Formal Chemoenzymatic Skeletal Editing of Fusicoccane Core to Access 6‑Aza-Fusicoccanes

Abstract

Fusicoccane (FC)-type diterpenoids constitute a family of cyclic diterpenoids featuring a 5/8/5-fused tricyclic carbon skeleton. These natural products have attracted considerable attention owing to their unique biological functions, including plant hormone-like properties and antiproliferative activities against cancer cells. Herein, we report the stereoselective synthesis of 6-aza-FC-type analogs via a chemoenzymatic skeletal editing approach using CotB2, a diterpene synthase responsible for constructing the FC skeleton. Based on the biosynthetic reaction mechanism, we designed an unnatural substrate bearing an amide moiety at the C6–C7 position of geranylgeranyl pyrophosphate. CotB2 promoted the cyclization of this substrate to afford an optically active 5/11-fused lactam in a stereoselective manner. Subsequent intramolecular cyclization of the resulting lactam through an epoxide intermediate furnished 6-aza-FC derivatives, in which the C6 carbon of FC skeleton is replaced with a nitrogen atom. Notably, one of the 6-aza-FC analogs exhibited selective cytotoxicity toward cancer cells, while displaying reduced activity toward HUVEC cells.

Introduction

Fusicoccane (FC)-type diterpenoids are a family of cyclic diterpenoids that share a 5/8/5-fused tricyclic carbon skeleton (Figure ). The characteristic FC-type carbon skeleton 2 is biosynthesized by the diterpene cyclase-mediated cyclization of the C₂₀ acyclic substrate, geranylgeranyl pyrophosphate (GGPP, 1). − More than 300 FCs have been isolated from bacteria, fungi, liverworts, and higher plants. Fusicoccin A (3), cotylenin A (4), and cyclooctatin (5) are representative members of this natural product family. Fusicoccin A (3) was isolated in 1964 from the plant-pathogenic fungus Phomopsis amygdali as a phytotoxin. Cotylenin A (4) was isolated from the fungus Cladosporium sp. 501–7W as a plant hormone-like diterpenoid by Sassa in 1970. These natural products have received significant attention due to their intriguing mode of action as a molecular glue , that stabilize protein–protein interactions , which are responsible for the characteristic biological functions of 3 and 4. − Cyclooctatin (5) was isolated from Streptomyces melanosporofaciens MI614–43F2 by Aoyagi et al. This natural product was discovered by a screening program of microbial natural products for lysophospholipase inhibitors. Cyclooctatin (5) is of potential utility as a selective lysophospholipase inhibitor. The diverse and potent biological properties of FC-type diterpenoids have stimulated extensive synthetic studies of FC-type natural products and their analogs.

1.

(a) Biosynthetic pathway of fusicoccane-type carbon skeleton 2 from GGPP (1). (b) Structures of fusicoccin A (3), cotylenin A (4), and cyclooctatin (5).

Nitrogen-containing heterocycles occupy a pivotal position as privileged scaffolds in medicinal chemistry. Recent statistical analyses have highlighted the growing importance of nitrogen-containing heterocycles in drug discovery. , In this context, rapid and facile synthesis of nitrogen-containing heterocycles has become an important subject in medicinal chemistry. Among various approaches, skeletal editing has garnered considerable interest as a transformative approach to remodeling carbocyclic drug leads to be more drug-like. , In this study, we focused on skeletal editing that enables the remodeling of naturally occurring terpenoid cores into nitrogen-containing heterocycles. Cyclic terpenoids occupy a distinct sp³ carbon-rich chemical space , compared with synthetic sp² carbon-rich planar molecules. , The replacement of one carbon atom in terpenoid skeletons with a nitrogen atom provides structurally unique terpenoid-like nitrogen-containing heterocycles. However, such skeletal editing requires scission of inert sp³ carbon–carbon bonds within rigid terpenoid cores, presenting a significant synthetic challenge.

Recently, the synthesis of heteroatom-incorporated terpenoid-like molecules using type I terpene cyclase-mediated cyclization has been achieved through chemoenzymatic synthesis (SI section 1). − Several structurally complex nitrogen-, oxygen- and sulfur-containing heterocycles have been synthesized in a single step from unnatural heteroatom-containing acyclic substrates. This approach, herein termed “chemoenzymatic skeletal editing of terpenoids,” holds significant synthetic potential for accessing unique terpenoid-like heterocycles. However, to the best of our knowledge, the synthesis of amide-containing polycyclic terpenoid-like molecules through chemoenzymatic skeletal editing has not yet been reported. Herein, we report the first synthesis of 6-aza-FC derivatives using CotB2, the diterpene synthase encoded by the cyclooctatin biosynthetic gene cluster as a biocatalyst (Scheme ). The approach exploits the substrate promiscuity of CotB2 to accept an amide-bearing unnatural substrate. A combination of CotB2-mediated cyclization and chemical cyclization afforded novel 6-aza-FC derivatives that retain the 5/8/5-fused tricyclic core of FCs.

