Suppression of the native hydrolytic activity of Pseudozyma antarctica lipase B (PalB) (formerly Candida antarctica lipase B) in water is demonstrated. By replacing the catalytic Ser 105 residue with an alanine unit, promiscuous Michael addition activity is favored. A Michael addition reaction between methyl acrylate and acetylacetone was explored as a model system. For the PalB Ser 105 Ala mutant, the hydrolytic activity was suppressed more than 1000 times and, at the same time, the Michael addition activity was increased by a factor of 100. Docking studies and molecular dynamics simulations revealed an increased ability of the PalB Ser 105 Ala mutant to harbor the substrates close to a catalytically competent conformation.
Applied enzyme catalysis is used industrially for the production of fine chemicals and intermediates for pharmaceuticals and is a key area within industrial biotechnology.1, 2 Enzymes are attractive biological catalysts that can catalyze organic reactions with high substrate specificity under mild conditions in aqueous solution at near neutral pH and at moderate temperatures. However, in the chemical industry, enzymes, such as lipases, are mainly used in organic solvents to avoid hydrolytic side reactions. Organic synthesis performed in water with enzymes is of interest in terms of cost, safety, and environmental aspects.3 It is, therefore, of high interest to understand enzymatic catalysis to be able to adapt enzymes to catalyze reactions other than their native ones.
Over the past few years, there has been significant progress in research related to enzyme catalytic promiscuity,4–8 that is, the capacity of enzymes to catalyze reactions other than the ones for which they are evolved. Promiscuous activities can provide clues on how to modify enzymes or reaction conditions to enhance desired qualities or suppress unwanted activities. The enzyme Pseudozyma antarctica lipase B (PalB; formerly known as Candida antarctica lipase B or CalB) displays promiscuous behavior. Besides its native hydrolytic activity, it is known to catalyze aldol addition reactions,9, 10 direct epoxidation reactions,11 and various conjugate addition reactions.12–20 The Michael addition reaction is the nucleophilic addition of carbanions to α,β-unsaturated carbonyl compounds and is one of the most useful methods to form carbon–carbon bonds in organic synthesis.21 The hydrolytic activity of PalB follows a ping-pong bi–bi type of kinetics that includes the formation of an intermediate acyl enzyme. PalB has a catalytic triad comprising Ser 105, Asp 187, and His 224 and an oxyanion hole consisting of Thr 40 and Gln 106, which can stabilize a negatively charged oxyanion intermediate formed during the reaction.22 The PalB mutant Ser 105 Ala has previously been shown to catalyze aldol addition reactions,9, 10 direct epoxidations,11 and conjugate addition reactions.12, 13 It is suggested that removal of the catalytically important Ser 105 would prevent the formation of an acyl enzyme and thus suppress hydrolytic activity without compromising the Michael addition activity. This approach could be used to perform Michael addition reactions on esters in water without hydrolyzing the ester functionality.
Herein, we explore the molecular interactions between the substrates and enzyme when using the PalB wild-type and Ser 105 Ala mutant. As a model system, we used the Michael addition of acetylacetone and methyl acrylate in aqueous solution (Scheme 1). First, an experimental study in the laboratory was performed to characterize the Michael addition and hydrolytic activities of both enzyme variants. Secondly, a molecular modeling study was undertaken to provide a molecular rationale for the suppressed hydrolytic activity and the exposed Michael addition activity.
