Download PDF
Synlett 2022; 33(09): 836-850
DOI: 10.1055/a-1748-4744
account

Biomimetic Diels–Alder Reactions in Natural Product Synthesis: A Personal Retrospect

Jingchun Liu 
,
Shuang Xi
,
Yefeng Tang  ∗

We acknowledge the financial supports from the National Natural Science Foundation of China (21772109 and 21971140) and the Natural Science Foundation of Beijing Municipality (M21011).
› Further Information
 
 PDF Download Permissions and Reprints All articles of this category


Abstract

Nature has been recognized for her super capability of constructing complex molecules with remarkable efficiency and elegancy. Among nature’s versatile synthetic toolkits, Diels–Alder reaction is particularly attractive since it allows for rapid generation of molecular complexity from simple precursors. For natural products biosynthetically formed through Diels–Alder reactions, the most straightforward way to access them should build on biomimetic Diels–Alder reactions. However, the implementation of biomimetic Diels–Alder reactions in a laboratory setting may encounter considerable challenges, particularly for those suffering from complicated reactivity and selectivity issues. Indeed, the translation of a biosynthetic hypothesis into a real biomimetic synthesis entails the orchestrated combination of nature’s inspiration and chemist’s rational design. In this Account, we will briefly summarize our recent progress on the application of biomimetic Diels–Alder reactions in natural product synthesis. As shown in the discussed stories, rational manipulation of the structures of biosynthetic precursors plays a crucial role for the successful implementation of biomimetic Diels–Alder reactions.

1 Introduction

2 Biomimetic Synthesis of Rossinone B

3 Biomimetic Synthesis of Homodimericin A

4 Biomimetic Synthesis of Polycyclic and Dimeric Xanthanolides

5 Biomimetic Synthesis of Periconiasins and Pericoannosins

6 Biomimetic Synthesis of Merocyctochalasans

7 Conclusion and Outlook


#

Key words

Diels–Alder reaction - natural product - biomimetic synthesis - reactivity - selectivity

Biographical Sketches

Zoom Image

Yefeng Tang was born in Xinjiang Province, P. R. of China in 1976. He received his BSc degree from Lanzhou University (1995–1999), and then his MSc degree from the Institute of Materia Medica, Chinese Academy of Medical Science (1999–2002). During the period of 2003–2006, he pursued his PhD degree under the supervision of Prof. Zhen Yang and Prof. Jiahua Chen at Peking University. He then joined Prof. K. C. Nicolaou’s group at The Scripps Research Institute as a postdoctoral associate (2006–2009). He started his independent career as a tenure-track assistant professor in School of Pharmaceutical Sciences, Tsinghua University in 2010, and now is a tenured associate professor. His major research interests include total synthesis of natural products, synthetic methodology development, bioorthogonal chemistry, and antiviral drug discovery.

Zoom Image

Jingchun Liu was born in Jiangsu Province, P. R. of China in 1996. He received his BSc degree from Lanzhou University in 2018, and then joined Prof. Yefeng Tang’s research group at Tsinghua University. He is presently a PhD candidate in Prof. Tang’s group. His current research focuses on total synthesis of complex natural products.

Zoom Image

Shuang Xi was born in Shandong Province of China in 1996. She received her BSc degree from Qingdao University of Science and Technology University (2014–2018), and then her MSc degree from LiaoNing Petrochemical University (2019–2021). Her major research interests include synthetic methodology development and medicinal chemistry.

1

Introduction

Since its first discovery in the 1920s,[1] Diels–Alder reaction has been standing at the center of modern organic synthesis and never been out of style. Owing to its extraordinarily efficient and versatile nature, Diels–Alder reaction unarguably ranks among the most important named reactions and has found widespread application in total synthesis of natural products.[2] Interestingly, long before its discovery in the laboratory settings, nature has already mastered this magic reaction over the course of million-years evolution. Indeed, it has been well-known that numerous natural products feature Diels–Alder reaction as the key element of their biosynthetic pathways.[3] Not surprisingly, owing to its prevalence in the biosynthesis of natural products, biomimetic Diels–Alder reaction has been frequently utilized as a key step in total synthesis of natural products. However, since biological Diels–Alder reactions generally occur under physiological conditions which are not easily mimicked in the laboratory, their realization may encounter considerable challenges in practice, particularly for those cases with intriguing reactivity and selectivity issues.[4]

Our group has shown keen interest in biomimetic synthesis of natural products, with over 50 targets having completed to date.[5] Interestingly, many of the syntheses involve biomimetic Diels–Alder reaction as the key steps. While several elegant reviews have been published on the topic of biomimetic Diels–Alder reactions,[6] in this Account we will share our own research experience on this subject, with a focus on dissecting the key rationales behind the biomimetic Diels–Alder reactions from a strategic point of view. As shown in the discussed cases, the successful realization of a biomimetic Diels–Alder reaction generally entails the synergistic interaction of nature’s inspiration and chemist’s rational design. Particularly, judicious manipulation of the biosynthetic precursors and reaction conditions proves to be very crucial for securing the feasibility and practicality of biomimetic Diels–Alder reactions.


