Crystal Structure of (S)-3-Hydroxybutyryl-CoA Dehydrogenase from Clostridium butyricum and Its Mutations that Enhance Reaction Kinetics

Abstract

3-Hydroxybutyryl-CoA dehydrogenase is an enzyme that catalyzes the second step in the biosynthesis of n-butanol from acetyl-CoA, in which acetoacetyl-CoA is reduced to 3-hydroxybutyryl-CoA. To understand the molecular mechanisms of n-butanol biosynthesis, we determined the crystal structure of 3-hydroxybutyryl-CoA dehydrogenase from Clostridium butyricum (CbHBD). The monomer structure of CbHBD exhibits a two-domain topology, with N- and C-terminal domains, and the dimerization of the enzyme was mostly constituted at the C-terminal domain. The mode of cofactor binding to CbHBD was elucidated by determining the crystal structure of the enzyme in complex with $NAD^+$. We also determined the enzyme's structure in complex with its acetoacetyl-CoA substrate, revealing that the adenosine diphosphate moiety was not highly stabilized compared with the remainder of the acetoacetyl-CoA molecule. Using this structural information, we performed a series of site-directed mutagenesis experiments on the enzyme, such as changing residues located near the substrate-binding site, and finally developed a highly efficient CbHBD K50A/K54A/L232Y triple mutant enzyme that exhibited approximately 5-fold higher enzyme activity than did the wild type. The increased enzyme activity of the mutant was confirmed by enzyme kinetic measurements. The highly efficient mutant enzyme should be useful for increasing the production rate of n-butanol.

Introduction

n-Butanol is one of the most promising biofuels produced by microbial fermentation. The widely known anaerobic bacterial strain Clostridium acetobutylicum efficiently produces n-butanol via a carbohydrate catabolic pathway [12,18]. Once the key intermediate, acetyl-CoA, has been generated, it can be metabolized via three different pathways, producing acetone, ethanol, and butanol in the so-called “acetone-butanol-ethanol (ABE)” fermentation [12,18]. The n-butanol synthetic pathway is composed of six tightly regulated steps that involve enzymes such as thiolase, dehydrogenases, and crotonase [13]. The main interest in the biosynthesis of n-butanol is that it has higher energy content, and is less corrosive, more water-soluble, and easier to blend with motor-vehicle fuels than bioethanol [7,8,16]. Before n-butanol can become a next-generation biofuel, however, a major improvement in synthetic efficiency is needed.

Since the 1990s, a large number of engineering efforts have been made, ranging from genetic modifications to optimizations of culture conditions, aiming to improve n-butanol production in ABE fermentation. Even with those efforts, n-butanol titers derived from clostridial fermentations are usually less than 20 g/l [18], which prohibits its utilization in industrial processes. An alternative approach to enhancing n-butanol production is to engineer the enzymes involved in the n-butanol biosynthetic pathway by using widely used industrial hosts, such as Escherichia coli, Pseudomonas putida, and Bacillus subtilis, because their genetic and physiological characteristics are well defined, and there are many genetic tools available for their modification [12]. In spite of this, the final n-butanol titers so far have been even lower than those obtained via clostridial strains. In heterologous host cells expressing the whole clostridial n-butanol biosynthetic machinery, n-butanol titers do not normally exceeded 1 g/l [7,8,12,16].

There are several reasons why engineering non-solventogenic microbes to produce large amounts of n-butanol is a challenging task. First, n-butanol can be toxic to bacterial cells [13]. For example, low concentrations of n-butanol inhibit E. coli growth, which ceases almost entirely in approximately 1% n-butanol [10]. Second, the additional pathways for n-butanol synthesis disrupt the balance of energy carriers such as NADH/NAD+, which lowers n-butanol production [2]. In addition, anaerobic NADH generation is not sufficient for n-butanol production in E. coli [1,7]. Third, the activities of the heterologous enzymes for n-butanol synthesis are host-cell specific, so each enzyme of the pathway should be optimized for the heterologous host. For example, compared with the performance of the clostridial pathway, n-butanol titers in E. coli are about 5-fold higher with a chimeric pathway that uses enzymes from three different species [20,23]. These observations demonstrate that it is important to understand in detail the reactions and regulatory mechanisms of the key enzymes in the n-butanol biosynthetic pathway [11,15]. This information would permit a rational approach to the optimization of the heterologous metabolic pathways, leading to a maximization of n-butanol yield from engineered non-solventogenic microbes [2,21].

