Synthetic nat- or ent-Steroids in as Few as Five Chemical Steps: An enantiospecific, convergent and flexible route from epichlorohydrin

Abstract

Steroids are of profound importance in human biology and medicine. Today, >100 FDA-approved steroidal agents are prescribed daily for indications that include heart failure, inflammation, pain, and cancer, among others. The scientific foundation of this now rich history of clinical success is in chemical synthesis, where advances provided a means to access meaningful quantities of these complex carbocyclic species. While triumphs in organic chemistry have, without question, enabled the establishment and sustained growth of the steroid pharmaceutical industry, production of highly functionalized synthetic steroids of widely varying substitution and stereochemistry (including unnatural ent-steroids) remains challenging, despite the numerous reports of elegant strategies for their de novo synthesis. In other words, broad exploration of >50% of this pharmaceutically privileged area of ‘chemical space’ remains challenging, labor intensive, and costly – potential factors that have contributed to the current reality that 100% of FDA-approved steroidal agents are of the natural antipode (absolute stereochemistry). Here, we describe an advance in chemical synthesis that has established an enantiospecific means to access novel steroidal compositions of matter with unprecedented facility and flexibility through the sequential use of two powerful ring-forming reactions: a modern metallacycle-mediated annulative cross-coupling and a new acid-catalyzed vinylcyclopropane rearrangement cascade. In addition to demonstrating the usefulness of this chemistry for accessing synthetic steroids of either enantiomeric series (nat- or ent-), steroidal products from this synthesis pathway have been selectively functionalized at sites within each of the four carbocyclic rings, a synthetic ent-steroid has been prepared on multi-gram scale, the enantiomer of a selective estrogen has been prepared, and a novel ent-steroid with growth inhibitory properties in three cancer cell lines has been discovered [GI₅₀ = 0.32 – 1.07 μg/mL (1.2 – 4 μM) in MDA-MB-231 (breast), U2OS (osteosarcoma), and AsPC-1 (pancreatic) human cancer cell lines].

Following the first correct structure elucidation of a steroid and the synthesis of equilenin in 1932 and 1939 (Fig. 1a,b), respectively, much attention has been given to developing chemistry capable of producing natural and synthetic steroids^1,2. Efforts in this area established a scientific foundation that has delivered >100 FDA-approved steroidal agents as therapeutics, leading to their current status as arguably the most well-studied and successful class of natural product-inspired pharmaceuticals³. While steroids have had a transformative impact on medicine and society, playing vital roles as oral contraceptives and treatments for cancer, heart failure, inflammation, pain, and traumatic brain injury, among others, there exist substantial barriers in preparative organic chemistry that continue to constrain our ability to freely explore the medicinal value of novel synthetic steroidal compositions of matter. Here, we describe a new approach to steroid synthesis that allows for preparation of partially aromatic steroids in as few as five chemical steps from a readily available chiral starting material (epichlorohydrin). The pathway is uniquely capable of accessing structurally diverse steroids in a straightforward fashion (housing modification in each of the four steroidal rings), proceeds in an enantiospecific fashion, and can be used to prepare multigram quantities of nat- or ent-steroidal compositions of matter with ease. This new steroid synthesis has been realized through capitalizing on the power of a modern metallacycle-mediated annulative cross-coupling reaction and a new acid-promoted vinylcyclopropane rearrangement cascade.

Figure 1.

a, The general steroid tetracyclic skeleton. b, Examples of natural steroids. c, Selected approaches to steroid synthesis using semisynthetic and de novo strategies.

