If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
School of Pharmaceutical Sciences, MOE Key Laboratory of Bioorganic Phosphorus Chemistry & Chemical Biology, Tsinghua University, Beijing, 100084, China
School of Pharmaceutical Sciences, MOE Key Laboratory of Bioorganic Phosphorus Chemistry & Chemical Biology, Tsinghua University, Beijing, 100084, China
School of Pharmaceutical Sciences, MOE Key Laboratory of Bioorganic Phosphorus Chemistry & Chemical Biology, Tsinghua University, Beijing, 100084, China
School of Pharmaceutical Sciences, MOE Key Laboratory of Bioorganic Phosphorus Chemistry & Chemical Biology, Tsinghua University, Beijing, 100084, China
School of Pharmaceutical Sciences, MOE Key Laboratory of Bioorganic Phosphorus Chemistry & Chemical Biology, Tsinghua University, Beijing, 100084, China
Cytochalasans are an important class of fungal metabolites displaying remarkable structural diversity and significant biological activity. Historically, considerable effort has been devoted toward the synthesis of cytochalasans. Comparably, merocytochalasans, an array of more complicated molecules biosynthetically derived from cytochalasans and epicoccine through hetero-dimerization, -trimerization or -tetramerization, have been less explored. In recent years, our group has shown keen interest on cytochalasans and merocytochalasans, which has culminated in the collective syntheses of a variety of representative molecules of this family. To this end, we first developed a highly concise, efficient and modular approach to access 11/6/5 tricyclic cytochalasans, featuring an intermolecular Diels-Alder reaction and a ring-closing metathesis reaction as key steps. Based on this synthetic route, several cytochalasan monomers, including aspochalasin B, aspochalasin P, aspochalasin D and aspergillin PZ, have been synthesized within 11–14 longest linear steps. Subsequently, a couple of hetero-dimeric and -trimeric merocytochalasans, including asperchalasines A-E and asperflavipine B, have been accessed from aspochalasin B and the suitable epicoccine precursors through a series of intriguing bio-inspired transformations such as Diels-Alder reaction and formal [5 + 2] cycloaddition. Moreover, the full profile of the regio- and endo/exo selectivity of the Diels-Alder heterodimerization has been explored through computational study. The present study not only notably enriches the synthetic chemistry of cytochalasans and merocytochalasans, but also affords venerable clue to decipher the biosynthetic origins of the newly identified merocytochalasans.
Cytochalasans represent a large family of polyketide–amino acid hybrid metabolites that have attracted broad interest from the chemical and biological communities [
]. As their name implies, this class of natural products has long been recognized for their capability of targeting the actin cytoskeleton, which, in turn, could interfere with various cellular processes such as cell adhesion, motility, signaling and cytokinesis [
]. Indeed, it has been proven that cytochalasans possess a wide range of biological activities such as cytotoxic, antimicrobial, antiparasitic, and antiviral effects. In addition, some cytochalasans could also inhibit cholesterol synthesis or interfere with glucose transport and hormone release. Owing to their appealing biological profiles, cytochalasans have been widely used as drug leads or chemical tools in biomedical research.
Structurally, most known cytochalasans are characterized by an n/6/5 tricyclic skeleton that consists of a multi-substituted perhydro-isoindolone moiety and a macrocyclic ring. The structural diversity of cytochalasans is mainly manifested by the versatile amino acid residues (e.g. phenylalanine, tyrosine, tryptophan, leucine, or alanine) incorporated into the perhydro-isoindolone moiety and the varied ring sizes (n = 7–16) and forms (e.g. carbocycle, lactone or cyclic carbonate) of the macrocycles [
]. Taking aspochalasins as examples, they generally share a common tricyclic skeleton with a leucine-derived perhydro-isoindolone moiety fused to an 11-membered carbocycle, as represented by the structures of 1–3 (Fig. 1A) [
Recently, the chemistry and biology of cytochalasans have been greatly enriched by the continuous discovery of a series of unprecedented merocytochalasans that display amazing structural complexity and novelty (Fig. 1B) [
]. For example, in 2015 Zhang and co-workers identified five cytochalasan-containing hetero-dimers or -trimer from the culture broth of Aspergillus flavipes [
]. Among them, asperchalasines B–E (5–8) feature a T-shaped hexacyclic framework consisting of one cytochalasan (red) and one epicoccine (blue) unit, and asperchalasine A (9) bears a more complicated, sandwich-shaped decacyclic framework containing one epicoccine and two cytochalasan units [
] all bear heterotrimeric skeletons that are composed of two epicoccine units and one cytochalasan unit (either 1 or 2); asperflavipine A (14), the first naturally occurring cytochalasan-epicoccine heterotetramer, exhibits an incredibly complicated, dumbbell-shaped tetradecacyclic ring system [
]. Besides their dazzling chemical structures, these newly identified merocytochalasans also display versatile biological profiles. For examples, asperchalasine A (9) could induce the arrest of G1-phase cell cycle by selectively inhibiting cyclin A, CDK2 and CDK6 in cancer cells [
]; both epicochalasines A (11) and B (12) could induce the arrest of G2/M-phase cell cycle and apoptosis in leukemia cells through the caspase-3 activation and PARP degradation [
From a synthetic point of view, the newly identified merocytochalasans represent more challenging targets than tricyclic cytochalasans. Naturally, they have attracted great attention from the synthetic community since their disclosure. Our group has shown keen interest on the chemical syntheses of cytochalasans in recent years. In 2016, we reported the collective syntheses of several tricyclic and polycyclic cytochalasans, namely periconiasins A–E [
]. Encouraged by this experience, we embarked on a program in 2016, with the ambition to address the unmet challenge associated with the newly discovered merocytochalasans. As part of this program, we successfully achieved the first total syntheses of asperchalasines A-E (5–9) in 2018 [
]. More recently, a number of other merocytochalasans and polycyclic cytochalasans have also been accomplished by the Deng and Trauner groups consecutively [
]. Promoted by these seminal works, herein we provide a comprehensive summary of our own progress on this long-term project, which has culminated in the collective syntheses of an array of tricyclic, polycyclic, hetero-dimeric and -trimeric cytochalasans and merocytochalasans.
