Structural and Computational Insight into the Catalytic Mechanism of

Dec 12, 2017 - The final Hessian was computed to check the existence of a single imaginary frequency (saddle point) or no imaginary frequency (minimum...
0 downloads 15 Views 1MB Size
Subscriber access provided by READING UNIV

Article

Structural and Computational Insight into the Catalytic Mechanism of Limonene Epoxide Hydrolase Mutants in Stereoselective Transformations Zhoutong Sun, Lian Wu, Marco Bocola, H.C. Stephen Chan, Richard Lonsdale, Xu-Dong Kong, Shuguang Yuan, Jiahai Zhou, and Manfred T. Reetz J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.7b10278 • Publication Date (Web): 12 Dec 2017 Downloaded from http://pubs.acs.org on December 12, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of the American Chemical Society is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Structural and Computational Insight into the Catalytic Mechanism of Limonene Epoxide Hydrolase Mutants in Stereoselective Transformations Zhoutong Sun†#, Lian Wu‡#, Marco Bocola§ξ, H.C. Stephen Chan ¶, Richard Lonsdale§ξ, Xu-Dong Kong ‡ , Shuguang Yuan ¶*, Jiahai Zhou‡*, and Manfred T. Reetz†§ξ* †

Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, 32 West7th Avenue, Tianjin Airport Economic Area, Tianjin 300308, China ‡ State Key Laboratory of Bio-organic and Natural Products Chemistry, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai 200032, China § Max-Planck-Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mülheiman der Ruhr, Germany ξ Fachbereich Chemie der Philipps-Universität, Hans-Meerwein-Strasse, 35032 Marburg, Germany ¶ Laboratory of Physical Chemistry of Polymers and Membranes, Ecole Polytechnique Fédérale de Lausanne (EPFL),CH B3 495 (Bâtiment CH) Station 6, CH-1015 Lausanne, Switzerland ABSTRACT: Directed evolution of limonene epoxide hydrolase (LEH), which catalyzes the hydrolytic desymmetrization reactions of cyclopentene oxide and cyclohexene oxide, results in (R,R)- and (S,S)-selective mutants. Their crystal structures combined with extensive theoretical computations shed light on the mechanistic intricacies of this widely used enzyme. From the computed activation energies of various pathways, we discover the underlying stereochemistry for favorable reactions. Surprisingly, some of the most enantioselective mutants that rapidly convert cyclohexene oxide do not catalyze the analogous transformation of the structurally similar cyclopentene oxide, as shown by additional X-ray structures of the variants harboring this slightly smaller substrate. We explain this puzzling observation on the basis of computational calculations which reveal a disrupted alignment between nucleophilic water and cyclopentene oxide due to the pronounced flexibility of the binding pocket. In contrast, in the stereoselective reactions of cyclohexene oxide, reactive conformations are easily reached. The unique combination of structural and computational data allows insight into mechanistic details of this epoxide hydrolase and provides guidance for future protein engineering in reactions of structurally different substrates.

INTRODUCTION In studies focusing on the elucidation of enzyme mechanisms, stereoselectivity has traditionally served as a unique probe.1 More recently, the rapidly growing area of directed evolution of stereoselective enzymes as practical catalysts in organic chemistry and biotechnology continues to generate a huge set of potentially useful data,2 which in principle can be used to deepen our knowledge of the intricacies of enzyme mechanisms. In endeavors of this kind, crystal structures of enantioselective mutants,3 ideally complemented by kinetic studies and molecular dynamics (MD) computations, are particularly revealing. Such an approach is especially powerful when the directed evolution study includes Xray structures of both (+)- and (-)-selective mutants which enable access to both enantiomeric products.3a Any notable increase or decrease in activity (in the extreme case complete shutdown) also constitutes a handle for studying mechanisms. Epoxide hydrolases (EHs) occur widely in many organisms and plants, catalyzing the hydrolysis of epoxides with formation of the corresponding 1,2-diols. The respective biological functions vary according to the organism of origin, which include biosynthesis of natural products, detoxification of toxic epoxides, and cellular signaling.4 LEHs have also been used extensively in organic chemistry for stereoselective synthesis of chiral vicinal diols (and also of epoxides).5 In the present study, we focus on limonene epoxide hydrolase LEH)6 and utilize extensive information obtained from stereoselective mutants, kinetics, X-ray analyses and MD computations

