Structure-Based Engineering of a Sesquiterpene Cyclase to Generate an Alcohol Product: Conversion of epi-Isozizaene Synthase into α-Bisabolol Synthase

Abstract

The sesquiterpene cyclase epi-isozizaene synthase (EIZS) from Streptomyces coelicolor catalyzes the metal-dependent conversion of farnesyl diphosphate (FPP) into the complex tricyclic product epi-isozizaene. This remarkable transformation is governed by an active site contour that serves as a template for catalysis, directing the conformations of multiple carbocation intermediates leading to the final product. Mutagenesis of residues defining the active site contour remolds its three-dimensional shape and reprograms the cyclization cascade to generate alternative cyclization products. In some cases, mutagenesis enables alternative chemistry to quench carbocation intermediates, e.g., through hydroxylation. Here, we combine structural and biochemical data from previously-characterized EIZS mutants to design and prepare F95S-F198S EIZS, which converts EIZS into an α-bisabolol synthase with moderate fidelity (65% at 18 ºC, 74% at 4 ºC). We report the complete biochemical characterization of this double mutant as well as the 1.47 Å resolution X-ray crystal structure of its complex with three Mg²⁺ ions, inorganic pyrophosphate, and the benzyltriethylammonium cation, which partially mimics a carbocation intermediate. Most notably, the two mutations together create an active site contour that stabilizes the bisabolyl carbocation intermediate and positions a water molecule for the hydroxylation reaction. Structural comparison with a naturally-occurring α-bisabolol synthase reveals common active site features that direct α-bisabolol generation. In showing that EIZS can be redesigned to generate a sesquiterpene alcohol product instead of a sesquiterpene hydrocarbon product, we have expanded the potential of EIZS as a platform for the development of designer cyclases that could be utilized in synthetic biology applications.

Graphical Abstract

Introduction

Terpene natural products are secondary (a.k.a. specialized) metabolites that are found in all domains of life.^1–3 Despite the extraordinarily diverse array of hydrocarbon skeletons that continue to be discovered in this family of more than 102,000 natural products,⁴ all derive from a mere handful of simple isoprenoid precursors.^5–8 This remarkable chemodiversity is largely attributed to the activity of terpene cyclases, enzymes that catalyze multi-step reaction sequences that transform linear isoprenoid substrates into structurally complex products typically containing one or more rings and stereocenters.^8–14 Notably, terpenoid natural products have been essential components of the pharmacopeia since times of antiquity. For example, myrrh was used by Hippocrates (470–377 B.C.E.) as a pain-killer.¹⁵ This analgesic effect was later determined to result from two polycyclic sesquiterpenes, furanoeudesma-1,3-diene and curzarene; since the effect was reversed by naloxone, the mode of action implicated sesquiterpene binding to brain opioid receptors.¹⁶ More recent examples of medicinally useful terpenoids include the sesquiterpene artemisinin, an antimalarial drug,^17,18 and the diterpene paclitaxel (Taxol), a drug used for cancer chemotherapy.^19–22

The three-dimensional contour of a terpene cyclase active site serves as a template for catalysis by chaperoning the conformation of a flexible isoprenoid substrate required to generate a specific product.^10–14 Following ionization of the enzyme-bound substrate, the hallmark of a terpene cyclase reaction is a highly-controlled progression of multiple high-energy carbocation intermediates leading to the final product. Aromatic residues largely define the active site contour and stabilize these intermediates and their flanking transition states through cation-π interactions. Carbocation intermediates are generally protected from premature quenching by bulk solvent due to protein conformational changes that fully enclose and encapsulate the hydrophobic active site. Active site closure is triggered by molecular recognition of the substrate diphosphate group through metal coordination and hydrogen bond interactions. While isoprenoid cyclization cascades are often quenched by proton elimination from the final carbocation intermediate, they can also be quenched by water to generate terpene alcohol products. In such cases, a water molecule is thought to be sequestered in the active site along with the substrate.

Remolding the active site contour of a terpene cyclase through mutagenesis can reprogram the cyclization cascade by diverting the reaction sequence at specific carbocation checkpoints to generate alternative cyclization products. Consider the class I sesquiterpene cyclase, epi-isozizaene synthase (EIZS) from Streptomyces coelicolor, which catalyzes the magnesium-dependent transformation of farnesyl diphosphate (FPP) into the tricyclic hydrocarbon epi-isozizaene in the biosynthesis of albaflavenone antibiotics.^23,24 EIZS is a high-fidelity cyclase, generating 99% epi-isozizaene at 4 °C (Figure 1); however, biosynthetic fidelity is compromised with increasing temperature, with 79% epi-isozizaene generated at 30 °C along with a mixture of alternative products.^25,26 The generation of alternative minor products hints at the potential of generating an even wider array of alternative minor and even major cyclization products through protein engineering.

