Download PDF
Synlett 2024; 35(05): 513-520
DOI: 10.1055/a-2066-1227
account
Biomimetic Synthesis

Embracing the Imperfectness of Nature using Highly Reactive N-Acyl Azahexatrienes

Kuan Zheng∗
a   CAS Key Laboratory of Synthetic Chemistry of Natural Substances, Center for Excellence in Molecular Synthesis, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai 200032, P. R. of China
,
Bingbing Zhang
a   CAS Key Laboratory of Synthetic Chemistry of Natural Substances, Center for Excellence in Molecular Synthesis, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai 200032, P. R. of China
b   University of Chinese Academy of Sciences, Beijing 101419, P. R. of China
,
Ran Hong∗
a   CAS Key Laboratory of Synthetic Chemistry of Natural Substances, Center for Excellence in Molecular Synthesis, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, Shanghai 200032, P. R. of China
b   University of Chinese Academy of Sciences, Beijing 101419, P. R. of China
› Author Affiliations

Financial support from the Youth Innovation Promotion Association of Chinese Academy of Sciences (2023267 to K.Z.) and the Science and Technology Commission of Shanghai Municipality (20XD1404700 to R.H.) is highly appreciated
› Further Information
 
 PDF Download Permissions and Reprints All articles of this category


Abstract

Incredible examples of controlling highly reactive functional groups to synthesize amazing architectures can be found in nature. N-Acyl azahexatriene, which is involved in biosynthesis, is clearly among them, despite the extremely limited number of examples disclosed in the literature. We explored the biomimetic synthesis of macrocarbocyclic natural products, chejuenolides A–C, as well as structural variants, to unveil the hidden stereochemical relationships between their biosynthesis and those of lankacidin antibiotics. This revealed the logic of the reaction pattern, which was likely influenced by catalytic promiscuity in nature.


#

Key words

antibiotic - biomimetic synthesis - chejuenolide - macrocyclization - Mannich

Biographical Sketches

Zoom Image

Kuan Zheng studied chemistry at Sichuan University. In 2018, he earned a Ph.D. at Shanghai Institute of Organic Chemistry (CAS) under the guidance of Prof. Ran Hong, focusing on the biomimetic synthesis of complex natural products. Currently he is working as a Research Associate in the same group. He was recently elected member of the Youth Innovation Promotion Association of Chinese Academy of Sciences. His research interests include natural products total synthesis and related synthetic methodology development.

Zoom Image

Bingbing Zhang was born in Anhui, China. He received his B.S. (2013) from Anhui Medical University and M.S. (2017) from Fudan University. He started his doctoral study in the Shanghai Institute of Organic Chemistry (SIOC) under the guidance of Prof. Ran Hong in 2019. His current research interests focus on the total synthesis of biologically active natural products.

Zoom Image

Ran Hong received his Ph.D. at Shanghai Institute of Organic Chemistry (Prof. Guo-Qiang Lin) and then had two postdoc stints in the laboratories of late Prof. Jay K. Kochi (University of Houston) and Prof. Li Deng (Brandeis University), respectively. Since 2007, his group at SIOC has been focused on the development of novel synthetic methods and biomimetic strategies toward complex natural products.

1

Introduction

Macrocyclic natural products have received increasing attention in drug discovery owing to their unique properties as chemical probes for biological applications.[1] Their densely functionalized macrocyclic frameworks have created great challenges for synthetic chemists and they are often resistant to comprehensive structure–activity relationship (SAR) analysis.[2] In general, the mainstream synthetic strategy is to construct the pivotal chiral centers in an acyclic precursor before performing the projected macrocyclization through C–O and C–N bond formations.[3] A less common approach relies on those reactions that can generate one or more stereogenic centers in a given macrocyclization step,[4] which has sometimes been illustrated as a crucial biosynthetic step toward numerous natural and nonnatural macrocyclic compounds.[5] Although many efficient successful total syntheses have been achieved by using the latter strategy, emulating the strategy of nature for selectively producing complex macrocycles can sometimes be challenging.[6]

We have a continuing interest in exploring novel cyclization methods to achieve pharmacologically active natural products, with a special focus on emulating the strategy of nature.[7] Lankacidin antibiotics, particularly the lactonic macrocyclic members isolated in the 1970s,[8] have attracted much interest from both academia and industry as a potential treatment for cancer[9] and multidrug-resistant bacterial infections.[10] In 2017, we expanded the macrocyclization reaction toolbox by developing an N,O-acetal-based biomimetic Mannich reaction to construct the lankacidin skeleton (Scheme [1]).[11] Nevertheless, a few questions remain unanswered. For example, will this novel macrocyclization apply to other stereo-variants? Additionally, will imperfect stereochemical control be an innocent consequence of biosynthesis? Herein, we present a further step in our investigation of the Mannich macrocyclization strategy for the biomimetic synthesis of chejuenolides.[12]

