Research Article pubs.acs.org/acscatalysis
Structural and Experimental Evidence for the Enantiomeric Recognition toward a Bulky sec-Alcohol by Candida antarctica Lipase B Areum Park,†,¶ Sunmin Kim,‡ Jeemin Park,§ Saerom Joe,†,¶ Bora Min,† Joonyoung Oh,∥ Jaekwang Song,∥ SangYoun Park,*,‡ Seongsoon Park,*,§ and Hyuk Lee*,†,¶ ACS Catal. 2016.6:7458-7465. Downloaded from pubs.acs.org by QUEEN MARY UNIV OF LONDON on 08/24/18. For personal use only.
†
Medicinal Chemistry Research Center, Korea Research Institute of Chemical Technology, 141 Gajeong-ro, Yuseong-gu, Daejeon 34114, Republic of Korea ¶ Department of Medicinal Chemistry and Pharmacology, University of Science and Technology, 217 Gajeong-ro, Yuseong-gu, Daejeon 34113, Republic of Korea ‡ School of Systems Biomedical Sciences, Soongsil University, Seoul 06978, Republic of Korea § Department of Chemistry, Center for NanoBio Applied Technology, Institute of Basic Sciences, Sungshin Women’s University, Seoul 01133, Republic of Korea ∥ Center for Bio-based Chemistry, Korea Research Institute of Chemical Technology, 141 Gajeong-ro, Yuseong-gu, Daejeon 35114, Republic of Korea S Supporting Information *
ABSTRACT: Candida antarctica lipase B (CAL-B) exhibits remarkable enantioselectivity for various chiral sec-alcohols, and the enantioselectivity is structurally well-understood. Two substituents at the chiral center of a sec-alcohol separately bind two pockets, namely, large and medium binding pockets. It has been believed that the medium pocket is too small to accommodate a large substituent (larger than an ethyl group), and thus, bulky secalcohols bearing two large substituents have been regarded as a poor substrate for CAL-B. However, we found that CAL-B can catalyze the transesterification of N-Boc-protected rac2-amino-1-phenylethanol (1a) enantioselectively with a moderate reaction rate. X-ray crystallography and computer modeling revealed that the rotation of the Leu278 side chain creates a space to accept the N-Boc-aminomethylene group of 1a. Moreover, a sec-alcohol substrate with less than one hydrogen atom at the γ-position from the hydroxyl group is required to achieve a moderate reaction rate. On the basis of this observation, we diversified bulky N-Boc-protected rac-2-amino-1-arylethanols for the transesterifications with high enantioselectivities (E > 200). KEYWORDS: 2-amino-1-phenylethanol, bulky sec-alcohol, lipase, molecular modeling, X-ray crystallography
■
INTRODUCTION Candida antarctica lipase B (CAL-B; also known as Pseudozyma antarctica lipase B) is classified as EC 3.1.1.3 and is one of the most widely used hydrolases in both academic and industrial research because it exhibits extraordinary thermal and chemical stabilities, far exceeding that of other hydrolases.1,2 In addition, CAL-B possesses remarkable enantioselectivity for sec-alcohols. These characteristics make CAL-B useful for the preparation of various chiral compounds.3 The molecular basis of CAL-B’s enantioselectivity for sec-alcohols during transesterification has been thoroughly documented since the crystal structure of CAL-B was revealed.4,5 Similar to other serine hydrolases, CALB possesses the catalytic triad composed of Ser105, His224, and Asp187. The catalytic serine attacks the carbonyl carbon atom of an ester substrate, and the tetrahedral intermediate is subsequently generated. The liberation of alcohol from the tetrahedral intermediate results in the formation of the acyl enzyme. Then, the reaction with sec-alcohol (e.g., 1-phenyl© 2016 American Chemical Society
ethanol) generates the second tetrahedral intermediate (Figure 1a). The oxyanion of the tetrahedral intermediate is stabilized through hydrogen bonding with the oxyanion hole composed of Thr40 and Gln106. Structurally, the acyl and the alcohol parts of the ester substrate bind to two different pockets, namely, the acyl-binding pocket and the alcohol-binding pocket. The alcohol-binding pocket is divided into two pockets: the large hydrophobic pocket and the medium pocket (also called the stereoselectivity pocket). The large pocket is located at the entrance, whereas the medium pocket is deep inside of CAL-B. The medium pocket is composed of Thr42, Ser47, and Trp104. The substituents at the chiral center of a sec-alcohol selectively bind to the pockets according to their sizes: larger substituents bind to the large hydrophobic pocket and smaller Received: August 1, 2016 Revised: September 19, 2016 Published: September 27, 2016 7458
DOI: 10.1021/acscatal.6b02192 ACS Catal. 2016, 6, 7458−7465
Research Article
ACS Catalysis