Currently, the physiological characteristics of natural n-butanol producers such as Clostridium spp. have been taken into account to increase n-butanol yields both from naturally occurring strains as well as non-solventogenic industrial microbes [14]. Consequentyly, CbHBD in clostridia catalyzes the reduction of acetoacetyl-CoA by NADH. This subprocess is the initial and necessary step toward the ultimate production of butyrate and butanol [5]. Here, we present the first report, to our knowledge, of the crystal structure of the Clostridium butyricum (S)-3-hydroxybutyryl-CoA dehydrogenase (CbHBD), an enzyme that catalyzes the second step in n-butanol bioysnthesis, converting acetoacetyl-CoA to 3-hydroxybutyryl-CoA. We also report the use of this structural information in the development of a highly efficient mutant CbHBD.

 

Materials and Methods

Preparation of CbHBD

The CbHBD coding gene (Met1-Lys282, MW 30.6 kDa) was amplified by polymerase chain reaction (PCR) using C. butyricum chromosomal DNA as a template. The PCR product was then subcloned into pET30a (Invitrogen) with 6-histag at the C-terminus. The expression construct was transformed into an Escherichia coli B834 strain, which was grown in 1 L of LB medium containing kanamycin (50 mg/ml) at 37℃. After induction via the addition of 1.0 mM IPTG, the culture medium was further maintained for 20 h at 18℃. The culture was harvested by centrifugation at 5,000 ×g at 4℃. The cell pellet was resuspended in buffer A (40 mM Tris–HCl at pH 8.0 and 5 mM β-mercaptoethanol) and then disrupted by ultrasonication. The cell debris was removed by centrifugation at 11,000 ×g for 1 h, and the lysate was bound to Ni-NTA agarose (Qiagen). After washing with buffer A containing 20 mM imidazole, the bound proteins were eluted with 300 mM imidazole in buffer A. A trace amount of contamination was removed by applying HiLoad 26/60 Superdex 200 prep grade (GE Healthcare) size-exclusion chromatography. The purified protein that showed ~95% purity on SDS-PAGE was concentrated to 30 mg/ml in 40 mM Tris-HCl, pH 8.0, containing 1 mM dithiothreitol. Site-directed mutagenesis experiments were performed using the QuikChange site-directed mutagenesis kit (Stratagene).

Crystallization, Data Collection, and Structure Determination of CbHBD

Initial screening for the crystallization of CbHBD was done using the hanging drop vapor diffusion method. Crystallization conditions obtained commercially from Emerald Biostructures (Wizard Screen I and II) were screened. Wizard screen I condition 33 produced crystals that diffracted to 1.8 Å resolution. Suitable crystals for diffraction experiments were obtained at 22℃ within 7 days from the precipitant of 0.2 M Li2SO4, 0.1 M CAPS (pH 10.5), and 2 M ammonium sulfate. CbHBD crystals in complex with NAD+ and with acetoacetyl-CoA were co-crystallized with the same crystallization condition supplemented with 20 mM each of NAD+ and acetoacetyl-CoA. The crystals were transferred to cryoprotectant solution containing 0.2 M Li2SO4, 0.1 M CAPS (pH 10.5), 2 M ammonium sulfate, and 30% glycerol, fished out with a loop larger than the crystals, and flash-frozen by immersion in liquid nitrogen at -173℃. The data were collected to a resolution of 1.8 Å at 7A beamline (SBI) of the Pohang Accelerator Laboratory (PAL, Pohang, Korea) using a Quantum 270 CCD detector (ADSC, USA). The data were then indexed, integrated, and scaled using the HKL2000 suite [21]. Crystals of an apo-form belonged to space group R3, with unit cell parameters of a = b = 146.48 Å, c = 202.29 Å. Assuming four molecules of CbHBD per asymmetric unit, the crystal volume per unit of protein mass was 3.41 Å3 Da-1 [17], which corresponds to a solvent content of approximately 63.61%. Crystals in complex with NAD+ belonged to the same space with the apo-form of CbHBD crystals with similar unit cell parameters. Crystals in complex with acetoacetyl-CoA belonged to space group R3 with unit cell parameters of a = b = 146.34 Å, c = 409.76 Å. Assuming eight molecules of CbHBD per asymmetric unit, the crystal volume per unit of protein mass was 3.3 Å3 Da-1 [17], which corresponds to a solvent content of approximately 63.05%. The structure of the apo-form of CbHBD was solved by molecular replacement using the Homo sapiens L-3-hydroxyacyl-CoA dehydrogenases (HuHAD) (PDB code 1F0Y) with the side chains converted to Ala as a search model [6]. The amino acid sequence identity between HuHAD and CbHBD was 43% [4]. Further model building was performed manually using the program WinCoot [9] and the refinement was performed with REFMAC5 [19]. The structures of CbHBD in complex with NAD+ and with acetoacetyl-CoA were solved by molecular replacement using the crystal structure of the apo-form of CbHBD. The refined model of the apo-form of CbHBD and those in complex with NAD+ and with acetoacetyl-CoA were deposited in the Protein Data Bank (PDB code 4KUE for the apo-form of CbHBD, and 4KUG and 4KUH for NAD+ and acetoacetyl-CoA bound forms of CbHBD, respectively).

