Discovery of Plant Sterol-Derived Gamma-Secretase Modulators by Molecular Editing of a Triterpenoid Glycoside

The term 'molecular editing' was coined by Danishefsky in reference to the synthetic chemist's unique ability to revise specific elements of a natural product's structure in order to modulate properties such as pharmacokinetic behavior or chemical stability. While natural products have been refined through cycles of selection/diversification, they are often suboptimal as orally bioavailable pharmacological agents and, thus, 'editing' particular metabolic or chemical 'soft spots' within a naturally occurring molecular structure can be beneficial to its development as a drug candidate. For example, the complex molecular framework of the antifungal triterpenoid glycoside, enfumafungin, was modified in the course of medicinal chemistry studies to excise undesirable functionality such as a cyclic hemiacetal and C3-O monosaccharide moiety (highlighted below in red and blue, respectively), both of which were found to be either chemically or enzymatically labile. These research efforts culminated in the development of the antifungal clinical candidate SCY-078, the first orally active glucan synthase inhibitor. Satori Pharmaceuticals have recently disclosed results from a similar natural products-based program aimed at the development of novel gamma-secretase modulators as therapeutics for Alzheimer's disease, starting from a triterpenoid glycoside (1) isolated from the black cohosh plant. The plant sterol natural product 1, like enfumafungin, exhibits exquisite and unique bioactivity but contains a few potential metabolic and chemical liabilities, which include a C3-O glycoside (blue) and C24 enol ether (red). Molecular editing through semisynthetic derivatization and medicinal chemistry optimization produced investigational lead compounds such as 2 that display improved potency and central nervous system (CNS) exposure.

The mode of action of 1 and 2 involves modulation of the peptidic site on amyloid precursor protein (APP) where gamma-secretase cleavage occurs. In brief, APP is cleaved at various sites by gamma-secretase to produce different amyloid-beta peptides. The relatively longer amyloid-beta fragments (which are around 42 amino acids long as opposed to 38 or 40) seem to be particularly neurotoxic with regard to initiation of the pathology of the Alzheimer's brain. Therefore, a small molecule that could lower the ratio of longer amyloid fragments relative to the shorter ones could represent a safe and tolerable therapeutic agent for Alzheimer's disease, devoid of the toxicities that arise from an overall blockade of the action of gamma-secretase (which include skin cancer). It should be noted as an aside that the steroidal lactone withanolide A also alters APP processing through an alternate mechanism that is discussed here.

The lead plant sterol derivative 2 is synthesized in four steps from the triterpenoid natural product 1. First, stereoselective reduction of the C16-C17 olefin is achieved by a two step sequence involving zirconium tetrachloride-catalyzed isomerization of the enol ether followed by reduction of the C15 ketone (dr > 95:5). Double oxidative cleavage of the C3-O glycoside by treatment of 4 with sodium periodate then secured a versatile dialdehyde intermediate (5) from which a variety of C3 morpholino derivatives could be obtained by application of a double reductive amination protocol. Derivative 2 effectively lowers levels of amyloid-beta 42 and 38 while preserving amyloid-beta 40 and total amyloid-beta. In addition, 2 exhibits improved CNS penetration relative to 1 and retains an excellent profile in standard transporter, off-target and safety assays. This natural products-based approach represents an exciting entry into gamma-secretase modulation (rather than inhibition) as a potential therapeutic approach to the treatment of Alzheimer's disease.

A Structural Revision of Neoveratrenone Prompted by Violation of Bredt's Rule

            In 2003, Crews and Clardy isolated a highly unique steroid derivative, isocyclocitrinol A, from the extract of a saltwater culture of sponge-derived Penicilliun citrinum. The compound exhibited rather uninteresting (weak) antibacterial activity but can be considered quite captivating in terms of its skeletal structure due to the presence of a rarely encountered caged bicyclic motif (highlighted above in blue). The 'citrinane' ring skeleton of isocyclocitrinol A encompasses the first example of a bicyclo[4.4.1] system contained within the A/B substructure of a rearranged steroid. Biosynthetically, this carbon connectivity likely arises via a 1,2-migration of the steroidal C5-C10 bond to generate a new C5-C18 bond. Isocyclocitrinol A is an example of a natural product that violates Bredt’s rule, an empirical observation made by Julius Bredt in the 1920s that states that a double bond cannot be placed at the bridgehead of a bridged bicyclic ring system. Bredt's rule is a consequence of the fact that having a double bond at a bridgehead position is comparable to the incorporation of a trans double bond into a small ring (fewer than eight atoms), which confers a considerable amount of unfavorable ring and angle strain.

