Construction of an artificial system for ambrein biosynthesis and investigation of some biological activities of ambrein

Ambergris, a sperm whale metabolite, has long been used as a fragrance and traditional medication, but it is now rarely available. The odor components of ambergris result from the photooxidative degradation of the major component, ambrein. The pharmacological activities of ambergris have also been attributed to ambrein. However, efficient production of ambrein and odor compounds has not been achieved. Here, we constructed a system for the synthesis of ambrein and odor components. First, we created a new triterpene synthase, “ambrein synthase,” for mass production of ambrein by redesigning a bacterial enzyme. The ambrein yields were approximately 20 times greater than those reported previously. Next, an efficient photooxidative conversion system from ambrein to a range of volatiles of ambergris was established. The yield of volatiles was 8–15%. Finally, two biological activities, promotion of osteoclast differentiation and prevention of amyloid β-induced apoptosis, were discovered using the synthesized ambrein.

Since the production efficiency of 1 and odor components was still too low for industrialization, we sought to construct an artificial synthesis system that is dramatically more efficient than the one used currently. First, we created an "ambrein (1) synthase" that produces a higher yield of 1 than that of the final products (8 and 9) produced by wild-type BmeTC. Next, we established an efficient photooxidative conversion system to transform 1 to volatile components of ambergris. Finally, the synthesized 1 was analyzed for its bioactivities that have not yet been explored, by using cell culture assays. In this study, we focused on the effects of 1 on bone cell differentiation and against amyloid β neurotoxicity, since ambergris has been used in traditional medicine to treat rheumatism and disease of the nervous system.

Results
Screening for variants suitable for the synthesis of 1. Mutation analysis of BmeTC and SHC suggested that the D373 mutation of BmeTC was important for the synthesis of 1. However, none of the studies have analyzed D373 variants other than BmeTC D373C . In this study, an Escherichia coli cell-free system expressing the D373 variants ( Supplementary Fig. 1) substituted with 11 amino acids (C, A, F, G, H, L, M, N, Q, S and W) was used to confirm their ability to convert 6 to 1. The results indicated that only the C mutant produced 1 (12.5% yield) (Fig. 3). This revealed that cysteine at position D373 is critical to form 1. BmeTC D373C accumulated large amounts of intermediates, bicycle 7 and monocycle 10, during the reaction with 6 ( Fig. 3), showing that the bulky bicyclic and monocyclic structures of intermediates (Fig. 4a) interfered in the second-step reaction.
Based on this working hypothesis, 6 residues (Y167, Y255, Y257, N302, L596, and F600) that were presumed to be located near the bicyclic or monocyclic structure during the second reaction were selected based on the modeling structure of BmeTC (Fig. 4) and replaced by a smaller Ala. Enzymatic reactions of 6 Ala mutants were performed using a cell-free system ( Supplementary Fig. 1). Y167A, Y257A, and N302A variants showed a higher level of activity on substrates 6 and 7 to produce 8 and 9, compared with BmeTC WT (Fig. 5a,b and Supplementary Figs. 2 and 3). Whereas the product of BmeTC Y167A containing a novel tricyclic compound was previously reported 20 , the enzyme activity was first revealed in this study.