# 2

Biomimetic Synthesis of Rossinone B

Our interest on biomimetic synthesis of natural products could be traced back to the first project completed in our laboratory at Tsinghua University. In 2009, a unique marine meroterpenoid, namely rossinone B (1), was disclosed by Copp and co-workers.[7] Structurally, rossinone B bears a congested 6/5/6/5-tetracyclic system embracing multiple continuous stereocenters, which represents an attractive target for total synthesis. Moreover, the preliminary biological study revealed that rossinone B exhibited potent antiviral, anti-inflammatory, and antileukemic activities, thus rendering it an excellent lead for the development of new drugs.

Interestingly, rossinone A (2), a monocyclic hydroquinone derivative, was also isolated along with rossinone B from the same source, which shed light on their biosynthetic relationship. As shown in Scheme [1], rossinone A (2) could convert into vinyl quinone 3 through Schenck ene reaction followed by oxidation. Once formed, an intramolecular Diels–Alder reaction could occur, transforming 3 into tricyclic compound 4. Subsequently, 4 could advance to compound 6 through double-bond migration followed by allylic oxidation. Finally, a tandem oxa-Michael addition/β-elimination could take place, giving rise to rossinone B (1).

Zoom Image
Scheme 1 Proposed biosynthetic origin of rossinone B and rationally designed biomimetic strategy

The proposed biosynthetic origin of 1 offered an inspiring clue to access this target. However, we anticipated that there might exist some challenges for its realization. First, the proposed transformation from 3 into 4 might be problematic for the reactivity issue. Although it has been reported that vinyl quinone unit could engage in the inverse-electron-demand Diels–Alder reaction with an electron-rich dienophile as the reactant partner,[8] in the current scenario, the dienophile is electron-deficient, and thus the proposed Diels–Alder reaction appears to be electronically mismatched. Secondly, even if the Diels–Alder reaction could work under very harsh conditions, the resulting product might readily undergo double-bond migration followed by oxa-Michael addition with the C-15 hydroxy groups. If this happens, the late-state introduction of the hydroxy group on the C-6 position would become challenging.

Keeping the above concerns in mind, we decided to modify the biosynthetic precursor 2 by introducing a methoxy group on the C-13 position. We anticipated that such a structural modification would offer three advantages by: 1) increasing the HOMO energy of the vinyl quinone unit of 3′, which renders the diene–dienophile pair more electronically matched for a normal Diels–Alder reaction; 2) serving as a blocking group to preclude the isomerization of the newly generated double bond (C5=C6) in the Diels–Alder reaction; 3) providing an opportunity directly transforming the Diels–Alder adduct 4′ into natural product 1 through double oxa-Michael addition/β-elimination reactions, thus avoiding extra steps for the introduction of C-6 hydroxy group in the late stage.

To our delight, the above design worked well in practice. While we will not discuss the details of the entire synthesis, some key elements are highlighted in Scheme [2]. Thus, the linear fragment 8 was first prepared from the known compound 7 in five steps. Subsequently, regioselective metalation of 9 followed by reaction with aldehyde 8 afforded 10 as a mixture of diastereoisomers. Upon treatment with 6N HNO3/Ag2O, 10 readily underwent one-pot allylic isomerization, deprotection and oxidation to give the desired precursor 3′. Next, the Diels–Alder reaction was attempted in toluene at elevated temperature (150 °C, sealed tube). Pleasingly, the reaction worked smoothly, furnishing the tricyclic compound 4′ as a pair of inseparable C-15 epimers. The harsh reaction conditions indicated that the Diels–Alder reaction bear a notably high energic barrier even with the C-13 methoxy group used as an activating group. Fortunately, the reaction displayed an excellent diastereoselectivity, only resulting in the endo-adduct 4′. Furthermore, we found that the two C-15 epimers could interconvert under various conditions, probably through a ketone–enol tautomerization equilibrium. Thus, they were directly used in the following reaction without purification. As the endgame, the double oxa-Michael addition/β-elimination cascade reaction was realized in MeOH/H2O with TsOH as the promoter, which delivered rossinone B (1) in 31% yield over two steps.[9]

Zoom Image
Scheme 2 Biomimetic total synthesis of rossinone B

The biomimetic synthesis of rossinone B is the first project accomplished in our group. While the target itself appears not very challenging, we would emphasize that this work was completed by a single undergraduate student in only six months! Apparently, key to the success relies on the adopt of a highly efficient biomimetic strategy. Besides, the judicious manipulation of the biosynthetic precursor also enables us to avoid some underlying traps along the synthetic adventure. Encouraged by this experience, we have applied the similar concept to many other total synthesis projects undertaken in our laboratory, some of which will be discussed in the following sections.


# 3

Biomimetic Synthesis of Homodimericin A

In 2017, homodimericin A (12) was identified by Clardy and co-workers from the soil fungus Trichoderma harzianum.[10] Structurally, it bears a highly congested hexacyclic framework containing eight contiguous stereogenic carbons, among which three are quaternary centers. Biologically, homodimericin A was produced by the fungus as a chemical defense to counter the exogenous oxidative stress imposed by a cocultured bacterium Streptomyces sp. 4231, and as a result, its molecular architecture displays a high level of oxidation state. Owing to its appealing chemical and biological profiles, homodimericin A (12) represents an attractive target for total synthesis.