Early scientific breakthroughs that established the steroid pharmaceutical industry were based on semisynthesis, where readily available natural steroids were chemically transformed to high-value therapeutic agents. Historical examples include Merck’s bile acid-to-cortisone process, and the Marker degradation (Fig. 1c) – an achievement that cleared the first synthetic pathway from a plant-derived steroid (diosgenin) to testosterone, estrone, estradiol, progesterone, and eventually cortisone^4–6. While continuing to play a valuable role in medicinal pursuits⁷, semisynthesis is often difficult to employ as a strategy to gain access to collections of highly oxygenated/functionalized steroidal targets, and is wholly unsuitable to explore the broad medicinal value of ‘ent-steroids’ (defined by an unnatural absolute stereochemistry of the tetracycle). This latter point deserves further comment, as decades of scientific triumph have positioned steroidal compositions of matter as ‘privileged’ in the pharmaceutical industry – a characterization that is based, in part, on physical properties that are critically important in medicinal chemistry (i.e. ‘drug like’ properties). While pairs of enantiomers share identical physical properties, and in the case of steroids ‘drug-like’ properties, it is surprising that 100% of FDA-approved steroidal drugs are of the natural (‘nat’) enantiomeric series. It is a growing sentiment that ent-steroids define fertile ground for discovery in biomedical science, boasting complementary three dimensional structures of potential relevance for targeting biology that is distinct from the nat-series^8–12, and retaining the extremely valuable drug-like properties of the parent class. In other words, ent-steroids are a privileged class of drug-like molecules that remain grossly underexplored in biology and medicine because they are not readily available from natural sources or chemical synthesis.

While important contributions in organic chemistry have established de novo synthesis pathways to synthetic steroids from non-steroidal starting materials, notably the Smith–Torgov synthesis of estranes¹³ and biomimetic cation–olefin cyclization¹⁴ processes (Fig. 1c), among others^15–21, semisynthesis remains the primary means by which pharmaceutically relevant steroids are prepared^7,22. This fact is humbling and reinforces the reality that substantial challenges persist with the de novo synthesis of complex carbocyclic structures, and that barriers continue to obstruct a more free exploration of this pharmaceutically privileged area of chemical space.

To address this gap at the interface of chemistry and medicine, efforts were focused on developing a conceptually unique approach to de novo steroid synthesis that would be: (1) convergent, to allow for ease in diversifying structure by changing the nature of coupling partners, (2) stereochemically flexible, (3) step economical²³ (proceeding in a small number of chemical steps), (4) enantiospecific, allowing access to either nat- or ent-steroidal systems with equal facility, and (5) capable of providing steroidal tetracycles that are well suited for a variety of subsequent functionalization processes (i.e. manipulation of functionality in each ring of tetracyclic nucleus).

As illustrated in Figure 2, it was thought that late-stage establishment of a steroidal tetracycle (1) may be possible through strategic formation of the C5–C6 bond by functionalization and activation of a tricyclic ACD-ring-containing system (2). Compound 2 would then surface as an intermediate of high value, and one that was recognized to be accessible from metallacycle-mediated annulative cross-coupling between a suitably functionalized alkyne (3) and a chiral enyne (4)²⁴. While not yet widely adopted in organic synthesis, metallacycle-mediated cross-coupling is an area of great recent growth^25–27. A variety of new convergent C–C bond-forming reactions that realize unique retrosynthetic relationships in organic chemistry have been described within this broad class of chemical reactivity, and applications in the context of complex molecule synthesis continue to surface²⁸. Here, a metallacycle-mediated annulative cross-coupling reaction was viewed as being particularly powerful, as it could provide a functionalized hydrindane (2) of great value for steroid synthesis from the stepwise coupling of simple starting materials.

Figure 2.

Retrosynthetic strategy for a convergent, flexible, and enantiospecific synthesis of partially aromatic (C19-nor) steroids by metallacycle-mediated annulative cross-coupling and late-stage formation of the steroidal B-ring.

The reduction of this chemical strategy to practice is depicted in Figure 3a. With initial focus on the nat-steroid series, conversion of epichlorohydrin (5) to the functionalized enyne 6, followed by metallacycle-mediated annulative cross-coupling with TMS-phenylacetylene 7 delivered hydrindane 8. This coupling process is unique in organic chemistry and particularly powerful here, delivering an angularly substituted trans-fused hydrindane in a convergent manner from acyclic precursors, and establishing three σ_C–C bonds and two stereocenters (one of which is quaternary). Our working mechanistic hypothesis for this annulation reaction is summarized in Fig. 3b, and includes: i) alkoxide-directed metallacycle-mediated cross-coupling to generate (A) (rs ≥ 20:1)²⁹, ii) stereoselective intramolecular [4+2] cycloaddition to furnish a bridged polycyclic metallacyclopentene (C) (ds ≥ 20:1), iii) elimination to provide a tertiary allylic metal species (D), iv) isomerization to the primary allylic organometallic (E), and v) regio- and stereoselective protonation to deliver hydrindane 8.

Figure 3.