Chejuenolides A–C, first identified from a culture extract of a gram-negative marine microorganism Hahella chejuensis,[13] are 17-membered carbocyclic tetraenes. They structurally resemble lankacyclinol but possess an acetylamido side chain, different stereochemistries at several stereocenters (C13/C2/C18), as well as a C16–C17 double bond, i.e., Z-configuration for chejuenolide C. The absolute configurations of two allylic alcohols were assigned using the modified Mosher method, while the relative stereochemistries of the other C2/C18 were deduced from two-dimensional nuclear Overhauser effect (2D-NOE) data. Chejuenolides were evaluated to be moderately active as protein tyrosine phosphatase 1B (PTP1B) inhibitors in the initial biological activity study. Interestingly, unlike the biosynthetically related lankacidin antibiotic family, not a single naturally occurring δ-lactone-containing congener of the chejuenolide family has been reported to date. Biosynthetic studies by Kim and co-workers[14] led to the successful construction of a mutant strain, O3KO, in which an amine oxidase gene was disrupted. It was found to produce a linear polyketide compound, O3P2, possessing a β-keto-δ-lactone segment with fully undetermined stereochemistry.[14a] Our synthetic project toward chejuenolides was launched with the primary goal of unveiling the missing link between chejuenolin (5),[15] a postulated cyclization product of the nascent polyketide intermediate O3P2 (4), and both its decarboxylated downstream metabolites (1–3).

Zoom Image
Scheme 1 (A) Representative chemical structures of lankacidin antibiotics. (B) Chemical structures of chejuenolides A–C. (C) Comparison of the proposed biogenesis involving a cryptic lactone from O3P2 and our biomimetic approach.

# 2

Preparation of Azahexatriene: An Underdeveloped Object

In the lankacidin or chejuenolide skeleton construction process, it is postulated that an azahexatriene-type neutral imine or corresponding iminium species is involved. Imines possessing different substituents on the nitrogen atom have been recognized as valuable intermediates in organic synthesis.[16] Unlike 1-azabuta-1,3-dienes,[17] the synthetic scope of electron-deficient N-acyl or N-alkoxycarbonyl-substituted 1-azahexa-1,3,5-trienes is much less limited. Because of the pronounced chemical instability of azahexatrienes, few studies on their preparation and transformation have appeared in the literature (Scheme [2]).[18] For example, Fowler and Wyle reported the synthesis of dihydropyridines 9 from hydroxylamine derivatives 7 under flash vacuum pyrolysis conditions (FVP), in which a highly reactive N-acyl-azahexatriene 8 was proposed as an intermediate without isolation.[18a] The isolation and characterization of N-methoxycarbonyl-azahexatriene 11, generated by low-temperature photolytic ring-opening of homopyrrole 10, was described by Kumagai and co-workers.[18b] Notably, when the authors attempted to purify its methanol adducts on silica or neutral aluminum oxide, only decomposition product 14 was obtained. Despite the enamide formation, electron-deficient 1-azapolyene species have a strong propensity to electrocyclize under reaction conditions.[19] In other cases, they (i.e., 16) might be accessible through transition-metal-catalyzed carbene insertion of 15 and subsequent rearrangement reactions.[18c]

During the synthetic journey toward lankacidin and chejuenolide, the major hurdle has been the identification of a suitable imine precursor, which should be synthetically accessible and, more importantly, should release the reactive N-acyl-azahexatriene under relatively mild conditions to preserve the sensitive functional groups. The scenario is further exacerbated because the Mannich adducts from β-keto-δ-lactone are known to be fragile, even under mildly acidic or basic conditions.[20] Through an exhaustive screening process, we demonstrated the effectiveness of a demethoxylative Mannich reaction under thermal conditions using a long-chain N,O-acetal as the macrocyclization precursor.[21] This azahexatriene equivalent can be easily synthesized via Stille coupling between a vinyl iodide and a truncated vinylstannane fragment with a preinstalled lactamide appendage. Intermolecular model studies revealed that the azahexatriene was reversibly generated in low concentrations as a transient species;[21b] the reaction was driven to completion by the kinetic nucleophilic capture. Another beneficial feature of this reaction is that no additional activation reagent is required and the only coproduct is methanol, thus making the reaction system simple and neutral.

