Substrate-Dependent Alteration in the C- and O-Prenylation Specificities of Cannabis Prenyltransferase

Abstract

CsPT4 is an aromatic prenyltransferase that synthesizes cannabigerolic acid (CBGA), the key intermediate of cannabinoid biosynthesis in Cannabis sativa, from olivetolic acid (OA) and geranyl diphosphate (GPP). CsPT4 has a catalytic potential to produce a variety of CBGA analogs via regioselective C-prenylation of aromatic substrates having resorcylic acid skeletons including bibenzyl 2,4-dihydroxy-6-phenylethylbenzoic acid (DPA). In this study, we further investigated the substrate specificity of CsPT4 using phlorocaprophenone (PCP) and 2′,4′,6′-trihydroxydihydrochalcone (THDC), the isomers of OA and DPA, respectively, and demonstrated that CsPT4 catalyzed both C-prenylation and O-prenylation reactions on PCP and THDC that share acylphloroglucinol substructures. Interestingly, the kinetic parameters of CsPT4 for these substrates differed depending on whether they underwent C-prenylation or O-prenylation, suggesting that this enzyme utilized different substrate-binding modes suitable for the respective reactions. Aromatic prenyltransferases that catalyze O-prenylation are rare in the plant kingdom, and CsPT4 was notable for altering the reaction specificity between C- and O-prenylations depending on the skeletons of aromatic substrates. We also demonstrated that enzymatically synthesized geranylated acylphloroglucinols had potent antiausterity activity against PANC-1 human pancreatic cancer cells, with 4′-O-geranyl THDC being the most effective. We suggest that CsPT4 is a valuable catalyst to generate biologically active C- and O-prenylated molecules that could be anticancer lead compounds.

INTRODUCTION

Cannabis sativa produces a number of cannabinoids, the pharmacologically active meroterpenoid metabolites composed of polyketide and monoterpene moieties.¹⁾ Among them, Δ⁹-tetrahydrocannabinol (THC) is the psychoactive principle of marijuana; however, this cannabinoid also has various therapeutic properties, including analgesic, anticonvulsant, and immunosuppressive effects, and has been formulated in Nabiximols (Sativex) for the treatment of multiple sclerosis and neuropathic pain.²⁾ Cannabidiol (CBD), a non-psychotropic THC isomer, has been studied for its effects on neuropsychiatric and neurodevelopmental conditions. A CBD-based medicine, Epidiolex, has been approved by the U.S. Food and Drug Administration for the treatment of rare and severe childhood epilepsy, and is currently undergoing clinical trials in Japan.³⁾ In addition, cannabigerol with a linear geranyl group has therapeutic potential in neurologic disorders and inflammatory bowel diseases by modulating several relevant receptors.⁴⁾ Furthermore, perrottetinene, a bibenzyl cannabinoid isolated from the liverwort Radula perrottetii,⁵⁾ is an intriguing natural product with a centrally-acting analgesic effect in a CB1 cannabinoid receptor-dependent manner.⁶⁾ However, unlike THC, perrottetinene significantly reduces the basal brain prostaglandin levels associated with the side effects of THC.⁶⁾ These studies indicate that cannabinoid-related compounds are beneficial medicinal resources with diverse therapeutic potential.

The biosynthetic enzymes and genes related to the cannabinoid pathway have been comprehensively identified and characterized. As shown in Fig. 1, Δ⁹-tetrahydrocannabinolic acid (THCA) and cannabidiolic acid (CBDA), the acidic precursors of THC and CBD, respectively, are synthesized by THCA synthase and CBDA synthase via oxidative cyclization from the common substrate cannabigerolic acid (CBGA).¹⁾ CBGA is synthesized through the formation of a regioselective C-C bond between olivetolic acid (OA) and geranyl diphosphate (GPP), catalyzed by the recently identified aromatic prenyltransferase (PT) named CsPT4.⁷⁾ CsPT4 is an important enzyme that plays a pivotal role in conjugating two distinct pathways, the polyketide pathway and the isoprenoid pathway, to produce CBGA, the central precursor of several bioactive cannabinoids.¹⁾ Our previous study has demonstrated that CsPT4 has an unusually broad substrate scope for a plant PT. CsPT4 accepts a variety of resorcylic acid derivatives including the bibenzyl cannabinoid precursor 2,4-dihydroxy-6-phenylethylbenzoic acid (DPA) and four different prenyl donor substrates, namely dimethylallyl diphosphate (DMAPP), GPP, farnesyl diphosphate (FPP), and geranylgeranyl diphosphate (GGPP).⁸⁾ In addition, enzymatically produced CBGA analogs show considerable preferential cytotoxic activity against PANC-1 human pancreatic cancer cells under nutrient-deprived conditions.⁸⁾

Fig. 1. Biosynthetic Pathway of the Major Cannabinoids in Cannabis sativa^a

^aTHC and CBD are formed by non-enzymatic decarboxylation.