3-Hydroxybutyryl-CoA Dehydrogenase Activity Measurement

We also investigated whether CbHBD could catalyze the reverse reaction, converting acetoacetyl-CoA to 3-hydroxybutyryl-CoA, using NADH as the cofactor. All assays were performed with a reaction mixture of 1 ml total volume. The reaction mixture contained 100 mM MOPS (pH 8.0), 100 µM NADH, 100 µM acetoacetyl-CoA, 1 mM DTT (dithiothreitol), and 32 µM CbHBD enzyme. After pre-incubation at 30℃ for 5 min, the reaction was initiated by the addition of enzyme. The decrease in NADH was then measured at 340 nm using an extinction coefficient of 6.3 × 103 [3,24]. The enzyme kinetics experiments were performed by addition of various concentrations of acetoacetyl-CoA substrate, namely 5, 10, 20, 30, 40, 60, 80, 100, and 200 µM. The kinetic parameters Km, Vmax and kcat for each substrate and cofactor were ascertained by varying their concentration and fitting the data to the Michaelis-Menten equation. Each reaction was performed in triplicate.

 

Results and Discussion

Overall Structure of CbHBD

To elucidate the enzymatic properties of the CbHBD protein, we determined the crystal structure at 1.8 Å resolution. The asymmetric unit of the crystal contained four CbHBD molecules, corresponding to two biologically active dimers (Fig. 1). Size-exclusion chromatography results also confirmed that CbHBD exists as a dimer (data not shown). As expected, the structure of CbHBD was highly homologous to that of human mitochondrial L-3-hydroxyacyl-CoA dehydrogenases (HuHAD) [3]. The amino acid sequence identity between HuHAD and CbHBD was 43%, with a root-mean-square deviation of 1.1 Å on 281 Ca atoms. The monomer structure of CbHBD exhibited a two-domain topology, with N- and C-terminal domains (Fig. 1C). The N-terminal domain (residues 1-182) showed a β-α-β fold similar to NAD(P)-binding enzymes. It consisted of a core eight-stranded β-sheet flanked by α-helices. As observed in a typical Rossmann fold, the first six strands (β1-β6) of the sheet were in a parallel conformation. The final two strands (β7, β8) were also parallel, but ran in the opposite direction relative to the first six strands. A large helix-turn-helix motif (α2 and α3) connected β2 and β3, and extended from the β-α-β core. Many charged residues were located in the motif, with some of them involved in binding the adenine diphosphate moiety of the acetoacetyl-CoA substrate; this is described later (Fig. 1C). The C-terminal domain (residues 183-282) consisted primarily of α-helices, and it is involved in subunit dimerization. It contained a bundle of five α-helices having orientations relative to each other, with the first two α-helices (α8, α9) of each subunit being involved in dimerization, mainly through hydrophobic interaction by residues such as Val186, Val187, Ile190, Leu191, Pro193, Met194, Ile201, Val206, Ile212, Met216, and Ala220.

Fig. 1.Overall shape of CbHBD. (A) Alignment of amino acid sequences of CbHBD and HuHAD. Amino acid sequences of CbHBD and HuHAD were aligned based on the structural information. Secondary structure elements are shown and labeled based on the structure of CbHBD. Identical and highly conserved residues are presented in red and blue colored characters, respectively. The G-x-G-x-x-G motif is indicated by rectangles of a yellow color, and the N- and C-terminal domains are identified and labeled. Residues involved in the NAD- and substrate-binding are marked with purple and red triangles, respectively. Mutational points for the increase of enzyme activity of CbHBD are marked with green-colored stars. (B) Dimer structure of CbHBD. The dimer structure of CbHBD is presented as a cartoon diagram with cyan and salmon colors for each of the monomers. NAD+ and acetoacetyl-CoA bound in the enzyme are shown as sphere models with light blue and magenta colors, respectively. (C) Monomer structure of CbHBD. The N- and C-terminal domains are distinguished with green and light blue colors, respectively. NAD+ and acetoacetyl-CoA bound in the enzyme are shown as stick models with light blue and magenta colors, respectively, and the helix-turn-helix motif involved in the binding of adenine ring of the substrate is indicated by the dotted red-colored circle.