Zoom Image
Scheme 2 Reported approaches to prepare N-alkoxycarbonyl and N-acyl-1-azahexa-1,3,5-trienes

# 3

Developing a Modular Synthetic Route

Our synthesis commenced with the total synthesis and structural confirmation of the biosynthetic precursor O3P2. Following a route similar to those of other acyclic lankacidins, enal 18, with the opposite configuration at C13, was secured on a multigram scale. Subsequent E-selective olefination with sulfone 19 and reduction of the resultant ester produced the branching point, aldehyde 21, after which all four δ-lactonic moieties could be divergently introduced in a diastereoselective manner (Scheme [3]). We next employed the Stille reaction to join vinyl iodide 23a–d with N-acetylated stannane 24. After the separate global desilylation of each coupling product, four analogs 4a–d were obtained. Only the spectra of 4R,5R-isomer 4a matched perfectly with the published data,[14a] revealing that the stereochemistries at C4/C5 of the hidden chejuenolide biosynthetic intermediates were identical to those in the lankacidin series. This result is not surprising given the reported similarity in the configuration and arrangement of the domains that constitute each PKS module of chejuenolide and lankacidin.

Zoom Image
Scheme 3 Synthesis and stereochemical determination of the biosynthetic precursor O3P2

# 4

Taming the Reactivity of Azahexatriene

The complex azahexatriene was then incorporated into the projected macrocyclization to yield chejuenolides A and B. Although truncated vinyl stannane fragment 27, with an acetyl substitution at the amide nitrogen, can be smoothly obtained by following the same carboxoimidoate-based protocol previously applied in lankacidinol synthesis,[21b] its coupling reaction with vinyl iodide 26 under previously optimized conditions met with limited success. Replacement of the O-TBS-protected lactoyl group using a structurally simpler acetyl group resulted in a greater lability of the coupling product. Severe decomposition of the long-chain N,O-acetal 28 was inhibited when the Buchwald precatalyst Xphos-Pd-G2 and sodium bicarbonate were used (Scheme [4]).

Zoom Image
Scheme 4 Imine precursor synthesis, intramolecular macrocyclic Mannich reaction and synthesis, structural confirmation of chejuenolides A (1), B (2), and C (3), chejuenolin (5), and other stereoisomers.

Much to our delight, the presumptive intermediate N-acetyl-1-azahexatriene was readily formed under thermal conditions[22] to produce three macrocycles (29a/29b/29c = ~6:1:2). Treatment of the predominant product 29a and the least predominant product 29b with basic fluoride reagent led to a ‘one-pot’ desilylation and decarboxylation, producing chejuenolide A. The spectral data obtained for the synthetic material matched well with those of the natural product. X-ray diffraction analysis of the synthetic chejuenolide A confirmed the absolute configurations of both C2 and C18 as reported in the original isolation paper.[13a] Chejuenolide B, which differs from chejuenolide A only at the C2 position, could also be accessed with chejuenolide A via a two-step process in which decarboxylation was conducted at a lower temperature in methanolic potassium carbonate. This biosynthetically inspired reaction sequence on the C13-epimeric substrate successfully incorporates the required C18-stereochemistry, which is opposite to that used in the lankacidin biosynthesis. Hence, the non-decarboxylative desilylation of major cyclization product 29a might yield chejuenolin (5), a plausible, undiscovered enzymatic metabolite of O3P2. In our hands, however, the synthetic chejuenolin (5) was much less chemically stable than lankacidinol or lankacidin C, both in neat form and in solutions of common organic solvents, indicating that the chejuenolin and other lactone-containing congeners synthesized in our work would be too fragile to preserve using conventional purification methods. As in the previous case, chejuenolides A and B may not be artifacts; however, it remains unclear at this stage whether an encoded enzyme participates in the downstream transformation of chejuenolin in the biosynthesis of chejuenolides A and B. Characterization of chejuenolin as the O3P2-derived macrocyclic product will be useful in the development of stabilized analogs of chejuenolides.

Furthermore, we achieved the first total synthesis of chejuenolide C (3), which possesses an intriguing Z-configured double bond at the C16/C17 position. Conjugated polyenals and related iminium derivatives are prone to isomerize under a variety of conditions. The thermal conditions for biomimetic macrocyclization do not involve the addition of an exogenous acid, base, or other transition-metal catalyst, and theoretically, the only by-product is methanol. The contamination of the dihydropyridine by-product during the construction of the lankacidin framework revealed that trans-to-cis isomerization of the ΔC16–C17 alkene is possible under thermal conditions.[21b] Notably, the Mannich addition products derived from this isomerized azatriene system were not observed in the crude reaction mixture during our previous synthesis of lankacidins. Under the optimized cyclization conditions, N,O-acetal 32, with the C16/C17 Z-configured double bond, resulted in a substantial amount of the 6π-azaelectrocyclization product. However, intramolecular Mannich ring-closing product 33 was also formed, albeit in a low isolated yield of 22%, with the newly generated stereocenters 2R,18R, as confirmed using single-crystal X-ray analysis. Toward this end, 33 was exposed to another optimized basic condition followed by mild silyl removal, which successfully afforded chejuenolide C.

