Introduction

Ambergris, a metabolic product of the sperm whale (Physeter catodon or Physeter macrocephalus), accumulates as gut concretions with a probability of about 1%, and is one of the most valuable scents of animal origin1,2,3. Ambergris, which exhibits various medicinal properties, has also been used worldwide as a traditional medication for maladies such as migraines, the common cold, constipation, disease of the nervous system, and rheumatism4,5,6. In addition, it is often used as an aphrodisiac7. However, ambergris is almost inaccessible because sperm whales are now protected under the Convention on the International Trade of Endangered Species of Wild Fauna and Flora. On rare occasions, jetsam ambergris is found on beaches around the world, and traded at a high price8. The odor components of ambergris result from the photooxidative degradation of the major component, ambrein (1)2,3,9,10,11, and the pharmacological activity is also believed to be due to 1 (Fig. 1)5,6. To date, efficient conversion to both 1 and odor components (25, Fig. 1) has never been achieved. The reported yields of chemical synthesis of 1 and conversion of 1 to odor components are 1.3–3.8% and approximately 1%, respectively11,12,13,14. In addition, since the biosynthetic pathway of 1 in sperm whales remains unclear, it is not possible to utilize the biosynthetic enzyme that produces 1.

Figure 1
figure 1

Photooxygenation of 1 to produce volatile compounds (25). Tricyclic triterpene, (+)-ambrein (1), is the primary source of ambergris-derived odor components2,3,4,9,10,11. Photooxidative degradation of 1 by1O2 around its central double bond produces volatile compounds with smaller molecular weights2,3,4,9,10,11. 2-5 are the main components identified. Tricyclic volatiles 35 originate from the bicyclic part of 1, while the smaller fragment 2 is related to the monocyclic part of 12,3,4,9,10,11.

Successful artificial enzymatic synthesis of 1 was reported in 2013 using onoceroid synthase, BmeTC, from Bacillus megaterium, which sequentially cyclizes both termini of squalene (6; Fig. 2a)12. Presumably, BmeTC first cyclizes one side end of 6, converts it into a bicycle 7, and then incorporates 7 into the same active site as a substrate and cyclizes the remaining terminal to produce onoceroids 8 and 9 (Fig. 2a). BmeTC catalyzes this two-step reaction (acycle [6] → bicycle [7] → bicycle-bicycle [8 and 9]; Fig. 2a)15. BmeTC can successfully form 1 using an abnormal monocyclic product (10), synthesized by a squalene-hopene cyclase variant (SHCD377C), as a substrate (Fig. 2b)15. Hence, enzymatic synthesis of 1 from inexpensive 6 was achieved via the route “acycle 6 → monocycle 10 → bicycle-monocycle 1” (Fig. 2b)15. Furthermore, the BmeTC variant BmeTCD373C, which is an equivalent point mutation as SHCD377C, could reportedly synthesize 1 from 6 both in vitro and in vivo (Fig. 2c)16,17,18. By acquiring a “new route” that forms a monocycle 10, in addition to the wild-type route (WT route) that forms bicycle 7, BmeTCD373C synthesizes 1 via the two pathways, 6 → 7 → 1 (WT route → new route) and 6 → 10 → 1 (new route → WT route) (Fig. 2c). In a bioreactor, the production of 1 by BmeTCD373C in yeast Pichia pastoris reached a maximum titer of 105 mg L-1 of culture medium17, which was higher than that produced using other host systems or with the two enzyme system involving SHCD377C and BmeTCWT (Supplementary Table 1)17,18,19.

Figure 2
figure 2

The enzymatic reaction pathway described in this study. (a) Mechanism of conversion of 6 to 8 and 9 by BmeTCWT. (b) Mechanism of conversion of 6 to 1 by SHCD377C and BmeTCWT. (c) Reaction pathway by BmeTCD373C or BmeTCY167A/D373C. BmeTCD373C or BmeTCY167A/D373C form final products (1, 8, 9 and 11) through a two-step reaction from 6 via intermediates (7 and 10) and may have catalytic functions in two pathways (WT and New route). (d) Two enzyme system (BmeTCWT and BmeTCY167A/D373C) constructed in this study.

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 BmeTCD373C. 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. BmeTCD373C 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.

Figure 3
figure 3

Yield of products produced by BmeTCX from substrate 6 in a cell-free system.