In the isolation paper,[10] Clardy and co-workers provided a hypothesis for the biosynthetic origin of homodimericin A (Scheme [3]). In their opinion, homodimericin A could be traced back to a monocyclic precursor 13a, although the latter has never been identified in nature. Featuring a highly substituted hydroquinone core, 13a readily undergoes oxidation to afford para-quinone 14a. Once formed, 14a could further advance to the tricyclic compound 18a through homodimerization. Bearing a hydroquinone moiety, 18a could be oxidized to quinone 19a which then undergoes an intramolecular Diels–Alder reaction to furnish compound 20a. Finally, a carbonyl-ene reaction could transform 20a into homodimericin A.

Zoom Image
Scheme 3 Proposed biosynthetic origin and rationally designed biomimetic strategy for homodimericin A

Although the above biosynthetic hypothesis seems to be inspiring, there exist considerable challenges for its realization. First, the assumed biosynthetic precursor 13a was not discovered along with homodimericin A, indicating that such type of compound could be highly unstable, thus casting a shadow on its accessibility. Second, even if such precursor could be prepared with careful synthetic manipulations, the following oxidation reaction might be problematic, since besides the desired para-quinone 14a, an ortho-quinone species (structure not shown) could also form in the reaction, which may arouse some undesired reaction pathways. Inspired by our previous experience on rossinone B,[9] we decided to adopt a modified synthetic precursor in the practical synthesis. We anticipated that introducing a methyl protecting group onto the C-2 phenolic hydroxy could improve the stability and synthetic accessibility of the biosynthetic precursor. Moreover, for such a precursor 13b, the desired para-quinone 14b could be generated in a controllable manner, which also has a beneficial effect on the subsequent homodimerization reaction. Of note, aside from the usage of a modified biosynthetic precursor, we proposed that the homodimerization reaction of 14b should proceed through an inverse bond-forming sequence in comparison to the hypothesis assumed by the Clardy group.

Our synthesis commenced with the preparation of the designed precursor 13b.[11] After analysis of the structural features of 13b, we developed a highly concise and modular approach to achieve this goal, featuring Moore rearrangement[12] as the key step. Thus, dibromopropene (22) was coupled with 2,4-hexadienal (23) through a Sn-mediated Barbier reaction,[13] and the resulting product was then converted into silyl ether 24 under conventional conditions. Next, 24 was coupled with 3-methoxy-4-methylcyclobutene-1,2-dione (25) through a lithium–bromide exchange followed by nucleophilic addition, which yielded 4-alkenyl 4-hydroxycyclobutenone derivative 26. Subsequently, Moore rearrangement of 26 was conducted in refluxing p-xylene, which gave ring-expansion product 27 in an excellent yield. After desilylation and oxidation, 27 was transformed to the desired para-quinone 13b smoothly. In align with our expectation, 13b displayed reasonable stability and existed predominantly in a ketone form 28. However, severe decomposition occurred after a long-term storage, indicting its fragile nature. Thus, the freshly prepared 28 was submitted to the dimerization reaction by treatment with catalytic amounts of sodium hydride in DCM, which led to tricyclic compound 18b in an acceptable yield. Further oxidation of 18b resulted in para-quinone 19b, thus setting the stage for exploring the crucial Diels–Alder reaction. Out of our expectation, the transformation appeared to be rather challenging. For example, no Diels–Alder adduct could be identified in the reaction conducted with EtAlCl2. Instead, compound 29 was generated as the major product, presumably through sequential [1,5]-H migrations.[14] To avoid this side reaction, we turned to examine the thermal conditions. After extensive trials, we found that the proposed biomimetic Diels–Alder reaction and carbonyl-ene cyclization could be achieved in a one-pot manner under harsh conditions (1,2-DCE, 180 °C, sealed tube), which delivered the hexacyclic compound 21 in 40% yield. Besides, a side product 30 was also isolated in the reaction, which should arise from 19b through tandem Diels–Alder reaction/ene cyclization. Notably, the intermediacy of Diels–Alder adduct was not observed in the reaction, implying that it readily underwent the following ene reaction under the harsh conditions. Finally, demethylation of 21 was achieved with magnesium iodide and quinoline, which gave homodimericin A (12) in an excellent yield (Scheme [4]). While the synthetic campaign seemed to arrive at the end at this moment, we disappointedly found that the NMR spectra of the synthetic sample were not fully consistent with those of natural product. After a discussion with the isolation team, we deduced that some unidentified impurity present in the originally natural sample might account for the observed discrepancy between the spectroscopic data of synthetic and natural products. Interestingly, several months later after we disclosed the synthesis work, we serendipitously obtained the crystal of the synthetic sample of homodimericin A, which offered an unequivocal evidence for its identity.