Conversion of synthetic 1 into volatile components.
Enzymatically synthesized 1 was converted to volatiles by 1 O 2, and these volatiles were compared with the volatiles present in ambergris. Ethanol tinctures of two ambergris samples ( Supplementary Fig. 7) mainly contain 4 known compounds (2-5) and 6 other unknown compounds (12)(13)(14)(15)(16)(17) (Fig. 7a,b). The ratio of volatile components was slightly different between the two ambergris samples obtained by us (Fig. 7b), suggesting that the slight difference between their odors could be due to differences in the oxidizing conditions of 1 in the environment. The conversion of 1 to volatile compounds was conducted via UV or visible light treatment using 3 photosensitizers (rose Bengal: RB, 5,10,15,20-tetraphenyl-  The yield of volatile components (ca. 1%) and the residual rate of 1 (ca. 68%) obtained from synthetic 1 via UV treatment were similar to those of ambergris (ca. 1% and ca. 68%, respectively) ( Fig. 7c,d), and also similar to previously reported results obtained by the conversion of 1 to volatile components via visible light using a photosensitizer 5,10,15,20-tetraphenyl-21H, 23H-porphine copper (II) (yield: ca. 1% and residual rate: ca. 50%) 11 . In the current study, visible light treatment using 3 photosensitizers (RB, TPP and MB) resulted in higher yields of volatile components (ca. 8-15%) and a lower retention of 1 (ca. 0-13%) (Fig. 7c,d). The formation rates of volatiles from UV and RB treated 1 were more similar to those of ambergris, while the TPP and MB treated samples had a much higher proportion of 14 (12.2%) and 3 (25.5%), respectively, than the ethanol tinctures of two ambergris samples (1.2-2.1% and 1.2-4.2%) (Fig. 7b). After wetting samples with filter papers and evaporating the solvent, we compared the scents and found that the scents associated with UV and RB treatments of 1 were similar to those obtained from the ethanol tinctures of two ambergris, whereas the scents associated with TPP and MB treatments were different. Notably, the yield of valuable fragrance compound 3, which is used as a substitute for ambergris samples 1-4 , was 6-21 times higher in MB treated 1 than that obtained from the ethanol tinctures of two ambergris (Fig. 7b). The artificial synthetic system for the major component 1 and odor components of ambergris achieved efficient enzymatic conversion of 6 to 1 as well as efficient conversion of 1 to volatile components by 1 O 2 .
Biological activity of 1. Ambergris was previously used as a traditional medicine for various maladies 4,7 .
However, the biological activities of natural 1, the main component of ambergris (ex. ambergris samples 1 and 2 contained 1 at ca. 68%; Fig. 7d) have not been assessed extensively due to its scarcity. To date, only its aphrodisiac, antinociceptive, and elastase release inhibitory activities are known 5,6,21 . Since the enzymatic synthesis in the current study enabled sufficient production of synthetic 1, its two biological activities were analyzed. First, we analyzed the effect of 1 on the differentiation of bone cells, osteoblasts and osteoclasts. Extracellular calcium deposited by mature osteoblasts was stained with alizarin red S after cells were incubated with or without 10 μM 1. However, a significant effect of 1 on the osteoblastic activity was not detected (Supplementary Fig. 13). In contrast, 1 enhanced osteoclastic differentiation at a concentration of 10 μM (Fig. 8). The results indicated that www.nature.com/scientificreports/ 1 significantly increased the number of mature osteoclasts in a concentration-dependent manner (Fig. 8). The effect of 50 μM 1 was similar to that of 5 μM kenpaullone 22 , which is a strong activator of osteoclastic differentiation. This result suggested that 1 may be a promising drug candidate for osteopetrosis, caused by defective osteoclast function. Next, we analyzed the protective effect of 1 against amyloid β (Aβ)-mediated neurotoxicity. Alzheimer's disease (AD), the most common type of dementia, is an age-related progressive neurodegenerative disorder characterized by depositions of amyloid β (Aβ), the primary component of senile plaques 23 . Aβ peptides elicit neurotoxicity, leading to neuronal loss and cognitive deficits. We examined the effect of 1 on Aβ-induced apoptotic cell death in human neuroblastoma SK-N-SH cells, in which Aβ 1-42 was used to induce cell death. Exposure of SK-N-SH cells to 1 µM Aβ 1-42 for 24 h led to a markedly increased percentage of early apoptotic cells (Annexin V + /7-AAD − ), as well as late apoptotic and dead cells (Annexin V + /7-AAD + , Fig. 9). Aβ 1-42 -induced apoptosis was significantly inhibited by pretreatment with 1 at concentrations of 1-20 µM for 24 h prior to Aβ 1-42 exposure (Fig. 9). These results implied that 1 possesses the potential to prevent Aβ neurotoxicity. Presently, we are investigating whether 1 modulates apoptosis signaling pathways and comparing the efficacy of 1 and other Alzheimer's drug candidates targeting Aβ neurotoxicity.