Results and Discussion
Determination of enzyme activities
Hydrolytic and Michael addition activities of PalB Ser 105 Ala and the PalB wild-type in aqueous solution were explored experimentally. Because Michael addition reactions can be catalyzed by weak bases,23 a low pH value is necessary to suppress background activity. Low pH values could possibly affect the activity of PalB. However, full activity has been shown within pH 5.5–10.24
The PalB wild-type only catalyzed a hydrolysis reaction, thus transforming methyl acrylate into acrylic acid according to the ping-pong bi-bi mechanism (Table 1 and Scheme 2). Acrylic acid (96 %) and no methyl 4-acetyl-5-oxohexanoate were detected after 48 hours. On the other hand, the hydrolytic activity was suppressed more than 1000 times by using PalB Ser 105 Ala. PalB Ser 105 Ala catalyzed the Michael addition of acetylacetone to methyl acrylate according to the proposed ordered bi–uni mechanism (Scheme 3).12, 13 Only methyl 4-acetyl-5-oxohexanoate (3 %) and no acrylic acid were formed after 72 hours. These data clearly show that a single mutation of the catalytic Ser 105 to an alanine residue changed the catalytic activity of PalB from the faster native hydrolysis to the promiscuous Michael addition activity in aqueous solution.
Table 1. Catalytic events per enzyme per minute for the hydrolysis and Michael addition activities measured in competition with PalB wild-type or PalB Ser 105 Ala as the catalyst.
Catalytic events per enzyme per minute [min−1]
[a] Measured after 14 days; [b] measured after 70 days.
PalB Ser 105 Ala
The kinetic parameters for the model Michael addition were calculated using pseudo-first-order kinetics for the reaction catalyzed by PalB Ser 105 Ala (Table 2). An apparent turnover number of 0.08 min−1 and a rate enhancement of kcat./knon=105M−1 were determined. The low apparent turnover number could be due to enzyme inhibition by acetylacetone (see the section ‘Exploring substrate inhibition’ and ref. 13). The Michaelis–Menten constant had a high value, which is not uncommon when using unnatural substrates. Control reactions with only the polypropylene carrier used for enzyme immobilization were evaluated for a possible contribution to the total reaction rate. A 50 % increase in the background reaction rate was shown in the presence of the carrier without the enzyme. The hydrophobic surface of the carrier may provide a localized substrate concentration of the hydrophobic substrates, thus resulting in an increased reaction rate.
Table 2. Apparent kinetic constants for the PalB Ser 105 Ala-catalyzed Michael addition of acetylacetone to methyl acrylate and the background reactions with or without a carrier as a reference.
These experiments revealed different reaction specificities of the two PalB variants. The PalB wild-type catalyzed the hydrolysis of methyl acrylate, whereas the PalB mutant catalyzed the promiscuous Michael addition of methyl acrylate and acetyl- acetone. The reaction mechanisms for the hydrolysis reaction in the PalB wild-type and the Michael addition in PalB Ser 105 Ala (Schemes 2 and 3, respectively) display the different binding modes of methyl acrylate that are required for both reaction types. In the hydrolysis reaction, Ser 105 binds to the carbonyl carbon atom to form an acyl enzyme, the ester oxygen atom is directed toward His 224 to achieve a hydrogen-bond coordination, and the alkene chain is directed in the opposite direction. In addition, the binding of the Ser 105 residue to the carbonyl carbon atom destroys the conjugated system required for a Michael addition reaction. As Ser 105 binds to the carbonyl carbon atom, the proton that is initially shared between Ser 105 and His 224 is passed over to His 224. This proton helps to direct the ester oxygen atom to His 224 by providing a hydrogen-bonding interaction. Of course, the Michael addition reaction can proceed in the wild-type enzyme, as shown in previous reports,12–20 but in the presence of an ester group and water, native hydrolysis is favored. As the nucleophilic Ser 105 residue is removed, the hydrolysis reaction is suppressed and the promiscuous Michael addition activity is revealed. Methyl acrylate is free and only coordinated to the oxyanion hole in PalB Ser 105 Ala.
Exploring the Michaelis complex by using computer simulations
The binding of methyl acrylate and acetylacetone to both PalB variants were investigated by 250 independent molecular docking simulations of each substrate using the AutoDock4 program package in combination with AutoDock Tools.25, 26 The reactivity of the resulting structures were evaluated according to the near attack conformer (NAC) definition (or productive mode), as defined in the Experimental Section and displayed in Figure 5. The results are summarized in Figures 1–2, 3 and Tables 3 and 4.