NAD+ Binding to CbHBD

To identify the NAD+-binding mode, we determined the crystal structure of CbHBD in complex with the NAD+ cofactor at 2.3 Å resolution. Superimposing the enzyme onto the NAD+-bound form of HuHAD revealed that although these two enzymes share a similar NAD+-binding mode, there are several differences in the residues observed to be involved in NAD+-binding by CbHBD. The NAD+-binding site was located at the G-x-G-x-x-G nucleotide-binding motif, comprising residues Gly8-Ala9-Gly10-Thr11-Met12-Gly13, and the hydroxyl groups of a phosphate moiety were hydrogen-bonded with the mainchain nitrogen atoms of Thr11 and Met12 (Fig. 2A). The nicotinamide and the two ribose rings of NAD+ were stabilized through hydrogen bond interactions mediated by the conserved Asp31, Glu90, Lys95, Asn115, Ser117, and Asn141 residues (Fig. 2A). The adenine moiety of NAD+ was positioned at the hydrophobic pocket formed by hydrophobic residues such as Leu7, Ile32, Ala88, Ile89, Ile94, and Ile98. One exception was Arg30, which assists the binding of the adenine moiety of NAD+ through a hydrogen bond.

Fig. 2.Cofactor and substrate-binding mode of CbHBD. (A) NAD+-binding mode of CbHBD. The CbHBD structure is shown as a cartoon model with gray color. The bound NAD+ and acetoacetyl-CoA are as stick models with light blue and magenta colors, respectively. Residues involved in the NAD+ stabilization through hydrogen bonds are shown as stick models with cyan color, and appropriately labeled, and the G-x-G-x-x-G motif is distinguished with a green color. (B) Substrate-binding mode of CbHBD. The CbHBD structure is shown as a cartoon model with cyan and salmon colors for each of two monomers, and labeled as Mol A and Mol B. The bound NAD+ and acetoacetyl-CoA are as stick models with light blue and magenta colors, respectively. Residues involved in the acetoacetyl-CoA stabilization through hydrogen bonds are shown as stick models, and appropriately labeled. The residues mutated for the increase of enzyme activity of CbHBD are shown as a stick model with green color. Domain shift upon the binding of substrate in CbHBD (C) and HuHBD (D). The apo- and the substrate-bound forms of HBD are presented as a cartoon model with grey and cyan colors, respectively. The bound acetoacetyl-CoA substrate is shown as a stick model and labeled.

Acetoacetyl-CoA Binding to CbHBD

The ternary complex of HuHAD with NAD+ and acetoacetyl-CoA showed that upon binding the substrate, significant shifting of the NAD+-binding domain relative to the C-terminal domain occurs, with the NAD+-binding domain rotating inward toward the C-terminal domain, resulting in a strong binding of the substrate [3]. To determine whether CbHBD undergoes a similar conformational change upon substrate binding, and to identify the substratebinding mode, we determined the crystal structure of CbHBD in complex with acetoacetyl-CoA substrate at 2.5 Å resolution (Fig. 2B). We did not observe the significant domain shifting found with HuHAD, although the acetoacetyl-CoA substrate was positioned well in its binding site (Fig. 2C). Instead, the positions of the two domains in the NAD+-bound form of CbHBD were almost identical to those of the apo-form of the enzyme (Fig. 2D). It should be noted that CbHBD is involved in n-butanol synthesis and accommodates only 3-hydroxylbutyryl-CoA as a substrate, whereas HuHAD is able to utilize 3-hydroxylacyl-CoA substrates of various lengths. We therefore suspect that the domain shifting in HuHAD is to accommodate 3-hydroxylacyl-CoAs with different numbers of carbons.

The crystal structure of CbHBD in complex with acetoacetyl-CoA also revealed the substrate-binding mode of the enzyme. Interestingly, the electron density map of the adenosine diphosphate moiety was not clear, whereas that of the rest of acetoactyl-CoA was strong, indicating that the adenosine diphosphate moiety is not highly stabilized. The acetoacetyl moiety of the substrate was located in its binding pocket, positioned near the conserved catalytic residues, Ser117, His138, and Asn188, corresponding to the Ser137, His158, and Asn208 residues of HuHAD. The pantothenic moiety of the substrate is stabilized by the main chain of Asn141 and the side-chain of Asn221 via hydrogen-bond interactions (Fig. 2B). The enzyme did not have a pocket for binding the adenosine diphosphate moiety, but two lysine residues, Lys50 and Lys54, located at a3, were positioned proximal to the adenine ring of the substrate (Fig. 2B). No obvious interaction between the two lysine residues and the adenine ring was observed. To confirm that these residues are involved in binding of the adenosine ring, we generated two mutants, K50A and K54A, and measured their enzymatic activity. Contrary to our expectations, the mutants exhibited approximately 2-fold higher activity than the wild type (Fig. 3). This result indicates that the Lys50 and Lys54 residues are not directly involved in the binding of the moiety, and we speculate that formation of a more hydrophobic surface by the two alanine residues (K50A and K54A) might provide a more favorable environment for binding of the adenine ring.