binds in the medium pocket (Figure 1a). The steric constraints do not allow a large substituent to be accepted in the medium pocket. Thus, it was thought that the enantioselectivity and catalytic activity of CAL-B for bulky sec-alcohols bearing two large substituents would be significantly lower than those for sec-alcohols with smaller substituents (Figure 1b).8 Indeed, Rotticci et al. examined the size effect of the substituents of chiral sec-alcohols on the enantiomeric ratio (E) of CAL-B during transesterification and concluded that the medium binding pocket encounters difficulty in accepting a substituent that is larger or longer than an ethyl group.8a Since the establishment of CAL-B expression systems, including Pichia pastoris, Escherichia coli, and Hansenula polymorpha,9 protein engineering has been employed to alter this enzyme’s functional properties, such as to improve the activity and enantioselectivity, enhance the thermal tolerance, and alter the reaction specificity.10 Moreover, protein engineering can be used to increase the size of its medium pocket. Enlarging the medium pocket would likely allow the accommodation of a larger substituent than an ethyl group.11 Magnusson et al. reported that increasing the size of the medium pocket by substituting Trp104 with a smaller amino acid, such as alanine, allows the pocket to accommodate a large substituent (i.e., a n-pentyl group in 5-nonanol).11a Our initial aim was to identify the best hydrolase to resolve a bulky sec-alcohol, such as rac-2-amino-1-arylethanol, because optically active 2-amino-1-arylethanols are important building blocks in the pharmaceutical industry. The moieties of 2-amino1-arylethanols provide the skeletal structures of various pharmaceutical substances, such as salbutamol, fenoterol, (−)-arbutamine, dexsotalol, and (−)-denopamine. A few methods have been reported for the preparation of enantiopure 2-amino-1-arylethanols, including the synthesis of optically active cyanohydrins followed by the reduction of nitriles,12 the asymmetric hydrogenation of α-amino ketones,13 and the asymmetric β-peroxidation of nitroalkenes followed by the reduction of nitro groups.14 However, most of the examples required multistep processes to obtain the desired products. In addition, the hydrolase-catalyzed resolution of rac-2-amino-1arylethanols has been reported, but the resolution process
Figure 1. (a) Tetrahedral intermediate in CAL-B-catalyzed transesterification of 1-phenylethanol. (b) Until now, sec-alcohols with a smaller group than propyl group as the medium substituent were only adequate for CAL-B-catalyzed transesterification. (c) In this study, we elucidate how CAL-B converts rac-1a to (S)-2a with high enantioselectivity.
ones to the medium pocket. Thus, CAL-B, similar to most lipases, exhibits R-stereoselectivity for chiral sec-alcohols in accordance with the “Kazlauskas rule”.6 For instance, CAL-B exclusively recognizes (R)-1-phenylethanol over the counter enantiomer (i.e., E > 200, where E is as defined by Chen et al.7). The larger substituent, the phenyl group, at the chiral center of (R)-1-phenylethanol likely binds in the large binding pocket, whereas the medium substituent, the methyl group,
Table 1. Selected List of Hydrolases for the Transesterification of N-Boc-protected rac-2-Amino-1-phenylethanol (1a) with Vinyl Butanoate (3a)a
enzyme
time (h)
% ee of (S)-2a
conv. (%)b
Ec
Thermomyces lanuginosus lipase (TLL) Burkholderia cepacia lipase (immobilized, BCL)d Candida antarctica lipase B (immobilized, CAL-B)e Candida antarctica lipase A (immobilized, CAL-A) Aspergillus melleus acylase (AMA) Mucor miehei lipase (immobilized, MML)
1 48 48 6 48 48
>99.9 >99.9 >99.9 63 (R) >99.9 >99.9
48 22 12 50 4 3
>200 >200 >200 8.4 >200 >200
Reaction conditions: a mixture of rac-1a (0.05 mmol), 3a (0.15 mmol), and hydrolase (2 mg) in t-BuOMe was shaken at 200 rpm at 30 °C. bGC conversion. cE (enantiomeric ratio) is defined as in ref 7. dAMANO PS-DI. eNovozym 435. a
7459
DOI: 10.1021/acscatal.6b02192 ACS Catal. 2016, 6, 7458−7465
Research Article
ACS Catalysis afforded unsatisfactory enantiomeric purities (65% ee).15 Hence, we explored an efficient and effective hydrolasecatalyzed resolution process. N-Boc-protected rac-2-amino-1phenylethanol (1a; Boc = t-butyloxycarbonyl) was chosen as a substrate for the initial screening of hydrolases. Seventeen hydrolases were screened for the enantioselective transesterification of 1a with vinyl butanoate (3a) (Table 1 and Table S1). During the screening, CAL-B (immobilized, Novozym 435) showed moderate reactivity and high enantioselectivity (E > 200) for 1a with 12% conversion in 48 h. The result was not expected because both substituents (i.e., the phenyl and N-Boc-aminomethylene groups) of 1a are larger than an ethyl group, which are predicted to be excluded from the medium pocket of CAL-B. In this study, we focus on understanding how CAL-B accepts such bulky sec-alcohols as substrates and distinguishes between substituents (Figure 1c), which could help expand the scope of compounds considered as substrates for CAL-B-catalyzed reactions. We initially conducted a modeling study, but it was unable to explain the reactivity and selectivity of CAL-B for 1a. Then, we obtained the crystal structure for a transition-state analogue of CAL-B-catalyzed transesterification of 1a and subsequently conducted molecular modeling based on the crystallographic analysis. These results revealed that Leu278 plays a critical role in CAL-B’s acceptance of such bulky substrates.