with the aim of learning more about this widely used enzyme. The hydrolytic desymmetrization reactions of the cyclic epoxides 1 (dubbed CYO1) and 3 (dubbed CYO3), respectively, were used as the model LEH-catalyzed transformations (Scheme 1). Wildtype (WT) LEH is a poor catalyst in both reactions (13% ee favoring (R,R)-2, and 4% ee favoring (S,S)-4). These transformations were previously employed as the experimental platform for developing

Scheme 1. LEH-catalyzed hydrolytic desymmetrization of epoxides 1 and 3.

Scheme 2. The proposed catalytic mechanism of LEH6a,11

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 9

Table 1. The best LEH variants as catalysts in the hydrolytic desymmetrization of substrates 1 and 3.

1 Variant

Sequence

WT

3

ee%

c%

ee%

c%

13(R,R)

99

4(S,S)

99

Source

SZ92

L74F/M78F/L103V/L114V/I116V/F139V/L147V

72(S,S)

6

92(S,S)

99

SZ338

L74F/M78V/I80V/L114F

-

-

96(R,R)

83

SZ348

I80Y/L114V/I116V

-

≤5

99(S,S)

97

TCSM7a

SZ529

M32V/M78V/I80V/L114F

10(R,R)

8

97(R,R)

99

TCSM/ISM7a

SZ718

L74F/I80F/L114V/I116V/F139V

92(S,S)

56

95(S,S)

97

ISM/

SZ719

L74F/M78F/I80F/L114V/I116V/F139V

92(S,S)

85

97(S,S)

98

This study

SCSM/ISM3a

Figure 1. The overall structure and flexibility of LEH. (A) Cartoon structure of LEH. The α-helixes and β-strands were colored by red and blue respectively. Yellow spheres indicate the location of catalytic site. (B) The sequence secondary structure of LEH. (C) The intrinsic flexibility of LEH calculated from PCA analysis. The cyan vector length correlates with the domain-motion scale. Yellow spheres indicate the location of catalytic site. optimal directed evolution strategies which ensure the generation of high-quality mutant libraries with minimal screening effort3a,7 (bottleneck of directed evolution).8 Among them, single code saturation mutagenesis (SCSM)3a based on only one amino acid as building block in saturation mutagenesis at relatively large randomization sites, and triple code saturation mutagenesis (TCSM)7a employing three amino acids, combined with iterative saturation mutagenesis (ISM),9 were systematically explored. TCSM proved to be the most efficient method,7a leading to high-quality mutant libraries which require limited screening.8 Relevant is a recent Rosetta-based study on computationally designed LEH mutants as catalysts in the desymmetrization of cyclopentene oxide (1) and cyclohexene oxide (3), requiring the screening of only 37 variants with the identification of mutants showing moderate to pronounced enantioselectivity.10 Surprisingly, in our study some of the evolved highly stereoselective LEH mutants were found to catalyze the rapid reaction of

cyclohexene oxide 3 (CYO3), but not of cyclopentene oxide 1 (CYO1).3a In principle, CYO1 is more strained than CYO3 and can be expected to be hydrolyzed faster. Although the crystal structures of WT LEH and several mutants have been determined, which in some cases help to explain the source of stereoselectivity qualitatively, they are of limited use in revealing the observed drastic switches in substrate specificity.3a As already pointed out, previous directed evolution studies led to moderate to excellent improvements in (R,R)- as well as (S,S)-selectivity for both substrates (Table 1). To understand why some LEH variants are specifically active toward CYO1, whereas others are selective for CYO3, and why stereoselectivity is achieved, we crystalized three LEH mutants (SZ338, SZ348 and SZ529) in complex with CYO1 or its diol product. Coupled with new mutagenesis experiments, molecular docking, MD simulation as well as quantum mechanical (QM) calculations, the overall results shed new light on mechanistic details of LEH-catalyzed

