Progress Toward the De Novo Asymmetric Synthesis of Euphanes

Abstract

Progress toward an asymmetric synthesis of euphanes is described. A C14-desmethyl euphane systempossessing five differentially substituted and electronically distinct alkenes has been prepared. The route employed is based on sequential metallacycle-mediated annulative cross-coupling, double asymmetric Brønsted acid-mediated intramolecular Friedel–Crafts alkylation, and an oxidative rearrangement to establish the requisite C10 quaternary center. These studies have also led to the discovery of a novel euphane-based modulator of the Liver X Receptor.

Graphical Abstract

Tetracyclic triterpenoids are a large and diverse class of stereochemically complex carbocyclic natural products that continue to stand as challenges for de novo asymmetric synthesis.¹ Perhaps most recognizable among such targets are those that possess the lanostane skeleton (Figure 1A),² as much attention has been directed toward understanding the biosynthesis of lanosterol and its conversion to cholesterol and steroid hormones.³ Classic studies in the area led to development of the Stork–Eschenmoser hypothesis^1c,4 and the establishment of an intellectual framework that has enabled the biomimetic synthesis of numerous terpenoid targets through stereocontrolled cation–olefin cyclization processes.^1c,5 Despite this rich history in organic chemistry, only a relatively small subset of tetracyclic triteprenoid natural products have yielded to asymmetric synthesis. The euphanes stand as a structurally interesting subclass,^2a,6 comprising a molecular skeleton of similar substitution to lanostanes albeit possessing unique stereochemistry. Most notably euphane natural products have distinct stereochemistry at the C13, C14 vicinal quaternary centers (the CD-ring fusion) and also different stereochemical relationships between C13, C17 and C20 that likely impact the preferred orientation of their C17 side chains in three-dimensional space (Figure 1A).

Figure 1.

Introduction to euphanes and a retrosynthetic strategy based on a sequence of alkoxide-directed metallacycle-mediated annulation, double asymmetric Friedel–Crafts cyclization and oxidative rearrangement.

Here, we describe efforts in chemical synthesis that were designed to address key elements of the unique stereodefined euphane system, particularly with respect to the relative and absolute stereochemistry at C10, C13, C17 and C20. As illustrated in Figure 1B, these studies targeted the construction of the tetracyclic polyunsaturated C14-desmethyl euphane-inspired intermedediate 1. Retrosynthetically, it was reasoned that this target might be accessible from phenol I through an oxidative dearomatization terminated by a group selective Wagner–Meerwein rearrangement.⁷ In turn, I was imagined as the product of a tandem Brønsted acid-mediated desilylation of the polyunsaturated hydrindane II and subsequent double asymmetric Freidel–Crafts cyclization to establish the C9 stereocenter.⁷ Finally, hydrindane II was targeted for synthesis from metallacycle-mediated annulative cross-coupling of enyne III with TMS-propyne.^7,8 While this coupling technology has been the topic of a number of recent studies,⁹ an enyne possessing the substitution and unsaturation of III has yet to be demonstrated to be compatible with this type of organometallic coupling process. In addition to describing the chemical synthesis activities that have resulted in the preparation of 1, these efforts have also resulted in the discovery that this C14-desmethyl euphane-like system is a structurally novel agonist of the Liver X Receptor (LXR). While LXR is known to be functionally modulated by a wide variety of ligand types,¹² it is notable that 1 has a unique stereodefined natural product-like structure that is structurally distinct from the typical cholestane skeleton found in numerous natural product and natural product-inspired LXR agonists.

As illustrated in Figure 2, the desired enyne (2) was prepared from the known allylic alcohol 3.¹³ Sharpless asymmetric epoxidation¹⁴ proceeded in 79% yield to deliver a stereodefined disubstituted epoxide with moderate levels of diastereoselection (dr = 7:1). This intermediate (not depicted) was then converted to the stereodefined monosubstituted epoxide 4 through a five-step sequence: (i) silylation of the primary alcohol by the action of TBSCl, (ii) nucleophilic addition of 2-lithiopropene mediated by CuCN, (iii) desilylation using TBAF, (iv) selective activation of the primary alcohol of the resulting diol as its corresponding tosylate, and (v) epoxide formation by treatment with K₂CO₃. This sequence of steps delivered epoxide 4 in 20% isolated yield over the last five steps of the sequence. Moving forward, coupling of epoxide 4 with alkyne 5¹⁵ proceeded uneventfully via initial conversion of 5 to its corresponding lithium acetylide and exposure to the combination of BF₃•OEt₂ and epoxide 4. This final step furnished the fully functionalized and polyunsaturated enyne coupling partner 2 in 51% isolated yield (70% based on recovered 4).