1. This work: Chemoenzymatic Skeletal Editing of FC Core.

The present approach is underpinned by a synthetic design based on the biosynthetic reaction mechanism of CotB2, a type-I bacterial diterpene synthase derived from S. melanosporofaciens MI614–43F2. CotB2 promotes the cyclization of GGPP (1) to form cyclooctat-9-en-7-ol (6), the 5/8/5-tricyclic skeleton of cyclooctatin (5) (Scheme ). X-ray crystallographic analysis of CotB2 , and computational studies of the biosynthetic reaction mechanism − have demonstrated that GGPP (1) adopts a chiral helical conformation A upon binding at the catalytic site of CotB2. This preorganized product-like conformation A is essential for stereoselective cyclization to form an optically active 5/11-fused bicyclic carbocation B (Step 1). The resulting cation B undergoes a subsequent cationic cascade via a 1,5-hydride shift to furnish 6 (Step 2). On the basis of the biosynthetic reaction mechanism, we envisioned the design and synthesis of 6-aza-FC derivatives from GGPP analog 7 via 5/11-fused bicyclic lactam 8. In the CotB2-catalyzed reaction, the trisubstituted olefin at C6–C7 is not involved in the formation of carbocation intermediate B. This mechanistic feature inspired our substrate design, in which the trisubstituted olefin at C6–C7 of GGPP (1) is replaced by an amide moiety. This substitution was expected to preserve productive binding at the CotB2 catalytic site while destabilizing the downstream 1,5-hydride shift pathway through formation of an unfavorable α-carbonyl cation. In addition, docking studies of the CotB2–amide complex C confirmed that the binding pose of 7 at the CotB2 catalytic site closely resembles the product-like conformer observed in the CotB2–GGPP complex A. On this basis, we postulated that 7 would undergo CotB2-mediated cyclization to generate the chiral 5/11-fused bicyclic cation D. A 1,5-hydride shift from cation D would produce unstable α-carbonyl cation E. Given the inherent instability of α-carbonyl cations, the cationic cascade would instead be terminated at cation D to afford 5/11-bicyclic lactam 8. The resulting lactam 8 could be converted to the desired 6-aza-FC derivative 9 via intramolecular C–N bond formation between N6 and the C2–C3 olefin.

2. (a) Biosynthesis of Cyclooctat-9-en-7-ol from GGPP (1); (b) Synthetic Plan of 6-Aza-FC Analog 9 from Amide Substrate 7 .

Results and Discussion

Amide substrate 7 was prepared via coupling of amine 14 and carboxylic acid 15 (Scheme a). Amine 14 was prepared from known ketone 10. Horner–Wadsworth–Emmons olefination of 10 gave unsaturated ester 12 (E/Z = 3/1). After removal of Z-12 in part, a 7.7:1 E/Z mixture was transformed to amine 14 by a series of transformations: (i) reduction with diisobutylaluminum hydride (DIBAL-H), (ii) silyl protection of the resulting alcohol with tert-butyldiphenylsilyl chloride (TBDPSCl), and (iii) removal of the tert-butoxycarbonyl (Boc) group using trimethylsilyl trifluoromethanesulfonate (TMSOTf) in the presence of 2,6-lutidine. , In these steps, most of the Z-isomers were removed by each column chromatography. Coupling of acid 15 with amine 14 in the presence of 1,1′-carbonyldiimidazole (CDI) gave amide 16. Deprotection of the TBDPS group of 16 afforded alcohol 17. After isolation of E-17, E-17 was converted to amide substrate 7 in two steps: (i) conversion to an allyl chloride and (ii) pyrophosphorylation with tris­(tetrabutylammonium) diphosphate (18).

3. (a) Chemoenzymatic Synthesis of 5/11-Fused Bicyclic Lactam 8 from 10; (b) ORTEP Diagram of 8 (Flack Parameter, −0.07(11)); the Numbering System for Fusicoccane-Type Diterpenes, (c) GC/MS Spectrum of the Crude CotB2 Product (SI-experimental section 5).