ACS Paragon Plus Environment

Page 3 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Figure 2. X-ray electron density map of LEH variants complex for (A) SZ529, (B) SZ348 and (C) SZ338. Black grid: 2Fo-Fc at 1.2σ level. Green grid: Fo-Fc at 3.2σ level. Red grid: Fo-Fc at -3.2σ level. White sticks: side chain of key residues in the catalytic site of LEH. The inactive substrate CYO1 (yellow balls-and-sticks) is in multiple-conformation state in the complex structures of SZ529-CYO1 and SZ348CYO1, as well as in the resolved complex structures of SZ338-CYO1. stereoselectivity transformations. The basic mechanistic features derived from an earlier structural investigation6a are shown in Scheme 2, which are supported by a recent QM study.11 RESULTS AND DISCUSSION. Newly Evolved LEH Mutants Improve Enantioselectivity in Reactions of Substrate CYO1. In order to answer the puzzling question why some LEH mutants are completely inactive against CYO1, yet drastically enhancing enantioselectivity in the desymmetrization of CYO3, we screened several mutant libraries previously evolved for CYO3 (Table S1). Several variants were found to be active against CYO1, and had already been identified in the original libraries using CYO3 as substrate, e.g., SZ80 (L114V/I116V/F139V). However, their enantioselectivities were found to be poor (66% ee in favor of (S,S)-2, 80% conversion in the case of SZ80). To optimize enantioselectivity, SZ80 was used as a template for iterative saturation mutagenesis (ISM)9 at the 5residue randomization site lining the binding pocket L74/M78/I80/L103/L147 using SCSM based on phenylalanine as the only building block. Two optimal mutants SZ718 (L74F/I80F/L114V/I116V/F139V) and SZ719 (L74F/M78F/I80F/L114V/I116V/F139V) were discovered, showing a selectivity of 92% ee (S,S) and conversion amounting to 56~85%. More importantly, these two variants can also hydrolyze the larger CYO3 with ee-values of 95% and 97% respectively, yielding (S,S)-4. For comparison purpose, the present and previous results are summarized in Table 1. Crystal Structures and MD Simulations Reveal Flexibility of LEH. In order to throw light on the origin of enhanced LEH enantioselectivity, we determined the crystal structures of mutants SZ529 and SZ348 (Table S2), each bound with inactive substrate CYO1 (pdb: 5YAO,5YNG). We also re-solved the previously reported complex structure of SZ338 with CYO1 as inert “guest” in the binding pocket (pdb: 5YQT)3a by removing all suspicious water molecules and adding multiple ligand conformations. The overall structure of LEH comprises four α-helices and six βstrands, which fold into a compact catalytic pocket (Figure 1). The catalytic trait of LEH is composed of Y53, N55, R99, D101 and D132 (Figure 2). Both Y53 and N55 are located in β1, whereas both R99 and D101 reside in β4. D132 is in β6. Moreover, we found three flexible regions within the vicinity of the substrate binding pocket, the C-terminal loop (C loop), helix 4 (H4) and loop A. Principal component analysis (PCA)12 on the