2. Results and discussion
Proposed biosynthetic origin of representative merocytochalasans. Biosynthetically, all of the newly identified merocytochalasans could be traced back to two fundamental units, the tricyclic aspochalasin B (5) or D (7) [
], through diverse hetero-dimerizations, -trimerizations or -tetramerization. Taking asperchalasines A-E (5–9) and asperflavipine B (10) as examples, they should be derived from aspochalasin B (1) and epicoccine (15) through a series of intriguing transformations (Scheme 1). Specifically, epcoccine (15) readily undergoes oxidation to yield the o-quinone species 15b and 15c. Theoretically, 15b/c could further isomerize into the isobenzofuran species 15a that can react with aspochalasin B (1) through intermolecular Diels-Alder reaction. Due to the underlying endo/exo selectivity and regioselectivity issues, four Diels-Alder adducts 16–19 could generate in the reaction, among which three have been discovered in nature, namely asperchalasines F–H (16–18) [
]. Starting from 16–19, asperchalasines B-E (5–8) could be obtained through regioselective methylation. Furthermore, the heterodimer 19 could also undergo oxidation to give o-quinone 20 that then reacted with the second unit of 1 through an intermolecular [5 + 2] cycloaddition, leading to trimeric asperchalasine A (9). In parallel, o-quinone 20 may exist in an equilibrium with its regioisomer 21, and the latter could further engage in an intermolecular [5 + 2] cycloaddition with another unit of epicoccine (15), giving rise to asperflavipine B (10). Apparently, bearing three different forms, the oxidized epicoccine unit could serve as either a [4C]-synthon (15a) in the heterodimerization with cytochalasans or a [5C]- or [2C]-synthon (15b or 15c) in the homodimerizations with itself. Apparently, nature adopts a wise way to produce skeletally diverse merocytochalasans from two fundamental monomers through diverted heterodimerization and -trimerization, presumably with the help of enzymes. While such bio-inspiration provides a valuable guidance to access related natural products, how to control the reactivity and selectivity of the above transformations might be rather challenging in practice.
Scheme 1The plausible biosynthetic origin of some representative asperchalasines.
Evolution of the synthetic strategy towards cytochalasans. According to our biosynthetic hypothesis, we need to obtain the requisite monomers, aspochalasin B (1), aspochalasin D (2) and epicoccine (15), in order to access those merocytochalasans. While epicoccine (15) could be readily prepared through the known protocol developed by Trauner and co-workers [
], our attention was mainly focused on the synthesis of 1 and 2. Historically, the considerable endeavor has been devoted toward the syntheses of tricyclic cytochalasans, including an elegant total synthesis of aspochalasin B (1) reported by the Trost group [
]. However, we noticed that these syntheses usually took long linear steps and resulted in low overall yields. Moreover, most of the known synthetic routes were only applied for specific targets. In this context, we decided to develop a more concise and flexible synthetic strategy to access 11/6/5 tricyclic cytochalasans. From a strategic point of view, the most straightforward approach to accessing the 11/6/5 tricyclic core of aspochalasins should be built on a biomimetic intramolecular Diels-Alder reaction. Indeed, this strategy has been frequently employed in the syntheses of various tricyclic cytochalasans [
]. However, most of these works suffer from a common issue, that is, the intramolecular Diels-Alder reactions only afford moderate yields (<50%). This could be attributed to two inherent challenges associated with the reaction: 1) 3-acylpyrrol-2(5H)-one moiety, the key structural element presented in the precursors of Diels-Alder reactions, is notorious for its fragile nature, thus serious decomposition of substrate will take place unavoidably in the reactions [
]; 2) in general, a strained macrocycle will generate in the intramolecular Diels-Alder reaction, which makes it unfavorable compared to other conventional Diels-Alder reactions. To address these problems, a complementary strategy hinging on intermolecular Diels-Alder reaction has also been applied to the syntheses of tricyclic cytochalasans [
]. In this scenario, the intermolecular Diels-Alder reactions generally give improved yields. However, as compensation, they mostly suffer from a moderate endo/exo selectivity. Besides, the regioselectivity might also be problematic in some cases.