Figure 1.

Mechanisms of EIZS-catalyzed cyclization of farnesyl diphosphate (FPP) forming epi-isozizaene or α-bisabolol. The (1R)-bisabolyl cation is a catalytic checkpoint directing the reaction sequence toward the (7S)-homobisabolyl cation (green arrow) and ultimately epi-isozizaene, or toward α-bisabolol (blue arrow).

To date, site-directed mutagenesis has been used to reprogram the EIZS-catalyzed cyclization cascade to generate a multitude of alternative products with varying degrees of fidelity.^25–30 Intriguingly, select polar amino acid substitutions in the active site introduce hydroxylation activity, in which a carbocation intermediate is quenched by the addition of a water molecule to generate a sesquiterpene alcohol. For example, F96H EIZS generates nerolidol, an acyclic sesquiterpene alcohol, with 73% fidelity; the crystal structure of this variant reveals how H96 can stabilize and position a water molecule in the active site that is likely utilized in catalysis.^27,30 Other single-point mutants – F95N, F95Q, F95C, and W203H – generate trace amounts of α-bisabolol through the hydroxylation of the bisabolyl carbocation (Figure 1).²⁷ α-Bisabolol is a monocyclic sesquiterpene alcohol with anti-inflammatory and anti-microbial properties.³¹ As the predominant sesquiterpene found in German chamomile, α-bisabolol is thought to be responsible for the calming effects of chamomile tea, and it is also used in cosmetic products due to its skin-healing properties.^31–34

We hypothesized that we could improve upon the production of α-bisabolol by F95 mutants by introducing a second mutation to expand the volume of the active site, which might facilitate the binding of a water molecule along with substrate FPP. We focused on the recently characterized³⁰ single-point mutants F95S EIZS and F198S EIZS. Neither mutant generates measurable quantities of α-bisabolol, but each retains the ability to generate the α-bisabolyl carbocation. We reasoned that the expanded active site volume of the double mutant, F95S-F198S EIZS, would enable simultaneous accommodation of the α-bisabolyl carbocation and a water molecule, which in turn might enable the generation of a sesquiterpene alcohol. To be clear, we did not design the F95S-F198S double mutant based on the product arrays generated by either single-point mutant, but instead designed the double mutant solely based on structural analysis of the active site contour. Pleasingly, F95S-F198S EIZS is an α-bisabolol synthase, generating 65% α-bisabolol as its major cyclization product; additionally generated are 15% nerolidol, 5% β-curcumene, and 12% uncharacterized sesquiterpene hydrocarbon products. The 1.47 Å resolution crystal structure of F95S-F198S EIZS complexed with 3 Mg²⁺ ions, inorganic pyrophosphate (PP_i), and the benzyltriethylammonium cation (BTAC) confirms the enlarged active site volume of the double mutant and reveals the binding of water that appears to be well-positioned for hydroxylation chemistry.

Materials and Methods

Reagents.

All reagents for cloning and protein preparation were purchased from Fisher, Sigma, or GoldBio. All crystallization reagents were purchased from Hampton Research, and ligands for co-crystallization were purchased from Fisher. FPP was purchased from Isoprenoids, LC. All reagents, unless noted otherwise, were used without additional preparation or purification.

Cloning.

