Introduction

Terpenoid natural products have long been valuable sources of inspiration for developing therapeutically useful molecules1,2. Labdanes represent a structurally unique diterpenoid subclass encompassing pharmaceutically significant metabolites3, including andrographolide, forskolin, and nimbolide. Its structural diversity stems from its assembly line-like biosynthesis, where the building block is first assembled, followed by structural diversification. Representative modes of scaffold diversification include skeletal rearrangement, redox processes, appendages adding, dimerization, or combinations thereof. Among these, skeletal rearrangement stands out as it can generate entirely distinct skeletal frameworks, often leading to new families of terpenoids. This biosynthetic logic has provided significant inspiration for chemists in the development of synthetic strategies4,5,6,7,8,9,10,11,12,13,14,15,16,17.

1,2-Migration is extensively involved in the biosynthetic skeletal rearrangement of terpenoids, the classic example of which is 1,2-H and 1,2-methyl migration. Carbocation-mediated 1,2-migration rearrangements, particularly Wagner-Meerwein, pinacol, and semipinacol rearrangements, have been effectively utilized, demonstrating their powerful potential in the synthesis of terpenoids18,19,20,21,22,23,24,25. In contrast, examples of late-stage 1,2-migration rearrangements of the polycyclic ring systems proceeding via unconventional reaction mechanisms, in which carbocations might not be involved, remain scarce26,27,28,29. The discovery and utilization of migration patterns, particularly carbanion-mediated 1,2-acyl migration (Fig. 1a)30,31,32,33,34,35, remain urgent pursuits, and elucidating these skeletal rearrangements is of perennial importance in natural products chemistry.

Fig. 1: Background and our synthesis plan of the structurally diverse Haplomitrium diterpenoids.
Fig. 1: Background and our synthesis plan of the structurally diverse Haplomitrium diterpenoids.
Full size image

a Typical 1,2-alkyl/aryl shift involved in natural product biomimetic synthesis. b Our bioinspired skeletal reorganization strategy for the synthesis of structurally diverse Haplomitrium diterpenoids. rearr. rearrangement.

Since 2016, we have islolated a series of extensively rearranged and highly oxidized labdane-type diterpenoids (1–4, 7, 34, and 36) from the Chinese liverwort Haplomitrium mnioides, which show potent anti-inflammatory and allelopathic activities36,37,38,39. Structurally, hapmnioide A (1), haploide C (3), and haplomitrin A (4) feature unique 1,1’-bicyclopentyl, bicyclo[3,2,1]octane, and tetracyclo[7.4.1.02,7.011,14]tetradecane ring systems, respectively. Despite great progress in labdane synthesis, however, the development of unified strategies toward the collective construction of biogenetically related labdane-type diterpenoid skeletons is still limited40,41,42,43,44,45. In this work, inspired by the skeletal rearrangement in their biosynthetic pathways and in connection with our interest in divergent synthesis46,47,48, we envisage that these related Haplomitrium diterpenoids can be forged through late-stage biomimetic skeletal reorganization of the common labdane scaffold, featuring previously unreported tandem 1,2-acyl migration (Fig. 1b). Further investigations of 1,2-acyl migrations on other terpenoid scaffolds demonstrate the applicability of this late-stage remodeling strategy for accessing complex molecules.

Results

Retrosynthetic analysis

Our synthesis plan, which is inspired by the biosynthetic pathways of Haplomitrium diterpenoids, is shown in Fig. 2. We envisioned that the 1,1’-bicyclopentyl carbon framework of hapmnioide A (1) could be derived successively from trans-decalin 5, employing two consecutive stereospecific 1,2-acyl migrations of the labdane-type core skeleton at C-1 and C-649,50. Similarly, haploide O (2) could also be accessed via one stereospecific 1,2-acyl migration at C-20 from 5. To construct the bicyclo[3,2,1]octane of haploide C (3), we proposed that the tandem C5-isomerization/aldol/retro-Claisen reactions of aldehyde 6 are responsible for the direct assembly of 3. Highly rigid haplomitrin A (4) can be straightforwardly synthesized via an intramolecular light-initiated [2+2] cycloaddition of enone 737,51. Compounds 5–7 all feature the same labdane core skeleton. Thus, the proposed biosynthetic precursor 8 could serve as a common advanced precursor for quickly synthesizing 5–7 via functional group interconversions. Additionally, δ-lactone 8 could be derived from γ-lactone 9 by the installation of the furan-containing side chain, and the trans-decalin framework of 9 could be accessed by an endo-selective intramolecular Diels–Alder (IMDA) reaction of trienolide 10, concomitantly establishing the pivotal C10 quaternary carbon stereogenic center52. Trienolide 10 can be rapidly generated by the coupling of aldehyde 11 with selenolactone 12, which should be readily prepared from known compounds 13 and 14, respectively53,54,55.