Figure 4
figure 4

Selection of residues, presumably located near a bicyclic or monocyclic structure, during the second step of the reaction. (a) Structures of substrates 6, 7, and 10. Positions corresponding to bicyclic or monocyclic structure on the substrates are shown in yellow. (b) Homology model of BmeTC. The 3D structure of BmeTC was modeled using SWISS-MODEL. The structure was constructed using the coordinates of SHC complexed with 2-azasqualene inhibitor [Protein Data Bank (PDB) code: 1UMP] as a template and displayed using PYMOL. Blue: 2-azasqualene; pink and green: residues within 4 Å of 2-azasqualene; pink: residues targeted in this study; yellow: positions corresponding to bicyclic or monocyclic structure on the substrates.

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 BmeTCWT (Fig. 5a,b and Supplementary Figs. 2 and 3). Whereas the product of BmeTCY167A containing a novel tricyclic compound was previously reported20, the enzyme activity was first revealed in this study.

Figure 5
figure 5

Yield of products produced by BmeTCX from substrates (6, 7, and 10) in a cell-free system.

On the other hand, the yield of 1 synthesized from 10 by the L596A variant was 8 times that synthesized by BmeTCWT (Fig. 5c and Supplementary Fig. 4). Thus, it may be useful to replace BmeTCWT with BmeTCL596A during the second step of the two-enzyme system consisting of SHCD377C and BmeTCWT (Fig. 2b). However, as the single enzyme system of BmeTCD373C has been shown to be more advantageous for producing of 1 in yeast (Supplementary Table 1)17,18, the present study further improved upon BmeTCD373C.

Construction of the system for artificial biosynthesis of 1

In order to improve its reactivity with monocyclic 10 and bicyclic substrate 7, 4 mutations (Y167A, Y257A, N302A, and L596A) were introduced into BmeTCD373C, following which the reactivity of these variants (Y167A/D373C, Y257A/D373C, N302A/D373C, and D373C/L596A) with 3 substrates (6, 7, and 10) was accurately analyzed using the purified enzyme (Fig. 6, Supplementary Figs. 5 and 6). The results indicated that of the double mutants, only BmeTCY167A/D373C produced 1 from 6, in which the yield of 1 (21.5%) was improved approximately tenfold over that of BmeTCD373C (2.2%) (Fig. 6a). If BmeTCY167A/D373C produces 1 in P. pastoris with the same efficacy as BmeTCD373C 17, an end yield of approximately 1 g 1 / L culture medium can be expected.

Figure 6
figure 6

Yield of products produced by BmeTCX from substrates (6, 7, and 10) in the system using purified enzymes.

The conversion of 6 and 7 by BmeTCY167A/D373C to yield 1 (21.5 and 62.0%, respectively) were 4.7 and 7.2 times greater than the conversion of 6 and 7 by BmeTCWT to 8 and 9 (4.6 and 8.6%, respectively) by BmeTCWT (Fig. 6a,b). This indicated that the double mutant enzyme displayed activity beyond its original function to form onoceroids (8 and 9). Since the activity of previously constructed BmeTCD373C on substrates 6 and 7 (producing 2.2 and 0% of 1, respectively) was lower than that of BmeTCWT (Fig. 6a,b), we named the novel enzyme created in this study, BmeTCY167A/D373C, as “ambrein synthase.”

BmeTCY167A/D373C reacted best with 7 rather than 6 (Fig. 6a,b), indicating that the reaction with 6 was slower in the first step (6 → 7) than in the second step (7 → 1) (Fig. 2c); thus, the first step would be rate limiting. In addition, BmeTCY167A/D373C did not react with 10 (Fig. 6c), suggesting that 1 was specifically synthesized from 7 when 6 was used as a substrate, and that BmeTCY167A/D373C mainly catalyzed the reaction 6 → 7 → 1 (Fig. 2c). Hence, to accelerate the reaction in the first step of 6 → 7 → 1 and reduce the overall quantity of by-product 10, 6 was converted to 1 by adding BmeTCWT to the reaction solution of BmeTCY167A/D373C (Fig. 2d). The yield of 1 was evaluated via a system, in which the total amount of enzymes BmeTCWT and BmeTCY167A/D373C was the same as that of a single enzyme. The yield of 1 (46.0%) obtained via the new system, comprising BmeTCWT and BmeTCY167A/D373C was approximately twice that obtained via BmeTCY167A/D373C alone (21.5%) and approximately 20 times that of BmeTCD373C (2.2%) (Fig. 6a). Since the 2 enzymes can be co-expressed in the yeast P. pastoris17, the new system should be applicable in vivo (calculated titer: approximately 2 g 1/L culture medium in the bioreactor).

Conversion of synthetic 1 into volatile components

Enzymatically synthesized 1 was converted to volatiles by 1O2, 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 (25) and 6 other unknown compounds (1217) (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-tetraphenylporphine: TPP, and methylene blue: MB). Volatile components similar to that of ambergris (25 and 1217) were detected (Fig. 7b and Supplementary Figs. 8–12).