As we previously described,³⁰ a pET28 vector containing the gene encoding wild-type EIZS, codon-optimized for expression in E. coli, was synthesized by and purchased from Genscript. The F95S mutation was introduced into the wild-type gene using mutagenic primers designed by the NEBaseChanger tool and a Q5 mutagenesis kit (New England Biolabs), per the manufacturer’s protocol as implemented in our previous study.³⁰ Here, the F198S mutation was introduced into the F95S EIZS gene in the pET28 plasmid following the same protocol, using mutagenic primers designed by the NEBaseChanger Tool (5’-GAACTGCGCCGCCTCACGAGCGCGCACTGGATCTGGA-3’ [forward] and 5’-GAGGTACTCCTCCACGCCGGGCACGA-3’ [reverse]) and a Q5 mutagenesis kit (New England Biolabs) according to the manufacturer’s protocol. Briefly, amplification of the template plasmid with the mutant primers yields a linear DNA product, which is then circularized by the KLD enzyme mix (New England Biolabs) provided in the mutagenesis kit. The ligated plasmid DNA from this reaction was used to transform NEB-5α cells. The final nucleotide sequence of the F95S-F198S EIZS gene used in this study is detailed in Figure S1. Transformed cells were plated on selective media and grown overnight at 37 °C. Individual colonies were picked and grown in 5 mL of LB media supplemented with 50 μg/mL kanamycin (overnight, 250 rpm, 37 °C), and plasmid DNA was subsequently extracted using a Qiagen MiniPrep kit. Plasmid DNA samples were evaluated for successful mutagenesis via sequencing performed at the DNA sequencing facility at the University of Pennsylvania.

Protein Expression and Purification.

BL21(DE3) cells, containing the sequence-confirmed F95S-F198S EIZS pET28 plasmid, were inoculated into 6 × 1-L LB media supplemented with 50 μg/mL kanamycin and grown in an orbital shaker (250 rpm, 37 °C) until each culture reached an OD₆₀₀ of ~0.6. Cultures were cooled at 4 °C for 30 min, and protein expression was then induced upon the addition of isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 0.1 mM. Cultures were returned to the orbital shaker (overnight, 250 rpm, 18 °C) for expression. Cells were pelleted by centrifugation (15 minutes, 6000 rpm, 4 °C) and the pellet was stored at −80 °C until purification.

The cell pellet was thawed in 75 mL of HisTrap buffer A [50 mM sodium phosphate (pH 7.8), 300 mM NaCl, 20% (v/v) glycerol, 2 mM β-mercaptoethanol]. A total of 100 mg lysozyme, 10 mg DNAse, and 2 EDTA-free cOmplete Mini Protease Inhibitor tablets (Roche) were added to the lysis solution, which was gently stirred for 60 min at 4 °C. Cells were disrupted by sonication (8 min total: 1 s on, 2 s off at 30% amplitude) and the homogenate was clarified by centrifugation (60 minutes, 18000 rpm, 4 °C). The supernatant was loaded at 3 mL/min onto a 5-mL TALON Cobalt column (GE Healthcare) pre-equilibrated with buffer A. Bound proteins were eluted from the column via a step to 10% buffer B [buffer A augmented with 200 mM imidazole] for 3 column volumes, followed by a 50 mL gradient to 100% buffer B. Fractions containing EIZS were pooled for subsequent steps.

Pooled fractions (10 mL) were filtered with a 0.22-μm PVDF syringe filter (Millipore) and loaded onto a HiLoad 26/600 Superdex 200 gel filtration column, pre-equilibrated with sizing buffer [20 mM Tris-HCl (pH 7.4), 300 mM NaCl, 10 mM MgCl₂, 1 mM TCEP, 10% (v/v) glycerol] for size-exclusion chromatography. Fractions containing EIZS were determined to be >95% pure by SDS-PAGE, pooled, and concentrated to 7.5 mg/mL with an Amicon 30-kDa molecular weight cut-off centrifugal concentrator. Pure, concentrated enzyme was aliquoted, flash-cooled in liquid nitrogen, and stored at −80 °C until further use.

Crystallization.

A seed stock of wild-type EIZS was used to facilitate crystallization of F95S-F198S EIZS, as detailed previously for crystallization of other EIZS variants.^29,30 To prepare the seed stock, wild-type EIZS (10 mg/mL) was co-crystallized with 2 mM benzyltriethylammonium chloride, 2 mM sodium pyrophosphate, and 10 mM MgCl₂ by the sitting drop vapor diffusion method. A 1-μL drop of protein solution was mixed with a 1-μL drop of precipitant solution [0.1 M BisTris (pH 6.5), 0.2 M (NH₄)₂SO₄, 25% (w/v) PEG 3,350] and incubated at 21 °C. Crystals appeared overnight and were smashed, diluted with 50 μL mother liquor, vortexed with a PTFE microseed bead (Hampton), and stored at −80 °C until needed.