obtained a 47% conversion of (S)-2a with >99.9% ee under the reaction conditions of 0.05 mmol of rac-1a, 3 equiv of 3a, and 5 mg of CAL-B in t-BuOMe at 60 °C in 24 h. This optimization conditions were used for further studies. Identification of the Possible Binding Mode of the Phenyl Substituent in CAL-B. The substituents at the chiral carbon atom of 1a are phenyl and N-Boc-aminomethylene groups. Because each substituent must fit into one of the binding pockets of CAL-B for the reaction to occur, we first attempted to identify the binding pocket of the phenyl substituent in (S)-1a. Two model reactions were conducted with rac-1-phenylethanol (rac-4a) and diphenylmethanol (6) under the optimized reaction conditions (Scheme 1). CAL-B Scheme 1. Model Reactions To Identify the Binding Mode of the Phenyl Substituent
■
RESULTS AND DISCUSSION Screening Hydrolases for the Transesterification of rac-1a. To identify hydrolases with high enantioselectivity in the transesterification of rac-1a, 17 hydrolases were screened. The initial screening reactions were performed by adding a hydrolase (2 mg) to a mixture of rac-1a (0.05 mmol) and vinyl butanoate (3a, 0.15 mmol) in t-butyl methyl ether (t-BuOMe; 1 mL) at 30 °C. Although most of the hydrolases exhibited low catalytic activities (≤4% conversion after 24−48 h in Table S1) under the reaction conditions, Thermomyces lanuginosus lipase (TLL), Burkholderia cepacia lipase (BCL, PS-DI), Candida antarctica lipase A (CAL-A), and CAL-B (Novozym 435) yielded considerable conversions (12−50%) (Table 1). Among these lipases, TLL exhibited the highest activity and enantioselectivity (E > 200) with almost 50% of the desired ester (S)-2a within 1 h. This result presumably occurred because of TLL’s excellent catalytic activity for bulky secalcohols.16 BCL showed a moderate reaction conversion (22%) after 48 h but afforded the enantiopure product ((S)-2a, > 99% ee). A similar result using N-Cbz-protected rac-2-amino-1phenylethanol with n-PrCO2CH2CF3 was previously reported, but the resulting ester had low enantiopurity with 65% ee at 53% conversion (E = ∼ 10).15 CAL-A was the sole hydrolase that exhibited the opposite enantioselectivity, providing (R)-2a with 63% ee at 50% conversion in 6 h. In addition, CAL-B converted rac-1a to (S)-2a with a conversion of 12% in 48 h. We were intrigued by the unexpected reactivity of CAL-B because this reaction could only occur if the medium pocket of CAL-B was able to accept one of the large substituents (the phenyl or the N-Boc-aminomethylene) of (S)-1a, and this seemed unlikely based on the known size of CAL-B’s medium pocket. Then we optimized the CAL-B-catalyzed transesterification of rac-1a with 3a by varying the amount of CAL-B, the reaction temperature, the reaction time, the amount of 3a, the acylating agent instead of 3a, and reaction media (Table S2). After all, we
(Novozym 435) transformed rac-4a to (R)-5a with 51% conversion in 48 h (Scheme 1a), whereas 6 was converted to 7 but yielded only 2% conversion after 48 h (Scheme 1b). These results imply that the phenyl substituent cannot enter the medium pocket and instead should fit into the large pocket. In addition, we conducted model reactions using a series of rac-1-phenyl-1-alkanols (rac-4a−f) and measured the initial velocities to identify the appropriate substituent chain length for binding in the medium pocket under the optimized reaction conditions (Table 2). Rotticci et al. reported that the reaction Table 2. Effect of the Carbon Chain Length on the Reaction Rate of rac-1-Phenyl-1-alkanol (4)
entrya 1 2 3 4 5 6
1-phenyl-1-alkanol (4) 4a (n = 0) 4b (n = 1) 4c (n = 2) 4d (n = 3)e 4e (n = 4)e 4f (n = 5)e
conv. (%) c
50 50d 23 11 7 4
initial velocity (M·h−1)b 1.3 1.9 7.4 1.8 9.2 8.4
× × × × × ×
10−1 10−2 10−4 10−4 10−5 10−5
a