The first-generation approach to aspochalasins. Based on the literature's works [
] on the syntheses of tricyclic cytochalasans, we decided to attempt the intramolecular Diels-Alder reaction-based strategy at the initial of our study. As depicted in Scheme 2, we envisioned that both aspochalasins B (1) and D (2) could be traced back to the tricyclic intermediate 22a/b through late-stage derivatization. According to our synthetic blueprint, the C17 = C18 double bond, either in a trans- or a cis configuration, could serve as an anchor to introduce different functionalities as needed. For the tricyclic compound 22a/b, they could be accessed from the linear precursor 23a/b through intramolecular Diels-Alder reaction. In turn, 23a/b should be readily obtained from 24a/b through dehydrogenation. Further disconnection of 24a/b led to lactam 25 [
] and ester 26a/b as the precursors, which could be assembled together through Claisen condensation. Finally, conjugated triene 26a/b could be derived from the readily accessible fragments 27 and 28a/b through Julia olefination [
] was treated with vinyl magnesium bromide to give allyl alcohol 30, which then underwent Johnson-Claisen rearrangement to yield 31. Removal of the acetyl group of 31 followed by oxidation provided aldehyde 28a smoothly. With 28a in hand, we next attempted to construct the triene moiety through Julia olefination. Based on the work reported by Thomas and co-workers [
], we first employed benzothiazol-2-yl sulfone 27 as the reaction pair. To our disappointment, the reaction resulted in a low yield (20%, E/Z = 10:1) under the reported conditions (LiHMDS, −78 °C to rt) [
]. To improve the reaction, we evaluated several key parameters of the reaction including Julia reagents, bases, and reaction temperatures (Table S1, Supporting Information). It was found that when the reaction was kept at −78 °C all along, an improved yield (ca. 50%) could be obtained, albeit accompanying a decreased E/Z selectivity (3:1). We also tried 1-phenyl-1H-tetrazol-5-yl sulfone 33 [
] in the reaction, which gave an acceptable yield (55%) but poor E/Z selectivity (1.2:1). Besides, the replacement of LiHMDS with KHMDS or NaHMDS did not affect the yield, but largely inverted the E/Z selectivity, with the Z isomer generated as the major component. Since the one-pot modified Julia olefination [
] failed to give a satisfactory result, we turned to utilize the classical Julia-Lythgoe olefination to prepare 26a in a stepwise manner, involving a sequence of the nucleophilic addition of phenyl sulfone 34 [
]. In this way, 26a was obtained in an acceptable overall yield (60%) and high E/Z selectivity (>20:1). Of note, featuring a conjugated triene moiety, compound 26a was not very stable and readily advanced into some unidentified products upon long-term storage.
Scheme 3The first-generation approach to the tricyclic core of aspochalasins.
With 26a in hand, we proceeded to install the 2-pyrrolidinone moiety. To this end, 26a was first hydrolyzed with NaOH, and the resulting carboxylic acid was converted to the corresponding carbonyl imidazole derivative that could further undergo Claisen condensation with lactam 25 to afford 24a as a mixture of unseparated diastereoisomers. Subsequently, phenylselenation of 24a followed by oxidative elimination furnished 3-pyrrolin-2-one derivative 23a. As we mentioned above, 23a was notoriously unstable and had to be proceeded immediately in the subsequent intramolecular Diels–Alder reaction without purification or long-term storage [
]. However, when we conducted the reaction under the thermal conditions (toluene, 90 °C, 3 h), only a poor yield of the product 22a (<10%) was obtained. The low yield of the reaction is mainly attributed to the serous decompose of the precursor 23a under the employed condition. We assumed that in the current scenario, a strained 11-membered macrocyclic ring bearing two trans double bonds would generate along with the 6-membered ring, which makes the Diels-Alder reaction rather challenging. Keeping this concern in mind, we attempted to improve the Diels-Alder reaction by adopting some modified precursors like 23b and 36. To our disappointment, they also failed to give notably improved results in practice.
Taking the above results into consideration, we thought that the intramolecular Diels-Alder reaction-based approaches seemed to be less optimistic for further exploration. Besides the low efficiency of the Diels-Alder reaction, the instability of some triene-containing intermediates along the synthetic route (e.g. 26a, 24a, and 23a) also compromises its applicability. Keeping this in mind, we turned to explore the intermolecular Diels-Alder reaction-based approach in the following study.
The second-generation approach to aspochalasins. The second-generation strategy for aspochalasins was depicted in Scheme 4. Simply, we envisioned that the tricyclic core of aspochalasins (22a/b) could be accessed from the precursor 37 through ring-closing metathesis (RCM) reaction. Based on the seminal works reported by the Stork's and Vedejs' groups [
], the perhydro-isoindolone derivative 37 could be assembled from two fragments 38 and 39 through a regio- and stereoselective intermolecular Diels-Alder reaction. Finally, both 38 and 39 could be accessed from some simple building blocks through the similar protocols adopted in the first generation of synthesis.
Scheme 4The second-generation strategy towards aspochalasins B and D.
