Synthetic PlanningWe envisioned that macrocalyxoformin A (5) and ludongnin C (7) could be respectively accessed from macroclyxoformin B (6) via stereoselective 1,2 and 1,4-reductions of the enone moiety on the D ring (Fig. 1c). Following topological principles for retrosynthetic analysis22, we prioritize disconnection of the most centrally located B lactone ring of 6 by cleaving the ester bond and the C9–C10 bond, the latter of which was anticipated to be forged via a decarboxylative Giese reaction23,24,25,26,27,28,29 between the tertiary acid 8 and radicophile 13. The E1 tetrahydrofuran ring of 8 was assumed to be accessible from 9 via an aldol reaction with formaldehyde followed by a Suárez modified Hofmann–Löffler–Freytag reaction30. To construct the β-keto ester 9, we conceived an approach involving a reductive decarboxylative cyclization/reductive radical-polar crossover (RRPCO)31,32,33,34,35,36,37,38/C-acylation cascade, starting from readily available redox-active ester (RAE) 11. In this process, the key carbon-carbon bond formations would result from an intermolecular radical conjugate addition (C5–C6 bond) and an acylation (C10–C9’ bond) reaction. Radicophile 13 could be obtained from 14 through a palladium-catalyzed oxidative cyclization39. Compound 14 could be readily traced back to the easily prepared compound (R)-1540.Although the proposed reductive decarboxylative cyclization/RRPCO/C-acylation cascade holds conceptual efficiency, it presents at least two considerable challenges: (i) a viable acylation reagent with proper reactivity has to be selected: if the acylation reagent is not reactive enough, then the carbanion 10 might be acylated by the starting material RAE 11; conversely, if the acylation reagent is too reactive, then could be reduced prior to the reduction of 11 or the radical precursor of 1041. (ii) the desired C-acylation has to override the O-acylation42. Approaching these challenges, we thought that CO2 would be a good choice43,44,45,46; however, considering the lability of the β-keto acid product, the additional esterification step, and the potential scalability issue in the early stage of the total synthesis, we prioritize the search for an appropriate acylation reagent.Preparation of the precursors and optimization of the reductive decarboxylative cyclization/RRPCO/C-acylation cascadeThe preparation of the RAE 11 commenced with the alcohol (R)-16 (Table 1), which could be easily prepared in 3 steps at decagram scale in 87% ee and 63% overall yield using Rawal’s procedure47. Etherification of alcohol 16 with tert-butyl bromoacetate (NaOH, TBAB, quant.) provided tert-butyl ester 17. Removal of the carboxyl tert-butyl protection (TFA) provided carboxylic acid, which can be converted to an array of RAEs in good-to-excellent yields.Table 1 Preparation of the RAEs and optimization of the decarboxylative cyclization/RRPCO/C-acylationWe began the investigation of the proposed reductive decarboxylative cyclization/RRPCO/C-acylation cascade using the canonical ester 11a as starting material (83% yield from tert-butyl ester 17), NiBr2•DME as catalyst, and Zn as reductant. After extensive screening of potential acylation reagents (entries 1–6, Table 1), we were pleased to find that the desired product 9d could be produced in 27% yield when di-tert-butyl decarbonate (Boc2O) was employed, while other tested acylation reagents (12a–f) performed sluggishly. An extensive investigation of bipyridine ligands (entries 7 and 8), alternative catalysts (entries 9–11), and the reaction temperature (entries 12 and 13) were unfruitful. Eventually, we were happy to find that changing N-hydroxyphthalimide (NHPI) ester 11a to N-(acyloxy)−1,8-naphthalimide 11d increased the yield of 9d to 53% (49% isolated yield, entry 16). Additionally, it is noteworthy that this cascade reaction can be performed at a decagram scale, easily producing multigram quantities of 9d in one pot (42% isolated yield).Mechanistic studiesAlthough our proposed reductive decarboxylative cyclization/RRPCO/C-acylation cascade showed good efficiency in the synthesis of the bicyclic β-keto ester 9d, the use of Boc2O as the C-acylation reagent seems counterintuitive because it is typically used to introduce the Boc protecting group to amine functionalities, with only a few examples as a C-acylation reagent reported to date48,49,50. We, therefore, performed a series of experiments to gain further insight into the mechanism of this cascade reaction (Fig. 2).Fig. 2: Mechanistic studies and the proposed mechanism.a, b Evidences for RRPCO. c Tracing the source of the C10 ester group. d Boc2O was used in the esterification of carboxylic acid 19. e