Crystallization of F95S-F198S EIZS was achieved by the sitting drop vapor diffusion method using a Mosquito crystallization robot. First, F95S-F198S EIZS (7.5 mg/mL) was incubated with 4 mM benzyltriethylammonium chloride, 4 mM sodium pyrophosphate, and 4 mM MgCl₂ on ice for 30 min. Subsequently, 300 nL of protein solution was mixed with 300 nL of precipitant solution [0.1 M BisTris (pH 6.4), 0.2 M (NH₄)₂SO₄, 50 mM NaF, 31% (w/v) PEG 3,350]; 25 nL of a 1:100 dilution of the wild-type EIZS seed stock was added to the crystallization drop. Crystallization drops were equilibrated at 21 °C against a 75-μL reservoir of precipitant solution. Individual hexagonal plates appeared overnight and reached maximum dimensions within 3 days. Crystals were looped from the drop and preserved in a cryosolution consisting of mother liquor augmented with 20% (v/v) glycerol prior to flash-cooling in liquid nitrogen.

Crystallographic Data Collection and Structure Determination.

Diffraction data were collected remotely at the 17-ID-1 AMX beamline at the National Synchrotron Light Source II (NSLS-II), Brookhaven National Laboratory (Upton, NY). Diffraction data were indexed, integrated, and scaled with XDS,³⁵ and then reduced with AIMLESS.³⁶ Initial electron density maps were phased using the PHASER³⁷ module of PHENIX,³⁸ with the search model consisting of the atomic coordinates of wild-type EIZS (PDB 3KB9) less ligands and solvent molecules. Iterative rounds of structure refinement and manual model building were performed using PHENIX and Coot, respectively.^38,39 Residue side chains showing limited or no electron density in 2mF_o – DF_c and composite omit maps were truncated or removed from the final model: R22, K40, R53, E61, E135, E188, E192, H264, H265, S266, L267, L269, E270, R277, R278, R296, K309. Water molecules were manually placed in peaks ≥ 3σ in the |F_o|-|F_c| map. Ligands and other nonproteogenic molecules were manually added to the model in the final stage of refinement. The final model was validated using Molprobity.⁴⁰ Data collection and refinement statistics are reported in Table 1.

Table 1.

Crystallographic Data Collection and Refinement Statistics

Structure	F95S-F198S EIZS–Mg2+3–PPi–BTAC
Unit Cell
space group	P21
a, b, c (Å)	53.11, 47.10, 75.31
α, β, γ (deg)	90, 95.35, 90
Data Collection
laboratory, beamline	NSLS-II, 17-ID-1 AMX
detector	EIGER 9M
resolution (Å)	1.47
total/unique no. of reflections	238,102/62,412
Rmergea,b	0.072 (0.74)
Rpima,c	0.042 (0.442)
CC1/2a,d	0.997 (0.703)
I/σ(I)a	9.7 (1.7)
Redundancya	3.8 (3.7)
completeness (%)a	99.7 (96.4)
Refinement
reflections used in refinement/test set	62,385/2,000
R work a,e	0.170 (0.277)
R free a,e	0.192 (0.302)
no. of protein chains	1
no. of nonhydrogen atoms	3,070
protein	2,714
ligand	43
solvent	313
average B factor (Å2)	21
protein	20
ligand	22
solvent	29
root-mean-square deviation from ideal geometry
bonds (Å)	0.014
angles (deg)	1.2
Ramachandran plotf
Favored (%)	99.41
Allowed (%)	0.59
Outliers (%)	0.00
Molprobity scoref	0.90
PDB entry	8V3K

^aValues in parentheses refer to the highest-resolution shell of data.

^b${\text{R}}_{\text{merge}}={\sum }_h{\sum }_i|{\text{I}}_{h,i}-{\langle \text{I}\rangle }_h|/{\sum }_h{\sum }_i{\text{I}}_{h,i}$, where $\langle {\text{I}}_h\rangle$ is the average intensity calculated for reflection h from i replicate measurements.

^c${\text{R}}_{\text{p.i.m.}}=({{\sum }_h(\text{l}/(\text{N}-1))}^{1/2}{\sum }_i|{\text{I}}_{h,i}-{\langle \text{I}\rangle }_h|)/{\sum }_h{\sum }_i{\text{I}}_{h,i}$, where N is the number of reflections and $\langle {\text{I}}_h\rangle$ is the average intensity calculated for reflection h from replicate measurements.

^dPearson correlation coefficient between random half-datasets.