In this study, we further investigated the substrate specificity of CsPT4 using acylphloroglucinol derivatives such as phlorocaprophenone (PCP, 1) and 2′,4′,6′-trihydroxydihydrochalcone (THDC, 2) as aromatic substrates. We describe the hidden catalytic potential of CsPT4 to catalyze C- and O-prenylation reactions to produce more potent antiausterity compounds.

RESULTS AND DISCUSSION

CsPT4 Reactivity against Acylphloroglucinols

We previously reported that CsPT4 reacts with a variety of resorcylic acid derivatives, but not with secondary metabolites having different skeletons, such as naringenin, resveratrol, p-coumaric acid, and umbelliferone.⁸⁾ However, for a better understanding of the substrate specificity of CsPT4, aromatic substrates with more structural similarity to resorcylic acids should be examined in enzymatic assays. We thus synthesized two acylphloroglucinol substrates, PCP 1 and THDC 2 (Fig. 2), the isomers of the typical substrates OA and DPA, respectively, and conducted enzyme assays in the presence of GPP. Because CsPT4 is a membrane-bound protein, the microsomal fraction of Pichia pastoris expressing the recombinant CsPT4 was used as the enzyme preparation.⁸⁾

Fig. 2. Structures of the Aromatic Substrates Used in This Study

HPLC and LC high-resolution electrospray ionization mass spectrometry (LC-HR-ESI-MS) analyses of the reaction products showed that CsPT4 unexpectedly generated two product peaks with molecular masses corresponding to monogeranylated compounds from both substrates 1 and 2 (Fig. 3). Comparison of retention times and MS spectra with chemically synthesized standard compounds demonstrated that the former peaks 1a and 2a corresponded to the C-geranylated products 3′-C-geranyl PCP and 3′-C-geranyl THDC, respectively, whereas the latter peaks 1b and 2b corresponded to the O-geranylated products 4′-O-geranyl PCP and 4′-O-geranyl THDC, respectively. These reaction products were novel compounds that have not been previously described.

Fig. 3. Analysis of the Reaction Products of the Recombinant CsPT4

(A) HPLC profile of the reaction with 1 and GPP. HPLC conditions: acetonitrile (85%) containing formic acid (0.1%). (B) HPLC profile of the reaction with 2 and GPP. HPLC conditions: acetonitrile (80%) containing formic acid (0.1%). (C) HR-ESI-MS (negative mode) of the standard and enzymatically synthesized 1a, 1b, 2a, and 2b.

CsPT4 accepted the acylphloroglucinol substrates 1 and 2, and catalyzed C- and O-prenylations; in contrast, this enzyme catalyzes regioselective C-prenylation at the C-3 position when resorcylic acid compounds such as OA and DPA are used as substrates.⁸⁾ These results suggest that the active site of CsPT4 does not bind acylphloroglucinol substrates tightly, which may explain why two different reactions occur simultaneously. In addition, CsPT4 did not accept phloroacetophenone with an acetyl group as an aromatic substrate, suggesting that the acyl groups should have suitable size and hydrophobicity to be recognized by CsPT4. We have previously reported a similar situation for alkyl groups attached to resorcylic acid substrates.⁸⁾

We also analyzed enzymatic reactions using prenyl donor substrates other than GPP, and confirmed that CsPT4 afforded two respective products from 1 and 2 when FPP was used as a prenyl substrate (Supplementary Figs. S23, S24). The structures of these products have not been fully characterized at this time, but LC-HR-ESI-MS analysis suggested that they were all monofarnesylated compounds. Therefore, this enzyme catalyzed multiproduct formation even when FPP was used as a substrate. In contrast, DMAPP and GGPP were not appropriate substrates when used with 1 and 2 because no reaction products were observed in enzymatic assays.