Fig. 2: Retrosynthetic analysis.
Fig. 2: Retrosynthetic analysis.
Full size image

The retrosynthetic analysis of Haplomitrium diterpenoids is based on a late-stage bioinspired skeletal reconstruction approach. FGI functional group interconversion, TBS tert-butyldimethylsilyl.

Synthesis of the common intermediate 8

Our synthesis commenced with the preparation of the chiral aldehyde 11 (Fig. 3), which was obtained via a three-step sequence of the known aldehyde 13 (prepared from geraniol in three steps53,54; see Supplementary Fig. 1a) in 60% overall yield involving epoxidation, Horner–Wadsworth–Emmons reaction, and oxidative cleavage. Subsequently, aldehyde 11 condensed with the lithium enolate of selenolactone 12 (prepared from L-glutamic acid in four steps55; see Supplementary Fig. 1b), followed by oxidation to afford trienolide 10 with 70% yield and 1:1 selectivity at C1. Trienolide 10 underwent a highly stereospecific IMDA reaction to form trans-decalin 9, the structure of which was confirmed by X-ray crystallography52,56,57,58,59,60. The selectivity was attributed to steric effects, which were mediated by the endo-selective transition state (TSendo). The isomeric product C1-iso-9, derived from the cyclization of C1-iso-10, could be converted to 9 via a two-step sequence involving oxidation and selective reduction using L-selectride in 90% yield. To introduce the C6 oxidation state, a singlet oxygen-mediated ene reaction of 9 was employed, yielding allyl alcohol 15 with moderate regioselectivity61. Notably, this transformation demonstrated a reduced reaction time and enhanced productivity when conducted in continuous-flow mode (see Supplementary Fig. 1c).

Fig. 3: Synthesis of the common intermediate 8.
Fig. 3: Synthesis of the common intermediate 8.
Full size image

m-CPBA 3-chloroperoxybenzoic acid, NaHMDS sodium bis(trimethylsilyl)amide, THF tetrahydrofuran, LDA lithium diisopropylamide, BHT 2,6-di-tert-butyl-4-methylphenol, MB methylene blue, TBSOTf tert-butyldimethylsilyl trifluoromethanesulfonate, DIPEA N, N-diisopropylethylamine, IBX 2-iodoxybenzoic acid, DMSO dimethyl sulfoxide, DMP Dess–Martin periodinane, TMSCHN2 (trimethylsilyl)diazomethane, Me-CBS 5,5-diphenyl-2-methyl-3,4-propano-1,3,2-oxazaborolidine.

We then introduced the furan-containing side chain to 15. Subsequent one-pot selective silyl protection of the C6- and C1-hydroxy groups of 15 furnished silyl ether 16. Triethylsilyl ether 16 was then subjected to IBX-mediated selective desilylation/oxidation to generate an unstable aldehyde intermediate62, which was treated with the C3’-lithiated furan, and further oxidation with DMP yielded furyl ketone 17. The base plays a key role in determining the regioselectivity of furan lithiation, where n-butyllithium predominantly facilitates lithiation at the C3’ position of the furan ring (see Supplementary Fig. 1d). The C–O bond adjacent to the ketone in 17 was reduced with SmI₂, and subsequent methylation of the resulting carboxylic acid using TMSCHN2 provided methyl ester 18. Notably, quenching SmI₂-mediated reduction with dilute HCl efficiently removed the C1-hydroxy triethylsilyl group in 17, resulting in fused cyclic ketal formation (see Supplementary Fig. 1e). Finally, ketone 18 was transformed into key precursor 8 via Corey-Bakshi-Shibata reduction followed by Otera lactonization63. Interestingly, during lactone ring closure attempts, degradation product 19 was observed under basic conditions (K₂CO₃/EtOH), arising from C1-C10 bond cleavage (see Supplementary Fig. 1f).