] first underwent acylation with 4-pentenoyl chloride (41) to give β-keto lactam 40 which, after sequential phenylselenation and oxidative elimination, could further convert into the 3-pyrrolin-2-one derivative 38. Not surprisingly, compound 38 was unstable and had to be used directly in the next step. In parallel, the other fragment 39 was easily accessed from α, β-unsaturated aldehyde 42 [
Having fragments 38 and 39 in hand, we moved to explore the key Diels-Alder reaction (Table 1). Initially, we attempted the reaction under the thermal conditions (neat, 80 °C) (entry 1). To our delight, the reaction worked as expected, affording a mixture of the endo (37) and exo (45) adducts in a 60% combined yield. Although the reaction showed a moderate endo/exo selectivity (2/1), its regioselectivity appears to be very high, which is in agreement with the result reported by Stork and co-workers [
]. The moderate yield was mainly attributed to the decomposition of dienophile 38. To improve the efficiency and endo/exo selectivity of the Diels-Alder reaction, we sought to attempt the reaction with Lewis acid as a promoter. It should be noted that although Diels-Alder reactions have been frequently employed in the syntheses of cytochalasans, they are mostly conducted under thermal conditions [
]. This is likely because the 3-pyrrolin-2-one moiety is prone to tautomerize into pyrrole ring in the presence of Lewis acid, and thus some undesired reaction pathways may occur.
Table 1Optimization of the intermolecular Diels-Alder reactiona.
Table 1Optimization of the intermolecular Diels-Alder reactiona.
a) General reaction condition: 44 (1.0 mmol), 39 (0.8 mmol), Lewis acid (0.15 mmol) in solvent (5.0 mL). b) Neat reaction. c) For those reactions with Et2AlCl, ZnBr2 and ZnCl2, CH2Cl2 was used as solvent; for those ones with Cu(OTf)2 and (CuOTf)2•PhMe, the mixed solvents (CH2Cl2: PhMe = 2:1) were used. d) The yield was calculated based on compound 39. e) The ratio of 37 and 45 was calculated on the basis of the 1H NMR spectra of the Diels-Alder products purified by column chromatography.
This assumption was partially supported by some results observed in our study. For example, when Et2AlCl was used in the reaction, the desired product could not be isolated. Instead, dienophile 38 decomposed completely (entry 2). Further screening showed that ZnBr2, a milder Lewis acid than Et2AlCl, was amenable to the reaction by affording a comparable yield and endo/exo selectivity to those of the thermal reaction (entries 3–5). Furthermore, decreasing the reaction temperature from 0 °C to −50 °C could notably improve the endo/exo selectivity, with little effect on the reaction yield. Encouraged by this observation, we further evaluated some other Lewis acids including ZnCl2 (entry 6), Cu(OTf)2 (entries 7–9) and (CuOTf)2·PhMe (entries 10 and 11), among which (CuOTf)2·PhMe [
] turned out to be the optimal choice by affording the best result (63% yield, endo/exo >20:1) (entry 10). To get deep insight into the unique reactivity of CuOTf)2·PhMe, we conducted a computational study. It was revealed that a tetracoordinated Cu species was involved in the optimized endo transition state TS-37, with the non-reacting C C bond of dienophile coordinated to the metal center. That means the unique coordination preference of Cu(I) species may contribute to the extra endo/exo selectivity [
]. The above result deserves emphasis, since it represents the first successful application of Lewis acid-promoted Diels-Alder reaction to forge the perhydro-isoindolone moiety of cytochalasans. Besides the notably improved endo/exo selectivity, equal equivalent diene and dienophile components were used in the present reaction. This is in sharp contrast with the thermal reactions that generally entail the usage of excess amounts of one of the reactant pairs to secure a satisfactory result.
With the 5/6 bicyclic ring system secured, we moved to construct the 11-membered macrocyclic ring through RCM reaction (Scheme 6). To our delight, upon the treatment of 37 with the Grubb's II catalyst, the expected reaction works smoothly, giving rise to tricyclic compound 22a in an excellent yield [
]. Interestingly, the RCM reaction displays remarkable E/Z selectivity, only resulting in (E)-isomer 22a. Our calculation study showed that E-isomer 22a was thermodynamically more stable than (Z)-isomer (ΔG = 2.2 kcal/mol), which may account for the observed experimental result. Notably, based on some inspiring works, we also attempted to invert the Z/E selectivity of the reaction by using the (Z)-selective Grubbs' catalysts 46 [
With compound 22a secured in a highly concise and efficient manner (4 longest linear steps from 25, 25% overall yield), the stage was set to access tricyclic aspochalasins through late-stage derivatization. At first, we focused our attention on aspochalasins B (1) and D (2), the requisite precursors for the syntheses of merocytochalasans (Scheme 7). To this end, we need to introduce the oxygenated functionalities onto the C17 = C18 double bond, which seems to be challenging since there exist three alkene units in the substrate 22a. After several tries, we found that OsO4-mediated dihydroxylation could serve this goal by delivering diol 47 in a moderate yield (45%) [
], although a substantial amount of product with both C17 = C18 and C6=C7 double bonds functionalized (structure not shown) was also detected in the reaction. We assumed that the higher olefin strain, as well as less steric hindrance associated with C17 = C18 double bond, renders it more reactive towards the dihydroxylation reaction. Subsequently, the N-benzoyl group was removed, giving rise to trans-diol 48. As we expected, 48 could further convert into aspochalasin P (3) through the chemoselective oxidation of the C-18 hydroxyl. However, no promising results were obtained after extensive tries. In most cases, the oxidative cleavage of vicinal diol occurred. Thus, we had to achieve this goal in a stepwise manner. As shown, selective protection of the C-17 hydroxyl was first achieved with BzCl/Et3N at −20 °C. The excellent chemoselectivity observed in this reaction should be attributed to the trivial difference between the steric hindrance associated with C-17 and C-18 hydroxyl groups. Besides, judicious manipulation of reaction conditions was also crucial for securing the high selectivity. Indeed, a substantial amount of double-protected product (structure not shown) was detected under the harsher conditions (5 equiv. of BzCl, 0 °C). Next, the remaining C-18 hydroxyl was readily oxidized to the corresponding ketone through Dess-Martin oxidation, which then underwent deprotection to give aspochalasin P (3). Finally, the C19 = C20 double bond was introduced through phenylselenylation followed by oxidative elimination, which afforded aspochalasin B (1) in 40% yield over two steps. Thus, after an exhaustive exploration of the potential approaches, we finally achieved the total synthesis of aspochalasin B (1) in 11 longest linear steps from the known compound 25.