^e${\text{R}}_{\text{work}}=\sum ||{\text{F}}_{\text{o}}|-|{\text{F}}_{\text{c}}||/\sum |{\text{F}}_{\text{o}}|$ for reflections contained in the working set. $|{\text{F}}_{\text{o}}|$ and $|{\text{F}}_{\text{c}}|$ are the observed and calculated structure factor amplitudes, respectively. R_free is calculated using the same expression for reflections contained in the test set held aside during refinement.

^fCalculated with MolProbity.

Structure Modeling and Active Site Cavity Determination.

Structure figures were prepared using PyMOL (PyMOL Molecular Graphics System, version 2.1, Schrödinger LLC). All superpositions and root-mean-square deviation (RMSD) calculations were performed using the Align feature in PyMOL. Active site cavities were contoured using the GetCleft feature of the PyMOL plug-in, NRGSuite.⁴¹

Enzyme activity.

The EnzCheck Pyrophosphate Assay Kit was used to monitor the kinetics of the enzyme-catalyzed cyclization reaction. The 100-μL reactions were performed at 18 ºC in reaction buffer [50 mM Tris-HCl (pH 7.5), 1.0 mM MgCl₂, 0.1 mM NaN₃]; added to each reaction was 200 μM 2-amino-6-mercapto-7-methylpurine ribonucleoside (MESG), 1 unit purine nucleoside phosphorylase, 0.03 unit inorganic pyrophosphatase, and 1.0 μM EIZS. Reactions were initiated upon the addition of FPP to a final concentration of 100 μM. The assay was performed in quadruplicate on a 96-well clear, flat-bottom plate and monitored at A₃₆₀ on a Tecan M1000 spectrophotometer. Reaction velocities were calculated using the ΔA₃₆₀ within the linear range.

GC-MS Product Analysis.

Cyclization products were analyzed by gas chromatography/mass-spectrometry using an Agilent 7890A/5976C GC-MS system in EI positive mode, running a temperature program of 60–240 °C with a gradient of 20 °C/min and a solvent delay of 3 min. A 30-mL reaction was performed in reaction buffer [50 mM Tris-HCl (pH 7.5), 1.0 mM MgCl₂, 0.1 mM NaN₃] with 5 μM enzyme. The reaction was initiated upon the addition of FPP to a final concentration of 500 μM and was immediately overlaid with 10 mL of hexanes to capture volatile products. The reaction vessel was capped before being incubated overnight at 18 °C; a second incubation at 4 ºC was also performed. Brine (5 mL) was added to the mixture and the reaction was extracted 3 times with a total of 30 mL hexanes. Extracts were dried over a column of anhydrous Na₂SO₄, filtered, and concentrated to ~500 μL under a stream of N₂ gas. Products were identified by comparison of individual chromatographic retention indices and mass spectra with those of authentic compounds in the NIST and MassFinder 4.0 databases, and product percentages were determined by integration of the GC peak.

Results and Discussion

The F95S-F198S EIZS double mutant was prepared and assayed to confirm catalytic activity and to identify alternative cyclization products. Although F95S-F198S EIZS exhibits an approximately 6-fold reduction in specific activity relative to the wild-type enzyme, appreciable activity is retained (Figure 2A). This observation is consistent with previous characterization of EIZS single-point mutants.^25–30

Figure 2. Catalytic activity and product diversity of F95S-F198S EIZS.

(A) Specific activities of 1.0 μM wild-type EIZS and F95S-F198S EIZS at 18 ºC. Reactions were monitored for the release of co-product PP_i over 60 min using a coupled enzyme assay. Error bars represent the standard deviation of four independent measurements. (B) Organic extracts of the enzymatic reaction were analyzed by GC-MS to identify alternative reaction products generated at 18 ºC. Products were identified by comparing retention indices and individual mass spectra to those of authentic standards in the NIST and MassFinder 4.0 databases. Of note, α-bisabolol generation increases to 74% at 4 ºC (Table 2). (C) Butterfly plot comparing the experimental mass spectrum of the GC-MS peak identified as α-bisabolol (top, red) to that of an authentic α-bisabolol standard in the NIST database (bottom, blue).