Figure 7
figure 7

Conversion of 1 into volatile components. (a) GC–MS chromatogram of ambergris volatiles. Compounds 25 were identified, while compounds 1217 were not. (b) Percentage of ambergris volatile components and volatile compounds produced by treating 1 with UV light and a photosensitizer (RB, TPP, or MB). The ratio of "others" was calculated using the total amount of compounds, except 25 and 1217. (c) Yield of ambergris volatile components and volatile compounds produced by treatment of 1 with UV light and a photosensitizer (RB, TPP, or MB). (d) Residual rate of 1 in ambergris and reaction solution exposed to UV light and photosensitizer (RB, TPP, or MB).

Figure 8
figure 8

Biological effect of 1 on the osteoclast differentiation. The number of TRAP-positive osteoclasts, differentiated from RAW264.7 cells, was counted in the absence or presence of different concentrations of 1. Kenpaullone is a positive control. The data are expressed as mean ± S.D. (n = 3, * = P < 0.01).

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 samples1,2,3,4, 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 1O2.

Biological activity of 1

Ambergris was previously used as a traditional medicine for various maladies4,7. 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 known5,6,21. 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 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 kenpaullone22, 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 plaques23. 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 Aβ1–42 was used to induce cell death. Exposure of SK-N-SH cells to 1 µM Aβ1-42 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). Aβ1-42-induced apoptosis was significantly inhibited by pretreatment with 1 at concentrations of 1–20 µM for 24 h prior to Aβ1-42 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.

Figure 9
figure 9

Protective effect of 1 on Aβ1-42-induced apoptosis in SK-N-SH cells. Cells were pretreated with different concentrations of 1 (1, 2, 5, 10 and 20 μM) for 24 h before being exposed to 1 μM Aβ1-42 for 24 h. Apoptotic cells were analyzed via flow cytometry using an Annexin V/7-AAD staining assay. (a) Dot plots of representative experiments. The number in the lower right quadrant signifies the percentage of early apoptotic cells (Annexin V+/7-AAD), and the upper right quadrant signifies the percentage of late apoptotic and dead (Annexin V+/7-AAD+) cells. (b) Percentages of early apoptotic cells (Annexin V+/7-AAD) and late apoptotic and dead cells (Annexin V+/7-AAD+). Data are expressed as mean ± SEM; (n = 3). * and refer to the comparison of Aβ1-42 alone group versus control group in Annexin V+/7-AAD and Annexin V+/7-AAD+ cells, respectively; * or  = P < 0.01. # and refer to the comparison of 1 pretreatment groups versus Aβ1-42 alone group in Annexin V+/7-AAD and Annexin V+/7-AAD+ cells, respectively; # or  = P < 0.01.

Discussion

The present study redesigned BmeTC to create a new enzyme named “ambrein (1) synthase” (BmeTCY167A/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 (BmeTCD373C)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 whales24. In addition, 2 enzymes are utilized to convert symmetric compounds to asymmetric fern onoceroids and carotenoids25,26, 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.

Although earlier studies on the photooxidation of 1 were aimed at mimicking the production of volatile components of natural ambergris and isolating the volatiles 9,10,11, none of the studies aimed to achieve the efficient 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 purposes9,10,11 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 1217 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.

Methods

General

E. coli JM109 (Takara, Shiga, Japan) was used for sequencing analysis, and E. coli BL21(DE3) (Takara), pColdTF (Takara), and pColdI (Takara) were used to express BmeTCX genes. NMR spectra were recorded using a Bruker DPX 400 spectrometer (Billerica, MA, USA) at 400 MHz for protons (1H) and 100 MHz for carbon (13C). GC–MS was performed on a JMS-T100GCV spectrometer (JEOL, Tokyo, Japan) equipped with a DB-1 capillary column (30 m × 0.25 mm × 0.25 µm; J&W Scientific. Inc., Folsom, CA, USA), using the EI mode operated at 70 eV. GC analyses were performed using a Shimadzu GC-2014 chromatograph equipped with a flame ionization detector and using a DB-1 capillary column (30 m × 0.25 mm × 0.25 µm; J&W Scientific, Inc.). GC and GC–MS conditions for the BmeTCX products were as follows: injection temperature = 300 °C, column temperature = 220–300 °C (1 °C min−1). GC and GC–MS conditions for the volatile compounds were as follows: injection temperature = 200 °C, column temperature = 40–300 °C (5 °C min−1) for GC and 30–300 °C (5 °C min-1) for GC–MS. Compound 6 was purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Two ambergris samples (NSMT M55020 and NSMT M55019; Supplementary Fig. 7) stored in the National Museum of Nature and Science (Japan) for more than 30 years were used for the analysis of volatile components.