Establishment of a new chemical pathway for the enantiospecific synthesis of steroids. a, Establishment of a synthesis pathway from optically active epichlorohydrin to an enantiodefined estrane by way of a metallacycle-mediated annulative cross-coupling and an acid-mediated vinylcyclopropane rearrangement cascade. b, Proposed mechanism for the metallacycle-mediated annulative cross-coupling reaction.

Moving forward, a “functionalization” protocol was sought that could install the steroidal C6 carbon in concert with revealing a reactive intermediate for B-ring formation. Cyclopropanation was selected as the means to accomplish such a process and, as depicted in Fig. 3a, can be employed to selectively engage the 1,1-disubstituted alkene of 8 en route to the vinylcyclopropane intermediate 10^30,31. While efficiency in this initial attempt at cyclopropanation was rather low (41% isolated yield), subsequent studies have concluded that higher yields are routinely encountered if protection of the secondary alcohol is conducted prior to cyclopropanation (vide infra). B-ring formation by way of the vinylcyclopropane 10 was initially thought possible from ionization and electrocyclic ring opening of the resulting cyclopropyl cation, followed by intramolecular Friedel–Crafts alkylation^32–35. Unfortuantely, all attempts to accomplish such a transformation were unsuccessful, ultimately leading to the triene 11, presumably through pentadienyl cation intermediate I. In an attempt to achieve cyclization without proceeding through this unstable cationic intermediate, it was found that treatment of 10 with TiCl₄ in nitromethane resulted in the formation of steroidal product 12 in 68% yield. Here, reaction of TiCl₄ with the D-ring hydroxy group (or with adventitious water) is thought to produce protic acid in situ and convert 10 to the reactive homoallylic cation intermediate II, presumably through a process that includes protodesilylation, protonation of the resulting styrenyl alkene, and regioselective cyclopropane fragmentation^36–39. Ring closure would then be possible by engaging the phenyl ring in an intramolecular Friedel–Crafts alkylation to deliver an intermediate that is transformed to 12 through loss of HBr^40,41.

As illustrated in Table 1, this chemical pathway is useful for generating a wide range of steroidal systems in an exceptionally concise and enantiospecific fashion. First, to demonstrate the facility with which this chemistry delivers ent-steroids, ent-6 (derived from (+)-epichlorohydrin) was smoothly converted to ent-12 (entry 1). As experienced with our first attempt, depicted earlier in Figure 3A, the efficiency of the two-step ring-closing process (ent-8 → ent-12) was fairly low due to the challenge of accomplishing cyclopropanation in the presence of the C16-OH. Entry 2 reveals that the efficiency of the overall process is substantially greater if the C16 hydroxy group is silylated prior to cyclopropanation and, in combination with entry 3, demonstrates the ease with which A-ring aromatic steroids containing varied oxygenation patterns can be preapared (15 and 18). Interestingly, while proceeding uneventfully in these two examples, it was later found that the acid-mediated vinylcyclopropane rearrangement can be more complex.

Table 1.

^ayield reported is for the combination of diene iosmers (trans + endo).

^bstereoselectivity for the annulation process (trans:cis) is typically ~6:1

^cyield reported is for the 2–3 step sequence.

^dcyclization sequence was conducted without protection of the C16 hdyroxy group.

^eregioselectivity based on ¹H NMR of the crude reaction product.

^fyield for this unselective process was not determined.

^gstructure confirmed by X-ray diffraction - See Supporting Information.

As illustrated in entry 4, use of the p-chloro-substituted phenylacetylene 19 led ultimately to the discovery that the final ring closure can proceed with additional rearrangement. Here, steroidal products 21a and 21b were produced in roughly equal proportions – an observation that is consistent with the mechanistic proposal depicted in Figure 4, where cyclization may proceed either through direct Friedel–Crafts alkylation (II → III → F), or by initial formation of a spirocyclic intermediate and subsequent rearrangement with selective migration of C9 (II → IV → G).

Figure 4.

Mechanistic divergence in the vinylcyclopropane rearrangement and intramolecular Friedel–Crafts alkylation process.