Zoom Image
Scheme 4 Total synthesis of homodimericin A

Beside our work, three other biomimetic syntheses of homodimericin A were also completed by the Wang,[15] Yang,[16] and Deng[17] groups in a short period of time, among which the one completed by Wang and co-workers is particularly notable. In their work, the authors achieved the homodimerization of the para-quinone 14a, the intramolecular Diels–Alder reaction, and the carbonyl-ene cyclization in one pot under mild conditions (PBS buffer, air, room temperature), albeit in a relatively low yield.[5] This result indicated that although the introduction of a methyl protecting group on the proposed biosynthetic precursor did improve its stability and synthetic accessibility, such option lowered its reactivity for some biomimetic transformations (e.g. Diels–Alder reaction). In this context, it remains a trick for synthetic chemists to balance the reactivity, stability, and accessibility of the precursor for a biomimetic reaction.


# 4

Biomimetic Synthesis of Polycyclic and Dimeric Xanthanolides

Xanthanolides are a family of natural products attracting considerable attention from our group over the past decade.[18] Our initial interest in this family of natural products stems from another long-term project directed towards the development of novel dyotropic rearrangements of β-lactones and their application in natural product synthesis.[18a] [19] In 2012, we reported a unique dyotropic rearrangement of 3,4-cis-β-lactones,[18a] which enabled the facile access of the characteristic 5/7-trans-bicyclic core of xanthanolides. Hinging on this methodology, a number of bicyclic xanthanolides including xanthatin (32) and 8-epi-xanthatin (33) were synthesized in an efficient and scalable manner. This work paved the way to access some more complicated xanthanolides biosynthetically derived from 32 and 33. For example, 4β,5β-epoxyxanthatin-1α,4α-5/7-endoperoxide (35), a tetracyclic xanthanolide bearing a characteristic endoperoxide moiety, could be derived from xanthatin (32) via sequential C2=C3 double-bond isomerization, 6π-electrocyclization, and singlet-oxygen-promoted [4+2] cycloaddition (Scheme [5]). Comparably, 8-epi-xanthatin (33) would undergo a tandem C2=C3 double-bond isomerization/6π-electrocyclization/intramolecular Diels–Alder reaction to yield the pentacyclic xanthipungolide (37). Furthermore, 8-epi-xanthatin (33) could also undergo homodimerization via a Diels–Alder reaction to produce pre-pungiolide (38) which, upon further oxidation and (or) cyclization, could lead to various other dimeric xanthanolides such as pungiolide A (39) and D (40).

Zoom Image
Scheme 5 Proposed biosynthetic origin of polycyclic and dimeric xanthanolides

Since the total syntheses of xanthanolides have been summarized elsewhere recently,[20] this Account will mainly concentrate on some key elements related with the biomimetic Diels–Alder reactions. Initially, we chose to investigate the biomimetic synthesis of 35 through the proposed tandem reaction. To this end, we treated xanthatin (32) in benzene with photoirradiation (Hg lamp) in the presence of dioxygen.[21] Unexpectedly, only a trace amount of 35 was obtained. Instead, substantial amounts of the dimeric compounds 42, 43, and 44 were identified in the reaction (Scheme [6]). Apparently, both 43 and 44 should stem from 42 through a photo [2+2]-cycloaddition, and the latter, in turn, appeared to arise from 32 through a head-to-head Diels–Alder dimerization.

Zoom Image
Scheme 6 Biomimetic synthesis of xanthatin-based polycyclic and dimeric xanthanolides

However, a careful analysis of the stereochemistry of 43 indicated that a direct dimerization of 32 through Diels–Alder reaction could be excluded, since the relative stereochemistry of C-1 and C-5 positions seemed incorrect. Based on some inspiring cases as well as additional experimental evidence, we deduced that the observed dimeric products should be generated from 32 through an unusual pathway. Thus, the photoinduced C1=C5 double-bond isomerization should take place first to generate a highly strained trans-cyclohepatene species 41 which then went through a tail-to-tail Diels–Alder reaction to afford the product 42. It should be noted that although the Diels–Alder reaction of trans-cyclohepatene species has been known for decades, its application in natural product synthesis has been rarely explored.

The above outcome indicates that although both C1=C5 and C2=C3 double bonds of 32 could undergo isomerization upon photoirradiation, the high reactivity of the trans-cyclohepatene species 41 drives the reaction towards the dimerization pathway. In this vein, to achieve the desired biomimetic transformation leading to 35, it is essential to inhibit the competing dimerization reaction. To this end, we choose to conduct the photoreaction in the presence of excess amounts of DBU, with the assumption that the trans-cycloheptene species 41, once generated, could be immediately trapped by DBU through 1,6-conjugate addition.[22] To our delight, this idea worked well, giving a mixture of the diastereoisomers 46a and 46b (dr = 5:1) in an excellent combined yield (Scheme [3]).

No dimeric products were detected in the reaction. Having 46a/46b in hand, we then submitted them to the singlet-oxygen-mediated [4+2] cycloaddition (Hg lamp, O2), which gave rise to 4β,5β-epoxyxanthatin-1α,4α-endoperoxide (35) and its diastereoisomer 47 in 58% and 12% yield, respectively.