Furthermore, we also completed aspochalasin D (2) and aspergillin PZ (4) from the common intermediate 47 (Scheme 8). As shown, selective protection of the C-17 hydroxyl, followed by oxidation of the remaining C-18 hydroxyl, enabled the access of ketone 51 in an excellent yield over 2 steps. Next, we attempted to introduce the C-18 hydroxyl through diastereoselective reduction. We expected that the steric effect of OBz group on the C-17 position will direct the reductant approaching from the other side. To test this idea, we examined a variety of reductants in the reactions. However, most of them gave the undesired diastereoisomer 50 as the sole or major product. Interestingly, we found that removal of the benzoyl group followed by reduction of the ketone with NaBH4 led to the desired diastereoisomer 53 as a single product. With the stereochemistry of C-18 stereogenic center secured, the next task was to introduce the C19 = C20 double bond. To this end, the vicinal diol of 53 was first protected in a form of isopropylidene acetal, and the resulting product underwent phenylselenation to give compound 54. Subsequently, removing the benzoyl and isopropylidene groups proceeded smoothly under the conventional conditions, furnishing 56 in a yield of 81% over two steps. At last, aspochalasin D (2) was accessed from 56 through oxidative elimination. Pleasingly, we found that, upon the treatment with 2% HCl solution, aspochalasin D (2) readily underwent a tandem transannular ene cyclization/etherification reaction to give the pentacyclic (+)-aspergillin PZ (4) in an excellent yield.
Scheme 8Total syntheses of aspochalasin D and aspergillin PZ.
Biomimetic syntheses of heterodimeric merocytochalasans. With aspochalasins B (1) and D (2) in hand, we moved to explore the biomimetic syntheses of the newly identified merocytochalasans. Initially, our attention was focused on asperchalasines A–E (5–9), the earliest discovered merocytochalasans in nature [
]. Based on our hypothesis, these targets could be accessed from aspochalasin B (1) and epicoccine (15) through either heterodimerization or -trimerization (Scheme 1). To test this hypothesis, we treated epicoccine (15) with K3Fe(CN)6 in the presence of aspochalasin B (1) [
], assuming that the in situ-generated isobenzofuran species 15a could be trapped by aspochalasin B (1) through Diels–Alder reaction. Unexpectedly, the desired reaction did not work. Instead, the homodimerization of epicoccine took place under the oxidative conditions, which was consistent with the result reported by Trauner and co-workers.22
Since the original design turned out to be problematic, we decided to adopt an alternative way to realize the heterodimerization under non-oxidative conditions (Scheme 9) [
]. As an inspiring case, Szmuszkovicz and co-workers have reported that isobenzofuran species could be obtained from 1-ethoxy-1,3-dihydroisobenzofuran through an acid-promoted 1,4-elimination reaction [
]. Moreover, the in situ-generated isobenzeoene species could be trapped by suitable dienophiles through Diels-Alder reactions. Based on this seminal work, we proposed that the desired heterodimerization could be achieved from the precursor 57 (Scheme S1, Supporting Information) through an acid-promoted 1,4-elimination reaction followed by the Diels-Alder reaction with aspochalasin B (1). Gratifyingly, this idea did work in practice. As shown, when we treated compounds 57 and 1 in toluene in the presence of CSA at 60 °C, some heterodimeric compounds could be detected in the reaction mixtures. However, we quickly found that these products were rather sensitive to air, which rendered them not amenable for conventional work-up operation. To address this issue, we directly submitted the reaction mixtures to acetylation, which led to two fully acetylated products 59a (exo) and 59b (endo) in a ratio of 5:1. Notably, 59b has ever been reported as an artificial natural product, namely spicarin B, by Zhu and co-workers [
Encouraged by this promising result, we moved to synthesize the heterodimers asperchalasines B-E (5–8) with the mono-methylated hemiacetal 60 (Scheme S2, Supporting Information) as the precursor. To our delight, 60 turned out to be a superior substrate for the reaction, with all of the four plausible Diels-Alder cycloadducts (5:8:6:7 = 10:2:1:1) obtained in an excellent combined yield (80%) (Scheme 10). This outcome appeared to be encouraging, since it implied that the introduction of a methyl group on epicoccine unit could notably increase the efficiency of the Diels-Alder reaction, presumably for two reasons. First, the in situ-generated isobenzeoene species in this case cannot advance to the corresponding ortho-quinone species 15b and 15c, and thus the homodimerization of epicoccine unit was completely inhibited. Second, the resulting Diels-Alder cycloadducts in the current scenario are more stable towards air oxidation, thus resulting in an improved isolated yield.