Proposed reaction mechanism. f Substrate scope of the reductive decarboxylative cyclization/RRPCO/C-acylation cascade. Abbreviations: DME, 1,2-dimethoxyethane; D2O, deuterium oxide; NMP, N-mehtyl-2-pyrrolidone; A4, 1,8-naphthalimide; Boc2O, di-tert-butyl decarbonate; Et, ethyl; Bn, benzyl; RRPCO, reductive radical-polar crossover; 13C, carbon-13; rsm, recovered starting material.The reductive radical cyclization/RRPCO sequence was inferred from our findings that (i) replacing Boc2O (6 equiv) with D2O (20 equiv) in the standard reaction conditions provided the only observed product 18 in 80% yield with >99% deuterium incorporation (Fig. 2a), indicating a sequence of decarboxylative 5-exo-trig radical cyclization and an RRPCO of C10 radical occurred; (ii) no 9d was detected when 18 was applied as the starting material instead of RAE 11d (Fig. 2b), excluding the possibility that 18 was the precursor of the acylation. Moreover, we found that Boc2O was not a good C-acylation reagent for 18, as treatment of 18 with a variety of bases (e.g. LDA, NaHMDS) followed by adding Boc2O only provided trace amounts of 9d and the C2 acylated isomer (Supplementary Fig. 6).We next focused on identifying the source of the C10 ester group of 9d (Fig. 2c) via a series of isotope labeling experiments. Initially, 13C labeled Boc2O was prepared [13CO2 (>99% 13C), tBuOK, MsCl, pyridine, Supplementary Fig. 7] and used instead of Boc2O. We obtained 9d with 69% 13C incorporation (35% yield, Fig. 2c, entry 1), which indicated that the C10 ester group was not fully derived from Boc2O. Consequently, we labeled the RAE group of 11d with 13C (see Supplementary Figs. 11 and 12 for its preparation) and subjected 13C-11d to the standard reaction conditions, we detected the product 9d with 27% 13C incorporation (47% yield, Fig. 2c, entry 2), which indicated that the C10 ester group is partially derived from the reaction of the C10 carbanion with CO2 released by the decarboxylation process. To further demonstrate this point, we performed the standard reaction under a 13CO2 (>99% 13C) atmosphere (Fig. 2c, entry 3), and a 44% 13C incorporation of product 9d was observed.On the basis of these lines of evidence, a putative mechanism for the acylation process is proposed (Fig. 2e). The carbanion 10 (was generated through the reductive decarboxylative cyclization/RRPCO process) reacts with CO2 produced by the decarboxylation, giving rise to the carboxylate 20. Esterification of 20 by Boc2O affords the desired β-keto esters 9d along with one tBuO– and two CO251. The released CO2 could also be involved in the carboxylation of carbanion 10, accounting for the high efficiency of the acylation and the relatively low 13C incorporation in the above isotope labeling experiments. Note that: (i) we detected >99% 13C incorporation of 9d when Boc2O and RAE 11d were replaced by 13C labeled Boc2O and 13C-11d simultaneously in standard reaction conditions (Fig. 2c, entry 4), (ii) Additionally, we found that treatment of the carboxylic acid 19 with our standard conditions (Fig. 2d) could afford the esterification product 9d in 37% yield. Both experiments support the plausibility of our proposed reaction mechanism.Armed with a comprehensive understanding of the mechanism and optimal conditions, we next briefly explored the generality of this cascade process (Fig. 2f). Intriguingly, displacing the methyl group on the existing C4 quaternary carbon to an ethyl group (9e), a benzyl group (9f), an allyl group (9g), or removal of the C4 methyl group (9h) did not compromise the yield. Furthermore, the substrates with substitutions at C3 and C2 were also amenable to this cascade reaction, delivering 9h–9j in synthetically useful yields. Notably, the incorporation of an aromatic ring at the α-position of the α, β-unsaturated ketones (C10) resulted in the formation of the O-acylated product 9k in high yield, rather than the expected C-acylated product, indicating a significant impact of steric hindrance on the C-acylation.Building the E1 ring and the radicophile 13: challenges in decarboxylative Giese reactionWith a reliable and scalable synthesis of 9d in hand, we set out to construct the E1 THF ring (Fig. 3a). Treatment of 9d with formalin in the presence of Yb(OTf)3 gave rise to the C10 aldol product 21 in a completely stereoselective manner52. The high stereoselectivity is plausibly attributable to the influence of the axial methyl group at C420. Subjection of 21 to the conditions reported by Suárez and co-workers (PIDA, I2, hν) forged the E1 THF ring smoothly30, producing 22 in good yield (61% from 21). Notably, a ring flip process occurred during the reaction. Reduction