As illustrated in entry 5, it is possible to achieve high levels of selectivity in favor of ring closure by rearrangement. Here, the p-methoxy-substituent of 23 plays a signficant role in biasing the course of reaction, presumably due to its electronic contribution to favoring the formation of a spirocyclic intermediate akin to IV (Figure 4). Here, production of the C2-methoxy-substituted steroid 24 was found to proceed with very high levels of seletivity (rs ≥ 20:1). Interestingly, as depicted in entry 6, further modification of A-ring substitution (26) was found to restore a preference for cyclization without rearrangement. Here, cyclization delivered the C3-methoxy-substituted steroid 27 as a single regioisomer in a modest 31% yield (over three steps) – an observation that is consistent with steric effects stemming from the chloride substituents that may dissuade oxonium ion formation in the spirocyclic intermediate (i.e. an oxonium ion intermediate would be destabilized by significant 1,5-interactions between the methyl group of the oxonium ion and one of the ortho chlorine substitutents). Additional examples that illustrate the facility with which this steroid synthesis provides access to products of varying A-ring structure and substitution are depicted in entries 7 and 8.

This enantiospecific entry to steroids can be used to gain access to a range of systems that boast additional substitution and varying stereochemistry (Figure 5a). In products 34 and 35, the C13 quaternary center at the junction of rings C and D was altered simply by changing the Grignard reagent used to prepare the initial enyne for annulation (R¹ = Et or Bn). In product 36, the stereochemistry at C14 was altered by using a variant of the metallacycle-mediated annulation process that furnishes the cis-fused isomer⁴². Finally, simple functional group manipulations can be used to gain access to steroidal compositions of matter possessing varied structure within each of the four rings of the tetracycle (37–41) – including a remarkably stable naphthoquinone methide (39)^43–50. Finally, as depicted in Figure 5b, this synthesis pathway is capable of producing multigram quantities of steroidal products with ease. Here, 4.0 g of ent-12 was prepared in just five steps from epoxide 42 with an overall isolated yield of 20%.

Figure 5.

a. Access to steroids with varying substitution and stereochemistry (see Supporting Information for experimental details associated with the syntheses of 34–41). b. Multigram-scale preparation of a synthetic ent-steroid. c. Preparation of ent-estra-1,3,5(10),6,8-pentaene-3,16α-diol. d. Discovery of an ent-steroid with cytotoxic properties: Cells were plated at 1000 cells/well of a 96 well plate. The following day, compound 39 was added in 2-fold dilutions (8 wells/concentration). After 7 days growth, cells were lyzed and analyzed for total DNA content as previously described⁵².

The synthetic pathway to ent-steroids that has been established here is useful for the preparation of antipodes medicinally relevant agents. For example, 16-hydroxyestratrienes have been identified as synthetic estrogens that have a dissociation in favor of their estrogenic action on bone rather than the uterus (Figure 5c)⁵¹. While estra-1,3,5(10),6,8-pentaene-3,16α-diol is a representative member of this class, its enantiomer has never been described. Because the eanantiomer of estradiol is known to have neurological activity of potential value for the treatment of traumatic brain injury, and lacks activity as an estrogenic compound, ent-estranes are beginning to emerge as a broader class with potentially useful “non-steroidal” pharmaceutical properties. Here, it is demonstrated that one of the products from this synthesis pathway (27) can be easily transformed to ent-estra-1,3,5(10),6,8-pentaene-3,16α-diol (43). We look forward to future efforts that explore the neurological and estrogenic activities of this and related synthetic ent-16-hydroxyestranes.

Finally, a novel synthetic ent-steroid that was prepared in these studies (39) induced growth inhibition in three cancer cell lines with 50% inhibitory concentrations of 1.2 – 4 μM (0.32 – 1.07 μg/mL) (Figure 5d).