With asperchalasines B-E (5–8) obtained, we turned our attention to the heterotrimer asperchalasine A (9). We noticed that the heterodimerizations of 1 with 57 and 60 led to the corresponding exo cycloadducts (e.g. 58a and 5) as major isomers; comparably, the proposed biosynthetic precursor of asperchalasine A (9) displays an endo configuration. In this context, we had to find a suitable method to invert the endo/exo selectivity of the heterodimerization. Serendipitously, we found that when compound 61 (Scheme S2, Supporting Information), a synthetic intermediate enroute to 60, was used as the epicoccine precursor, a mixture of cycloadducts could be detected, which, after debenzylation, gave asperchalasines E (8) and D (7) in a ratio of 3:2. The above results indicated that the protecting group of epicoccine unit exerted a profound influence on the selectivity of the Diels-Alder reaction. To get a deep insight into the observed selectivity, a computational chemistry study was conducted using Gaussian 16 (C01) program [
]. To begin with, a systematic conformational search was conducted to get the most stable conformation of aspochalasin B (1) in toluene, as shown in Fig. 2. FMO analysis was performed based on the optimized structures of aspochalasin B (1) as well as 60a and 61a. Further orbital composition analysis was done by SCPA method [
] on the FMOs (LUMO of aspochalasin B (1) and HOMO of 60a and 61a, since the distribution of FMOs will exert a profound influence on the energy of transition states (TSs) and the regioselectivity of reaction. For aspochalasin B (1), canonical LUMO has more distribution on C19 than on C20 (19.1% vs. 14.4%). For 60a, canonical HOMO has more distribution on C8’ than on C1’ (20.7% vs. 15.7%). However, it is interesting to find that when C3’-OH and C5'-OH were protected by the Bn group, the difference of canonical HOMO distribution on C8’ and C1’ is greatly reduced. The distribution of HOMO on these two carbon atoms tends to be average (18.7% vs. 17.5%). This result is consistent with experimental results. When aspochalasin B (1) reacts with 60a, the two major products (5 and 8) exhibit the same regioselectivity, with C19–C8'/C20–C1’ being connected preferentially. Comparably, for the reaction between aspochalasin B (1) and 61a, both two types of regioisomers (C19–C8'/C20–C1’vs C19–C1’ and C20–C8’) generate in similar proportions.
Fig. 2Rationalization of the regioselectivity of Diels-Alder reactions through calculated FMOs.
Besides regioselectivity, the endo/exo selectivity of the above reactions also deserves further analysis. For the reactions with partially protected epicoccine precursor 60, exo-adduct 5 was obtained as the major component; in sharp contrast, only endo-adducts 62 and 63 were observed when the fully protected precursor 61 was employed. In general, it is expected that endo TSs will be more favorable than exo TSs due to the favorable second-order orbital interaction. So, it seemed to be a surprise that exo-adduct 5 was obtained predominantly in the Diels-Alder reaction of 1 and 60. To explain, all four transition states of the reactions were optimized at the level of IEFPCM(Toluene)-B3LYP-D3(BJ)/def-TZVP//6-31G × level (Fig. 3) [
]. The calculated Gibbs energy barriers well match the experimental outcomes. For 61, the two endo transition states TS61-endo-1 (△G‡Relative = 0.0 kcal/mol) and TS61-endo-2 (△G‡Relative = 2.6 kcal/mol) have a lower barrier than the two exo transition states TS61-exo-1 (△G‡Relative = 3.9 kcal/mol) and TS61-exo-2 (△G‡Relative = 4.5 kcal/mol). However, for the case of 60, TS60-exo-1 becomes the most stable transition state (△G‡Relative = 0.0 kcal/mol), followed by TS60-endo-1 (△G‡Relative = 3.0 kcal/mol) and TS60-endo-2 (△G‡Relative = 3.9 kcal/mol). TS60-exo-2 is the most unfavorable one (△G‡Relative = 6.0 kcal/mol). The elusive low barrier for TS60-exo-1 could be attributed to the strong hydrogen-bond interaction between the free phenolic hydroxyl at C-3′ position of the epicoccine unit and the C-1 carbonyl of aspochalasin B. The distance between the corresponding H and O is only 1.695 Å. RDG analysis [
] shows that there is a deep blue region between OH and carbonyl, suggesting a very strong hydrogen bond. The favorable hydrogen-bonding effect greatly stabilizes TS60-exo-1, thus delivering 5 as the major component. The above results indicated that the existing forms of epicoccine precursors played a crucial role in controlling the endo/exo-selectivity of Diels–Alder reactions. Taken together, the computational study not only helps to rationalize the observed selectivity of Diels-Alder reactions, but also provides deep insight into the biosynthetic origins of naturally occurring dimeric merocytochalasans.