of the C1 ketone of 22 with a sterically hindered reducing reagent LiAl(OtBu)3H afforded the desired alcohol 23 with excellent diastereoselectivity (9:1, 80%), while other less hindered reagents (e.g., NaBH4, DIBAL-H) proved to be unviable. Having secured access to 23, we rapidly prepared two precursors (24 and 25), which can be used to construct the C9–C10 bond via the decarboxylative Giese reaction: deprotection of the tBu ester of 23 with trifluoroacetic acid (TFA), protection of the C1 alcohol and carboxylic acid with tert-butyldimethylsilyl (TBS) group and hydrolysis of TBS ester produced acid 24 (93%), which was sequentially activated with NHPI to afford the RAE 25 (66%).Fig. 3: Preparation of precursors and attempts for the decarboxylative Giese reaction.a Stereoselective construction of the E1 ring and the C1–OH. b Asymmetric synthesis of radicophile 13. c Attempts with audiophile enone 13. d Attempts with radicophile (S)−31. e Attempts at the decarboxylative intramolecular Giese reaction. Abbreviations: tBu tert-butyl, PIDA phenyliodine diacetate, hν light irradiation, DIBAL-H diisobutylaluminium hydride, TFA trifluoroacetic acid, TBSOTf tert-butyldimethylsilyl trifluoromethanesulfonate, NHPI N-hydroxyphthalimide, DIC N,N’-diisopropylcarbodiimide, DMAP 4-dimethylaminopyridine, Ac acetyl, TBDPS tert-butyldiphenylsilyl, MsCl methanesulfonyl chloride, DBU 1,8-diazabicyclo[5,4,0]undec-7-ene, TBHP tert-butyl hydroperoxide, THF tetrahydrofuran, DMSO dimethyl sulfoxide, d.r. diastereomer ratio, e.e. enantiomeric excess, [H] hydrogenation, [O] oxidation, [3,3] 3,3-sigmatropic rearrangement, ND not detected, RA radical acceptor.Preparation of radicophile 13 began with an asymmetric α-allylation of β-ketoester 2640, giving (R)-15 in 96% yield with 96% e.e. (Fig. 3b). Treatment of (R)-15 with NaBH4, followed by mesylation and elimination of the resulting mesyl ester, produced alkene 27, which underwent an allylic oxidation (CrO3, TBHP), affording enone 14 (27% over 4 steps). Exposure of 14 with tert-butyldimethylsilyl trifluoromethanesulfonate (TBSOTf) in the presence of Et3N and subjection of the resulting silyl enol ether with Pd(OAc)2 furnished radicophile 13 (66% over 2 steps)39, whose absolute configuration was confirmed by X-ray crystallography of its acid derivative 28.In the subsequent assembly process utilizing decarboxylative Giese reaction, despite our diligent efforts, employing the radicophile enone 13 in conjunction with various photoredox decarboxylation conditions using acid 24 (see Supplementary Fig. 31 for details)23,24 or different reductive decarboxylation conditions using RAE 25 (see Supplementary Fig. 32 for details)25,26,27 only resulted in direct decarboxylation rather than the desired coupling product 29 (Fig. 3c). We initially ascribed this outcome to the significant steric hindrance arising from the γ-quaternary carbon of α, β-unsaturated 13. Notably, to our knowledge, no prior reports exist on the decarboxylative Giese reaction of tertiary radicals with α, β-unsaturated acceptors containing a γ-quaternary carbon.In an effort to mitigate the steric hindrance associated with the radicophile, we chose to employ α, β-unsaturated lactone (S)-3153 (Fig. 3d, see Supplementary Fig. 30 for its enantioselective preparation), which lacks substitution on the carbon adjacent to the reacting carbon centers of the radicophile. Notably, this approach could capitalize on the intrinsic configurations of C13 in (S)-31, potentially leading to the desired stereochemistry at C9. Subsequent B lactone ring construction could employ a lactonic Ireland–Claisen rearrangement54. However, none of the decarboxylative conditions utilizing acid 24 and RAE 25 yielded the desired coupling product with (S)-31, underscoring the pivotal role of steric effects caused by the substituents at the reacting carbon centers of the α, β-unsaturated ester.We also investigated the intramolecular Giese reaction using acids 34–36 (Fig. 3e, see Supplementary Figs. 34–36 for their preparations) and the RAEs thereof 37–39 (see Supplementary Fig. 38 for their preparations); unfortunately, we could not detect the desired coupling products. We speculated that the failures were due to (i) the C1-ester groups not favoring the cis configuration needed in the reaction transition state and (ii) the relatively high barrier of rotation around the ester C–O bond55 to the desired reaction transition state. Furthermore, attempts to use the C10-acyl telluride as the tertiary radical precursor56 also proved unsuccessful in both intermolecular and intramolecular Giese reactions (see Supplementary Figs. 33 and 40 for details).Completion of the total