A mixture of rac-4 (0.05 mmol), 3a (0.15 mmol), and CAL-B (5 mg, Novozym 435) in t-BuOMe (1 mL) was shaken at 200 rpm at 60 °C. b For measuring initial velocities for 4a−f, the reaction scale was increased to 0.5 mmol of rac-4. c50% conversion of 4a after 1 h. d50% conversion of 4b after 3 h. eThe configuration of 4d, 4e, and 4f was confirmed by 1H NMR chiral analysis in comparison to the chemical shift of (R)-4c with (R,R)-KIM-Na.17 7460
DOI: 10.1021/acscatal.6b02192 ACS Catal. 2016, 6, 7458−7465
Research Article
ACS Catalysis rate dramatically decreased when compounds bearing a substituent longer than an ethyl group were used.8 In our case, the reaction with the substrate 4c bearing an n-propyl (n = 2) group exhibited a 25-fold lower rate than the substrate 4b containing an ethyl group at the chiral center. When the substrate contained an n-butyl (4d, n = 3), n-pentyl (4e, n = 4), or n-hexyl (4f, n = 5) group, the reaction rate dramatically decreased by factors of 100, 200, and 230, respectively. We also measured the initial velocity (2.0 × 10−3 M·h−1) for rac-1a under the same reaction conditions. It is noteworthy that the reaction of rac-1a was 24 times faster than that of 4f, although the length of the N-Boc-aminomethylene chain in rac1a is six atoms, which is similar in length as the n-hexyl group in 4f. This implies that the N-Boc-aminomethylene group binds in the medium pocket much better than the hexyl group. However, these results cannot explain how the binding pockets of CAL-B can accept the phenyl and N-Boc-aminomethylene group of 1a. Thus, how CAL-B accommodates rac-1a still remained unclear. Computer Modeling. We conducted computer modeling to understand the binding mode of rac-1a in CAL-B. The CALB-catalyzed transesterification of rac-1a proceeds through two tetrahedral intermediate steps as discussed above. Presumably, the enantioselectivity is determined at the second step.18 A critical feature for the reaction to proceed is the stabilization of the transition state, which is likely similar to the second tetrahedral intermediate. It is thought that CAL-B stabilizes the transition state through five key hydrogen bonds with the substrate 1a (a−e in Figure 2a); when these hydrogen bonds
Table 3. Key Hydrogen Bonds in a Proposed Tetrahedral Intermediate for the CAL-B-Catalyzed Transesterification of N-Boc-Protected (S)-2-Amino-1-phenylethanol ((S)-1a)
a
bonda
distance (Å)
angle
a b c d e
3.1 2.8 2.9 2.6 4.1
124.1° 155.3° 168.9° 162.3° 147.1°
The bonds are defined as in Figure 2a.
probably pushes the substrate and interferes with the productive binding of (S)-1a. Therefore, the current molecular modeling was unable to explain the productive binding of the substituents of (S)-1a. Design and Synthesis of a Transition-State Analogue. Because neither the model reactions nor the molecular modeling clearly revealed the binding mode of 1a in CAL-B, we decided to perform crystallographic analysis of a transitionstate analogue bound to CAL-B. To identify the binding mode of (S)-1a in the transition state within the medium pocket of CAL-B, the propyl phosphonate inhibitor (S)-8 was designed, and (S)-8a was successfully synthesized by incorporating (S)-1a and 4-nitrophenol with n-propylphosphonic dichloride (Scheme 2a).19 However, (S)-8a lacked sufficient solubility in Scheme 2. Transition-State Mimicking (a) Synthesis of (S)8. (b) Transition-State Analogue of CAL-B with the Phosphonate Inhibitor (S)-8
Figure 2. Proposed tetrahedral intermediate of CAL-B-catalyzed transesterification of N-Boc-protected rac-2-amino-1-phenylethanol (rac-1a). (a) A diagram of the tetrahedral intermediate. (b) An energy-minimized tetrahedral intermediate of the fast-reacting enantiomer, (S)-1a. Ser105, His224, Thr40, Leu278, and the fastreacting enantiomer represented by a stick model. The dotted lines represent hydrogen bonds.