Most plant aromatic PTs are C-PTs,⁹⁾ and only a few O-PTs have been identified and characterized.^10,11) As the only PT that catalyzes both C- and O-prenylations, MePT1 from Murraya exotica synthesizes a trace amount of 7-O-geranylumbelliferone along with C-geranylated major products.¹²⁾ In addition, plant aromatic PTs are characterized by their high substrate specificity, and in general recognize aromatic substrates with specific skeletons.⁹⁾ Therefore, CsPT4 is a notable enzyme that can accept both resorcylic acid and acylphloroglucinol substrates and alters the C-/O-prenylation patterns depending on the substrate skeleton. These unique features of CsPT4 are similar to those of several microbial PTs,¹³⁾ such as NphB, which has been applied to the synthetic biology of various prenylated aromatic compounds.¹⁴⁾

Kinetic Evaluation of CsPT4 against Acylphloroglucinols

To evaluate the apparent affinity and reactivity of CsPT4 against 1 and 2, we performed the kinetic analysis of CsPT4 for respective C-/O-prenylation reactions in the presence of GPP. For comparison, kinetic parameters of C-prenylation reactions with OA and DPA, to synthesize CBGA and 3-geranyl DPA, respectively, were also analyzed using the same microsomal fraction.

CsPT4 showed comparable apparent K_m values for C-prenylation reactions with all aromatic substrates tested (Table 1). The apparent K_m value for O-prenylation with 1 was also within a similar range. On the other hand, it was unexpectedly observed that CsPT4 showed a higher apparent K_m value for O-prenylation with 2 (259 ± 49 µM) than that for C-prenylation with the same substrate (11.0 ± 0.2 µM), as if a different substrate was used in each reaction (Table 1). One possible explanation is that CsPT4 utilized two different substrate-binding modes suitable for the respective prenylation reactions with acylphloroglucinol substrates. If this hypothesis is correct, it may be difficult for 2 to bind the enzyme in the docking mode suitable to undergo O-prenylation, presumably owing to its bulky aromatic side chain.

Table 1. Steady-State Kinetic Parameters of the Recombinant CsPT4^a)

Substrates	Products (prenylation pattern)	Km [µM]	Vmax [pmol s−1]	Vmax/Km [fmol s−1 µM−1]
OA	CBGA (C)	12.7 ± 0.9	8.27 ± 0.20	651 ± 32
DPA	3-Geranyl DPA (C)	16.9 ± 0.7	8.31 ± 0.13	492 ± 19
1	1a (C)	14.2 ± 2.0	0.944 ± 0.037	67.1 ± 7.5
1	1b (O)	19.0 ± 1.5	7.81 ± 0.09	412 ± 29
2	2a (C)	11.0 ± 0.2	6.00 ± 0.29	546 ± 30
2	2b (O)	259 ± 49	0.612 ± 0.067	2.37 ± 0.21

a) All reactions were performed with the same microsomal fraction as the crude enzyme.

Also interestingly, CsPT4 showed higher V_max values for O-prenylation with 1 (7.81 ± 0.09 fmol s⁻¹) and C-prenylation with 2 (6.00 ± 0.29 fmol s⁻¹) than those for alternative prenylation patterns with the same substrates (Table 1). The V_max values for the respective priority reactions with 1 and 2 were comparable with those for C-prenylation with OA and DPA. The V_max/K_m values, a measure of catalytic efficiency, showed a similar tendency for each substrate. These results suggested that CsPT4 preferentially catalyzed O-prenylation with 1 to synthesize 1b and C-prenylation with 2 to produce 2a. The catalytic efficiency of CsPT4 may depend on how suitably the geranyl cation intermediate, which is derived from GPP, is positioned relative to the aromatic substrate, as in the case of the structurally characterized microbial PT.¹⁵⁾ It is nevertheless a unique property of CsPT4 that the priority of prenylation patterns changes depending on the substrate. Analysis of the tertiary structure of CsPT4 is essential to elucidate the structural basis for the unique properties of this enzyme.