In addition to monitoring the kinetics of the cyclization reaction, we also characterized the unique array of sesquiterpene products generated by F95S-F198S EIZS by GC-MS (Figure 2B). In contrast with the wild-type enzyme, which generates 79% epi-isozizaene along with a mixture of other hydrocarbon and no sesquiterpene alcohol products, and compared to the individual F95S EIZS and F198S EIZS mutants, which generate predominantly β-curcumene and β-cedrene hydrocarbon products and no sesquiterpene alcohol products, F95S-F198S EIZS generates 65% α-bisabolol as its major product at 18 ºC; at 4 ºC, 74% α-bisabolol is generated (Table 2). Product verification was achieved by comparison of its mass spectrum to that of authentic α-bisabolol in the NIST and MassFinder 4.0 Databases (Figure 2C). F95S-F198S EIZS also generates 18% and 20% (E)-nerolidol at 18 ºC and 4 ºC, respectively, which reflects premature quenching of the nerolidyl carbocation prior to the ring-closure step yielding the α-bisabolyl carbocation intermediate. Additional minor products are generated at 18 ºC, including 5% β-curcumene and 12% unknown/trace compounds consisting of three different hydrocarbons. At 4 ºC, only 3% β-curcumene and 3% unknown/trace compounds are generated. Improved cyclization fidelity at lower temperature presumably reflects higher activation barriers that become more inaccessible for generating off-pathway products.

Table 2.

Sesquiterpene products generated by EIZS mutants^a

Product	RIb	wild-typec	F95Sd	F198Sd	F95S-F198S, 18 ºC	F95S-F198S, 4 ºC
α-cedrene	1418	2
β-cedrene	1424			58
sesquisabinene A	1435	2
epi-isozizaene	1444	79	2
(E)-β-farnesene	1446	5
(+)-zizaene	1456	9		20
γ-curcumene	1475		38
β-curcumene	1503		47	2	5	3
(Z)-γ-bisabolene	1505		13	9
β-sesquiphellandrene	1516	1
(E)-nerolidol	1553				18	20
α-bisabolol	1673				65	74
unknown				11	12	3

^aProducts are reported as a percentage of total sesquiterpene products.

^bRetention indices as reported in MassFinder 4.0 database.

^cRef. 25.

^dRef. 30.

Notably, epi-isozizaene generation is completely abolished in F95S-F198S EIZS. These results demonstrate that polar amino acid substitutions in the active site contour reprogram the cyclization cascade and introduce a new catalytic function, i.e., hydroxylation of a carbocation intermediate. This suggests that the increased active site volume resulting from two phenylalanine-to-serine substitutions enables the binding of at least one water molecule along with the substrate, and this water molecule is properly positioned and restrained so as to preferentially react with the α-bisabolyl carbocation intermediate.

To explore the structural basis of reprogrammed cyclization activity, the X-ray crystal structure of the F95S-F198S EIZS–Mg²⁺₃–PP_i–BTAC complex was determined at 1.47 Å resolution. The refined structure reveals a characteristic class I terpene cyclase fold stabilized on a closed active site conformation through the binding of Mg²⁺₃-PP_i (Figure 3A). No major structural differences are observed in comparison with the corresponding wild-type enzyme complex (PDB 3KB9),²⁵ and the root-mean-square deviation (RMSD) is 0.107 Å over 291 Cα atoms. A single molecule of BTAC is encapsulated in the active site, and its positively charged quaternary ammonium group engages in a cation-π interaction with the side chain of F96 (Figure 3B). BTAC is nearly equal in volume to and is a partial mimic of the α-bisabolyl carbocation intermediate in terms of its size and charge.

Figure 3. F95S-F198S-EIZS–Mg²⁺₃–PPi–BTAC Complex.

(A) Overlay of wild-type EIZS (purple ribbon) and F95S-F198S EIZS (grey ribbon) monomers, which exhibit the characteristic type I terpene cyclase α-fold. (B) Close-up view of the active site of F95S-F198S EIZS. Ligands and Polder omit maps are shown as follows: Mg²⁺ (green spheres, green mesh contoured at 4σ); PP_i (red/orange sticks, orange mesh contoured at 4σ); BTAC (gold sticks, gold mesh contoured at 5σ); serine residues (maroon sticks, maroon mesh contoured at 4σ); water molecules (blue spheres, blue mesh contoured at 4σ). Hydrogen bonds are indicated by black dashed lines and the cation-π interaction is indicated by a dashed red line. Selected active site residues are shown as grey sticks. (C) Active site contour (grey surface) of F95S-F198S EIZS shows how each mutation creates a unique bulge in the active site cavity that accommodates a solvent molecule.