Table 3. Fraction of correctly positioned structures that fulfill the NAC conditions from docking simulations of the substrates and intermediate to the PalB wild-type and PalB Ser 105 Ala.
Fraction of correctly positioned structures [%]
PalB Ser 105 Ala
[a] Empty active site; [b] methyl acrylate already in the oxyanion hole.
Table 4. Fraction of correctly positioned structures that fulfill the NAC conditions from molecular dynamics simulations of the Michaelis complex to the PalB wild-type and PalB Ser 105 Ala.[a]
[a] Average distances are given for both enzyme variants; [b] see Figure 5 for definition of distances D; [c] no coordination of acetylacetone to His 224.
PalB Ser 105 Ala
According to our proposed reaction mechanism (Scheme 3), methyl acrylate should bind to the oxyanion hole and is thus assumed to be the first substrate to enter the enzyme. Consequently, methyl acrylate was docked to both the PalB wild-type and PalB Ser 105 Ala. When using the PalB wild-type, all the docking simulations resulted in methyl acrylate in the active site, but none of the docked structures showed a productive mode for the reaction according to the criteria defined in the Experimental Section. In 99 % of the simulations, the ester oxygen atom of methyl acrylate formed hydrogen bonds with Thr 40 and the β-carbon atom was directed away from His 224. The remaining 1 % of cases were close to a reactive conformation but still lacked the proper hydrogen-bond coordination to the enzyme (Figure 1). Because AutoDock4 failed to position methyl acrylate in the correct orientation in the oxyanion hole for the hydrolysis reaction, a second docking simulation was made by using AutoDock in the YASARA software.25, 27 In that case, all the docking structures coordinated the carbonyl oxygen atom to the hydroxy group of Thr 40 with one hydrogen bond. The ester oxygen atom was directed towards His 224 and the vinyl group towards Gln 157.
PalB Ser 105 Ala, however, coordinated the carbonyl oxygen atom of methyl acrylate to the backbone amide groups of Thr 40 and Gln 106 by two hydrogen bonds in all of the dockings (Figure 2). In addition, the conformation and position of methyl acrylate in the active site was in a productive mode.
Second, acetylacetone was docked into both PalB variants with the oxyanion hole already occupied by methyl acrylate in a conformation closest to fulfill the NAC requirements. When using the PalB wild-type, 32 % of the docked acetylacetone molecules ended up in the active site. However, in PalB Ser 105 Ala, almost all the acetylacetone structures bound in the active site, but with acetylacetone positioned above methyl acrylate (Figure 3). The coordination of one α-hydrogen atom of acetylacetone to the Nε2 atom of His 224 was not the most preferred position because the two carbonyl oxygen atoms rather coordinated to the oxyanion hole. The formation of a NAC could not be confirmed, which could be due to the lack of attraction between the C3 carbon atom of acetyl- acetone and the Nε2 atom of His 224.
Molecular dynamics simulations were then performed on the Michaelis complex to further explore the enzyme–substrate interactions. The fraction of analyzed snapshots with fulfilled NAC criteria and average distances between the substrate and enzyme are summarized in Tables 3 and 4.
For the PalB wild-type, the molecular dynamics simulation of the Michaelis complex showed no hydrogen-bond coordination from the enzyme to methyl acrylate or acetylacetone. During the simulation, methyl acrylate slowly drifted away from the oxyanion hole. Because no productive binding was found during the simulation, no further molecular dynamics simulations of the transition-state analogue in the wild-type were performed.