are missing, no reaction occurs. Thus, we created a tetrahedral intermediate model of CAL-B with 1a for molecular modeling. Only a tetrahedral intermediate for the fast-reacting enantiomer (S)-1a was modeled because CAL-B showed exclusive enantioselectivity for the fast-reacting enantiomer. The model structure was minimized in terms of its energy (Figure 2b), and the distances and angles of the hydrogen bonds were then analyzed (Table 3). The distance (4.1 Å) between the oxyanion and the amide nitrogen of the Thr40 residue was longer than a typical H-bond distance (≤3.2 Å). Our model structure revealed that the t-butyl group of the NBoc-aminomethylene group points toward the Leu278 residue, which is located on the surface of CAL-B, and has close contact (∼3.6 Å) with the residue. In the model structure, Leu278
buffer solutions. To overcome this solubility issue, we substituted the N-Boc group of (S)-8a with the methoxycarbonyl (CH3OCO−) group by employing (S)-1g to form (S)-8g in 13% yield. When CAL-B is mixed with (S)-8 in a buffer solution, the hydroxyl group of the catalytic serine (Ser105) attacks the phosphorus atom in (S)-8 to expel the 4-nitrophenolic substituent and forms an O−P covalent bond between CALB and (S)-8 (Scheme 2b). The crystal structure of this protein−inhibitor complex will provide a transition-state analogue of (S)-1a with 3a by mimicking the tetrahedral intermediate, where the propyl phosphonic group in (S)-8 takes the place of the propyl carbonyl group of 3a. Hence, the 7461
DOI: 10.1021/acscatal.6b02192 ACS Catal. 2016, 6, 7458−7465
Research Article
ACS Catalysis
tances of 3.3 Å with C1 and 3.2 Å with C2), His224 (average carbon−carbon distance of 3.8 Å with C2), and Leu278 (average carbon−carbon distance of 3.3 Å with C3) and hydrogen bonding between the Nε-hydrogen of His224 (average nitrogen−oxygen distance of 3.3 Å with the oxygen of C2O). Surprisingly, compared with the native CAL-B (PDB code 1TCC), the side chain of Leu278 in the inhibitor-bound CAL-B is rotated away to accommodate the bulky CH3OCONHCH2− group in the medium pocket. This rotation allows the hydrophobic side chain of Leu278 to interact with (S)-8g’s phenyl group in the large pocket. Hence, notably, Leu278 mediates the interactions in both the large and medium pockets, and its rotation allows the bulky sec-alcohol to be a suitable CAL-B substrate. This observation affords the hypothesis that the rotation of the Leu278 residue is critical for the reaction to occur. Modeling the Molecular Basis of the Enantioselectivity of CAL-B Based on the Alternative Configuration of the Leu278 Residue. The crystal structure suggested that the side chain of Leu278 can rotate to create room to accommodate the N-Boc-aminomethylene group. A comparison of the pockets for substrate binding in CAL-B before and after the rotation of Leu278 explicitly showed that the rotation of the side chain of Leu278 produced an extra space to accommodate the substituent (Figure S2). Indeed, Leu278 in CAL-B has been known as one of the important residues for determining the catalytic activities of CAL-B.20 We again modeled the binding structures of 1a because the crystal structure contained (S)-N-methoxycarbonyl-2-amino-1-phenylethanol ((S)-1g), instead of 1a. Using the crystal structure as a guide, we first rotated the side chain of Leu278 and then minimized the tetrahedral intermediates for both enantiomers (Figure 4 and Table 4).
formation of the CAL-B/(S)-8g complex and its crystallization were attempted. Crystal Structure of the Transition-State Analogue of CAL-B. The crystal structure of CAL-B with the phosphonate inhibitor (S)-8g was determined (Tables S4 and S5) and analyzed (Figure 3). In the eight molecules of CAL-B found in
Figure 3. X-ray crystallographic analysis of the active pocket in CAL-B with the phosphonate inhibitor, (S)-8g. The active pocket structure of CAL-B with its modified catalytic Ser105 is shown. Important residues for inhibitor interactions are indicated. Inhibitor-containing CAL-B is illustrated in green via ribbon and surface diagrams, and CAL-B residues and the inhibitor are shown in yellow using stick representations. The rotation at the side chain of Leu278 relative to the native CAL-B (from PDB code 1TCC in cyan) is necessary to accommodate the inhibitor and, hence, the bulky sec-alcohol as a substrate.
