Robert E.
Arnold
a,
Jan
Saska
a,
Raquel
Mesquita-Ribeiro
b,
Federico
Dajas-Bailador
b,
Laurence
Taylor
a,
William
Lewis
a,
Stephen
Argent
a,
Huiling
Shao
c,
Kendall N.
Houk
*c and
Ross M.
Denton
*a
aThe GlaxoSmithKline Carbon Neutral Laboratories for Sustainable Chemistry, University of Nottingham, Jubilee Campus Triumph Road, Nottingham, NG7 2TU, UK. E-mail: ross.denton@nottingham.ac.uk
bSchool of Life Sciences, University of Nottingham, NG7 2UH, UK
cUniversity of California, Department of Chemistry and Biochemistry, 607 Charles E. Young Drive East, Box 951569, Los Angeles, CA 90095-1569, UK
First published on 19th June 2024
We report the first total syntheses of simonsol F (3), simonsinol (5), fargenin (4), and macranthol (6) in addition to syntheses of simonsol C (2), simonsol G (1), and honokiol (14). The syntheses are based upon a phosphonium ylide-mediated cascade reaction and upon natural product isomerization reactions which proceed through Cope rearrangements of putative biosynthetic dienone intermediates. As a corollary of the natural product isomerization reactions, we propose an alternative biosynthesis of honokiol (14), simonsinol (5), and macranthol (6) which unites the natural products in this family under a single common precursor, chavicol (7). Finally, we demonstrate that simonsol C (2) and simonsol F (3) promote axonal growth in primary mouse cortical neurons.
In contrast simonsinol (5) and macranthol (6) (Scheme 1B) are prima facie heterooligomers derived from chavicol (7) and the additional phenolic precursor (15) through a presumed oxidative coupling which generates the natural product honokiol (14).23–28 From this point macranthol (6) and simonsinol (5) can be reached following further oxidative coupling with chavicol (7) as indicated (Scheme 1B).
However, while chavicol (7) and its methylated counterpart estragole have been isolated from numerous plants, including basil, fennel, and star anise (Illicum verum),29 phenol 15 is very scarce indeed.30 Furthermore, honokiol (14) is typically isolated with the homodimer magnolol (10) suggesting a common origin from chavicol (7) which undermines the presumed oxidative coupling of 7 and 15 depicted in Scheme 1B.31
Herein we report total syntheses of simonsol C (2), simonsol F (3) and simonsol G (1) based upon the successful implementation of this general approach. Additional syntheses of simonsinol (5), fargenin (4), macranthol (6), and honokiol (14) through isomerization processes provide evidence for a new biosynthesis proposal which unites the neolignans in Fig. 1 under a single common precursor. Finally, using neuronal models, the potential connection between neolignans and galantamine alkaloids is explored and tetrahydrodibenzofuran-containing neolignans are shown to promote axonal growth.
Suzuki–Miyaura cross-coupling of this mixture with bromide 26 then gave biaryl 27 which was brominated to afford intermediate 28 from which acetal 29 was prepared by reaction with bromine and ethyl vinyl ether. A dilute solution of 29 was then treated with cesium fluoride, which resulted in silyl ether cleavage and spirocyclization to afford dienone 30, the structure of which was confirmed by X-ray analysis.
Having reached this stage we sought hydrolysis conditions that would convert 30 into the corresponding hemi-acetal 33 required for the cascade reaction (Scheme 1C). Following some experimentation, we found that treatment of 30 with aqueous hydrochloric acid gave 28% of the desired hemiacetal 33 along with enone 32 (6%) and inseparable ketone diastereomers 31 (50%) which had arisen through post hydrolysis oxa-1,4-addition reactions. While hemiacetal 33 was not obtained as the major product it was available following chromatography and so we proceeded to examine the planned cascade reaction. To this end 33 was treated with methylenetriphenylphosphorane and the desired tricyclic intermediate 34 was obtained in an excellent 90% isolated yield and the structure of this material was confirmed by X-ray crystallography (Scheme 2C).
Although our synthesis plan had now been validated through the successful conversion of 33 into 34, we had to face the problem that hemiacetal 33 was only available as the minor product from the hydrolysis of acetal 30. To develop a viable synthesis of simonsol C (2) a means of recycling or, ideally, progressing the other hydrolysis products 31 and 32 was required. Since these by-products had arisen through conjugate addition, we reasoned it might be possible to effect β-elimination reactions under the basic conditions of the cascade reaction and thereby develop a process under which the hydrolysis products converge. Fortunately, this proved to be the case, and when a mixture of 31, 32 and 33 were treated with the phosphonium ylide we were delighted to isolate product 34 in 57% yield. We believe that this convergent process occurs as shown in Scheme 1B. It is likely that dienone 33 and enone 32 converge to give aldehyde 36 through deprotonation and ring-opening in the case of 33 and β-elimination/ring-opening in the case of 32. Subsequent conjugate addition of 36 and methylenation then gives product 34 after work up. In the case of the inseparable ketone diastereoisomers 31, deprotonation results in ring-opening to give alkoxide 37 which then forms alkoxyphosphorane intermediate 38 (ref. 37) after alkoxide addition to the phosphonium salt. Elimination of this intermediate to generate 39 may then proceed by an E1cB mechanism resulting in the generation of methyldiphenylphosphine oxide and benzene.