After accomplishing the syntheses of all reported chejuenolides, Liu and co-workers also identified these three natural products from a newly discovered H. chejuensis strain, NBU794, by applying genome mining and metabolomics.[23] Similar to the original isolation protocol,[13] the relative abundance of chejuenolide C in the fermented extracts was remarkably lower than the other two congeners (A and B). Some minor compounds sharing similar spectra of known chejuenolides were also mentioned in the original report, although their detailed structures were not disclosed because of their extremely limited quantities. One might expect the natural existence of several other C2/C18-diastereomeric chejuenolides (i.e., 2-epi-chejuenolide C, 2,18-bis-epi-­chejuenolide A), as in the case of our previous natural product discoveries in the lankacidin series.[21b] [c]


# 5

Stereochemistry in Macrocyclization: ­Insights for Biosynthesis

The degree of diastereoselection in the pivotal biomimetic macrocyclization was found to depend on the stereochemistry of the linear substrate, reaction media, and reaction temperature. Therefore, the experimental results described above raise an important question: which is the dominant stereo-controlling factor, the chiral lactoyl substitution on nitrogen or the C13 stereogenic center configuration? With this in mind, a hybrid cyclization precursor 38 bearing the identical O-TBS-(S)-lactoyl substitution was synthesized (Scheme [5]). The product distribution was similar to that of the macrocyclization reaction using N-acetyl substrate 28, indicating a negligible induction effect between the chiral lactoyl group and the acetyl group. Next, we prepared a C4/C5-bisepimeric hybrid. The C4/C5-epimer of chejuenolide-yielding precursor 41 was cyclized smoothly within 4 h. The prevalent macrocycle 42a in the latter crude reaction mixture was found to have 2R,18S-stereochemistry identical to that of chejuenolin (5), as indicated by the detailed 2D-NOE analysis and by its facile conversion into chejuenolide A.

Zoom Image
Scheme 5 Exploration of the stereo-controlling factors in the macrocyclic Mannich reaction. (A) Macrocyclization in the biomimetic synthesis of lankacidin; (B) macrocyclization in the biomimetic synthesis of chejuenolide; (C) macrocyclization of the hybrid cyclization precursor (38); and (D) macrocyclization of C4/C5-bisepimer of chejuenolide (41).

Considering the all-trans tetraenic lankacidinol and chejuenolide-ring-closing systems overall, an improved understanding of the stereo-controlling issues on the macrocyclization process has emerged. Among all four possible diastereomeric Mannich adducts, 2S,18R-macrocycle and 2R,18S-macrocycle always prevailed over 2S,18S-macrocycle or 2R,18R-macrocycle in the crude reaction mixture. Most importantly, the configuration at C13 has a crucial impact on the dominant stereochemistry of the product. For the 13R-configured lankacidinol series (36), the 2R,18S-adduct was the dominant product relative to the 2S,18R-adduct, whereas this ratio was dramatically reversed in the current 13S-configured chejuenolide series (i.e., 28 and 38). Currently, the authors could only take an empirical approach to rationalize how the remote C13-chiral center dictates both facial preferences of two reactive sites through attenuating macrocyclization constraints.[12] To minimize several 1,3-allylic strains associated with the bulky silyloxy substitute at C13 as well as the methyl-bearing conjugated C14-C17 diene system, the C13–H bond axis may have a preference to be co-planar with the C14-C18 azahexatrienic plane, resulting in a configurationally conservative unit that is responsible for the observed C18 stereochemistry. The stereochemistry of C2 may be attributed to the conformer of the lactonic part, in which a visually outside-orientated enol form (TS-I), being more capable of attenuating the macrocyclization constraints in the 13S-configured precursor, gives 29a, while an inside form TS-II yields 29b (Scheme [6]). However, the involvement of s-trans of the C17–C18 bond and s-cis of the C15–C16 bond in an alternative stereo-directing conformer may not be excluded at the current stage, and further investigations are required.