Discussion
The present study redesigned BmeTC to create a new enzyme named "ambrein (1) synthase" (BmeTC Y167A/D373C ), which displays activity beyond its wild-type function (production of 8 and 9). We also constructed an efficient in vitro artificial biosynthetic pathway, which can be used for mass production of 1 in vivo. The new two-enzyme system (Fig. 2d) gives approximately 20 times more yield of 1 than the most efficient system currently known (BmeTC D373C ) 17 (Fig. 6a) and is expected to produce 2 g 1/L culture medium in yeast P. pastoris. Recently, it was hypothesized that 1 is biosynthesized via the pathway 6 → 7 → 1 in sperm whales 24 . In addition, 2 enzymes are utilized to convert symmetric compounds to asymmetric fern onoceroids and carotenoids 25,26 , via a strategy similar to the one we finally adopted. It is interesting that the pathways adopted by nature are similar to the artificial pathways (Fig. 2d) we have developed.  www.nature.com/scientificreports/ conversion of 1 to volatiles. In this study, we were able to obtain a yield of 8-15% (Fig. 7c), which was higher than the yields obtained previously for different purposes [9][10][11] and the content in the natural ambergris analyzed by us (Fig. 7c). The synthetic system of volatiles constructed by us could change odor depending on the type of photosensitizers used (Fig. 7b). In the future, a variety of odors may be created by examining various reaction conditions including use of different photosensitizers. In addition, although unknown volatile compounds 12-17 were detected in this study ( Fig. 7a and Supplementary Fig. 12), their structures could not be determined. New odor compounds may be identified in the future if a large amount of 1 is photooxidized. Further, we identified two biological activities of 1: promotion of osteoclast differentiation and prevention of Aβ neurotoxicity (Figs. 8 and 9). However, it remains unclear how this compound performs these activities. Identification of an intracellular target molecule of 1 may allow the discovery of therapeutic agents for osteopetrosis and Alzheimer's disease in the future. This study differed from the conventional biosynthesis studies that have aimed to reconstruct natural biosynthetic pathways. It was a challenge to synthesize a rare natural product (1) whose biosynthetic pathway remains unclear, with an artificial biosynthetic route using an enzyme created in the laboratory. Many new natural products have been discovered by genome mining. However, if the biosynthetic enzyme is of a new type or if the natural producer of the product is unknown, genome mining cannot be performed. Therefore, it will be important in the future to synthesize desired compounds by artificially creating new biosynthetic enzymes. In addition, the system constructed in this study can be synthesize 1 analogues and fragrance analogues by redesigning enzymes and using substrate analogues, and will lead to the creation of compounds with numerous odors and biological activities beyond those found in nature in the future.
Isolation and structural analysis of 1, 7, 10, and 11 synthesized from substrate 6 by BmeTC X . Compound 11, biosynthesized in the yeast P. pastoris, was identified previously by MS analysis 17 .
However, no NMR data were available for 11. Therefore, isolation and structural analysis of 11 was performed in the present study. Compounds 1, 7, and 10 were isolated for use as substrates for enzymatic reactions and material for conversion into volatile components. As a typical example, the method used to isolate 1, 7, 10, and 11 produced by BmeTC D373C/L596A is described below. Compounds 1, 7, and 10 were synthesized and isolated via a method similar to BmeTC D373C/L596A , mainly using BmeTC Y167A/D373C , BmeTC WT , and BmeTC D373C , respectively.  20 . The pellet was discarded, and the resulting supernatant was used as the cell-free extract.
To analyze BmeTC X products, the substrate (6, 7, or 10; 0.1 mg) was emulsified with Tween 80 (20 mg) in buffer A (1 mL), and incubated with the cell-free extracts containing BmeTC X (4 mL) at 37 °C for 64 h. After 15% KOH/MeOH solution (6 mL) was added to the reaction mixture, the lipophilic products were extracted with n-hexane (10 mL × 3) and concentrated. Next, the n-hexane extract containing the products and residual substrate was analyzed by GC and GC-MS. Standard deviations were calculated from the results of 3 replicates.
Analysis of BmeTC X products using purified enzymes. In order to analyze the correct enzyme activity, BmeTC X (X: WT, D373C, Y167A/D373C, Y257A/D373C, N302A/D373C and D373C/L596A) was expressed without the fused TF-tag. The BmeTC WT gene was excised from the NdeI and XhoI sites of pColdTF-BmeTC WT , and introduced into the same site of pColdI to construct pColdI-BmeTC WT . Next, pColdI-BmeTC D373C and pColdI-BmeTC Y167A/D373C were synthesized by Genewiz (Morrisville, NC, USA) using codon-optimized sequences for E. coli, following which pColdI-BmeTC Y257A/ D373C and pColdI-BmeTC N302A/ D373C were prepared via a Quik Change Site-directed Mutagenesis Kit (Stratagene) using pColdI-BmeTC D373C as a template and the primers listed in Supplementary Table 3. Then pColdI-BmeTC X was introduced into BL21(DE3) together with pGro7, and soluble BmeTC X was expressed. The expression and purification of BmeTC X was basically the same as in Ref. 29 as described below. After culturing as in the case of pColdTF-BmeTC WT , cells expressing the recombinant protein were harvested by centrifugation and disrupted by sonication in buffer B [20 mM Tris-HCl (pH 7.9) and 300 mM NaCl] (30 mL/L cultured cells) containing 10 mM imidazole and 0.1% Tween 80 at 4 °C 29 . The homogenate was centrifuged at 18,270 × g for 20 min to prepare the supernatant containing soluble Histagged fusion protein, which was loaded into a Ni-NTA agarose column (0.2 mL; Qiagen, Hilden, Germany), followed by washing with 10 mL of buffer B containing 10 mM imidazole and then by 12 mL of buffer B containing 50 mM imidazole and 0.1% Tween 80 29 . The purified protein was eluted with 3 mL buffer B containing 250 mM imidazole and 0.1% Tween 80, and buffer-exchanged into 3 mL buffer C [50 mM Tris-HCl (pH 7.5), 2.5 mM dithiothreitol, 1 mM EDTA, 300 mM NaCl and 0.1% Tween-80] by gel-filtration chromatography using a Sephadex G-10 column (GE Healthcare, Pittsburgh, PA, USA) 29 . The expression and purification of BmeTC X were analyzed via 10% SDS-PAGE ( Supplementary Fig. 5).
The reaction mixture used to analyze the BmeTC X products in buffer C (total volume: 1 mL) contained 8.2 mM (50 μg) substrate (6, 7 or 10) emulsified with 1 mg Tween 80 and 1.4 μM (50 μg) purified BmeTC X . Reactions were performed at 37 °C for 64 h. As shown in the representative example in Supplementary Fig. 22, the time-dependent activities of BmeTC X were linear at 64 h. After 15% KOH/MeOH solution (1.2 mL) was added to the reaction mixture, the lipophilic products were extracted using n-hexane (2 mL × 3) and the products and the residual substrate was analyzed using GC and GC-MS. Standard deviations were calculated from the results of 3 replicates. www.nature.com/scientificreports/ differentiation, the cells were incubated in α-MEM complete medium supplemented with 10 mM β-glycerol phosphate, 50 μg/mL ascorbic acid, 10 nM dexamethasone, and 10 μM 1 or 0.1% DMSO as a control; the medium was exchanged with fresh medium every 3 d. Murine macrophage-like pre-osteoclastic RAW264.7 cells (ATCC, Manassas, VA, USA) were cultured in α-MEM medium supplemented with 10% FBS, 2 mM L-glutamine, 100 U/ mL penicillin, and 100 μg/mL streptomycin. In order to induce osteoclastic differentiation, 100 ng/mL sRANKL (Oriental Yeast, Japan) was added to the medium with different concentrations of 1 or 0.1% DMSO as a control and incubated for 4 d with 5% CO 2 at 37 °C. Human neuroblastoma SK-N-SH cells were obtained from Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University (Sendai, Japan) and cultured in DMEM supplemented with 10% FBS, 1 mM sodium pyruvate and 1% penicillin/ streptomycin. SK-N-SH cells were seeded at a density of 1 × 10 6 cells/mL, and following overnight incubation, treated with or without 1 (1-20 μM) for 24 h. Subsequently, culture supernatants containing 1 were removed, and the cells were exposed to 1 µM Aβ 1-42 (Wako, Osaka, Japan) for an additional 24 h to induce Amyloid β (Aβ)-mediated neurotoxicity.