Synthesis of hapmnioide A (1) and haploide O (2)

Next, we proceeded to synthesize hapmnioide A (1) and haploide O (2) using the common labdane-type intermediate 8 (Fig. 4). We first modulated the oxidation state of 8 to access the key substrate 5, which is required for the 1,2-acyl migration reaction. Tetrabutylammonium fluoride (TBAF) was employed to remove all three silyl protecting groups in 8, exposing the hydroxyl groups at the C1, C6, and C18 positions in one step. Surprisingly, the major product obtained was the C1 hydroxyl isomer 20, which was likely formed via a retro-aldol/aldol process (see Supplementary Fig. 2b)64, analogous to the C1–C10 bond cleavage observed in the formation of degradation product 19. This further highlights the sterically congested environment at C10, which predisposes the C1–C10 bond to cleavage. Treatment of triol 20 with Jones reagent, followed by one-pot methylation of the resulting carboxylic acid, afforded methyl ester 5 in 65% yield, with its structure unequivocally confirmed by X-ray crystallographic analysis.

Fig. 4: Synthesis of hapmnioide A (1) and haploide O (2).
Fig. 4: Synthesis of hapmnioide A (1) and haploide O (2).
Full size image

a Divergent total synthesis of hapmnioide A (1) and haploide O (2) from the common intermediate 8. b Optimization of reaction conditions for pivotal 1,2-acyl migration. TBAF tetrabutylammonium fluoride, KHMDS potassium bis(trimethylsilyl)amide, rt room temperature.

With the proposed biogenetic precursor 5 synthesized, we then systematically investigated its conversion to hapmnioide A (1) and haploide O (2), with the optimal conditions summarized in Fig. 4b49,50. To our delight, treatment of 5 with KHMDS at –70 °C induced stereospecific 1,2-acyl migration at C-20 through intermediate A2, affording haploide O (2) in 50% yield (Fig. 4b, entry 1). Interestingly, increasing the temperature to ambient conditions facilitated sequential stereospecific 1,2-acyl migrations through intermediates A3 and A4, yielding hapmnioide A (1) together with aromatized side products 21 and C4-iso-21 (Fig. 4b, entry 2, the structure of 21 was confirmed by X-ray crystallography). The formation of 21 and C4-iso-21 proceeds through a C-1 1,2-acyl migration followed by retro-Michael cleavage of the C4-C5 bond (see Supplementary Fig. 2c). Higher reaction temperatures are needed, presumably due to the high TS energy barrier of A365. Owing to competing reaction pathways, efforts to increase the yield of hapmnioide A (1) failed (Fig. 4b, entries 3 and 4). The spectroscopic data for synthesized hapmnioide A (1) and haploide O (2) matched those reported in the literature36,38. Notably, β-ketoester 5 was unstable, and underwent retro-Claisen cleavage to yield product 22 during reduced-pressure distillation (Fig. 4b, entry 5). Treatment of 5 with (PhO)2PO2H provided the C5-isomerized product C5-iso-5 in 70% yield (Fig. 4b, entry 6)42. Photochemical strategies have been extensively employed in acyl migration processes35,66,67, prompting our investigation of diverse irradiation conditions. Interestingly, visible light irradiation (395 nm) of compound 5 with triethylamine triggered decarboxylative degradation, affording product 23 (structure unequivocally confirmed by X-ray analysis). Notably, light irradiation is indispensable for this transformation. The higher-energy UV light (302 nm) caused substantial decomposition of 5 (Fig. 4b, entries 7 and 8).