the crystal asymmetric unit, all eight catalytic Ser105 residues were covalently modified by the phosphonate compound. In addition, Asn74, which undergoes glycosylation, was also identified with either two or four covalently attached sugar units in all cases. The sugar units were modeled as N-acetyl glucosamine (NAG). The eight molecules of CAL-B containing the phosphonate inhibitor were nearly identical with 0.47 Å rms deviation in all main-chain atoms. The inhibitor-containing CAL-B showed an average rms deviation of 0.37 Å (in Cα) compared with the native CAL-B (PDB code 1TCC), indicating that no gross change occurred in the CAL-B fold. However, the extra density in the Fo-Fc map clearly indicates the location of the transitionstate phosphonate inhibitor bound to Ser105 in the active pocket (Figure S3). Extensive networks of hydrogen and hydrophobic interactions mediate the association of CAL-B and the inhibitor, as expected for a productive binding mode. For instance, the Nε-hydrogen of His244, which is the critical residue of the His-Ser-Asp triad, interacts with the two phosphodiester oxygens via hydrogen bonding (average distances of 3.1 Å from Nε to the side-chain oxygen of Ser105 and of 3.0 Å from Nε to the phosphonate P−O oxygen). The main-chain amide nitrogens of Gln106 and Thr40 and the side-chain hydroxyl oxygen of Thr40 are within hydrogenbonding distances (average distance of 3.2, 2.8, and 2.6 Å, respectively) to the phosphonate PO oxygen. The large pocket, which is responsible for binding the phenyl group of (S)-8g, is formed by the hydrophobic residues of Ile189, Leu278, and Ile285 with average carbon−carbon distances of 3.9, 4.0, and 3.3 Å, respectively, to the phenyl group. The medium pocket recognizes the bulky C3H3OC2ONHC1H2− group in (S)-8g through hydrophobic interactions between Trp104 (average carbon−carbon dis-
Figure 4. Proposed tetrahedral intermediates: (a) for the fast-reacting enantiomer, (S)-1a, (b) for the slow-reacting enantiomer, (R)-1a. Ser105, His224, Thr40, Leu278, and the substrate are represented by a stick model. The dotted lines represent hydrogen bonds. The distance (bond e) (3.5 Å) between the oxyanion and the amide nitrogen atom of Thr40 in the model of the slow-reacting enantiomer is longer than a typical hydrogen bond length (≤3.2 Å).
This revealed that the N-Boc group could be accepted without hindrance from Leu278. The minimized structure included all the key hydrogen bonds for the fast-reacting enantiomer (S)-1a, but one hydrogen bond (e in Figure 4 and bond e in Table 4) was missing for the slow-reacting enantiomer, (R)-1a. These modeling data explained the high enantioselectivity of CAL-B for rac-1a and supported the hypothesis that the rotation of Leu278 is essential for the NBoc-aminomethylene group to bind in the medium binding pocket. In addition, the results suggest that the substitution of 7462
DOI: 10.1021/acscatal.6b02192 ACS Catal. 2016, 6, 7458−7465
Research Article
ACS Catalysis Table 4. Key Hydrogen Bonds in Possible Transition State Models of the CAL-B-Catalyzed Transesterification of (S)-1a and (R)-1a fast enantiomer, (S)-1a bonds a b c d e a
a
Table 5. Comparison of the Reaction Conversions in the Transesterification of rac-4a
slow enantiomer, (R)-1a
distance (Å)
angle
distance (Å)
angle
3.0 2.9 3.0 2.5 3.2
135.5° 158.3° 170.4° 173.5° 163.8°
2.7 3.1 3.1 2.5 3.5
151.5° 131.8° 167.6° 173.4° 156.2°
CAL-B
time (h)
% ee of (R)-5a
conv. (%)
E
Novozym 435 wild-typea L278Aa
3 12 12
>99.9 >99.9 >99.9
33 28 23
>200 >200 >200
a
The enzymes were immobilized onto porous spherical polymethacrylate beads with a protein loading of 10 mg/g of beads.
The bonds are defined as in Figure 2a.