The side chains of the newly introduced serine residues, S95 and S198, each adopt two partially populated conformations, while the conformations of F96 and W203 remain largely unchanged. Two water molecules are sequestered into the active site: water #38 forms a 2.7 Å hydrogen bond with the newly introduced hydroxyl group of S198, and water #158 forms a 2.7 Å hydrogen bond with the hydroxyl group of Y172 (Figure 3B). Each phenylalanine-to-serine substitution creates a cavity that accommodates a water molecule: the S198 cavity accommodates water #38 and the S95 cavity accommodates water #158 (Figure 3C).

We recently reported the X-ray crystal structures of single-point mutants F95S EIZS and F198S EIZS complexed with Mg²⁺₃–PP_i–BTAC,³⁰ so structural comparisons with the double mutant F95S-F198S EIZS–Mg²⁺₃–PP_i–BTAC complex are highly informative. The double mutant exhibits similar overall architecture to that of each corresponding single mutant, with RMSD values of 0.143 Å over 299 Cα atoms and 0.081 Å over 301 Cα atoms compared with F95S EIZS and F198S EIZS, respectively.

The two simultaneous mutations appear to increase the volume of the active site synergistically, in that the increased active site volume of the double mutant is greater than the sum of the increased volumes for each single-point mutant (active site volumes are as follows: wild-type EIZS, 279 ų; F95S EIZS, 319 ų; F198S EIZS, 333 ų; F95S-F198S EIZS, 391 ų). New active site crevices resulting from each phenylalanine-to-serine mutant separately are conserved in the double mutant. Noteworthy differences between the structures of single and double mutants include the number of conformations for engineered serine side chains (S95 adopts 3 conformations in F95S EIZS and S198 adopts one conformation in F198S EIZS) and the binding conformation of BTAC (Figure 4). As noted previously,³⁰ the loss of the bulky phenyl ring of F198, which appears to clamp the phenyl group of BTAC along with F96, induces a rotation of the phenyl moiety of BTAC. This subtle shift, along with the new cavity formed by the F95S substitution, enables the binding of water molecule #158 (Figure 4).

Figure 4. Comparison of BTAC Binding Modes.

(A) Overlay of BTAC (magenta sticks) from the wild-type EIZS complex (PDB 3KB9) onto the F95S-F198S EIZS-BTAC complex. Newly introduced water molecule #158, which occupies the void resulting from the F198S substitution, is indicated. (B) Overlay of BTAC (turquoise sticks) from the F95S-EIZS complex (PDB 8SU3) onto the F95S-F198S EIZS-BTAC complex. (C) Overlay of BTAC (blue sticks) from the F198S-EIZS complex (PDB 8SU4) onto the F95S-F198S EIZS-BTAC complex.

Comparisons of the active site contours of wild-type and mutant enzymes illustrate how each mutation alters the active site contour to synergistically enable the generation of α-bisabolol (Figure 5). F95 appears to control the bisabolyl carbocation checkpoint in the reaction sequence (Figure 1), and the F95S substitution remolds the active site contour so as to halt the cyclization cascade at the bisabolyl carbocation. The elongated shape of the active site contour in F95S EIZS (Figure 5B) compared with the wild-type enzyme (Figure 5A) would favor an extended conformation for the bisabolyl carbocation, which in turn would disfavor formation of the acorenyl carbocation (Figure 1). A more compressed conformation of the homobisabolyl carbocation is required to hold the pendant prenyl group close to the cationic center to form the acorenyl carbocation, and such a compressed conformation is better enforced by the more constricted active site contour of the wild-type enzyme (Figure 5A).

Figure 5. Comparison of Active Site Contours.

In all panels, inorganic pyrophosphate is shown as orange/red sticks, Mg²⁺ ions are shown as green spheres, and the active site contour is shown as a grey surface. Residues which frame the active site of each enzyme are shown as sticks (A), wild-type EIZS, magenta; (B), F95S EIZS, turquoise; (C), F198S EIZS, blue; (D), F95S-F198S EIZS, grey/brown). Water molecules are shown as red (C) or blue (D) spheres, and hydrogen bonds are indicated by black dashed lines.

The F198S substitution results in an active site contour that retains a partially-constricted appearance despite an increase in active site volume (Figure 5C); based on the generation of β-cedrene and (+)-zizaene by F198S EIZS (Table 2), the remolded active site contour can still enforce a compressed conformation for the homobisabolyl carbocation intermediate that leads to the acorenyl carbocation intermediate and beyond. Together, however, the F95S and F198S substitutions elongate the active site sufficiently to halt the cyclization cascade at the bisabolyl carbocation intermediate and increase the active site volume so as to accommodate water along with the substrate (Figure 5D).