Conversion of 1 into volatile components.
Alizarin red S staining. MC3T3 31 . TRAP-positive osteoclasts with more than 3 nuclei were considered as mature osteoclasts and counted using a light microscope (Olympus IX73, Tokyo, Japan).

Apoptosis assay by flow cytometry.
For the apoptosis assay, flow cytometric analysis was performed using FITC-Annexin V (Biolegend, San Diego, CA, USA) and 7-Amino-Actinomycin (7-AAD; Biolegend). Cells were pretreated with or without 1 for 24 h, followed by 24 h exposure to 1 µM Aβ 1-42 . Subsequently, cells were harvested, washed twice with phosphate-buffered saline containing 0.1% bovine serum albumin, and resuspended in Annexin V binding buffer at 1 × 10 6 cells/mL. Thereafter, cells were stained with FITC-Annexin V (5 µg/mL) and 7-AAD (0.5 μg/mL) for 15 min in the dark. Finally, cells were analyzed using FACSCalibur flow cytometry system (BD Bioscience, San Jose, CA, USA). A minimum of 10,000 events was collected and the percentage of apoptotic cells was calculated using the CellQuest software (BD Bioscience).
Statistical analysis. Statistical significance was determined by one-way analysis of variance followed by the Tukey-Kramer test for multiple comparisons at P < 0.01.

Data availability
Data supporting the findings of this study are available within the article and the Supplementary Information files, and from the corresponding author upon reasonable request.