Notably, over the course of investigating the photoreaction of xanthatin (32), we serendipitously found that direct irradiation of the crystallized 32 led to another dimeric compound 49 that bears a pentacyclic skeleton distinct from 42 (Scheme [7]). Mechanistically, 49 should arise from 32 through a formal [4+2] cycloaddition involving a diradical intermediate. The observed regio- and stereochemical outcomes of the dimerization should originate from the restricted molecular motions and conformations in the solid state.[23] The type II dimerization mode deserved further analysis, since the resulting product bears a tail-to-tail dimeric framework that was similar to that of natural product pungiolide A (39). However, we noticed that the stereochemistry of the C-5 and C-11′ stereocenters was contrary to those of pungiolide A (39), which indicated that the latter should arise from its biosynthetic precursor 33 through another type of dimerization mode. After careful analysis of the biosynthetic origin of pungiolides, we deduced that compound 49 arose from 32 through a concerted Diels–Alder reaction.

Zoom Image
Scheme 7 Further exploration of the dimerization of xanthatin

Keeping this assumption in hand, we further screened the dimerization of 32 under various conditions. Gratifyingly, after extensive trials, we finally obtained a new dimeric product 48 with the treatment of 32 in refluxing water for a long reaction time. Importantly, this dimer shares the same skeleton and stereochemistry with those of pungiolides, implying that the naturally occurring dimeric xanthanolides such as pungiolide A and D should arouse from 33 through the type III dimerization mode.

When the program arrived at this stage, we could not help asking ourselves one question: could the dimeric compounds 43, 44, 48, and 49 be some natural products yet-to-be discovered? This question is interesting, given that xanthatin (32) is one of the most abundant members in the family of xanthanolides, but its dimer had never been identified in nature. Stimulated by this curiosity, we collaborated with the Hu group to search for the underlying xanthatin-type dimer in the Xanthium mogolium Kitag, a plant source rich in xanthatin (32) and related monomers. To our delight, we did find several unprecedented xanthanolide dimers in this plant, albeit in low abundance. More importantly, two of them bear the identical structures with compounds 43 and 48, which were named mogolide A and B by us, respectively.[18b] In this context, our study provided a sweet case of biomimetic-synthesis-guided discovery of new natural products.

The above work paved the way to access another group of polycyclic and dimeric xanthanolides derived from 8-epi-xanthatin (33). For example, xanthipungiolide 37, one of the most complicated xanthanolide monomers, was obtained from 33 through a one-pot reaction involving the photoinduced double-bond isomerization, 6π-electrocyclization, and intramolecular Diels–Alder reaction (Scheme [8]). It should be noted that the Diels–Alder reaction had to be conducted under the thermal conditions (45 °C) in order to secure an acceptable overall yield. In addition, we also realized that the biomimetic synthesis of a variety of dimeric xanthanolides such as pungiolide A (39) and D (40) through the Diels–Alder dimerization followed by further cyclization or functionality elaborations. Similarly, the Diels–Alder dimerization of 33 was also performed with a high reaction temperature (110 °C) that is not readily accessible in living systems. Therefore, we suspect that the naturally occurring xanthanolide dimers might be generated from the corresponding precursors through some enzymatic processes.

Zoom Image
Scheme 8 Biomimetic synthesis of pungiolides

To sum up, we would say that although the biosynthetic origin of polycyclic and dimeric xanthanolides seems to be straightforward, their translation into the above biomimetic synthesis is still an adventure full of unexpected results. Fortunately, some of the results, albeit frustrating at the first glance, have guided us to discover some exciting landscapes that we have never conceived at the beginning. In this context, the present story showcases the great value of natural products as the engine for discovery and development of new chemistry.


# 5

Biomimetic Synthesis of Periconiasins and Pericoannosins

Cytochalasans represent a fascinating class of natural products that have attracted great attention from chemical and biological communities.[24] Structurally, most cytochalasans feature a typical 5/6/n-tricyclic framework that contains a relatively conserved perhydroisoindolone core (5/6-ring system) and a varied macrocyclic ring (n = 11–16). Comparably, cyctochalasans carrying a 9/6/5-tricyclic core have not been discovered until 2013 when periconiasins A–C (53–55, Scheme [9]) were identified by Dai and co-workers[25] from the endophytic fungus Periconia sp. F-31. Besides their unique molecular architectures, we were also attracted by periconiasins for their prominent biological activity.

Zoom Image
Scheme 9 Proposed biosynthetic origin of pericoannosins and periconiasins

Preliminary biological study revealed that periconiasin A and B display selective and potent cytotoxicity against the HCT-8 and BGC-823 cell lines, which could serve as promising leads for the development of anticancer drugs.

Interestingly, several other congeners related to periconiasin A–C were also disclosed by the same research group in 2015,[26] among which pericoannosin A (57) is particularly notable, since it displays an unusual oxadecalin 3-pyrrolidin-2-one skeleton. After a further analysis of the biosynthetic connection between periconiasins and pericoannosin A, we realized that they might arise from a common biosynthetic precursor 56 through divergent Diels–Alder reactions. In accordance with the biosynthetic pathways leading to typical tricyclic cytochalasans, 56 could undergo an intramolecular Diels–Alder reaction with the engagement of C10=C11–C12=C13 diene moiety and C2=C15 dienophile moiety, giving rise to the 9/6/5-tricyclic skeleton (I) presented in periconiasins A–C (path a). Alternatively, a hetero-Diels–Alder reaction could also take place between the C5=C4–C3=O2′ diene moiety and C10=C11 dienophile moiety (path b), leading to the oxadecalin-containing tricyclic skeleton (II) related to pericoannosin A. Interestingly, besides these two pathways, another Diels–Alder reaction could also be imagined for the precursor 56, with C10=C11–C12=C13 and C4=C5 as the diene–dienophile combination. As result, a decalin-3-pyrrolidin-2-one framework (III) would generate in the reaction (path c). Of note, although the natural products correlated to the type III skeleton have not been disclosed by then, the last reaction pathway seems to be highly possible from a chemical point of view.