Fig. 3Rationalization of the endo/exo selectivity of Diels-Alder reactions with 60 and 61.
Biomimetic syntheses of heterotrimeric merocytochalasans. The above results paved the way to access other more challenging congeners within this family of natural products such as asperchalasine A (9) and asperflavipine B (10). As a proof-of-concept case, we first focused our attention on asperchalasine A (9). To this end, we choose the fully protected hemiacetal 66 (Scheme S1, Supporting Information) as the precursor of epicoccine. As expected, only two endo-type products were detected, which turned out to be 67 and 68 (80%, 67:68 = 3:2) (Scheme 11). With the pivotal heterodimer 67 secured, we proceeded to try the subsequent heterotrimerization reaction. Thus, debenzylation of 67 (Raney Ni, H2, EtOH) [
Selective hydrogenolysis of the benzyl protecting group for hydroxy function with raney-nickel in the presence of the mpm (4-methoxybenzyl) and dmpm (3,4-dimethoxybenzyl) protecting groups.
] afforded the intermediate 19, which without further purification, was directly submitted to the proposed biomimetic oxidative [5 + 2] cycloaddition [
] with aspochalasin B (1). After screening a variety of oxidants (e.g. DDQ, o-chloranil, K3Fe(CN)6, Ag2O), we finally found that the desired transformation worked smoothly by directly mixing the intermediate 19 with aspochalasin B (1) in the presence of air, which delivered asperchalasine A (9) in 57% yield over two steps. In this way, the asymmetric total synthesis of asperchalasine A (9), one of the representative A/B/A-type heterotrimers within the newly identified merocytochalasans, was achieved from the commercially available compound 25 in a linear sequence of 13 steps.
Encouraged by the above work, we moved to explore the biomimetic synthesis of another heterotrimer, namely asperflavipine B (10). Different from asperchalasine A (9), asperflavipine B (10) bear an A/B/B-type heterotrimeric skeleton which, as we assumed, should be generated from the heterodimer 19 and another unit of epicoccine through an oxidative [5 + 2] cycloaddition followed by further dehydrogenative aromatization (Scheme 1). To test this hypothesis, the Diels-Alder heterodimerization was first conducted with aspochalasin B (1) and the fully protected hemiacetal 69 as the reactant pairs. Of note, we intentionally introduced a MOM protecting group on the C-5′ phenol of epicoccine unit, with the assumption of that such maneuver could help to circumvent some underlying side reactions associated with the proposed biomimetic [5 + 2]-cycloaddition leading to asperflavipine B (10). Not surprisingly, the heterodimerization worked smoothly under the standard conditions, leading to two endo-adducts 70 and 71 in an excellent combined yield of 83% (Scheme 12). Subsequently, debenzylation of 71 with Raney Ni/H2 afforded 72 which, without purification, was directly exposed to the oxidative conditions [K3Fe(CN)3, NaHCO3] in the presence of 15. To our delight, the oxidative [5 + 2] cycloaddition did work, leading to a heterotrimer in 54% isolated yield. However, we found that this [5 + 2]-adduct was reluctant to undergo the proposed dehydrogenative aromatization under various conditions [
Synthesis of 3,4-disubstituted 2,5-dihydrofurans starting from the Baylis-Hillman adducts via consecutive radical cyclization, halolactonization, and decarboxylation strategy.
], which prompted us to reconsider the experimental outcomes. After a careful analysis of the spectroscopic data of the [5 + 2]-adduct, we came to realize that it should be assigned as 74, a regioisomer of the expected product 73. Mechanistically, although the epicoccine unit, upon treatment with oxidant, may engage in the [5 + 2]-cycloaddition in two different forms as represented by 15b and 15c, it appears that the former played a dominant role in the current scenario, most likely because the corresponding reaction pathway bears a more favorable energetic barrier.
Scheme 12Preliminary result of the [5 + 2] cycloaddition leading to asperflavipine B.
The above result prompted us to revise the strategy for the biomimetic synthesis of asperflavipine B (10). As shown in Scheme 13, we envisioned that the target molecule could be obtained from the corresponding precursor 75 through dehydration. Compared to the previously designed oxidative dehydrogenation approach, we anticipated that the current one would be more easily achieved in practice. Besides, given that an undesired regioisomer was obtained in the preliminary study of the [5 + 2]-cycloaddition, we decided to replace the partially protected precursor 72 with the protection-group-free one (19), assuming that such operation may partially alter the regioselectivity of biomimetic [5 + 2]-cycloaddition.
Scheme 13Revised synthetic strategy towards asperflavipine B.