Isolation and structural analysis of 1, 7, 10, and 11 synthesized from substrate 6 by BmeTCX

Compound 11, biosynthesized in the yeast P. pastoris, was identified previously by MS analysis17. 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 BmeTCD373C/L596A is described below. Compounds 1, 7, and 10 were synthesized and isolated via a method similar to BmeTCD373C/L596A, mainly using BmeTCY167A/D373C, BmeTCWT, and BmeTCD373C, respectively.

The expression and preparation of the a cell-free extract was basically the same as in Ref. 20 as described below. E. coli BL21(DE3) harboring pColdTF-BmeTCD373C/L596A was grown at 37 °C in LB medium (1 L) with 100 μg mL-1 ampicillin20. Expression of the recombinant protein was induced by adding 0.1 mM IPTG when OD600 reached ~ 0.620. Further cultivation of BL21(DE3) recombinants was performed for 24 h at 15°C20. E. coli cells expressing recombinant BmeTCD373C/L596A were harvested by centrifugation and resuspended in buffer A (15 mL/5 g) containing 50 mM Tris–HCl (pH 7.5), 2.5 mM dithiothreitol, 1 mM EDTA, 0.1% ascorbic acid and 0.1% Tween-8020. Cells were disrupted by sonication with UP200s (Hielscher Ultrasonics GmbH) at 4–10 °C for 15 min20. The resulting suspension was centrifuged at 12,300 × g for 20 min (twice)20. The pellet was discarded, and the resulting supernatant was used as the cell-free extract.

To isolate product 11 formed by BmeTCD373C/L596A, 6 (7.5 mg) was emulsified with Tween 80 (150 mg) in buffer A (75 mL), and incubated with the cell-free extract (300 mL) at 30 °C for 112 h. Subsequently, 15% KOH/MeOH solution (450 mL) was added to the reaction mixture and lipophilic products were extracted with n-hexane (400 mL × 3) and concentrated. The gas chromatogram of n-hexane extract is shown (Supplementary Fig. 14). The n-hexane extract (5.29 g) was partially purified using silica gel (250 g) column chromatography with n-hexane and n-hexane/EtOAc (100:20). The fraction (Fra. A: 5.0 mg) eluted with n-hexane contained substrates 6, 8, 10, and 11, whereas the fraction (Fra. B: 188.0 mg) eluted with n-hexane/EtOAc (100:20) contained 1, 7, and 9. Pure 11 (oil; 1.5 mg) and 10 (oil; 0.1 mg) were obtained by SiO2 HPLC (Inertsil 100A, 7.6 × 250 mm; GL Science) with n-hexane from Fra. A, and pure 1 (oil; 2.1 mg) and 7 (oil; 0.6 mg) were obtained by SiO2 HPLC (Inertsil 100A, 7.6 × 250 mm; GL Science) with n-hexane:THF (100:2) from Fra. B.

The structure of compound 11 was determined using MS (Supplementary Fig. 15) and NMR (Supplementary Figs. 16–21). HR-EI-MS detected m/z 410.3904 [M]+ (calculated 410.3913 for C30H50). The structures and purity (> 99%) of 1, 7, and 10 were confirmed by 1H NMR to be consistent with those of previous reports15,27,28.

Analysis of BmeTCX products using a cell-free system

Construction of pColdTF-BmeTCWT and pColdTF-BmeTCY167A has been previously reported20,28. Site-directed mutagenesis of pColdTF-BmeTCWT and pColdTF-BmeTCD373C was performed using the Quik Change Site-directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA) and primers which are listed in Supplementary Table 2. The pColdTF-BmeTCY257A was synthesized by Genewiz (Morrisville, NC, USA) using codon-optimized sequences for E. coli. E. coli BL21(DE3) harboring pColdTF-BmeTCX (X: WT, D373C/A/F/G/H/L/M/N/Q/S/W, Y167A, Y255A, Y257A, N302A, L596A and F600A) was grown at 37 °C in LB medium (1 L) with 100 μg mL-1 ampicillin. The expression and preparation of the a cell-free extract was basically the same as in Ref. 20 as described below. Expression of the recombinant protein was induced by adding 0.1 mM IPTG when OD600 reached ~ 0.6. BL21(DE3) recombinants were further cultured for 24 h at 15°C20. E. coli cells expressing recombinant BmeTCX were harvested by centrifugation and resuspended in buffer A (3 mL/g cells) containing 50 mM Tris–HCl (pH 7.5), 2.5 mM dithiothreitol, 1 mM EDTA, 0.1% ascorbic acid and 0.1% Tween-8020. Cells were disrupted via sonication using an UP200s (Hielscher Ultrasonics GmbH) at 4–10 °C for 15 min20. The resulting suspension was centrifuged at 12,300 × g for 20 min (twice)20. The pellet was discarded, and the resulting supernatant was used as cell-free extracts. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) using 10% gel confirmed that all cell-free extracts contained approximately the same amount of BmeTCX (Supplementary Fig. 1).