Fig. 3.Relative enzyme activities of CbHBD mutants. The reduction of acetoacetyl-CoA to 3-hydroxybutyryl-CoA was measured under the same conditions, and the enzyme activities of the CbHBD mutants were compared with that of the wild type. The activity is an average value of three independently performed experiments. Each mutant is labeled at the bottom of the graph.

Structure-Based Protein Engineering of CbHBD

The increased enzymatic activity of the CbHBD K50A and K54A mutants motivated us to develop highly efficient mutants of the enzyme. First, we measured the enzyme kinetics of the mutants and compared them with those of the wild type. The kcat/Km values of the K50A and K54A mutants were respectively 1.3 and 1.6 times higher than that of the wild type, which is consistent with the relative activity comparisons (Figs. 4B and 4C, Table 2). We then replaced Leu232 with a tyrosine residue, because Tyr252 in HuHAD is located at a position corresponding to that of Leu232 in CbHBD, and is involved in the stabilization of the pantothenic acid moiety of the substrate (Fig. 2B). We hypothesized that the replaced tyrosine residue might aid the stabilization of the substrate, not only by providing a more complementary structure for binding of the pantothenic acid moiety, but also by forming a hydrogen bond with the hydroxyl- or carbonyl-group of the moiety. Interestingly, the L232Y mutant showed 2.3 times higher activity than the wild type (Fig. 3), and enzyme kinetic measurements confirmed the increased enzyme activity, with a kcat/Km value 1.7 times higher that of the wild type (Fig. 4D, Table 2). Next, we generated three combinational double mutants based on the K50A, K54A, and L232Y single mutants, and measured their enzyme activities. The three double mutants, K50A/K54A, K50A/L232Y, and K54A/L232Y, respectively exhibited 3.2, 3.0, and 3.6 times higher activities than did the wild type, all of which were higher than those of the single mutants (Fig. 3). Enzyme kinetic measurements of these mutants determined kcat/Km values that were also higher (2.5, 1.7, and 3.3 times, respectively) than those of the wild type (Figs. 4E, 4F, and 4G, Table 2). Finally, we generated a K50A/K54A/L232Y triple mutant and measured its enzyme activity. The triple mutant had 4.3 times the activity and 4.9 times the kcat/Km value of the wild type (Figs. 3, and 4H, Table 2). In conclusion, by several rounds of structure-based mutational experiments, we succeeded in developing a highly efficient CbHBD K50A/K54A/L232Y mutant enzyme that exhibited approximately 5-fold higher enzyme activity than the wild type. The increased activity of the mutant was confirmed by enzyme kinetic measurements that showed lower Km and higher kcat values than those of the wild type. Because we observed that the expression levels of these mutants in E. coli were quite similar compared with that of the wild type, we are confident that the highly efficient mutant enzyme can be used to increase the rate of n-butanol production, and this work might be an example of a platform technology that can be applied to other enzymes involved in the production of high-value bioproducts.

Fig. 4.Enzyme kinetics of CbHBD mutants. (A-H) Lineweaver-Burk plots of the CbHBD wild type and seven mutants. Each mutant is labeled appropriately. The reaction velocity versus substrate concentrations were measured with various substrate concentrations, namely 5, 10, 20, 30, 40, 60, 80, 100, and 200 µM.

Table 1.aThe numbers in parentheses are statistics from the highest resolution shell. bRsym = Σ|Iobs - Iavg| / Iobs, where Iobs is the observed intensity of individual reflections and Iavg is the average over symmetry equivalents. cRwork = Σ||Fo| - |Fc| / Σ |Fo |, where |Fo | and |Fc | are the observed and calculated structure factor amplitudes, respectively. Rfree was calculated with 5% of the data.

Table 2.The values of Km, Vmax, and kcat of CbHBD wild-type and seven mutants were calculated based on the enzyme kinetic experiments shown in Fig. 4. aThe values of the CbHBD mutants were calculated based on the kcat /Km value of the CbHBD wild type as 1.