins with the ‘hot dog’ fold: prediction of function and identification of catalytic residues of hypothetical proteins
.
BMC Struct. Biol.
9
,
37
27
Thoden
,
J.B.
,
Holden
,
H.M.
,
Zhuang
,
Z.
and
Dunaway-Mariano
,
D.
(
2002
)
X-ray crystallographic analyses of inhibitor and substrate complexes of wild-type and mutant 4-hydroxybenzoyl-CoA thioesterase
.
J. Biol. Chem.
277
,
27468
27476
28
Thoden
,
J.B.
,
Zhuang
,
Z.
,
Dunaway-Mariano
,
D.
and
Holden
,
H.M.
(
2003
)
The structure of 4-hydroxybenzoyl-CoA thioesterase from Arthrobacter sp. strain SU
.
J. Biol. Chem.
278
,
43709
43716
29
Zhuang
,
Z.
,
Latham
,
J.
,
Song
,
F.
,
Zhang
,
W.
,
Trujillo
,
M.
and
Dunaway-Mariano
,
D.
(
2012
)
Investigation of the catalytic mechanism of the hotdog-fold enzyme superfamily Pseudomonas sp. strain CBS3 4-hydroxybenzoyl-CoA thioesterase
.
Biochemistry
51
,
786
794
30
Song
,
F.
,
Thoden
,
J.B.
,
Zhuang
,
Z.
,
Latham
,
J.
,
Trujillo
,
M.
,
Holden
,
H.M.
et al. 
(
2012
)
The catalytic mechanism of the hotdog-fold enzyme superfamily 4-hydroxybenzoyl-CoA thioesterase from Arthrobacter sp. strain SU
.
Biochemistry
51
,
7000
7016

Supplementary data