To test the above strategy, we first obtained the key intermediate 19 following the same procedure applied for the synthesis of asperchalasine A (9). Once formed, 19 was directly treated with K3Fe(CN)6 in the presence of hemiacetal 57. The biomimetic oxidative [5 + 2] cycloaddition also worked smoothly. However, the purification of the resulting products seemed to be challenging, mostly because of their fragile nature. After extensive tries, we successfully isolated the major product (54%) by preparative TLC with MeOH/DCM as the eluent solvents, which was determined to be 77 (a 5:3 mixture of C-8′ diastereoisomers) based on the extensive NMR study (Scheme 14). Apparently, 77 was derived from the putative [5 + 2]-adduct 76 through hemiketal exchange (H2O⟶MeOH) during the purification process with MeOH-DCM as eluting solvents. Moreover, careful analysis of the crude 1H NMR of the reaction mixtures indicated that besides 76, the desired regioisomer 75 could also be detected, albeit as the minor product. Encouraged by this clue, we directly submitted the resulting mixtures of [5 + 2]-cycloaddition to dehydration conditions (10% HCl, THF) [
]. Gratifying, by this way asperflavipine B (10) was isolated in 5% overall yield for two steps. Of note, the other regioisomer 76 could not be identified after the acidic treatment, indicating that it decomposed under the conditions. Interestingly, a similar observation was also documented in the seminal work reported by the Deng group [
]. Based on these results, we deduce that the [5 + 2]-adduct 76 could also be a naturally occurring compound, although its identification may be very challenging due to its fragile nature.
To sum up, in the present work we describe our continuous progress on the chemical synthesis of cytochalasans and merocytochalasans, which has culminated in the collective total syntheses of more than ten biosynthetically related natural targets. The highlight of the entire synthesis campaign include: 1) a concise, modular and scalable approach has been established to access the 11/6/5 tricyclic aspochalasins, featuring a Cu(I)-catalyzed intermolecular Diels-Alder reaction and a RCM reaction as the key steps; 2) the biomimetic Diels-Alder heterodimerizaiton of the tricyclic aspochalasin B with different bicyclic epicoccine precursors has been explored systematically, which enabled the access of the heterodimers asperchalasines B-E (5–8) as well as related congeners in a collective manner; 3) a computational study has been conducted to illustrate the full profile of the regio- and endo/exo selectivity in the biomimetic heterodimerizations; 4) the biomimetic synthesis of [A + B + A]-type heterotrimer asperchalasine A (9) has been achieved from two aspochalasin B units and one epicoccine unit through a biomimetic Diels-Alder reaction followed by an oxidative [5 + 2] cycloaddition; 5) asperflavipine B (10), a representative [A + B + B]-type heterotrimer, has also been synthesized from one aspochalasin B unit and two epicoccine units through a biomimetic Diels-Alder reaction followed by an oxidative [5 + 2] cycloaddition and dehydration reaction.
Taken all, the present work largely enriches the synthetic chemistry of both classic cytochalasans and the newly identified merocytochalasans. Moreover, it also provides venerable information to decipher the intriguing biosynthetic origins of the chased targets and related congeners. For instance, our study reveals that the isobenzofuran species (e.g. 15a) involved in the biosynthesis of merocytochalasans is more likely derived from a precursor like 57 instead of the originally proposed 15. Moreover, given that both the Diels-Alder heterodimerizatons and the oxidative [5 + 2] cycloadditions discussed above have been achieved under very mild and nature-like conditions, we deduce that most of the discovered merocytochalasans, if not all, are probably produced in nature through non-enzymatic processes.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We acknowledge the financial support from National Natural Science Foundation of China (21572112, 21772109, 21971140) and Beijing Natural Science Foundation (2172026, M21011). We thank Dr. Ning Xu at the BioNMR facility, Tsinghua University Branch of China National Center for Protein Sciences (Beijing) for his assistance with the NMR data collection. We also thank Professor Jun Deng (Nankai University) for the helpful discussion.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Selective hydrogenolysis of the benzyl protecting group for hydroxy function with raney-nickel in the presence of the mpm (4-methoxybenzyl) and dmpm (3,4-dimethoxybenzyl) protecting groups.
Synthesis of 3,4-disubstituted 2,5-dihydrofurans starting from the Baylis-Hillman adducts via consecutive radical cyclization, halolactonization, and decarboxylation strategy.
00018-3&id=ga1.jpg){kind=link}
00018-3&id=gr1.jpg){kind=link}
00018-3&id=sc1.jpg){kind=link}
00018-3&id=sc2.jpg){kind=link}
00018-3&id=sc3.jpg){kind=link}
00018-3&id=sc4.jpg){kind=link}
00018-3&id=sc5.jpg){kind=link}
00018-3&id=sc6.jpg){kind=link}
00018-3&id=sc7.jpg){kind=link}
00018-3&id=sc8.jpg){kind=link}
00018-3&id=sc9.jpg){kind=link}
00018-3&id=sc10.jpg){kind=link}
00018-3&id=gr2.jpg){kind=link}
00018-3&id=gr3.jpg){kind=link}
00018-3&id=sc11.jpg){kind=link}
00018-3&id=sc12.jpg){kind=link}
00018-3&id=sc13.jpg){kind=link}
00018-3&id=sc14.jpg){kind=link}