MD trajectory of Apo WT LEH (pdb: 1NU3)6a indicates that these three motifs can undergo noticeable movements, that change both the shape and volume of the catalytic zone (Figure 1 and movie S1). This phenomenon was also confirmed by the root-meansquare fluctuation (RMSF) of LEH, which was calculated from the MD simulation trajectory (Figure S1). In the complex structures of SZ529-CYO1 and SZ348-CYO1 (Figure 2), we found that the inactive substrate CYO1 is quite flexible in the binding site, adopting two different conformations which fit both 2Fo-Fc and Fo-Fc map very well. Interestingly, a conserved water molecule between D132 and Y53 (Figure 2B and 2C), which is considered to be essential for the catalytic step,6a has been expelled from the catalytic center (Figure 2A). In our previously published crystal structures of SZ338 and SZ348, the Osubstrate-OD101 and Osubstrate-OY53 distances appear too short (2.2-2.4Å) for a typical hydrogen-bond interaction (2.7-3.2Å).3a,6a Considering the low occupancy of CYO1 in the asymmetric units of SZ338, we therefore re-refined the crystal structures of SZ338CYO1 (Figure 2C) introducing multiple orientations for the CYO1 molecule. In all crystal structures, the CYO1 molecules in various orientations fit the electron density maps perfectly in both SZ338-CYO1 and SZ348-CYO1. No close contact is observed for both crystal water and CYO1 molecules in the catalytic regions. Moreover, the highly conserved water molecules in both SZ338CYO1 and SZ348-CYO1 structures are maintained between D132 and Y53 (Figure 2B and 2C). Furthermore, the orientations of CYO1 in SZ338-CYO1, SZ348-CYO1 and SZ529-CYO1 are different from each other, confirming that CYO1 rotates freely in the binding pocket. Such high flexibility of CYO1 hinders perfect alignment of activated water and substrate in the transition state, as also indicated in the crystal structures, is responsible for its inactivity towards SZ338, SZ348 and SZ529. To further confirm these assessments, we performed extensive all-atom MD simulations for the complexes SZ338-CYO1, SZ338-CYO3, SZ348-CYO1, SZ348-CYO3, SZ529-CYO1, SZ529-CYO3, SZ719-CYO1 and SZ719-CYO3. The CYO3 complexes were constructed by molecular docking. It is worth noting that mutants SZ338, SZ348 and SZ529 are either inactive or show extremely low activity toward CYO1. In all cases, CYO1 re-orientates

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3. The Substrate binding modes and protein-ligand interaction fingerprint for (A) SZ529-CYO3, (B) SZ719-CYO1 and (C) SZ719CYO3. Dash: hydrogen-bond interactions. Yellow balls-and-sticks: substrate molecule. White sticks: side chains of key residues in the catalytic pocket. Protein-ligand interaction fingerprint analysis (IFP) are indicated by cyan radar plots. The statistical interaction frequency in the IFP plot was obtained from calculating the protein-ligand interaction frequency in the MD simulations. Interestingly, we also found that the size and shape of the catalytic site in each mutant are noticeably different. The volumes of the catalytic regions are 107 ± 1 Å3, 98 ± 1 Å3 and 102 ± 1 Å3 for mutants SZ338, SZ348 and SZ529 respectively, whereas the volume of CYO1 is only 62 ± 1 Å3. Such noticeable differences are responsible for the CYO1 flexibilities in both crystal structures and all-atom MD simulations (Table S3). However, the volume of SZ719 is much more limited, with a value of 86 ± 1 Å3 which can accommodate CYO1 in a more stable conformation. The computed volume of the CYO3 molecule amounts to ~75 ± 1 Å3, which results in a much more stable interaction within all mutants. Combining the observations from both crystal structures and MD simulations, we conclude that the volume of the inactive LEH variants are too big to stabilize a proper near attack conformation (NAC)13 for the smaller CYO1 substrate prior to reaching the transition state. In such an environment, CYO1 continuously alters its conformations in the catalytic site. However, with a Figure 4. The calculated activation energy barrier and substratelimited space, as in the case of mutant SZ719, both CYO1 and water distances of SZ719. (A) The activation energy barrier of CYO3 are held in active poses, leading to (S,S)-2 and (S,S)-4, SZ719-CYO1. Distance in blue was calculated by QM methods. respectively. (B) Distances of C1-O2 and C2-O2 along MD simulations of Shedding Further Light on the Active Substrate Binding SZ719-CYO1. (C) The activation energy barrier of SZ719-CYO3. Mode of LEH. In order to elaborate the atomic details of active Distance in blue was calculated by QM methods. (D) Distances of substrate-LEH interactions, we performed molecular docking and C1-O2 and C2-O2 along MD simulations of SZ719-CYO3 all-atom MD simulations and protein-ligand interaction fingerprint analysis (IFP) (Figure 3 and Figure S4) for active subitself freely in the trajectories. This is suggested by the higher strates in complex with LEH, including the complexes SZ338root-mean-square deviations (RMSD) of both protein and that of CYO3, SZ348-CYO3, SZ529-CYO3, SZ719-CYO1 and SZ719CYO1 (Figure S2 and S3, movie S2-S4). In contrast, SZ719 is CYO3. In each case, the oxygen atom in the epoxide consistently active on CYO1, which remains stable in the catalytic zone with a establishes a strong hydrogen-bond interaction with the protonatvery low RMSD value (Figure S3, movie S5). Identically, all ed studied mutants are active on CYO3, which also exhibits a more confined orientation in each case (Figure S3, movies S6-9).