For PalB Ser 105 Ala, the molecular dynamics simulation of the Michaelis complex revealed a coordination of the carbonyl oxygen atom to the oxyanion hole. However, the NAC criteria were never fulfilled because no activation of acetylacetone by His 224 was seen. The required hydrogen bonds were formed between methyl acrylate and the backbone amino groups of Thr 40 and Gln 106, but the side-chain hydroxy group of Thr 40 was rotated away from the carbonyl carbon atom of methyl acrylate during most of the simulation. Acetylacetone stayed in the active site during the whole simulation, but without any coordination to the enzyme. According to the proposed reaction mechanism, acetylacetone would be activated and coordinated for the Michael addition reaction to occur by the Nε2 atom of His 224. This behavior was earlier shown by molecular dynamics simulations when using PalB Ser 105 Ala for the direct epoxidation with hydrogen peroxide11 and conjugate addition with methanethiol,12 but could not be confirmed by this dynamics simulation.
In summary, no productive binding of the Michaelis complex was established with either PalB variant and the activation of acetylacetone by His 224 was unclear. However, the geometry of the Michaelis complex in PalB Ser 105 Ala was significantly closer to a reactive conformation, thus implying an increased likelihood of catalytic events compared to the PalB wild-type.
Exploring the transition state by computer simulations
The binding of the Michael addition transition-state analogue to the active sites of the PalB wild-type and Ser 105 Ala mutant was explored by computer simulations. The reactivities of the resulting structures were evaluated according to the NAC definition (or productive mode) as defined in the Experimental Section and displayed in Figure 6. The results are displayed in Figure 4 and Table 5.
Table 5. Fraction of correctly positioned structures that fulfill the NAC conditions from molecular dynamics simulations of the transition-state analogue to PalB Ser 105 Ala.[a]
[a] Average distances are given. [b] See Figure 6 for definition of distances D.
PalB Ser 105 Ala
In the molecular docking simulations with the PalB wild-type, all 250 dockings of the transition-state analogue in the Michael addition reaction ended up in the active site, but the NAC criteria were never fulfilled. Dockings to PalB Ser 105 Ala revealed the oxyanion hole as the most preferred binding site in 93 % of the docking simulations, and 10 % of those also fulfilled the NAC conditions (Figure 4).
Molecular dynamics simulations were then performed on the transition-state analogue to explore the ability of the substrates to form NACs in PalB Ser 105 Ala. The fraction of correctly positioned structures that fulfilled the NAC criteria and the average distances between the transition-state analogue and the enzyme are summarized in Table 5. The molecular dynamics simulation of the transition-state analogue showed coordination to Thr 40 in PalB Ser 105 Ala. The distance between the transition-state analogue and Gln 106 became longer and the binding to Thr 40 became tighter during the dynamics simulation. The distance between His 224 and the α-carbon atom of the Michael addition intermediate was within the van der Waals distance (2.25 Å on average) and indicated the possibility that His 224 could return the proton to the α-carbon atom of the transition-state analogue to form the final Michael adduct.
Exploring substrate inhibition
The inhibitory effect of acetylacetone was explored for both PalB variants. Docking simulations with the PalB wild-type showed that acetylacetone ended up in the active site in 96 % of cases (Table 3). The docked structures were coordinated to the enzyme by one hydrogen bond from Thr 40 in 62 % of these cases, whereas the bound acetylacetone was coordinated to the oxyanion hole by three hydrogen bonds in the case of PalB Ser 105 Ala. This outcome correlates well with the previously found inhibitory effect of acetylacetone in PalB Ser 105 Ala.13
A mutant of P. antarctica lipase B was experimentally and theoretically explored for the feasibility of a promiscuous Michael addition reaction with an ester substrate in water with focus on the interactions between substrates and enzyme. The Ser 105 Ala mutant lacked the nucleophilic Ser 105 residue and the ability to form an acyl enzyme. Consequently, the mutation suppressed the native hydrolytic enzyme activity to favor the promiscuous Michael addition activity. The wild-type enzyme, on the other hand, did not show Michael addition activity in water, only hydrolysis. These findings show the possibility of using “hydrolases” as biocatalysts in water while avoiding hydrolysis reactions.