With a viable cascade sequence now developed we examined the Suzuki–Miyaura coupling reaction to access simonsol C (2). As expected, this proved challenging and was complicated by: (a) the free hydroxyl group present within the boronic acid; (b) isomerization of the allyl groups; and (c) additional reactions of the natural product which occurred post cross-coupling (vide infra). After numerous experiments, in which a wide range of palladium pre-catalysts were examined, an efficient cross-coupling was realized using Buchwald's third generation palladacycle XPhos Pd G3 (ref. 38) which gave simonsol C (2) in a 72% isolated yield. The structure of our synthetic material was confirmed by X-ray crystallography (Scheme 2C).
Having now fully validated our synthesis strategy we next turned our attention to simonsol F (3) which, prior to our work, had not been synthesized (Scheme 3A). To this end biaryl 40 was prepared by Suzuki–Miyaura coupling of the appropriate aryl halide and a mixture of boronic acids 24 and 25 (see ESI†). The biaryl then underwent bromoacetalization to give 41 followed by desilylation and spirocyclization to give an inconsequential mixture of bromodienone diastereoisomers 42, whose structures were confirmed by X-ray crystallography (Scheme 3C). Unfortunately, progressing either the single dienone isomers or the mixture through the required acetal hydrolysis proved to be extremely challenging. A range of conditions were examined, all of which gave intractable mixtures of debrominated material. As a result, we were forced to explore Suzuki–Miyaura cross-coupling prior to hydrolysis of the cyclic acetal. In our initial experiments we established that separate diastereoisomers of 42 underwent efficient coupling with boronic acid ester 43 (not depicted) before developing a more efficient method for cross-coupling of the mixture (Scheme 3A) which gave 44 as a mixture of diastereoisomers in excellent yield.
We now had to develop hydrolysis conditions to obtain hemiacetal 45 for deployment in the cascade reaction. In this instance, when mixture 45 was heated with aqueous hydrochloric acid under the conditions used for the synthesis of simonsol C (2) a complex mixture of products was obtained. However, acid-catalyzed hydrolysis using camphorsulfonic acid gave hemiacetal 45 as the major product in an acceptable yield of 66%. Treatment of this mixture with methylenetriphenylphosphorane again proceeded smoothly to give methyl-protected simonsol F (47) establishing this sequence as a general method for the construction of functionalized tetrahydrodibenzofurans. In this iteration of the cascade reaction, the presumed dienone phenolate intermediate 46, or its methylenated equivalent, can undergo conjugate 1,4-addition at two sites (Scheme 3B). Experimentally, formation of simonsol F (3) via intermediate 47 was observed and QM calculations suggested that the site selectivity is thermodynamically controlled. The alternative conjugate addition products 49 and 50 are 2.3 and 5.4 kcal mol−1 higher in energy respectively than 47. This is consistent with nucleophilic attack on the more electron-deficient enone, not substituted by the electron-rich aromatic substituent. This reaction also preserves the conjugation of this substituent to the double bond, stabilizing the product.
As with simonsol C (2) the final step of the synthesis proved to be challenging as a result of the sensitive allyl groups and the reactivity of the natural product post deprotection. After considerable experimentation we identified demethylation conditions that involved treatment of 48 with BCl3 at 0 °C for 5.5 hours which gave simonsol F (3) in an excellent isolated yield of 85% (Scheme 3A). The structure of the natural product was confirmed by X-ray crystallography (Scheme 3C).
As depicted in Scheme 3A the formation of these products can be rationalized by invoking retro conjugate addition to generate dienone 11. Fargenin (4) is then obtained through 1,2-addition of the phenolic hydroxyl group followed by dehydration, as indicated by the blue label on intermediate 11. An alternative oxidative 1,4-addition reaction (pink-labelled carbon) gives rise to benzofuran 48, and a Cope rearrangement (green label) is responsible for the generation of macranthol (6). The Cope rearrangement is of potential biosynthetic significance as it provides a hitherto unknown pathway to macranthol (6) which we had previously assumed to be a heterooligomer originating from the oxidative coupling of chavicol (7) and 2-allylphenol (15) (Scheme 1B).
Given these observations we next sought to effect isomerization of other homooligomeric natural products into their presumed heterooligomeric aromatic counterparts (Scheme 4A). In agreement with the results described above we discovered that simonsol G (1), which was prepared using our cascade strategy (see ESI† for the synthesis), could be converted into honokiol (18) under mild basic conditions. Similarly, simonsol C (2) and simonsol F (3) were converted into simonsinol (5) and macranthol (6) respectively. In the case of simoninsol (5), permethylation to give ether 52 was carried out for comparison with the literature compound derived from natural simonsinol (5).