Figure 2.

Construction of enyne 2.

With a stereoselective route to the desired enyne coupling partner in hand, attention was then focused on the key ring-forming reactions presented in the retrosynthetic pathway previously discussed. As depicted in Figure 3, alkoxide-directed metallacycle-mediated [2+2+2] annulation between enyne 2 and TMS-propyne proved effective, delivering a highly functionalized hydrindane intermediate (B) presumably by way of the metallacyclopentadiene A that is thought to undergo chemo- and stereoselective intramolecular [4+2] cycloaddition with the 1,1-disubstituted alkene and chelotropic extrusion of the metal from the resulting organometallic intermediate.^8a–b From a practical perspective, the decision was made to move the crude hydrindane product of this annulation reaction (B) forward through the Brønsted acid-mediated tandem protodesilylation and double asymmetric Friedel–Crafts cyclization. As illustrated, exposure to the complex formed from (S)-BINOL and SnCl₄ resulted in regio- and stereoselective cyclization and delivered the tetracyclic species 6 in 44% isolated yield (no evidence was found for the production of a stereoisomeric product from this sequence of reactions). Surprisingly, but uneventfully, a small amount of the C24–C25 hydration product of 6 was also isolated from this two-step sequence (6’; 9% – not depicted).

Figure 3.

Coupling of enyne 2 with TMS-propyne and conversion to the euphane-inspired system 1.

Moving forward, we selected to employ what has become standard conditions in our laboratory for demethylation of the C3 methyl ether in numerous synthesis campaigns targeting tetracyclic terpenoid structures — DIBAL in PhMe at reflux.^7,15 Unfortunately, in the present case, this procedure for demethylation resulted in the generation of a significant amount of product where the trisubstituted alkene formerly present in the C17 side chain was reduced (7:8 = 1:1.5). While numerous other methods for demethylation are available that could avoid this undesired reduction of the side chain,¹⁶ the mixture of products from this reductive demethylation was advanced through the oxidative rearrangement reaction. As depicted at the bottom of Figure 3, the final two-step sequence proceeded with reasonable efficiency and delivered a mixture of dienone products (1 and 9) in 54% yield.

Compound 1 was separated from the overreduced species 9 and fully characterized. Notably, 1 possesses a variety of molecular features of potential interest in total synthesis campaigns: (1) it contains the desired stereodefined and unsaturated side chain at C17 that is found in scores of terpenoid natural products, (2) it has the anti- relative stereochemistry between the C10 and C13 quaternary centers that is typical of euphane and limonoid natural products, and (3) it possesses five distinct olefins that are expected to be capable of undergoing selective reactions with a wide variety of reagents. Regarding future efforts for accomplishing de novo asymmetric syntheses of euphanes based on the foundation of chemistry described here, the primary challenge remains the selective installation of the requisite quaternary center at C14. Future studies will be focused on this problem and will be reported in due course.

Notably, the C14-desmethyl euphane 1, while not a known natural product, bears some structural resemblance to numerous natural and semisynthetic modulators of the Liver X Receptor (LXR), as well as numerous sterol intermediates in the mammalian biosynthetic pathway to cholesterol (e.g., FF-MAS, T-MAS, and zymosterol, among others),³ despite having fundamentally distinct stereochemistry at C13 and C17, and possessing a C3 ketone rather than the typical β-OH.^12,17 LXR has been a target of great interest in the pharmaceutical industry for nearly twenty years, with selective agonism thought to be a viable means to achieve novel therapeutic interventions for a wide variety of indications that include atherosclerosis, inflammation and Alzheimer’s disease. Unfortunately, potent small molecule agonists of this receptor (not sterol in nature) typically result in triggering undesired cellular pathways that cause lipogenesis and hepatic steatosis.^10,11 As such, no pharmaceutical agent targeting LXR has been successfully advanced through clinical trials to approval.