Preferential Cytotoxic Activity of C- and O-prenylated Products against PANC-1 Human Pancreatic Cancer Cells

Cancer cells adapt to the low nutritional and hypoxic conditions in the microenvironment by altering their energy metabolism. In particular, human pancreatic cancer cells such as PANC-1 have acquired this ability, which is called austerity.¹⁶⁾ Thus, antiausterity compounds that preferentially kill cancer cells under nutrient-deprived conditions are considered promising targets for novel anticancer drugs. To date, the antiausterity activity of diverse natural and synthetic compounds, including prenylated polyphenols, has been investigated.^17,18)

We tested the preferential cytotoxicity of non-prenylated (1 and 2) and four prenylated acylphloroglucinols (1a, 1b, 2a, and 2b) against PANC-1 cells cultured in either nutrient-deprived medium (NDM) or nutrient-rich Dulbecco’s modified Eagle’s medium (DMEM). The antiausterity activity was presented as a PC₅₀ value, representing the concentration that kills 50% of cells cultured in NDM, whereas the IC₅₀ value, measured for cells cultured in DMEM, indicates conventional cytotoxicity.

Compound 1 displayed considerable preferential cytotoxicity despite lacking a prenyl moiety, with a PC₅₀ of 4.4 µM, whereas the activities of prenylated compounds 1a and 1b were marginally enhanced, with PC₅₀ values of 2.4 and 2.7 µM, respectively (Table 2). This result was somewhat unexpected because OA, the isomer of 1, acquires cytotoxicity only after being prenylated and becoming CBGA analogs.⁸⁾ Meanwhile, compounds 2a and 2b showed significantly higher activity than that of 2, with PC₅₀ values of 4.8 and 0.99 µM, respectively (Table 2). Although all tested compounds showed some cytotoxicity, the most notable compound was 2b, which showed a preferential cytotoxic profile (Fig. 4) with a potency comparable to that of arctigenin (Table 2), a well-studied antipancreatic cancer natural product that underwent Phase 1 clinical trials.^19,20) In addition, the PC₅₀ value of 2b was several-fold lower than those of CBGA analogs previously obtained as reaction products of CsPT4.⁸⁾ Therefore, the catalytic potential of CsPT4 facilitated the discovery of anticancer lead compound 2b, an O-prenylated chalcone with potent antiausterity activity. Prenylated chalcone derivatives such as nicolaioidesin C and panduratin A from Boesenbergia pandurata also exhibit high antiausterity activities,^17,18) and we consider that this category of natural products are potential medicinal resources.

Table 2. Preferential Cytotoxic Activity of CsPT4 Products against PANC-1 Human Pancreatic Cancer Cells

Compound	PC50 (µM) in NDM	IC50 (µM) in DMEM
1	4.4	>100
2	14	>100
1a	2.4	51
1b	2.7	94
2a	4.8	29
2b	0.99	93
Arctigenina)	0.82	>100

a) Data for arctigenin were from Ref. 8.

Fig. 4. Preferential Cytotoxic Activity of 2b against PANC-1 Cells in NDM (Red) and DMEM (Blue)

Data are means ± standard deviation from triplicate measurements.

CONCLUSION

In this study, we demonstrated that CsPT4 not only had broad substrate specificity but also showed the remarkable catalytic property of changing its prenylation patterns depending on the substrate structures. In addition, O-prenylated chalcone 2b emerged as a potent antiausterity compound. 2b is a minor product in the enzymatic reaction catalyzed by CsPT4, but it would be possible to redesign the enzymatic function to produce 2b efficiently once the structural analysis of CsPT4 is performed. Furthermore, studies on structure–activity relationships of 2b-related prenylated chalcones may lead to the discovery of more potent anticancer compounds.

Acknowledgments

We sincerely thank Dr. S. Sun (University of Toyama) for technical assistance in biological activity experiments. This work was supported by JSPS KAKENHI Grant Number: JP21K06627 (to F.T.), JST SPRING Grant Number: JPMJSP2145 (to R.T.), Kobayashi International Scholarship Foundation (to F.T.), and Takahashi Industrial and Economic Research Foundation (to F.T.).