In conclusion, steroids are unquestionably the most appreciated and well-developed class of natural product-inspired pharmaceutical agents. While originally pursued as treatments for inflammation and fertility, they continue to define a rich platform for drug discovery and development across a diverse therapeutic landscape. However, the greater potential of steroids as privileged drug-like skeletons has been, and will continue to be, limited by the chemistry that is available to access their fused polycyclic skeletons. While accomplishments in drug development have been primarily fueled by semisynthesis, a dependence on this chemical strategy will continue to greatly constrain our ability to take advantage of the favorable drug-like properties of diverse steroidal compositions across a broad therapeutic landscape. Chemical strategies based on semisynthesis are inherently limited in scope, proceeding from a subset of abundant natural steroidal starting materials of only a single antipode (chirality) and typically requiring significant effort to install or remove functionality through uneconomical means. De novo synthesis of steroids has a rich history in organic chemistry, yet crowning achievements in this area do not routinely surface as enabling technology in discovery or development. Such a reality is not uncommon in natural product-inspired drug discovery, and attention to this gap between accomplishments in total synthesis and the persistent difficulties associated with discovery and development of natural product-inspired drugs, will continue to be an important consideration in the ongoing evolution of synthetic organic chemistry⁵³. Here, a step-economical (concise), flexible, convergent, and enantiospecific synthesis of partially aromatic synthetic steroids has been established from a readily available chiral starting material (epichlorohydrin). The route proceeds by way of a complex metallacycle-mediated annulative cross-coupling reaction and a new acid-catalyzed vinylcyclopropane rearrangement cascade, and can be employed to gain access to novel steroidal compositions of varying stereochemistry and substitution. Given the value of steroidal structures as platforms for drug discovery, and the growing appreciation of the potential therapeutic value of ent-steroids, this chemical advance can play an important, even enabling, role in the future identification of novel and biomedically relevant steroidal compositions of matter.

Methods

A representative set of experimental procedures is provided below for the multigram synthesis of ent-12.

To a stirring solution of epoxide 42 (14.8 g, 0.079 mol, 1 equiv) in 300 mL THF under N₂ atmosphere at −78 °C was added CuI (3.0 g, 0.016 mol, 0.2 equiv) followed by 2-propenylmagnesium bromide (189 mL, 0.094 mol, 0.5 M, 1.2 equiv). The resulting yellow suspension was stirred for 1 h at −78 °C, warmed to rt, and then stirred until the reaction was judged to be complete by TLC analysis. The reaction was quenched by adding 200 mL sat. NH₄Cl (aq) to the reaction mixture. The organic layer was separated, and the aqueous layer was extracted with EtOAc (200 mL × 3). The combined organic layers were dried over anhydrous MgSO₄, filtered, and the solvents were removed in vacuo. Purification of the crude product by SiO₂ flash column chromatography (20% EtOAc/80% Hexane) afforded 14.8 g of the enyne ent-6 as yellow oil (81%).

To a stirring solution of alkyne 7 (30 g 0.17 mol, 2.7 equiv) in 1.2 L of dry toluene at rt under N₂ atmosphere was added Ti(Oi-Pr)₄ (49.1 g, 0.17 mmol 2.7 equiv). The resulting mixture was cooled to −78 °C, and n-BuLi (2.7 M, 135 mL, 0.364 mol, 5.7 equiv) was added dropwise. The resulting black mixture was warmed first to rt, then heated to 50 °C and stirred at that temperature for 1 h (no reflux condenser required). In a separate flask under N₂ atmosphere, enyne ent-6 (14.7 g, 0.064 mol, 1.0 equiv) was dissolved in 300 mL of dry toluene, cooled to −78 °C, and treated with n-BuLi (2.5 M, 25 mL, 0.064 mmol, 1.0 equiv); added dropwise at −78 °C. The resulting mixture was warmed to rt, and then transferred by cannula to the black Ti-alkyne complex at −78 °C. The mixture was slowly warmed to rt overnight (approx. 17 h). After this period, 1.5 L of dry MeOH was cooled to −78 °C in a separate flask under N₂ atmosphere, and the reaction mixture was transferred by cannula to the pre-cooled MeOH. Once the addition was complete, the reaction mixture was warmed to rt, and 500 mL of DI H₂O was added. 500 mL of EtOAc was added, the organic phase was separated, and the aqueous layer was extracted with EtOAc (500 mL × 3). The combined organic layers were dried over MgSO₄, the drying agent was removed by filtration, and the resulting solution was concentrated in vacuo. Purification of the crude product by flash column chromatography afforded 12.4 g of the title hydrindane ent-8 as yellow film (60%, isolated as 4:1 mixture of ent-8: ent-9).