To gain further insight into the Cope rearrangements we conducted DFT calculations to study the isomerization process that converts simonsol G (1) into honokiol (14). In the presence of base, this likely occurs through the generation of an equilibrium concentration of enolate 53 which could proceed through either Path A or Path B (Scheme 4B), depending on the order of the retro conjugate addition and Cope rearrangement. Path A, as invoked in Scheme 4A, involves a retro-conjugate addition of simonsol G enolate 53 to form the dienone intermediate 54, which undergoes Cope rearrangement and tautomerization to form anionic honokiol 57. Alternatively, along Path B, Cope rearrangement of 53 forms the dienolate intermediate 55, which then undergoes retro-conjugate addition and tautomerization to form 56. For completeness, we also investigated the corresponding neutral Cope rearrangements leading to honokiol (14) and macranthol (6) (see ESI†). Our computed reaction coordinate profile of Path A suggests that the formation of the dienone intermediate 54 from the simonsol G enolate 53 is exergonic by 8.8 kcal mol−1 (Scheme 4B). The subsequent Cope rearrangement transition state (Cope-TS1) has a relatively low free energy barrier of 25.5 kcal mol−1, and formation of the rearrangement product 56 is exergonic by 1.9 kcal mol−1. Subsequent tautomerization forms the honokinol phenolate 57, which is exogonic by 18.6 kcal mol−1.
Alternatively, Path B (Scheme 4B) via Cope-TS2 has a high free energy barrier of 35.9 kcal mol−1, which is kinetically unlikely at the reaction temperature (85 °C). Both Cope transition states adopt the favored chair-like conformation (Scheme 4B). As a result of the cross conjugated nature of dienone 54 the Cope-TS1 is favored over Cope-TS2 by 19.2 kcal mol−1. Based on these calculations we conclude that the natural product conversions take place through the proposed biosynthetic dienones 51, 13 and 11, or corresponding anions depending on the reaction conditions (Scheme 4A).
Alternatively, honokiol (14), simonsinol (5) and macranthol (6) could arise from the isomerization processes that we have demonstrated experimentally (Scheme 4A) with the formation of an additional aromatic ring as the thermodynamic driving force. Although the calculated barrier of the Cope rearrangement of 54 is relatively high (t1/2 6 days at 25 °C), it is possible that this could be accelerated through enzyme catalysis. Examples of enzymes catalyzing pericyclic reactions are rare;39 however, there is evidence to support biosynthetic Cope rearrangements in alkaloid biosynthesis40–42 and there are biogenetically inspired syntheses of other Illicium natural products which feature Claisen and Cope rearrangements.43
The fact that both magnolol (10), which is contained within simonsol C (2), and honokiol (14) have been shown to possess neuroprotective and restorative effects44–46 lends credence to this notion. As does the additional observations that a variety of structurally related neolignans and sesquiterpene neoligans exhibit neurite outgrowth properties.14–19,47,48 Given that axonal functional loss and degeneration is known to occur at the early stages of neurodegenerative diseases which precedes cell death and the onset of clinical symptoms49 we opted to screen compounds in an axonal growth assay. We began by examining magnolol (10), simonsinol (5), simsonol C (2), and simonsol F (3) along with selected synthetic intermediates 58, 59 and 25 which contained phenolic, phenylpropenoid and ether moieties (Scheme 5A). The compounds were applied as DMSO solutions to primary embryonic mouse cortical neurons that had been cultured for two days on poly-L-ornithine coated glass coverslips. Following a forty-eight-hour incubation, neurons were fixed and immunostained for the axonally-enriched acetylated tubulin to reveal overall neuron morphology. Each coverslip was imaged at 20 random points and the axonal length of neurons was measured computationally (Scheme 5B).
This study established that magnolol (10), simonsinol (5), simonsol F (3) and simonsol C (2) increased the length of mouse primary cortical axons at 1 μM. Furthermore, simonsol F (3) and simonsol C (2) promoted axonal growth at a level above that of the dimeric and trimeric aromatic natural products magnolol (10) and simonsinol (5) (Scheme 5B). Therefore, the galantamine alkaloid-like tetrahydrodibenzofuran core is biologically relevant in the promotion of axonal growth.
To gain further insight, we prepared a range of novel analogs (see SI for synthesis) in order to further probe the role of the tetrahydrodibenzofuran core (Scheme 5C). We began by examining simonsol F (3) and its reduced counterpart 60 at a concentration of 1 μM (Scheme 5D). The allylic alcohol 60 did not promote axonal growth, which illustrates the requirement for the enone within the tetrahydrodibenzofuran ring system. A second comparison between simonsol C-(2) and derivatives 61, 62 and 63 was then made (Scheme 5C and D). These results show that, in the case of simonsol C (2), a free phenol and the enone are required for the significant axonal growth which is observed. Taken together these data support Banwell's suggestion that simonsol C (2), and analogs thereof, may be promising starting points for the development of novel small molecules of therapeutic relevance in neurodevelopmental and neurodegenerative diseases.
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2327688–2327695. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4sc03232b |
This journal is © The Royal Society of Chemistry 2024 |