Interestingly, while the tetracycles prepared here are structurally related to cholestane-based LXR agonists, their stereochemistries, and hence, three-dimensional shapes are distinct. These structural differences manifest from the unique C10-C13 “anti-” relative stereochemistry that alters the shape of the carbocyclic core skeleton, and the stereochemistry at C17 that is opposite to that encountered in cholestane-based LXR agonists. With curiosity leading the way, a number of the tetracyclic sterol-like agents synthesized were evaluated for their ability to modulate LXR.

The nuclear receptor LXR exists as two isoforms (LXRα [NR1H3] and LXRβ [NR1H2]). LXRβ is very widely expressed while LXRα is much more limited to tissues such as the liver. We assessed the ability of some of the compounds prepared in this effort to bind to the ligand binding domains (LBDs) of both LXRα and LXRβ using a radioligand binding assay with the tritiated synthetic non-steroidal LXR agonist ³H-T0901317¹⁰ (Figure 4A). As illustrated in Figure 4B, unlabeled T0901317 displayed the highest affinity with a K_i of 0.13 μM (LXRα) and 0.10 μM (LXRβ). 25-hydroxycholesterol (25-OHC), a naturally occurring steroidal LXR agonist,^11d displayed a K_i of 4.0 and 6.9 μM at LXRα and LXRβ, respectively. Notably, several of the novel tetracyclic compounds prepared here also displayed similar potency to 25-hydroxycholesterol with LXR potencies ranging from 2.1 to 13 μM (Figure 4B).

Figure 4.

Evaluation of synthetic tetracycles as ligands to LXRα and LXRβ, and functional evaluation of 1 on gene expression induced by selective agonism of LXR. A) Chemical structures of compounds evaluated for binding to LXRα and LXRβ. B) Results of radioligand binding assays for LXRα and LXRβ using 3H-T0901317 as a radioligand and compounds from (A) as cold competitors. Ki values are indicated in μM. C) Evaluation of the ability of 1 to modulate LXR target genes in the human THP-1 monocytic cell line – concentrations compounds tested: T0901317 (10 μM), 25-hydroxycholesterol (10 μM), 1 (0.01 μM, 0.1 μM, and 1 μM). ABCA1, ATP binding cassette subfamily A Member 1.

The most potent compound, tetracycle 1 (2.1 μM (LXRα) and 2.4 μM (LXRβ)), was assessed for cell based activity in the human monocytic THP-1 cell line for its ability to modulate expression of a well-characterized LXR target gene, ABCA1, which is a critical regulator of cholesterol efflux. As has been previously described, the non-steroidal LXR agonist T0901317 (at 10 μM) is an effective LXR agonist increasing the expression of ABCA1 (Figure 4C). As anticipated, the steroidal LXR agonist 25-hydroxycholesterol (25-OHC; at 10 μM) increased the expression of ABCA1 as well, albeit at lower levels than T0901317. Importantly, compound 1 also increased ABCA1 expression in a dose-dependent manner (administered at 0.01 μM, 0.1 μM, and 1 μM) consistent with this compound functioning as an agonist of LXR.

Overall, these studies describe progress toward establishing a de novo asymmetric synthesis pathway to euphanes. Efforts have resulted in demonstrating that metallacycle-mediated annulative cross-coupling is effective with an enyne substrate that contains a sterol-like unsaturated side chain. The highly functionalized hydrindane intermediate generated from the metallacycle-mediated annulative cross-coupling was successfully advanced by way of tandem Brønsted acid-mediated protodesilylation and double asymmetric Freidel–Crafts cyclization followed by oxidative dearomatization/Wagner–Meerwein rearrangement. Overall, this sequence of chemical transformations furnished the complex tetracyclic dienone 1 in just four steps from the enyne coupling partner 2. It is understood that lingering challenges associated with the use of this foundation of chemistry to access euphanes include installation of the C14 quaternary center, and accomplishing this goal is the focus of ongoing studies. Nevertheless, it was recognized that the stereochemically defined tetracyclic compounds produced in these studies have molecular features related to steroidal LXR agonists, albeit having subtly different three-dimensional shapes. Following up on this recognition, the C14-desmethyl euphanes produced in this study were studied as potential functional ligands of LXR. Notably, 1 emerged as the most potent synthetic natural product-like LXR agonist that, while possessing only modest potency, has similar properties to 25-hydroxycholesterol. To our knowledge, this is the first example of a selective modulator of LXR based on structural features of a euphane.