To analyze BmeTCX 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 BmeTCX (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 BmeTCX products using purified enzymes

In order to analyze the correct enzyme activity, BmeTCX (X: WT, D373C, Y167A/D373C, Y257A/D373C, N302A/D373C and D373C/L596A) was expressed without the fused TF-tag. The BmeTCWT gene was excised from the NdeI and XhoI sites of pColdTF-BmeTCWT, and introduced into the same site of pColdI to construct pColdI-BmeTCWT. Next, pColdI-BmeTCD373C and pColdI-BmeTCY167A/D373C were synthesized by Genewiz (Morrisville, NC, USA) using codon-optimized sequences for E. coli, following which pColdI-BmeTCY257A/ D373C and pColdI-BmeTCN302A/ D373C were prepared via a Quik Change Site-directed Mutagenesis Kit (Stratagene) using pColdI-BmeTCD373C as a template and the primers listed in Supplementary Table 3. Then pColdI-BmeTCX was introduced into BL21(DE3) together with pGro7, and soluble BmeTCX was expressed. The expression and purification of BmeTCX was basically the same as in Ref. 29 as described below. After culturing as in the case of pColdTF-BmeTCWT, 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 °C29. The homogenate was centrifuged at 18,270 × g for 20 min to prepare the supernatant containing soluble His-tagged 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 8029. 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 BmeTCX were analyzed via 10% SDS-PAGE (Supplementary Fig. 5).

The reaction mixture used to analyze the BmeTCX 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 BmeTCX. Reactions were performed at 37 °C for 64 h. As shown in the representative example in Supplementary Fig. 22, the time-dependent activities of BmeTCX 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.

Conversion of 1 into volatile components

Two ambergris tinctures were prepared with 1 mg ambergris/ 200 µL (95% EtOH). UV treatment of 1 was performed by irradiating a sample [1 mg 1/200 µL (95% EtOH)] in a glass vial with a UV lamp (15 W) at 26 °C for 6 weeks. Visible light treatment of 1 using a photosensitizer was carried out by irradiating a sample [1 mg 1/200 µL (RB and MB: 95% EtOH; TPP: dichloromethane)] in a glass vial with LED visible light lamp (60 W) at 26 °C for 4 h, followed by adding the photosensitizer (RB and MB: 100 µM; TPP: 50 µM) every hour and stirring. The samples were directly injected for analysis of volatile compounds by GC and GC–MS. The compounds 25 were identified by comparing the EIMS spectra of 25 with those of the NIST library (Supplementary Figs. 8–11). Quantification of volatile compounds by GC was performed by comparison with the peak area of authentic 3 (Kao, Tokyo, Japan). Standard deviations were calculated from the results of 3 replicates.

Cell culture and treatments

Murine pre-osteoblastic MC3T3-E1 cells (RIKEN Cell Bank, Tsukuba, Japan) were cultured in Minimum Essential Medium Eagle-alpha modification (α-MEM) supplemented with 10% fetal bovine serum (FBS), 100 μg/mL streptomycin, and 100 U/mL penicillin. For induction of osteoblastic 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% CO2 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 × 106 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 Aβ1-42 (Wako, Osaka, Japan) for an additional 24 h to induce Amyloid β (Aβ)-mediated neurotoxicity.

Alizarin red S staining

MC3T3-E1 cells cultured in the osteoblastic differentiation medium for 3 weeks and were subsequently fixed with 4% PFA for 30 min at 24 °C. Finally, cells were incubated for 45 min at 24 °C with 1% alizarin red S (FUJIFILM Wako Pure Chemical Corp., Osaka, Japan), and then washed by distilled water30.

TRAP staining

Matured osteoclasts were fixed with 10% glutaraldehyde for 15 min at 37 °C and subsequently incubated for 10 min at 37 °C in TRAP staining buffer containing 10 mg/mL naphthol AS-MX phosphate, 0.3 mg/mL Fast Red Violet LB Salt, 0.1 M sodium acetate, 0.3 M potassium tartrate, 0.1% Triton X-100, and 0.1 M acetic acid31. 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 Aβ1-42. 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 × 106 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.