The following three-step procedure was used to convert 9.9 g of the trans-fused hydrindane ent-8 to the steroidal product ent-12 – a process that proceeded with an overall 40% isolated yield. This yield is based on the amount of hydrindane ent-8 present in a 4:1 mixture with the unreactive “endo” diene isomer: To a solution of 12.4 g of the 4:1 mixture [containing 9.9 g of hydrindane ent-8 (0.030 mol, 1.0 equiv) and its corresponding endo diene isomer ent-9 (2.5 g, 0.008 mol)] in 500 mL THF was added TBSCl (11.4 g, 0.076 mol, 2.5 equiv) and imidazole (5.2 g, 0.076 mmol, 2.5 equiv). The reaction mixture was stirred at rt under N₂ atmosphere overnight (approx. 17 h), then partitioned between a solution of 200 mL sat. NaHCO₃(aq) and 200 mL ethyl acetate. The organic layer was separated, and the aqueous layer was extracted with ethyl acetate (200 mL × 3). The combined organic layers were washed with brine and dried over anhydrous MgSO₄. The resulting suspension was filtered through a course glass fritted funnel to remove solid material, and the filtrate was concentrated in vacuo. The resulting residue was filtered through a pad of silica gel using a hexanes and ethyl acetate solvent mixture as eluent (hexanes/EtOAc = 80/1). Concentration of the resulting dilute solution afforded 15.8 g of a yellow oil. This crude product was used in the subsequent step without further purification.

To a solution of the above crude product (15.8 g) and triethylbenzylammonium chloride (1.7 g, 7.3 mmol) in 73 mL CHBr₃ at rt was added KOH (24.6 g, 0.44 mol) in water (25 mL). The reaction mixture was stirred at 45 °C overnight (approx. 17 h), cooled to rt, then partitioned between a solution of 100 mL water and 200 mL CH₂Cl₂. The organic layer was separated, and the aqueous layer was extracted with CH₂Cl₂ (200 mL × 3). The combined organic layers were washed with brine and dried over anhydrous MgSO₄. The resulting suspension was filtered through a course glass fritted funnel to remove solid material, and the filtrate was concentrated in vacuo to afford 19.6 g of a crude light brown oil that was used in the subsequent step without further purification.

To a solution of the above crude product (19.6 g) in 600 mL nitromethane was added i-PrOH (20.4 g, 0.34 mol) and TiCl₄ (15.9 g, 0.084 mol). The resulting mixture was stirred at rt for 1 h under N₂ atmosphere and a second aliquot of i-PrOH (20.4 g, 0.34 mol) and TiCl₄ (15.9 g, 0.084 mol) was added. The mixture was stirred for another 1 h at rt under N₂ atmosphere, then partitioned between a solution of 200 mL sat. NaHCO₃(aq) and 200 mL CH₂Cl₂. The organic layer was separated, and the aqueous layer was extracted with CH₂Cl₂ (200 mL × 3). The combined organic layers were washed with brine and dried over anhydrous MgSO₄. The resulting suspension was filtered through a course glass fritted funnel to remove solid material, and the filtrate was concentrated in vacuo. Purification of the crude product by flash column chromatography afforded 4.0 g of the title compound as yellow solid (40% over 3 steps).

Spectral data for ent-12: ¹H NMR (600 MHz, CD₂Cl₂) δ 8.22–8.18 (m, 1H), 8.00–7.96 (m, 1H), 7.57–7.52 (m, 3H), 4.72–4.63 (app. m, 1H), 3.27 (app. dd, J = 18.5, 8.0 Hz, 1H), 3.22–3.14 (m, 2H), 2.29 (dd, J = 12.4, 7.3 Hz, 1H), 2.19 – 2.15 (app. m, 2H), 2.13 (app. dd, J = 12.7, 7.8 Hz, 1H), 1.91 (td, J = 11.8, 7.8 Hz, 1H), 1.68 (s, 1H), 1.40 (dd, J = 12.4, 5.7 Hz, 1H), 0.64 (s, 3H); ¹³C NMR (150 MHz, CD₂Cl₂) δ 138.3, 134.0, 131.7, 130.8, 129.6, 128.0, 127.1, 126.5, 124.2, 120.9, 72.5, 51.1, 48.3, 41.9, 37.8, 36.3, 25.0, 18.0; IR (thin film): 3335, 2930, 2856, 1584, 1266, 1200, 1044, 753, 666 cm^–1; HRMS (EI+) Calculated for C₁₈H₁₉OBr [M⁺]330.0619, found 330.0610; ${[\alpha ]}_{\mathrm{D}}^{23}$ −21 (c 6.8, CHCl₃)