Chemicals: Acetylacetone (Fluka) and methyl acrylate (Fluka) were used as substrates for the kinetic and thermodynamic measurements. 1,4-Dioxane (Labora) was used as an internal standard for GC measurements. Methyl 4-acetyl-5-oxohexanoate (Aldrich) was used to identify the Michael addition product by using GC and GC-MS. Sodium acetate buffer (20 mM, pH 5.0) was used as the reaction solvent. Diethyl ether (SDS) was used for the preparation of samples for GC analysis. Samples for GC analysis were dried over anhydrous MgSO4 (Merck).
Protein production and immobilization: PalB Ser 105 Ala was created by site-directed mutagenesis.9 Both the wild-type and Ser 105 Ala variant were expressed and produced in Pichia pastoris, purified by hydrophobic interaction chromatography and gel filtration, and freeze dried.10, 28 The amount of active immobilized PalB wild-type was determined by active-site titration with methyl para-nitrophenyl n-hexylphosphonate on freeze-dried enzyme before immobilization.10, 29 The PalB wild-type and PalB Ser 105 Ala were immobilized on a polypropylene carrier Accurel MP1000 (<1500 μm) in potassium phosphate buffer (50 mM, pH 7.6) and equilibrated to a water activity of 0.11 by saturated lithium chloride.28, 29 The amount of protein adsorbed on the carrier was determined to be 1.6 % w/w for the PalB wild-type and 1.8 % w/w for PalB Ser 105 Ala.
Enzyme kinetics: Pseudo-first-order kinetic constants were determined by measuring the initial reaction rates at a constant concentration of methyl acrylate (160 mM) and varied concentration of acetylacetone (10, 20, 30, 40, 50, 80, 120, and 160 mM). The concentrations of the substrates were chosen according to the solubility limit of methyl acrylate in water. Immobilized enzyme PalB wild-type or PalB Ser 105 Ala (20 mg) and the internal standard 1,4-dioxane (40 mM) were used in a 1 mL reaction mixture in sodium acetate buffer (20 mM, pH 5.0). The reactions were incubated at 21 °C in an end-over-end rotator. The initial reaction rates were determined by taking 4–6 samples below 5 % conversion. The kinetic constants and were determined from a least-square regression of the Michaelis–Menten equation. Background reactions were run in parallel. The background reaction rate constant knon was determined from v=knon[S]1[S]2, where [S]1 and [S]2 are the concentrations of methyl acrylate and acetylacetone, respectively. The initial rate from the background reaction containing only the carrier without the enzyme was subtracted from the initial enzyme-catalyzed reaction rates.
Product identification and quantification: Samples (100 μL) were extracted with diethyl ether (1–3×100 μl) and dried over MgSO4 in home-made columns with cotton as the filter. Compound identification was made on a GC analyzer HP6890 (Hewlett Packard) with a capillary column (J&W CycloSil-B, 30 m×0.32 mm i.d., df=0.25 μm) equipped with a mass-spectrometric detector. The Michael addition product was identical to the MS spectrum of the commercial available methyl 4-acetyl-5-oxohexanoate. The quantitative analyzes were carried out on a GC HP5890 Series II (Hewlett Packard) with a FID detector and a capillary column (CP-Chirasil Dex CB, 25 m×0.32 mm i.d., df=0.25 μm).