CotB2-mediated cyclization of amide substrate 7 was initially tested in a small-scale experiment. CotB2 was produced by heterologous expression in E. coli and purified by Ni-affinity column chromatography. The cyclization was carried out by treatment of 7 with 5 mol % of CotB2 in the presence of Mg²⁺ ions in buffer. Gas chromatography–mass spectrometry (GC-MS) analysis of the crude ethyl acetate extract showed a major peak at 15.9 min with m/z 275, consistent with a cyclization product (Scheme c). Blank experiments suggested that Mg²⁺ ions are indispensable for the conversion (SI Figure S2). The enzymatic cyclization was performed on a 400 mg scale of 7 using 1 mol % of CotB2, affording 8 (136 mg, 62%). Nuclear magnetic resonance (NMR) analysis of the product indicated that 8 was a 5/11-fused bicyclic lactam (Table S4). The absolute stereochemistry of 8 was unambiguously determined by single crystal X-ray diffraction analysis to be 10S, 11R, 14S (Scheme b). Notably, the conversion efficiency was comparable to that of the CotB2-mediated cyclization of GGPP (1) to give cyclooctat-9-en-7-ol (6) in 68% yield (SI section 7-5). These results indicated the remarkable substrate promiscuity of CotB2, enabling efficient conversion of unnatural substrate 7 to 8. The stereochemical outcome of the cyclization was governed by CotB2 via adoption of the preorganized product-like conformer C (Scheme b). Furthermore, the amide group directed the cyclization pathway toward preferential formation of 5/11-fused bicyclic lactam 8 (Scheme ).

X-ray analysis of 8 revealed that the C2–C3 olefin stands perpendicular to the 11-membered ring plane to minimize transannular nonbonded interactions. In this conformation, the amide nitrogen (N6) and the C2–C3 olefin are in close proximity (N6···C2, 3.2 Å). We anticipated that 5/11-fused bicyclic lactam 8 could be transformed to the target tricyclic lactam 9 stereoselectively through nucleophilic addition of the amide nitrogen of 8 to an electrophilic counterpart such as a C2–C3 epoxide (Scheme ). Toward this end, the key C–N bond formation was attempted by transannular S_N2 substitution of epoxide 19. Treatment of 8 with one equivalent of triazox gave monoepoxide 19 as a single stereoisomer. The resulting epoxide 19 smoothly underwent base-promoted intramolecular C–N bond formation to furnish 6-aza-FC derivative 9. Single crystal X-ray diffraction analysis established the absolute stereochemistry of the newly created stereogenic centers in 9 to be 2R, 3S. These results revealed that (i) the trisubstituted C2–C3 olefin embedded in the strained 11-membered ring was more reactive toward the electrophilic oxidant than the C15–C19 olefin, (ii) the addition of triazox occurred exclusively from the less hindered peripheral Si-face of the C2–C3 olefin of 8 (Scheme b), and (iii) transannular nucleophilic addition of the amide nitrogen to epoxide 19 proceeded in an S_N2 fashion to furnish 6-aza-FC derivative 9 bearing the same 5/8/5-fused tricyclic skeleton as FCs.

4. (a) Conversion of 8 to 6-aza-FC Analogs 9 and 21ab; ORTEP Diagram of 9, 21a, and 21b; Flack Parameter for 9 Refining to −0.03(19). (b) A Proposed Model for the Facial Selective Epoxidation of Lactam 8 .

Treatment of 8 with excess amounts of triazox provided an inseparable 4:1 mixture of bis-epoxide 20. Without separation, 20 was subjected to the base-promoted intramolecular C–N bond formation to afford a mixture of 5/8/5-fused tricyclic lactams 21a and 21b. These products were separated by reversed-phase HPLC. The stereochemistry of 21a and 21b was determined by single crystal X-ray diffraction analysis to be 15R for minor diastereomer 21a and 15S for major diastereomer 21b, respectively. The stereoselectivity of the epoxidation is rationalized based on the crystal structure of 8 (Scheme b). According to this model, triazox preferentially reacted from the sterically less hindered Re-face of the C15–C19 olefin to give 21b as the major diastereomer (Scheme b).

GGPP (1) was converted to the 5/8/5-fused tricyclic skeleton via the 1,5-hydride shift from cation intermediate B, whereas amide substrate 7 underwent termination at cation D to afford 5/11-fused bicyclic lactam 8 (Scheme ). This mechanistic difference was further supported by density functional theory (DFT) calculations. In the biosynthetic pathway leading to cyclooctat-9-en-7-ol (6), the activation energy (IM1 to TS_1–2) and relative Gibbs free energy (IM2 relative to IM1) were estimated to be +4.5 kcal/mol and −3.9 kcal/mol, respectively (Figure a). In contrast, the activation energy (IM1_amide to TS_1–2_amide) and relative Gibbs free energies (IM2_amide relative to IM1_amide) were calculated to be +32.8 kcal/mol and +22.8 kcal/mol, respectively (Figure b). These results demonstrate that formation of α-carbonyl cation, IM2_amide, is highly disfavored. Instead, cation cascade termination at IM1_amide to afford 8 is governed by the electron-withdrawing effect of the amide carbonyl, which destabilizes the adjacent carbocation.