            On the basis of Bredt’s rule, a recently reported 'anti-Bredt' natural product, neoveratrenone, was reassigned by the laboratory of Craig Williams at the University of Queensland. Williams and co-workers embarked on a rigorous reinterpretation of the 1D and 2D NMR data presented in the original structure elucidation paper and concluded that, indeed, the putative bicyclo[3.3.1]nonenone-containing neoveratrenone was more accurately characterized by a ‘6-6-5-6’ C-nor-D-homo steroidal skeleton that does not violate Bredt’s rule. The most notable member of the C-nor-D-homo rearranged steroid family is the steroidal alkaloid cyclopamine, which is pharmacologically infamous due its ability to cause fatal birth defects including cyclopia (this link is not for the squeamish). The fascinating chemical synthesis and biochemistry associated with cyclopamine have been the subjects of previous posts at Modern Steroid Science. Williams and co-workers adeptly noticed that a close structural derivative of cyclopamine, verapatuline, was co-isolated with veratrenone, further bolstering the impetus for the structural revision. The contribution by the Williams laboratory provides a great service to the synthetic community in that it will prevent the allocation of resources to synthesize the initially proposed, incorrect, structure.

Baran's Expedient Chemical Synthesis of (+)-Ingenol

          The phorbol esters are well-known tigliane natural products derived from Croton tiglium, the source of croton oil, as well as from other plants of the family Euphorbiaceae. Phorbol esters such as 12-O-tetradecanoylphorbol-13-acetate (TPA) are useful pharmacological tool compounds in models of carcinogenesis due to their potent activity as mouse skin tumor promotors. Various esters of phorbol such as TPA, which mimic the chemical structure of diacylglycerol, bind to and activate protein kinase C. This type of modulation of one of the major signal transduction mechanisms within cells enables the phorbol esters to elicit a broad range of biological activities. TPA has been used extensively in biomedical research as a probe to identify the physiological systems in which protein kinase C is involved. Chemical syntheses of phorbol are notoriously challenging and have required between 36 and 52 total steps.

          Ingenol mebutate (trade name Picato) is the ester derived from a related ingenane diterpene, ingenol, and angelic acid. This drug was recently approved by the FDA as a first-in-class treatment for actinic keratosis, a precancerous skin condition. Unlike the tigliane/phorbol-type diterpenoids, the ingenane carbocyclic framework consists of a unique in,out-[4.4.1]bicycloundecane core. This ring skeleton is quite strained due to the "inside-outside" intrabridgehead stereochemistry of the BC ring system, wherein the C8 and C10 configurations are trans to one another. This renders ingenol a formidable synthetic target for total synthesis. The cis-triol of the AB ring fragment is also difficult to elaborate and often requires lengthy synthetic sequences. The recent chemical synthesis of ingenol by Baran's group at the Scripps Research Institute will be the subject of this post.

          Baran's route, like many previous synthetic studies targeting ingenol, starts from a readily available terpene, (+)-carene. The cyclohexanone intermediate 2 is assembled by means of a stereocontrolled aldol condensation between an elaborated carene derivative and the allene-containing aldehyde 1. Treatment of 2 with ethynyl magnesium bromide gave the diol 3  (d.r. 10:1) which was subsequently protected as the bis-silyl ether 4. Baran then beautifully applied Kay Brummond's (University of Pittsburgh) rhodium(I)-catalyzed allenic Pauson-Khand-type chemistry to the intricate substrate 4, which fashioned the carbocyclic tigliane carbon skeleton in an extremely concise manner compared to previously demonstrated synthetic strategies.

          The key step in Baran's ingenol synthesis is the vinylogous 1,2-pinacol rearrangement of the advanced tigliane/phorbol-type system 6, which sets the requisite inside-outside (trans) intrabridgehead stereochemistry of 7. This fascinating skeletal transformation, speculated to be of biosynthetic relevance, involves a 1,2-alkyl shift and is reminiscent of the Tsuchihashi-Suzuki rearrangement of 2,3-epoxy alcohols. In order to be successful, the C9-C11 bond of 6 must selectively undergo the desired migration to the developing partial positive charge at C10. In the event, exposure of intermediate 6 to Lewis acidic conditions at low temperature induced the projected vinylogous pinacol-type rearrangement which established the quaternary center at C10 in a highly stereocontrolled fashion. The advanced intermediate 7 then underwent sequential, chemoselective oxidation reactions that culminated in fully synthetic ingenol in only 14 total operations.