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

REFERENCES

• 1) Gülck T, Møller BL. Phytocannabinoids: origins and biosynthesis. Trends Plant Sci., 25, 985–1004 (2020).
• 2) MacCallum CA, Russo EB. Practical considerations in medical cannabis administration and dosing. Eur. J. Intern. Med., 49, 12–19 (2018).
• 3) Masataka Y, Sugiyama T, Takumi I, Yamamoto H. Cannabidiol supplement reduces epileptic seizures in the Japanese population: cross-sectional study for intractable epilepsy patients. Neurol. Asia, 27, 891–898 (2022).
• 4) Nachnani R, Raup-Konsavage WM, Vrana KE. The pharmacological case for cannabigerol. J. Pharmacol. Exp. Ther., 376, 204–212 (2021).
• 5) Toyota M, Kinugawa T, Asakawa Y. Bibenzyl cannabinoid and cisbibenzyl derivative from the liverwort Radula perrottetii. Phytochemistry, 37, 859–862 (1994).
• 6) Chicca A, Schafroth MA, Reynoso-Moreno I, Erni R, Petrucci V, Carreira EM, Gertsch J. Uncovering the psychoactivity of a cannabinoid from liverworts associated with a legal high. Sci. Adv., 4, eaat2166 (2018).
• 7) Luo X, Reiter MA, d’Espaux L, et al. Complete biosynthesis of cannabinoids and their unnatural analogues in yeast. Nature, 567, 123–126 (2019).
• 8) Tanaya R, Kodama T, Lee YE, Yasuno Y, Shinada T, Takahashi H, Ito T, Morita H, Awale S, Taura F. Catalytic potential of Cannabis prenyltransferase to expand cannabinoid scaffold diversity. Org. Lett., 25, 8601–8605 (2023).
• 9) de Bruijn WJC, Levisson M, Beekwilder J, van Berkel WJH, Vincken JP. Plant aromatic prenyltransferases: tools for microbial cell factories. Trends Biotechnol., 38, 917–934 (2020).
• 10) Munakata R, Olry A, Takemura T, et al. Parallel evolution of UbiA superfamily proteins into aromatic O-prenyltransferases in plants. Proc. Natl. Acad. Sci. U.S.A., 118, e2022294118 (2021).
• 11) Han L, Zhang L, He Y, Liao L, Li J, Xu S, Zhao Y, Bian X, Xia Y. Three carbon-/oxygen-prenyltransferases responsible for furanocoumarin synthesis in Angelica dahurica. Ind. Crops Prod., 200, 116814 (2023).
• 12) Li N, Liu X, Zhang M, Zhang Z, Zhang B, Wang X, Wang J, Tu P, Jiang Y, Shi SP. Characterization of a coumarin C-/O-prenyltransferase and a quinolone C-prenyltransferase from Murraya exotica. Org. Biomol. Chem., 20, 5535–5542 (2022).
• 13) Mori T. Enzymatic studies on aromatic prenyltransferases. J. Nat. Med., 74, 501–512 (2020).
• 14) Qian S, Clomburg JM, Gonzalez R. Engineering Escherichia coli as a platform for the in vivo synthesis of prenylated aromatics. Biotechnol. Bioeng., 116, 1116–1127 (2019).
• 15) Cheng W, Li W. Structural insights into ubiquinone biosynthesis in membranes. Science, 343, 878–881 (2014).
• 16) Izuishi K, Kato K, Ogura T, Kinoshita T, Esumi H. Remarkable tolerance of tumor cells to nutrient deprivation: possible new biochemical target for cancer therapy. Cancer Res., 60, 6201–6207 (2000).
• 17) Phan ND, Omar AM, Takahashi I, Baba H, Okumura T, Imura J, Okada T, Toyooka N, Fujii T, Awale S. Nicolaioidesin C: an antiausterity agent shows promising antitumor activity in a pancreatic cancer xenograft mouse model. J. Nat. Prod., 86, 1402–1410 (2023).
• 18) Sun S, Kim MJ, Omar AM, Phan ND, Awale S. (+)-Panduratin A induces PANC-1 human pancreatic cancer cell death preferentially under nutrient starvation by inhibiting PI3K/Akt/mTOR/autophagy signaling pathway. Phytomed. Plus, 1, 100101 (2021).
• 19) Awale S, Lu J, Kalauni SK, Kurashima Y, Tezuka Y, Kadota S, Esumi H. Identification of arctigenin as an antitumor agent having activity to eliminate cancer cells tolerance to nutrient starvation. Cancer Res., 66, 1751–1757 (2006).
• 20) Ikeda M, Sato A, Mochizuki N, Toyosaki K, Miyoshi C, Fujioka R, Mitsunaga S, Ohno I, Hashimoto Y, Takahashi H, Hasegawa H, Nomura S, Takahashi R, Yomoda S, Tsuchihara K, Kishino S, Esumi H. Phase I trial of GBS-01 for advanced pancreatic cancer refractory to gemcitabine. Cancer Sci., 107, 1818–1824 (2016).