2.

(a) Energy diagrams for the putative 1,5-rearrangement pathways from IM1to IM2 via TS_1-2. (b) Energy diagrams for the putative 1,5-rearrangement pathways from IM1_amide to IM2_amide via TS_1-2_amide. Gibbs free energies of intermediates and transition structures (kcal/mol) are shown. These are calculated at the mPW1PW91/6–31+G­(d,p) level of theory based on B3LYP/6–31+G­(d,p) geometries. ColabReaction was used for transition state searches, and further optimization was performed by Gaussian16.

We next investigated the cytotoxic activities of 5/11-fused lactams 8 and 20, and 5/8/5-fused lactams 21a and 21b against three cancer cell lines, human pancreatic cancer cells (PANC-1), human colorectal cancer cells (HT-29), and human myeloma cells (KMS-11) (Table ). Human umbilical vein endothelial cells (HUVEC) were also employed for screening of the selective anticancer agents. The cytotoxic activity of each compound against each cell line was evaluated using three independent biological replicates (n = 3). For comparison, the cytotoxic activities of cyclooctat-9-en-7-ol (6) were also evaluated. Interestingly, 6 exhibited moderate cytotoxic activity against all cell lines. In contrast, 5/11-bicyclic lactam 8 and its bis-epoxide analog 20 showed no or negligible cytotoxic activity. Notably, 6-aza-FC analogs 21a and 21b showed selective cytotoxic activities against KMS-11 and/or HT-29. Aza-FC analog 21b with 15S-epoxide moiety was more potent than 15R-isomer 21a. Although the mechanistic details are currently under investigation, 21a and 21b represent promising lead compounds for the development of selective anticancer agents. Further structure activity relationship experiments to identify the key functional groups and improve the potency are ongoing.

1. Cytotoxicity of Cyclooctat-9-en-7-ol (6), and Lactams 8, 20, and 21ab .

	IC50 (μM)
Cell lines	6	8	20	21a	21b
PANC-1	30	86	>300	>300	>300
HT-29	30	225	>300	>300	55
KMS-11	28	>300	>300	141	30
HUVEC	19	171	>300	>300	>300

^aHuman pancreatic cancer cells (PANC-1).

^bHuman colorectal cancer cells (HT-29).

^cHuman myeloma cells (KMS-11).

^dHuman umbilical vein endothelial cells (HUVEC).

In summary, we have achieved the chemoenzymatic synthesis of novel nitrogen-containing 5/11-fused bicyclic lactam 8 and 5/8/5-aza-FC derivatives in a stereoselective manner. This approach complements the chemical single carbon atom replacement of structurally complex carbocycles which has posed a significant synthetic challenge, particularly for polycyclic terpenoids. We found that one of the 6-aza-FC derivatives displayed selective cytotoxicity against cancer cells. These results demonstrate that chemoenzymatic skeletal editing of biologically active terpenoids is a promising strategy for the discovery of new drug leads. Expansion of the present chemoenzymatic approach for the development of novel biologically active aza-terpenoid-like compounds is currently underway.

Glossary

Abbreviations

FC

fusicoccane

CotB2

cyclooctatin biosynthetic gene encoded diterpene synthase

GGPP

geranylgeranyl pyrophosphate

DIBAL-H

diisobutylaluminum hydride

TBDPSCl

tert-butyldiphenylsilyl chloride

Boc

tert-butoxycarbonyl

TMSOTf

trimethylsilyl trifluoromethanesulfonate

CDI

1,1′-carbonyldiimidazole

GC-MS

gas chromatography–mass spectrometry

NMR

nuclear magnetic resonance

DFT

density functional theory

PANC-1

human pancreatic cancer cells

HT-29

human colorectal cancer cells

KMS-11

human myeloma cells

HUVEC

human umbilical vein endothelial cells

IM

intermediate

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.6c00833.

• Experimental procedures, analytic data of new compounds (PDF)

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. S.E. and E.F. contributed equally to this work.

This work was supported by JSPS KAKENHI (22H05120 to T.K., 25K01905 and 23H04560 to T.S.) and a Grant-in-Aid for JSPS Fellows (24KJ1908 to E.F.).

The authors declare no competing financial interest.

Published as part of JACS Au special issue “Molecular Editing in Natural Products and Drug Molecules”.