Zoom Image
Scheme 6 Stereochemical rationale in the macrocyclization of the linear precursor 28

# 6

Conclusions and Outlook

It can be difficult to realize a given transformation in the presence of complex and sensitive functional groups. Unexpected inertia of a functional group conversion, failure to obtain a high degree of stereo-control, or reluctant ring closure can be laborious in daily laboratory work, especially for reaction systems involving polycyclic or macrocyclic substrates. Conversely, the synthetic machinery of nature has developed the unique capability to create several wonderfully complex substances and has thus long been a source of inspiration for synthetic organic chemists.[24] [25]

At the beginning of our endeavor to mimic the enzymatic macrocycle-forming process in the lankacidin synthesis, the oxidative Mannich approach for emulating the dual function of monoamine oxidase[26] was abandoned because it was necessary to chemically protect the oxidation-sensitive β-keto lactone moiety. Instead, we anticipated that a cyclization precursor with a higher C18-oxidation state might be a more feasible surrogate. Meanwhile, to minimize the manipulation of additional functional groups on the nitrogen atom, the preinstallation of a lactoyl group on this linear imine precursor was also pursued. Before starting the synthetic campaign, the major concerns were (a) whether we could obtain the macrocyclization adduct with the desired flagship C2/C18-stereochemistry without enzymatic assistance, and (b) how to redesign the reaction system to reverse the intrinsic bias if it became necessary to correct an undesired stereochemical event. However, the finalization of 36 later proved to be a highly orchestrated and substrate-specific event, in which considerable time was required to find a solution. The acid- and base-sensitive nature of both the reactant and Mannich adduct severely limits the chemical reagent system. An exhaustive evaluation ultimately led to the development of a thermal demethoxylation methodology that was uniquely effective in this macrocyclic construction. Fortunately, as a special bonus for executing a biomimetic strategy, the stereochemical outcome of the Mannich reaction to produce either a lankacidin or a chejuenolide macrocycle was found to be quite satisfactory. Further investigation unveiled that every cyclized diastereomer could be utilized to access other congeners or some undiscovered natural products, although the perfect substrate or reagent-controlled selectivity was not successfully achieved in our hands. However, the imperfection of nature may leverage catalytic promiscuity as a compromise for the diversity of chemical space.[27] [28]

Herein, we have described an overview of our biomimetic endeavor toward the carbocyclic natural products, chejuenolides. A previously disclosed method to generate 1-acyl-1-azahexatriene, which belongs to a useful yet rarely pursued class of highly reactive intermediates, was further developed. Total synthesis enabled the stereochemical determination of O3P2, allowing the full structural assignments of chejuenolin, the presumed biogenetic parent of chejuenolides. In addition to shedding light on the previously underexplored biosynthetic relationships between lankacyclinol and chejuenolides, these results contributed to establishing a broad stereo-controlling landscape for the pivotal macrocyclization reaction, enabling rapid access to a variety of analogs with predictable C2/C18-diastereoselectivity. We hope that presenting these achievements will inspire further research to expand the chemistry toolbox in the process of executing biomimetic strategies.


#
#

Conflict of Interest

The authors declare no conflict of interest.


Corresponding Authors

Kuan Zheng
CAS Key Laboratory of Synthetic Chemistry of Natural Substances, Center for Excellence in Molecular Synthesis, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences
Shanghai 200032
P. R. of China   

Ran Hong
CAS Key Laboratory of Synthetic Chemistry of Natural Substances, Center for Excellence in Molecular Synthesis, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences
Shanghai 200032
P. R. of China   

Publication History

Received: 25 February 2023

Accepted after revision: 30 March 2023

Accepted Manuscript online:
30 March 2023

Article published online:
05 May 2023

© 2023. 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 (A) Representative chemical structures of lankacidin antibiotics. (B) Chemical structures of chejuenolides A–C. (C) Comparison of the proposed biogenesis involving a cryptic lactone from O3P2 and our biomimetic approach.
Zoom Image
Scheme 2 Reported approaches to prepare N-alkoxycarbonyl and N-acyl-1-azahexa-1,3,5-trienes
Zoom Image
Scheme 3 Synthesis and stereochemical determination of the biosynthetic precursor O3P2
Zoom Image
Scheme 4 Imine precursor synthesis, intramolecular macrocyclic Mannich reaction and synthesis, structural confirmation of chejuenolides A (1), B (2), and C (3), chejuenolin (5), and other stereoisomers.
Zoom Image
Scheme 5 Exploration of the stereo-controlling factors in the macrocyclic Mannich reaction. (A) Macrocyclization in the biomimetic synthesis of lankacidin; (B) macrocyclization in the biomimetic synthesis of chejuenolide; (C) macrocyclization of the hybrid cyclization precursor (38); and (D) macrocyclization of C4/C5-bisepimer of chejuenolide (41).
Zoom Image
Scheme 6 Stereochemical rationale in the macrocyclization of the linear precursor 28