          It should be noted that pinacol-type rearrangements involving 1,2-alkyl migrations have been previously applied to the synthesis of ingenol. For example, in the second total synthesis of ingenol, Kuwajima and co-workers utilized this type of skeletal reorganization with a ring system (11) that was derived from a functionalized trans-decalin (10). In a related study, Cha's group assembled the Tsuchihashi-Suzuki rearrangement substrate 14 using olefin metathesis. In these examples, stereoelectronic requirements dictate that an antiperiplanar alignment of the C9-C11 bond with respect to the C10-O is required for the desired migration to occur. While the intended rearrangements in both examples were quite efficient from a yield perspective, both Kuwajima's and Cha's substrates lacked the intact cis-triol of the AB ring fragment and, thus, multistep oxidative elaboration following the skeletal reorganization was required. In comparison, Baran's substrate (6) is expertly designed. The relative stereochemistry of 6 satisfies all of the stereoelectronic requirements for bond migration and the vinylogous SN2' nature of the transformation directly installs the olefin between C1 and C2 of the A-ring. Baran clearly benefited from a discerning knowledge of the literature precedent, but also designed and executed a virtuosic refinement to the existing technology. His chemical synthesis compares favorably to isolated yields of ingenol from plant material and may facilitate the commercial manufacture of Picato.

Baran's Total Synthesis of Racemic Steviol, the Aglycone of Stevia's Sweet Glycosides

A Steviol Glycoside.

          The prominent and detail-oriented 'Lydia' character from the AMC series 'Breaking Bad' was wont to partake of chamomile tea sweetened with a packet of 'Stevia' during cafe meetings. Walter White eventually took advantage of her fastidious nature to replace the single packet of Stevia on her table with his long ago-isolated ricin, a potent toxin derived from Castor beans. Steviol glycosides (a representative structure is depicted above) are responsible for the sweet taste of the leaves of the stevia plant (Stevia rebaudiana Bertoni), which range in sweetness from 40 to 300 times sweeter than sucrose. Moreover, steviol glycosides do not induce a glycemic response when ingested, rendering them attractive as natural sweeteners for diabetics. The diterpene steviol, the aglycone of Stevia's sweet glycosides, was the target of recent synthetic studies led by the recently crowned MacArthur fellow Phil Baran at the Scripps Research Institute. The Baran laboratory's campaign, which culminated in the development of a practical total synthesis of (+/-)-steviol, is the subject of this post.

          Steviol is an oxidized congener of ent-kaurene, the biosynthetic precursor to the well-known  diterpene plant hormone gibberellic acid. Baran's group constructed the key early-stage tricyclic intermediate 2 in a biomimetic fashion that drew inspiration from Nature's fascinating enzymatic conversion of geranylgeranyl pyrophosphate (GGDP) into ent-kaurene. The enone 3 was then obtained from 2 by a three-step sequence involving elimination of a secondary alcohol, hydrogenolysis and Birch reduction/isomerization. In the subsequent operation, a critical allene [2 + 2] photocycloaddition installed the hindered C8 quaternary center of the advanced cyclobutane intermediate 4. Several alternate methods to install the C8 stereocenter had failed.

          Ozonolysis of 4, when conducted in methanol, induced fragmentation of the strained cyclobutane framework to generate the intermediary methyl ester 5. Next, the [2.2.2]bicyclic system of 6 was fashioned by exposure of 5 to forcing acidic conditions and subsequent reductive cyclopropanation in the presence of acetic anhydride led to the advanced diacetate 7. Finally, controlled fragmentation of 7 with methanolic hydrochloric acid, followed by an expedient methylenation/oxidation endgame sequence produced fully synthetic steviol in only 17 total steps starting from geranyl acetate.

          Notably, an intriguing skeletal rearrangement of steviol, induced by HBr, provides access to the related beyerane diterpene natural product isosteviol, which is apparently thermodynamically favored compared to its kinetic precursor. This impressive work establishes a rapid and efficient synthetic route to access minimally oxidized members of the ent-kaurane and beyerane class of diterpenes. As noted above, members of this natural product family are commercially valuable in the flavor industry as precursors natural and/or semisynthetic sweeteners.