synthesisRecognizing the pivotal role of the steric effects of the radicophile in constructing the C9–C10 bond, we chose to use the simplest acyclic unsaturated esters with no substitution on the reacting carbon centers20,57,58. As shown in Fig. 4, the protection of the C1–OH of 23 with a benzyl group (NaH, BnBr, 84%) followed by deprotection of the C10-tert-butyl ester (TFA) and activation of the resulting acid with NHPI (DIC, DMAP) provided RAE 40 in an excellent yield (99%). The benzyl-protected RAE 40 was not used in the investigate of the aforementioned intermolecular decarboxylative Giese reaction (Fig. 3), due to the presence of potential coupling products containing alkene group, which would be incompatible with the benzyl deprotection. Subjection of RAE 40 with 2,2,2-trifluoroethyl acrylate under Baran’s decarboxylative Giese reaction conditions [Ni(ClO4)2·6H2O, Zn, LiCl] successfully afforded the desired coupling product 41 (75% NMR yield)26. Subsequent removal of the benzyl group under hydrogenation conditions (H2, Pd/C), followed by a spontaneous lactonization, yielded lactone 42, whose structure was confirmed by X-ray crystallography.Fig. 4: Total syntheses of (–)-macrocalyxoformins A and B and (–)-Ludongnin C.All yields were determined by isolation. Abbreviations: BnBr benzyl bromide, TFA trifluoroacetic acid, NHPI N-hydroxyphthalimide, DIC N,N’-diisopropylcarbodiimide, DMS dimethyl sulfide, LDA lithium diisopropylamide, HMPA hexamethylphosphoramide, LiHMDS lithium bis(trimethylsilyl)amide, nBu n-butyl, TBHP tert-butyl hydroperoxide, DMP Dess–Martin periodinane, brsm based on recovered starting material, d.r. diastereomer ratio.Oxidative dehydrogenation of lactone 42 [(PhSeO)2O, 49% from 40]59, followed by a conjugate addition of allyl cuprate species (allylMgBr, CuBr·DMS, LiBr) to the resulting α, β-unsaturated lactone afforded 43 (96%) with complete diastereoselectivity. Note that the conformation of the lactone ring was converted from a twist-boat to a half-chair conformation in this process. Deprotonation of lactone 43 (LDA, HMPA), followed by allylation with 2,3-dibromopropene produced 44 (43%, 57% brsm, d.r. = 10:1). Interestingly, the conformation of the lactone ring was transformed back to a twist-boat, and the newly installed allyl group occupied an equatorial position. The stereoselectivity of this allylation and ring flip process is presumably caused by the steric repulsion of the vicinal C9–allyl group and the 1,3-diaxial effect of the C10–C20 bond.Allylation of the lactone ring of 44 (LiHMDS, allyl iodide, 88%) followed by ring-closing metathesis (RCM) reaction of the resulting triene using Grubbs 2nd catalyst proceeded with excellent stereo- and chemoselectivity, giving rise to 45 in an excellent yield (89%). It is noteworthy that (i) the 1,3-diaxial effect caused by the C10–C20 bond probably overrode the steric influence exerted by the vicinal C9–allyl group, resulting in the superb stereoselectivity of this allylation; (ii) the vinyl bromide moiety was not disturbed during the RCM process60. Finally, the D ring of macroclyxoformin B (6) was smoothly constructed from 45 via a classic sequence involving a 5-exo radical annulation (Et3B, nBu3SnH, quant.)17,61, an allylic oxidation (SeO2, TBHP), and a Dess-Martin oxidation (63% over 2 steps). The structure of macroclyxoformin B (6) was confirmed by X-ray crystallography. Consequentially, respective Luche reduction (NaBH4, CeCl3·7H2O) and hydrogenation (H2, Pd/C) of macrocalyxoformin B (6) produced macrocalyxoformin A (5, 89%) and ludongnin C (7, 92%), both as single diastereomers.Anticancer activity evaluation of the synthetic natural productsThe α-methylenecyclopentanone system (D ring) in ent-kauranoids is recognized as a crucial pharmacophore for their antitumour activity1. Our synthesized natural products 5–7 exemplify this role effectively. We evaluated their impact on cell viability across nine cancer cell lines from five human tissues. Macrocalyxformin B (6) displayed significant broad-spectrum anticancer activity at the micromolar level, as shown in Table 2. Conversely, macrocalyxformin A (5) and ludongnin C (7) exhibited negligible and weak activity, respectively, across the tested cancer cell lines, reaffirming the essential role of the α-methylenecyclopentanone moiety in anti-tumor activity. Notably, the comparatively stronger anti-tumor activity observed in compound 7, in contrast to compound 5, implies a potential non-covalent interaction facilitated by the ketone group, thus highlighting a promising direction for future exploration.Table 2 IC50 values (μM) of synthetic natural products against 9 different cancer cell lines