Late-stage 1,2-acyl migrations of polycyclic terpenoids

We further validated the late-stage 1,2-acyl migration rearrangements of labdane-type terpenoids or analogs to skeletal reorganization. As exemplified in Fig. 5, treatment of γ-lactone 24 (prepared from 15 in one step; see Supplementary Fig. 3a), whose five-membered lactone moiety resisted 1,2-ester migration, with KHMDS at room temperature afforded isomerized product 25 as the major species. Consistent with δ-lactone 5, the reaction pathway demonstrated significant temperature sensitivity, impacting the reaction selectivity. When the temperature was increased to 60 °C, two consecutive stereospecific 1,2-acyl migrations predominated, resulting in 1,1’-bicyclopentyl 26 as the major product. To further investigate the impact of γ-lactones on these rearrangement processes, cis-decalin 27 was prepared from 4,4-dimethyl cyclohexanone in four steps to obtain the 1,2-ester migration product (see Supplementary Fig. 3b)68. Interestingly, exposure of the open-chain methyl ester 27 to KHMDS conditions yielded only the isomerized product 28, even at elevated temperatures. We further incorporated the ester moiety into the bicyclic framework, obtaining trans-[6.6]-bicyclic δ-lactones 29 from forskolin in three steps (see Supplementary Fig. 3c)69. While treatment of 29 with KHMDS at room temperature provided the 1,2-ester migration product 30 in 33% yield. All the rearranged molecules (25, 26, 28, and 30) were unambiguously characterized by comprehensive NMR studies, and their structures were further confirmed by X-ray crystallographic analysis.

Fig. 5: Late-stage 1,2-acyl migration rearrangements of polycyclic terpenoids.
Fig. 5: Late-stage 1,2-acyl migration rearrangements of polycyclic terpenoids.
Full size image

KHMDS potassium bis(trimethylsilyl)amide, rt room temperature, THF tetrahydrofuran.

Synthesis of haploide C (3)

We next pursued the biomimetic synthesis of haploide C (3), featuring a bicyclo[3.2.1]octane scaffold (Fig. 6a). The key aldehyde 6 was synthesized from triol 20 via DMP-mediated oxidation, enabling exploration of the tandem C5-isomerization/aldol/retro-Claisen cascade. Notably, under TBD catalysis, B-ring aromatized product 31 was obtained, presumably via a retro-Claisen/retro-Michael/aromatization cascade reaction (Fig. 6a, entry 1; see Supplementary Fig. 3d). Employing potassium tert-butoxide afforded C10-iso-3 (60% yield) and haploide C (3) (5% yield), along with the C7-C18 cyclized product 32, the structure of which was established by X-ray crystallography (Fig. 6a, entry 2). The proposed formation pathways for compounds C10-iso-3 and 32 are detailed in Supplementary Fig. 3e. Attempts to isomerize the C10 position of C10-iso-3 were unsuccessful. Systematic evaluation of base-mediated condensation protocols revealed limited efficiency, prompting us to explore an acid-catalyzed strategy. Fortunately, treatment of aldehyde 6 with the Brønsted acid (PhO)2PO2H successfully afforded haploide C (3) in 55% yield42. The observed facial selectivity at C18 could be attributed to the double hydrogen bonding coordination of the phosphinic acid catalysts (see Supplementary Fig. 3f)70. The spectroscopic data of the synthesized haploide C (3) were consistent with the literature reports38.

Fig. 6: Synthesis of haploide C (3), haplomitrenolides C/D (7/34), and haplomitrins A/C (4/36).
Fig. 6: Synthesis of haploide C (3), haplomitrenolides C/D (7/34), and haplomitrins A/C (4/36).
Full size image

a Total synthesis of haploide C (3) enabled by the tandem C5-isomerization/aldol/retro-Claisen reaction of 6. b Divergent total syntheses haplomitrenolides C/D (7/34) and haplomitrins A/C (4/36) from the common intermediate 8. DMP Dess–Martin periodinane, TBD 1,5,7-triazabicyclo[4.4.0]dec-5-ene, DMSO dimethyl sulfoxide, TBAF tetrabutylammonium fluoride, MsCl methanesulfonyl chloride, DMF N,N-dimethylformamide, Py pyridine, NBS N-bromosuccinimide, BPO dibenzoyl peroxide.