Nature seemingly adopts a wise way to access periconiasins and pericoannosin A from a common precursor in a divergent manner with the aid of enzymes; however, how to achieve these pathways in a controllable fashion poses a formidable challenge to synthetic chemists. Keeping this concern in mind, we decided to develop a practical and controllable strategy to access periconiasins from a modified biosynthetic precursor, in which the C4=C5 double bond was tentatively masked as a single bond. By this way, both the pathway b and c in Scheme [9] could be excluded in principle.

To test this idea, the polyketide–amino acid precursor 61a was first prepared from three building blocks 58–60 through five steps, featuring tandem aldol condensation/Grob fragmentation as the key step.[27] It was found that 61a was extremely unstable and had to be immediately submitted to the Diels–Alder reaction under the thermal temperature (CHCl3, 90 °C). To our delight, the reaction did work, resulting in a mixture of the endo (62a) and exo (63a) adducts in a ratio of 3:1 (Scheme [10]). However, this reaction efficiency appeared to be moderate, which should be attributed to the severe decomposition of 61a under the employed conditions. We conducted an extensive condition optimization but failed to obtain a satisfactory result. After struggling for a while, we found that replacing the N-protecting group of the Diels–Alder precursor from Bz to o-Me-Bz could notably improve the efficiency of the reaction, with the endo (62b) and exo (63b) adducts obtained in 38% and 12% yields, respectively. Although the exact role of the N-protecting groups in this transformation remains unclear, it appeared that 61b was more stable than 61a, and thus the decomposition of 61b notably attenuated over the course of reaction. With the tricyclic core of periconiasins A–C secured, we moved to introduce the C4=C5 double bond through sequential selenation and oxidative elimination. Of note, only the Z-isomer 64 was obtained, indicating that the E-isomer bears considerable ring strain and is difficult to generate in the reaction. This result suggested that the proposed biosynthetic transformation from 56 to the 9/6/5-tricyclic skeleton (I) should be very challenging for both selectivity and reactivity issues. Bearing the requisite functionality for further elaboration, 64 was readily converted into periconiasin A (53) through oxa-Michael addition. Finally, several other congeners including periconiasins B–E (54, 55, 66, and 67) could also be obtained from 53 through late-stage functionality elaboration and (or) further cyclization.

Zoom Image
Scheme 10 Total synthesis of periconiasins

Shortly after we disclosed the total synthesis of periconiasins A–E, we got an opportunity to visit the Dai group, the isolation team that discovered periconiasins A–F and pericoannosin A. Through discussion with the isolation team, we were informed that some compounds were also isolated along with periconiasins A–E, whose structures, albeit not fully established yet, may feature the decalin 3-pyrrolidin-2-one as core skeletons. This clue is valuable, indicating that the proposed path c in Scheme [9] could also occur in nature. Inspired by our previous experience on xanthanolides, we initiated a collaborative project with the Dai group, aiming to identify the underlying decalin-containing compounds from the fermentation broth of endophytic fungus Periconia sp. F-31. To our delight, we did discover four new compounds bearing the proposed decalin 3-pyrrolidin-2-one skeleton, which were named pericoannosins C–F by the Dai group.[28] Furthermore, we also completed the total synthesis of pericoannosins C–F (68–70) (Scheme [11]), which allowed us to unambiguously confirm the structures of the newly identified natural products. The synthetic route towards 68–70 is outlined in Scheme [11]. To avoid the underlying selectivity issue, we chose to introduce the 3-pyrrolidin-2-one moiety in the late state of the synthesis. Thus, the truncated linear substrate 72, which was accessed from 71 through sequential selenation and oxidative elimination, readily underwent Lewis acid promoted Diels–Alder reaction to yield the 6/6-bicyclic compounds 73a/73b in a 9.5:1 endo/exo ratio. The reaction also worked under the thermal conditions (toluene, 150 ℃), although an inferior combined yield (52%) and endo/exo (2:1) selectivity were observed. With endo-adduct 73a as the precursor, we completed the total synthesis of 68–70 through sequential introduction of the pyrrolidinone moiety and the hydroxy group on the C-2 position.

Zoom Image
Scheme 11 Total synthesis of pericoannosins

The completion of the total synthesis of periconiasins and pericoannosins promoted us to reconsider the biosynthetic hypothesis outlined in Scheme [9]. Based on our experimental results as well as mechanistic analysis, it is less likely that both pericoannosins and periconiasins are derived from the common polyene precursor 56 through divergent Diels–Alder reactions. Instead, a more reasonable hypothesis is suggested by us, which involves the polyketide–amino acid derivative 75 as the common biosynthetic precursor for pericoannosins and periconiasins.