Leu278 with a smaller residue would increase the reaction rate if the rotation of Leu278 is involved in the rate-determining step. Preparation of Leu278Ala CAL-B Mutant and Comparison of Its Reaction Conversion with That of the Wild-Type CAL-B. The X-ray crystallographic analyses and the modeling studies supported the hypothesis that the rotation of Leu278 can generate sufficient space to accept the N-Bocaminomethylene substituent of (S)-1a in the medium pocket. This suggests that the substitution of leucine with a smaller amino acid would result in a higher reaction conversion or enhanced reaction rate if the rotation of Leu278 is important in the rate-determining step. We prepared a Leu278Ala (L278A) CAL-B mutant that contained a smaller amino acid than the Leu278 residue. The wild-type CAL-B and L278A mutant were expressed in Pichia pastoris, and the relative temperature stabilities of the wild-type and L278A mutant enzymes were determined using 4-nitrophenyl palmitate as a substrate (Figure S5 and see Supporting Information). The L278A mutant enzyme exhibited lower thermal stability than the wild-type CAL-B. After immobilization of the wild-type and L278A mutant enzymes on porous spherical polymethacrylate beads with a protein loading of 10 mg/g of beads, we performed the transesterification using the immobilized forms of the wild-type CAL-B, the L278A mutant, and Novozym 435 on a substrate of rac-4a with phenyl and methyl substituents at the chiral center. Because the methyl group of (R)-4a is small enough to avoid contact with the hydrophobic side chain of Leu278, the reaction conversion should directly correlate with the enzyme’s intrinsic catalytic activity. Both the wild-type and L278A mutant enzymes showed lower reaction conversions than Novozym 435 (Table 5). The L278A mutant resulted in slightly lower conversion (23%) than the wild-type CAL-B (28%). These results imply some loss of intrinsic catalytic activity in L278A mutant. We next conducted the transesterification of rac-1a also using the immobilized forms of the wild-type CAL-B, the L278A mutant, and Novozym 435. The experiment with rac-1a differed from that with rac-4a because the N-Boc-aminomethylene substituent of rac-1a binds in the medium pocket. In such case, the reaction conversion of the L278A mutant should exceed that of the wild-type CAL-B because the rotation of Leu278 is not necessary for the L278A mutant to accept (S)-1a. As anticipated, the reaction conversion by the mutant was slightly higher (Table 6) but did not dramatically improve. However, if we consider the lower thermal stability and the reduced intrinsic catalytic activity of the L278A mutant enzyme compared to that of the wild-type CAL-B, the activity of the mutant for rac-1a might be higher than that of the wild-type
Table 6. Comparison of the Reaction Conversions in the Transesterification of rac-1a
CAL-Ba
temp (°C)
% ee of (S)-2a
conv. (%)
E
wild-type wild-type L278A L278A
30 60 30 60
>99.9 >99.9 >99.9 >99.9
10 32 12 34
>200 >200 >200 >200
a
The enzymes were immobilized onto porous spherical polymethacrylate beads with a protein loading of 10 mg/g of beads.
CAL-B. Thus, we conclude that the rotation of Leu278 plays an important role in the transesterification of a bulky sec-alcohol such as 1a. Additional Investigation into the Relatively High Activity of 1a Compared to That of 4f. The L278A mutant experiment, the crystal structure of the transition-state analogue, and the molecular modeling indicated that the rotation of Leu278 is crucial for the reaction with rac-1a to proceed. However, the reason for the lower reactivity of CAL-B for 4f (4% conversion after 48 h, entry 6 in Table 2) is unclear because the tetrahedral intermediate model structure of (R)-4f with the rotated side chain of Leu278 includes all key hydrogen bonds (Figure S1 and Table S3). The difference in the reactivity between 1a and 4f might relate to steric energies. Both substrates 1a and 4f may undergo distortion and bending at the γ-carbon of the hydroxyl group to bind in the active site of CAL-B. Therefore, the distorted structures presumably require additional steric energy with increased activation energy. When both substrates are distorted in the transitionstate, the γ-positioned −CH2− of 4f would be more crowded because of the presence of two hydrogen atoms, whereas the γpositioned −NH− of 1a would be less hindered because only one hydrogen atom exists at the γ-position. In addition, the conformation of the β- and γ-carbons of 4f should exist as eclipsed. Hence, the presence of the two hydrogen atoms at the γ-position would cause severe steric hindrance in the case of 4f. As a result, the binding of substrate 4f in the active site of CALB is expected to be slower than that of 1a with the reaction proceeding slowly. To prove this assumption, we prepared substrate rac-9 possessing the same number of atoms on both substituents (i.e., the n-hexyl and 2-methoxyethoxymethylene groups) (Scheme 3) and compared the reaction conversion with tridecan-7-ol, 11, bearing two identical n-hexyl substituents. Although the sizes of both substituents in the substrate rac-9 are very similar, the 2-methoxyethoxymethylene 7463
DOI: 10.1021/acscatal.6b02192 ACS Catal. 2016, 6, 7458−7465
Research Article
ACS Catalysis
located in the large pocket of CAL-B during enantioselective resolution with 3a. We tested the enantioselective resolution of various 2-amino-1-arylethanols (1) (Table 7). When a 4methoxy or 4-trifluoromethoxy substituent was introduced into the phenyl group of 1a, the reaction times to 50% conversion increased to 42 and 36 h, respectively (entries 2−3). In addition, heteroaryl substrates (1d−f) with the pyridinyl, furanyl, and thiophenyl groups showed similar activities (entries 4−6). The substrate 1g with a methoxycarbonylamino group instead of an N-Boc of 1a exhibited similar reactivity in the resolution reaction (entry 7). In contrast, the reaction 3-amino1-arylethanol (1h), which includes an additional methylene group between the hydroxyl and amino groups, was severely retarded, resulting in only 20% conversion after 48 h (entry 8). This slow conversion is probably attributable to the steric hindrance at the γ-position because the γ-carbon bears two hydrogen atoms, as in 4f.