Synthesis of haplomitrins A/C (4/36)

Given the highly rigid tetracyclic framework of haplomitrins, we first prepared haplomitrenolide C (7), bearing a C1-methylene group, for [2 + 2] cycloaddition attempts37,51,71 (Fig. 6b). Selective C1 triethylsilyl ether deprotection using 1 equiv. TBAF, followed by mesylation and NaBH4 reduction to convert the hydroxyl group to a methylene group, formed 33 in 54% yield over three steps. Sequential deprotection of the C6 and C18 silyl groups using HF•Py, followed by Jones oxidation and methyl esterification, afforded haplomitrenolide C (7). To our delight, irradiation of haplomitrenolide C (7) with 365 nm light in MeCN efficiently formed haplomitrin A (4) in 85% yield (ring opening of the δ-lactone moiety occurred in ethanol). The synthesis of haplomitrin C (36) requires hydroxylation at C9. Exposure of 7 to tBuOK/P(OEt)3 under O₂ directly provided haplomitrenolide D (34) in 40% yield72. Alternatively, this enone allylic hydroxylation could be accomplished via a two-step bromination/hydrolysis sequence in 21% yield. Notably, the choice of silver salt and solvent is crucial, and the C9-brominated intermediate was efficiently converted to compound 35, constituting the haploide O (2) skeleton via carbocation-mediated 1,2-acyl migration in AgBF4/dichloromethane (see Supplementary Fig. 4a)73,74. Similarly, irradiation of haplomitrenolide D (34) with 365 nm light afforded haplomitrin C (36) in 84% yield. The NMR spectroscopic data of these synthetic natural products matched those reported in the literature38.

Discussion

In summary, we have developed a bioinspired and divergent synthesis strategy that enables total synthesis of seven biogenetically related haplomitrium diterpenoids: hapmnioide A, haploides C/O, haplomitrenolides C/D, and haplomitrins A/C. The key features of this route include a diastereoselective IMDA reaction to rapidly assemble the labdane core skeleton. 1,2-Acyl migration of the labdane core skeleton was employed to synthesize hapmnioide A and haploide O. Two consecutive stereospecific 1,2-acyl migrations have been observed. Furthermore, the tandem C5-isomerization/aldol/retro-Claisen reactions and light-initiated [2+2] cyclization of the common labdane skeleton enable the construction of haploide C and haplomitrins A/C, respectively. Further investigations of 1,2-acyl migration on other labdane-type terpenoids or analogs demonstrated the applicability of such late-stage rearrangements for the synthesis of structurally diverse small molecules. The synthetic work provides insights into in labdane-type terpenoid assembly, revealing that aldol-like reaction play a significant role in the biosynthesis of haplomitrium diterpenoids. The strategies described to prepare complex molecular architectures will be applicable to the assembly of other structurally related terpenoids.

Methods

Synthesis of hapmnioide A (1)

Under argon atmosphere, to a solution of compound 5 (7.8 mg, 20.2 μmol, 1.0 equiv.) in anhydrous THF (2.0 mL) was added KHMDS (10.1 μL, 1 M in anhydrous THF, 10.1 μmol, 0.5 equiv.) at 0 °C, followed by 5 min stirring at the same temperature. The reaction mixture was warmed to room temperature and stirred for 24 h before quenched with saturated aqueous ammonium chloride (2 mL). The aqueous layer was extracted with ethyl acetate (5 mL ×3). The combined organic layers were washed with brine (10 mL), dried over anhydrous sodium sulfate, filtered, concentrated and purified by flash column chromatography (dichloromethane/methanol = 50/1) to yield 1 (2.2 mg, 28%) and 21 & C4-iso-21 (2.1 mg, 27%) as white solids.

Synthesis of haploide C (3)

To a solution of compound 6 (6.0 mg, 16.8 μmol, 1.0 equiv.) in anhydrous toluene (1.7 mL) was added (PhO)2PO2H (5.1 mg, 20.2  μmol, 1.2 equiv.) under argon atmosphere. The reaction was stirred at 90 °C overnight. The mixture was concentrated in vacuo and purified by flash column chromatography (petroleum ether/ethyl acetate = 1/2) to yield 3 (3.3 mg, 55%) as a white solid.