In the scenario, 75 could undergo regioselective dehydration reaction to yield compounds 76 and 77 with the aid of specific dehydrases. In the absence of C4=C5 bond, 76 could specifically advance to periconiasins through the type I Diels–Alder reaction (Scheme [12]). Similarly, without the interference of C2=C15 double bond, only the two types of Diels–Alder reactions outlined in the path b and c (Scheme [9]) may take place, leading to the 6/6/5-tricyclic skeleton associated with pericoannosins.

Zoom Image
Scheme 12 Revised hypothesis for the biosynthetic origin of periconiasins and pericoannosins

# 6

Biomimetic Synthesis of Merocyctochalasans

Since 2015, a series of unprecedented polycyclic natural products, collectively referred to as merocyctochalasans, have been identified by Zhang and co-workers from the fermentation broth of Aspergillus flavipes.[29] Just as implied by their names, this family of natural products are biosynthetically derived from two subunits of natural origin, the typical tricyclic cytochalasin and the polyphenol derivative epicoccine, through either heterodimerization, -trimerization or -tetramerization. Asperchalasin A–E (78–82) are the first group of merocyctochalasans disclosed by the isolation team, among them, asperchalasins B–E (79–82) featuring a heterodimeric skeleton derived from one unit of cytochalasan (83) and one unit of oxidized epicoccine (84) through an intermolecular Diels–Alder reaction (Scheme [13]). Specifically, epicoccine (84), upon oxidation, could readily convert into benzoquinone 85a and 85b, both of which could further isomerize to isobenzofuran 85c. Once formed, 85c could react with the cytochalasin unit (83) through Diels–Alder reaction. Owing to the underlying regio- and endo/exo selectivity issues, four different cycloadducts (86–89) might be generated, which could advance to asperchalasins B–E (78–82), respectively, through selective methylation. Asperchalasin A (78) possesses an even intricate decacyclic framework comprised of one unit of epicoccine and two units of cyctochalasin moieties. Biosynthetically, it should stem from the heterodimer 89 and the second cytochalasan unit (83) through an oxidative [5+2] cycloaddition. Apparently, the newly identified merocyctochalasans notably enrich the chemistry of cyctochalasan-type of natural products. Particularly, their extremely complicated molecular architectures, coupled with the high degree of functionalities and densely arranged stereogenic centers, render them daunting challenges for total synthesis.

Zoom Image
Scheme 13 Proposed biosynthetic origin of asperchalasins A–E

Since our group has shown keen interest on the chemical synthesis of cyctochalasans,[27] [28] we decided to initiate a project to conquer these newly discovered targets. Our synthetic blueprint was built on the proposed biosynthetic pathways leading to asperchalasins A–E. At the initial of this project, our primary task was to access the requisite cytochalasin and epicoccine monomers. While the details of the synthesis of these monomers have been described in the original research paper,[30] we will focus our attention on the Diels–Alder heterodimerization reaction on this occasion. According to our proposal, we first attempted the reaction with aspochalasin B (83) and epicoccine (84) in the presence of K3Fe(CN)6. We envisioned that under these conditions epicoccine (79) could be oxidized into the corresponding ortho-quinone species 85a and 85b which may exist in an equilibrium with the isobenzofuran species 85c. As a reactive diene, the isobenzofuran species 85c could readily react with aspochalasin B (83) through an intermolecular Diels–Alder reaction. However, to our disappointment, we failed to detect the expected Diels–Alder cycloadducts; instead, some epicoccine dimer (structural not shown) was obtained in the reaction.

This observation was in accordance with the results reported by the Trauner group, which showed that epicoccine readily underwent homodimerization under similar conditions with the ortho-quinone species 85a or 85b involved as intermediates.[31] Apparently, to achieve the desired heterodimerization, we need to find a suitable way to generate the requisite isobenzofuran species without the interference of the ortho-quinone species. To our delight, after a survey of the related literatures, we found an alternative method to generate the isobenzofuran species 85c from the precursor 93 through an acid-promoted 1,4-elimination reaction (Scheme [14]).[32] As shown, direct treatment of a mixture of 83 and 93 in the presence of CSA under the elevated temperature led to two heterodimers which were tentatively assigned to be asperchalasins G (84) and H (88) on the basis of their crude 1H NMR data. The above products proved to be rather unstable, and thus they were directly transformed into the corresponding fully acetylated products 94 and 95 (5:1). Interestingly, one of them, compound 95, was reported as an artifact by Gu and co-workers.[33] This result was encouraging, since it showed that the proposed biomimetic Diels–Alder reaction did work. However, there remained some problems to be addressed. First, the overall yield of the reaction appeared moderate, mostly because both the precursor 93 and the resulting products are highly sensitive to oxygen and readily advance to some undesired products.