Scheme 3. Comparison of CAL-B-Catalyzed Transesterification of a Substrate 9 Bearing No Hydrogen Atom at the γ-Position with Tridecan-7-ol (11)
■
substituent possesses no hydrogen atom at the γ-position. CALB (Novozym 435) can distinguish the substituents of rac-9 and the enantiomeric ratio (E) was determined as 8. This enantioselectivity must be from the binding of 2-methoxyethoxymethylene group to the medium pocket. It is noteworthy that the reaction conversion (47% for 48 h) was much higher than that in the reaction with 11 ( 200). Moreover, we presumed that a sec-alcohol substrate with less than one hydrogen atom at the γ-position is required to achieve moderate activity using CAL-B. This finding will expand the scope of substrates in the CAL-Bcatalyzed reactions.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.6b02192. General information, screening data, kinetic data, crystallographic analysis, enzyme mutagenesis and production, molecular modeling method, preparation of bulky sec-alcohols, and biotransformation data (PDF)
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail:
[email protected] (S. Y. Park). *E-mail:
[email protected] (S. Park). *E-mail:
[email protected] (H. Lee). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2013R1A1A2005276 and 2014R1A2A2A01004836). Experiments at the Pohang Light Source (PLS) were supported in part by MEST and the Pohang University of Science and Technology. H.L. acknowledges the Korea Research Institute of Chemical Technology (SI1605). The authors also thank Prof. H. Kim (KAIST) for the kind donation of the (R,R)-KIM-Na. This paper is dedicated to Prof. Romas Kazlauskas for the occasion of his 60th birthday.
■
REFERENCES
(1) (a) Anderson, E. M.; Larsson, K. M.; Kirk, O. Biocatal. Biotransform. 1998, 16, 181−204. (b) Jaeger, K. E.; Dijkstra, B. W.; Reetz, M. T. Annu. Rev. Microbiol. 1999, 53, 315−351. (c) Kirk, O.; Christensen, M. W. Org. Process Res. Dev. 2002, 6, 446−451. (d) Houde, A.; Kademi, A.; Leblanc, D. Appl. Biochem. Biotechnol. 2004, 118, 155−170. (e) Bornscheuer, U. T.; Kazlauskas, R. J. Hydrolases in Organic Synthesis: Regio- and Stereoselective Biotransformations, 2nd ed.; Wiley-VCH: Weinheim, 2006. (f) Naik, S.; Basu, A.; Saikia, R.; Madan, B.; Paul, P.; Chaterjee, R.; Brask, J.; Svendsen, A. J. Mol. Catal. B: Enzym. 2010, 65, 18−23. (g) Sharma, D.; Sharma, B.; Shukla, A. Biotechnology 2011, 10, 23−40. (h) Andualema, B.; Gessesse, A. Biotechnology 2012, 11, 100−118. (2) (a) Rotticci, D.; Ottosson, J.; Norin, T.; HuIt, K. In Methods in Biotechnology: Enzymes in Nonaqueous Solvents; Vulfson, E. N., Halling, P. J., Holland, H. L., Eds.; Humana Press: Totowa, NJ, 2001; Vol. 15, pp 261−276. (b) Park, S.; Kazlauskas, R. J. J. Org. Chem. 2001, 66, 8395−8401. (3) (a) Santaniello, E.; Casati, S.; Ciuffreda, P. Curr. Org. Chem. 2006, 10, 1095−1123. (b) Patterson, L. D.; Miller, M. J. J. Org. Chem. 2010, 75, 1289−1292. (c) Alatorre-Santamaría, S.; Gotor-Fernández, V.; Gotor, V. Eur. J. Org. Chem. 2011, No. 6, 1057−1063. (4) (a) Uppenberg, J.; Hansen, M. T.; Patkar, S.; Jones, T. A. Structure 1994, 2, 293−308. (b) Uppenberg, J.; Oehrner, N.; Norin, 7465
DOI: 10.1021/acscatal.6b02192 ACS Catal. 2016, 6, 7458−7465