Molecular docking simulations: Binding of substrates and the Michael addition intermediate to the PalB wild-type and Ser 105 Ala variant was investigated by molecular docking simulations by using the AutoDock4 and AutoGrid4 program package in combination with AutoDock Tools.25, 26 Structures of the PalB wild-type and PalB Ser 105 Ala variant were prepared from the 1TCA crystal structure22 from the Protein Data Bank30 by using the YASARA software.27 All of the water molecules of crystallization were removed and polar hydrogen atoms were added. The Nε2 atom of His 224 was unprotonated for the substrate dockings and protonated for the docking of the Michael addition intermediate. The protons of the side-chain hydroxy groups of Thr 40 and Ser 105 were oriented to make three possible hydrogen-bond donors available to build up the oxyanion hole and to provide a hydrogen bond between Ser 105 and His 224. This was performed by energy minimization of the enzyme structure in which all atoms but the side-chain hydroxy groups of Thr 40 and Ser 105 were kept fixed. Structures of methyl acrylate, acetylacetone, and the Michael addition intermediate were constructed by using pyMol software.31 Gasteiger charges were added and a distance-dependent dielectric constant was used. All the possible torsion angles of the heavy atoms were set to active. A 60×60×60-point grid box with a grid spacing of 0.375 Å was centered on the Nε2 atom of His 224. The Lamarckian genetic algorithm was used with a maximum of 2 500 000 energy evaluations and a maximum of 27 000 generations. Each docking simulation contained 250 independent dockings with a starting population of 150 individuals. All the other options were set on program defaults.
Molecular dynamics simulations: Molecular dynamics simulations were performed by using the YASARA software,28 with all the starting structures originating from AutoDock4 simulation structures. Nonpolar hydrogen atoms were added and fractional bond orders adjusted to the corresponding pH value of the system (pH 5.0). A simulation cell with periodic boundaries was set to extend 10 Å at each side of the enzyme. To remove bumps and correct the covalent geometry, the structure was energy-minimized with the AMBER99 force field32 with a 7.86-Å force cut-off point and the particle mesh Ewald algorithm33 to treat long-range electrostatic interactions. The force-field parameters of the substrates/intermediate were assigned using AutoSMILES.34 After removal of conformational stress by a short steepest descent minimization, the procedure continued by simulated annealing (time step: 2 fs, atom velocities were scaled down by 0.9 every tenth step) until convergence was reached, that is, the energy improved by less than 0.1 % during 200 steps. The cell was filled with water (density: 1 g cm−3). Cell neutralization and prediction of the pKa values were made by using the “Neutralization” function in the YASARA software.35 Counterions were added (0.9 %) and ionizable groups were protonated according to pH 5.0. The Nε2 atom of the His 224 residue was kept unprotonated for the substrate simulations and protonated for the intermediate simulations. The solvent and enzyme structure with substrates/intermediates were then relaxed as above. Molecular dynamics simulations of 600 ps were performed at 298 K by using the built-in molecular dynamics simulation protocol with time steps of 1 fs to update intramolecular forces and 2 fs to update intermolecular forces.36 Snapshots were saved every 5 ps.
Definition of the near attack conformer (NAC): The original definition of a NAC was stated by Lightstone and Bruice in 1996;37 molecules must be within the van der Waals distance and within proper angles before bonds can start to make and break.37 In our definition of the NAC, for the Michael addition to take place in both lipase variants, methyl acrylate must be coordinated to the oxyanion hole (Thr 40 and Gln 106) by at least two hydrogen bonds to stabilize the negatively charged oxygen atom of an enolate (Figure 5, 1), acetylacetone should be activated by having one of its two α-protons within the van der Waals distance (2.75 Å) of the Nε2 atom of His 224 (Figure 5, 2), and acetylacetone should also be within the van der Waals distance (3.4 Å) of the β-carbon atom of methyl acrylate for nucleophilic attack (Figure 5, 3). Finally, the Michael addition intermediate must be coordinated by at least two hydrogen bonds to the oxyanion hole to stabilize the enolate (Figure 6, 1) and the α-carbon atom of the Michael addition intermediate must be within van der Waals distance (2.9 Å) of the H1 atom of His 224 to ensure the addition of the proton to the α-carbon atom of the Michael addition intermediate (Figure 6, 2). These requirements were based on the proposed reaction mechanism in Scheme 3.
Prof. Kalle Hult and Prof. Tore Brinck are gratefully acknowledged for fruitful discussion and suggestions.