The α-bisabolol cyclization product can be modeled into the active site of F95S-F198S EIZS to provide additional clues regarding water management for catalysis (Figure 6A). Water #158 is reasonably close to the position of the hydroxyl group of α-bisabolol. We hypothesize that water #158 would be further away from the cationic centers of the farnesyl or nerolidyl carbocations and it is stabilized in position by a hydrogen bond with Y172, so it would be unable to quench these initially formed carbocations as easily. This may account for the much lower percentage of nerolidol generated by the double mutant (Table 2). Accordingly, we hypothesize that water #158 is responsible for hydroxylation of the bisabolyl carbocation to yield α-bisabolol.

Figure 6. Active site contour of F95S-F198S-EIZS accommodates α-bisabolol and shares key structural features with a native α-bisabolol synthase.

(A) Stereoview showing α-(1R,7S)-bisabolol (green sticks) modeled into the active site contour (grey surface) of F95S-F198S EIZS. Active site residues are shown as grey and brown sticks, Mg²⁺ ions are shown as green spheres, and inorganic pyrophosphate is represented by an orange/red stick-figure. Water molecules are shown as blue spheres and hydrogen bonds are indicated by black dashed lines. The diastereomer α-(1R,7R)-bisabolol could also be modeled in the active site, but the fit appeared to be slightly better for the (1R,7S)-diastereomer. Moreover, the hydroxyl group of α-(1R,7S)-bisabolol is on the same side of the active site contour as water #158, hydrogen bonded to Y172. (B) Active site contour (green surface) of AaBOS (PDB 4FJQ). Proposed active site residues are shown as grey sticks, and ordered residues of the metal-binding motifs are shown as magenta sticks.

Finally, it is instructive to compare the active site contour of our newly engineered α-bisabolol synthase with that of a naturally occurring α-bisabolol synthase from the sweet wormwood plant, Artemisia annua (AaBOS).⁴² Only the structure of unliganded AaBOS with an open active site conformation is available for analysis, but even so, its active site contour is similar to that of F95S-F198S EIZS. Of particular note, both contours contain a “double dip” at the base of each active site around a tryptophan residue (W203 in EIZS, W271 in AaBOS) (Figure 6B). This structural feature may stabilize the extended conformation of the bisabolyl carbocation intermediate prior to quenching by a trapped water molecule in each enzyme active site.

At 37 °C, the naturally occurring α-bisabolol synthase exhibits modest catalytic efficiency with k_cat/K_M = 9.05 × 10⁻² μM⁻¹·s⁻¹, generating 92.8% α-bisabolol.⁴² Trace amounts of other monocyclic sesquiterpene products, including α-, β-, and γ-bisabolene, are also observed and result from proton elimination from the bisabolyl carbocation intermediate. Interestingly, despite the lower fidelity of α-bisabolol generation by F95S-F198S EIZS (65% at 18 °C and 74% at 4 °C), it appears that both the naturally occurring α-bisabolol synthase and F95S-F198S EIZS are capable of generating elimination products. Neither enzyme can generate a polycyclic product, suggesting that the extended active site contour with the “double dip” at the bottom (Figure 6) disfavors further cyclization of the bisabolyl carbocation intermediate in both enzymes.

Concluding Remarks

The present study represents the first biochemical and structural characterization of a double mutant of EIZS, building on the foundation of a previously characterized library of ~50 single-site mutants. Active site mutations work synergistically to remold the active site contour and introduce sufficient new volume to accommodate a water molecule along with the substrate. Terpene cyclases that catalyze hydroxylation reactions must be able to control the reactivity of trapped water to avoid premature termination of the reaction sequence; F95S-F198S EIZS appears to do so by holding a water molecule in a position where it preferentially reacts with the α-bisabolyl carbocation to generate α-bisabolol. Comparison with a naturally occurring α-bisabolol synthase reveals common active site features, thereby demonstrating significant potential for structure-based approaches to reprogramming isoprenoid cyclization cascades. These results reinforce the idea that one need not be limited to naturally occurring cyclases for the generation of desired terpene products, e.g., as applications in synthetic biology are considered.^14,43,44 Our continuing efforts to expand the product repertoire of EIZS will be reported in due course.