Zoom Image
Scheme 14 Total synthesis of asperchalasins A–E

Second, while four plausible isomers could be generated in the reaction, only two of them were detected in the current scenario. Unfortunately, the major product 94 proved to be the exo adduct, which could not be used for the synthesis of the heterotrimers like asperchalasine A. In this context, how to tune the regio- and endo/exo selectivity of the Diels–Alder reaction became the major challenge at this stage. Based on our previous experience, we assumed that both the reactivity and selectivity of the Diels–Alder reaction could be adjusted by the protecting groups of the epicoccine precursors. To test this idea, several partial and fully protected epicoccine precursors were prepared and submitted to the reaction. As shown, when the monomethylated hemiacetal 96 was used, a mixture of four diastereoisomers corresponding to asperchalasins B–E (79–82) were obtained in 80% combined yield in a ratio of 10:1:1:2. Among them, the exo-adduct asperchalasin B (79) was isolated as the major component. Apparently, introducing a methyl group on the epicoccine precursors could notably improve the efficiency of the reaction, despite showing little effect on the endo/exo and regioselectivity. Comparably, when we attempted the fully protected precursor 98 in the reaction, only a pair of endo products were detected in the reaction, which, after deprotection, were converted into asperchalasins E (82) and D (81) (3:2 ratio) in a 57% overall yield over two steps. This finding was encouraging, since it suggested that increasing the steric effect of the isobenzofuran species might invert the endo/exo selectivity of the Diels–Alder reaction. To further validate this assumption, the fully benzylated isobenzofuran precursor 98 was also attempted in the reaction. As expected, a pair of the endo-adducts 99 and 100 were obtained in an excellent combined yield (80%). The above results showed that it was feasible to obtain specific heterodimer as the major product in the Diels–Alder reactions by judiciously manipulating the protecting group of the epicoccine precursors. Of note, a systematic computational study was conducted by us recently, which enabled us to get a deep insight into the observed regio- and endo/exo selectivity of the Diels–Alder reactions. While the details will be disclosed elsewhere soon, we would like to say that the present discovery paves the way to access other more complicated heterotrimers and heterotetramer in a controllable fashion. As a proof-of-concept case, we have completed the total synthesis of the heterotrimer asperchalasin A with 99 as the precursor, featuring a biomimetic oxidative [5+2] cycloaddition as the key step.


# 7

Conclusion and Outlook

To conclude, in this Account we summarize a series of total syntheses of natural products accomplished in our group, all of which feature the biomimetic Diels–Alder reaction as the key step. As shown in these cases, the successful applications of these of biomimetic Diels–Alder reactions not only allow us to access the targets with remarkable conciseness and efficiency, but also provide venerable evidence to decipher their underlying biosynthetic origins. On the other hand, it should be emphasized that the translation of a postulated biosynthetic proposal into a real biomimetic synthesis is generally not as simple as expectation. Under most circumstances, it is fairly relied on the harmonious combination of insightful analysis of bio-inspiration, rational design of synthetic blueprint, continuous devotion of hard work, and sometimes even a little serendipity.

Looking forwards, we believe that biomimetic Diels–Alder reaction will continue to serve as a powerful tool in the total synthesis of natural products. Besides, it could also serve as an enabling platform to identify the underlying Diels–Alderase in the biosynthesis of natural products, as witnessed by some elegant works disclosed recently.[34] In turn, with the great advance in the field of synthetic biology and directed enzyme evolution, biomimetic Diels–Alder reactions may enter a new stage in which the reaction could be achieved in a more predictable and controllable manner with the aid of either natural or artificial Diels–Alderase. Without doubt, many more landscapes could be anticipated in this field in the foreseeable future.


#
#

Conflict of Interest

The authors declare no conflict of interest.

Acknowledgment

We sincerely thank all the former and current group members for their continuous contribution to the subject discussed in this Account.


Corresponding Author

Yefeng Tang
School of Pharmaceutical Sciences, MOE Key Laboratory of Bioorganic Phosphorus Chemistry & Chemical Biology, Tsinghua University
Beijing 100084
P. R. of China   

Publication History

Received: 31 December 2021

Accepted after revision: 22 January 2022

Accepted Manuscript online:
22 January 2022

Article published online:
28 February 2022

© 2022. Thieme. All rights reserved

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany


   Permissions and Reprints All articles of this category
Zoom Image
Zoom Image
Zoom Image
Zoom Image
Scheme 1 Proposed biosynthetic origin of rossinone B and rationally designed biomimetic strategy
Zoom Image
Scheme 2 Biomimetic total synthesis of rossinone B
Zoom Image
Scheme 3 Proposed biosynthetic origin and rationally designed biomimetic strategy for homodimericin A
Zoom Image
Scheme 4 Total synthesis of homodimericin A
Zoom Image
Scheme 5 Proposed biosynthetic origin of polycyclic and dimeric xanthanolides
Zoom Image
Scheme 6 Biomimetic synthesis of xanthatin-based polycyclic and dimeric xanthanolides
Zoom Image
Scheme 7 Further exploration of the dimerization of xanthatin
Zoom Image
Scheme 8 Biomimetic synthesis of pungiolides
Zoom Image
Scheme 9 Proposed biosynthetic origin of pericoannosins and periconiasins
Zoom Image
Scheme 10 Total synthesis of periconiasins
Zoom Image
Scheme 11 Total synthesis of pericoannosins
Zoom Image
Scheme 12 Revised hypothesis for the biosynthetic origin of periconiasins and pericoannosins
Zoom Image
Scheme 13 Proposed biosynthetic origin of asperchalasins A–E
Zoom Image
Scheme 14 Total synthesis of asperchalasins A–E