ACS Paragon Plus Environment

Page 4 of 9

Page 5 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Figure 5. The basis of LEH catalytic mechanism and stereoselectivity. (A) When the substrate approaches LEH, domain H4, C loop and loop A can undergo noticeable change, opening a space for the substrate entrance. A highly conserved water molecule is stabilized by a dedicated hydrogen-bond network, formed by Y53, N55 and D132. (B) Once the substrate sits in the catalytic site, the conserved water molecule can attack the epoxide ring in two different ways: at carbon C1 and C2. (C) If the distance between the water oxygen atom and C1 (d1) is shorter than that of C2 (d2), formation of (R,R)-products is favored, due to the lower activation energy. If d1>d2, LEH will produce (S,S)-products. D101,6a which is involved in a dedicated hydrogen-bond network with R99, D132, W130, Y53, N55 and a conserved water molecule W1. Among the LEH variants, different mutations in the vicinity of catalytic sites alter the hydrophobic interaction profiles. In SZ338, hydrophobic interactions are found between CYO3 and L35-L103-W130-F134, whereas CYO3 interacts with Y80-L103-L114-F134 in SZ348. In SZ529, identical interactions with L103-I116-F134-F139 are also observed. Both CYO1 and CYO3 are hydrolyzed in SZ719. The IFPs of SZ719-CYO1 and SZ719-CYO3 are similar to each other, both of them interact with F78-F80-L103-V116-F134-V139, although the interaction frequency of each residue varies slightly. Interestingly, the number of residues involved in the IFP of SZ719 (7 residues) is much higher than that of SZ338, SZ348 and SZ529 (5 residues). This is mainly because SZ719 has a much smaller catalytic site (volume = 86 ± 1 Å3) than any other variants (volumes = 98 - 107 ± 1 Å3). This enables SZ719 to undergo more compact interactions with substrates. Revealing the Source of Stereoselectivity. Quantum mechanics, especially the density functional theory (DFT), is an accurate and efficient method for the study of enzyme mechanisms.14 In order to understand the origin of evolved enantioselectivity, a large active-site model (280-300 atoms) based on crystal structures was designed for a QM study. In this model, we include a highly conserved water molecule, the substrate, and residues R99, W130, Y53, D132, V116, F134, N55, D101, V114, F80, L103, F78, V143, V139 as well as F74 in the corresponding positions of mutant SZ719. The geometry of each catalytic site was submitted to geometry optimization by QM method prior to transition state scan. Mutant SZ719 is capable of hydrolyzing both CYO1 and CYO3, yielding (S,S)-2 and (S,S)-4 molecules, respectively (Scheme 1 and Figure 4). All LEH variants, in principle, can generate two different stereoisomers (Schemes 1 and 2). If the highly conserved water molecule W1 (Figure 4) attacks the C2 carbon of the substrate, (S,S)-2 or (S,S)-4 are generated for CYO1 and CYO3, respectively. In contrast, if W1 (Figure 4) attacks the C1 carbon of the substrate, the enantiomers (R,R)-2 or (R,R)-4 are generated. We succeeded in gaining insight into the origin of stereoselectivity as follows: In both complexes SZ179-CYO1 and SZ719-CYO3, QM calculations and statistics on the distances of d1 (O2-C1) (corresponding to formation of (R,R)-2) and d2 (O2C2) (corresponding to formation of (S,S)-4) from MD simulations indicate that d1 is noticeably longer than d2, with statistical p-

values