The Sodium-Potassium Pump, “Neoglycosylation” and The Death of Emil Fischer

            The effects of cardenolides and related cardiotonic steroids on cardiac contractility is caused by a specific interaction with the sodium-potassium pump (Na⁺,K⁺-ATPase), which maintains normal gradients of sodium and potassium across the plasma membrane of eukaryotic cells. Partial inhibition of the ion-pumping function of the enzyme leads to increased strength of myocardial muscle contraction and this so-called positive inotropic pharmacological action is the basis of digoxin’s clinical utility in the treatment of congestive heart failure. More recently, the sodium-potassium pump has been shown to participate in protein-protein interactions that stimulate growth-related signal transduction pathways that are also essential to increased myocardial contractility. This Na⁺,K⁺-ATPase-induced second messenger signaling is thought to have different downstream consequences in various cell types (e.g. cancer versus non-malignant). Indeed, up-regulation of the sodium-potassium pump has been observed in a variety of cancers including ovarian, pancreatic and melanoma. This has led to the notion that cardiotonic steroids, as inhibitors of the Na⁺,K⁺-ATPase catalytic alpha subunit, represent viable lead compounds for the chemotherapeutic treatment of cancer. In a small cohort of breast cancer patients, it was reported that women who were taking a cardiotonic steroid at the time of their breast cancer diagnosis had tumors with less aggressive phenotypes than the breast tumors of women not taking a steroid such as digoxin. The same authors later reported a higher recurrence rate of cancer among women not taking a cardiotonic steroid drug. However, subsequent epidemiological studies of the association between cardenolide use and breast cancer incidence gave conflicting results. 

            Drug discovery research targeting the Na⁺,K⁺-ATPase is now facilitated by the recently reported crystallographic studies depicting the enzyme (from pig kidney) in complex with ouabain and digoxin, extended to 3.4 angstrom-resolution. The crystal structures clearly illustrate the extracellular region of the sodium-potassium pump to which cardiotonic steroids bind. A relatively small set of amino acids in the steroid-binding pocket serve as the primary contributors to binding the sugar substructure of the ligand. Interestingly, a handful of mutations near the binding site confer protection from poisons such as ouabain to a diverse range of insects, amphibians, reptiles and mammals, illustrating that similar selection pressures have resulted in convergent evolution across the animal kingdom.

            Based on its in vitro anticancer properties, and in conjunction with selected patient profiling data suggesting that the survival rate of cancer patients taking digitalis-derived drugs is statistically enhanced, digitoxin has been identified as a lead compound for oncology treatment applications. Jon Thorson’s research group at the University of Kentucky has used digitoxin as a model system for chemical derivatization by a technique that he refers to as ‘neoglycorandomization.’ The method applies a chemoselective glycosylation reaction developed in the late 1990’s to derivatized terpenoid aglycone substrates to produce ‘glycodiverse’ libraries via a one-step divergent process. The protocol circumvents the need for protecting group strategies, selective anomeric activation and stereochemical control over carbohydrate coupling steps. The neoglycosylation reacton, itself, is a chemoselective glycosylation between an N,O-dialkylhydroxyamine (a nucleophilic alkoxyamine) and an unprotected reducing aldose (hemiacetal). The adduct that is formed between the reacting partners, a ‘neoglycoside,’ exists as a cyclized saccharide containing an intact N-O bond. Recently, Thorson and co-workers used the neoglycorandomization technique to probe the structure-activity relationships associated with the sugar/amine regiochemistry of a set of digitoxigenin neoglycosides. They quickly identified a 3-amino-substitution on the sugar to be most advantageous, affording a digitoxigenin monosaccaride derivative (structure depicted above) that is equipotent to digitoxin in an in vitro assay against the non-small lung cancer cell line A549. The authors rationalize the cytotoxicity data using a docking model of the sodium-potassium pump derived from human Na⁺,K⁺-ATPase ligand-bound crystal structures. In this case, molecular modeling revealed a correlation between the determined anticancer activity with ligand-binding site occupancies wherein the polar sugar/amine moiety is fully solvent-exposed.

            It must be stated that when a chemist conducts drug discovery research for a living, it is indoctrinated at a relatively early stage of one’s career that molecules containing two consecutive heteroatoms linked by a sigma bond should not show up in screening libraries. So, compounds that contain a heteroatom-heteroatom bond that is not part of a heteroaryl ring system, for example hydrazides and oximes, are strongly discouraged as lead generation starting points in spite of their synthetic accessibility. When you talk to career medicinal chemists about why this should be the case, they usually say something like, “Well that’s what killed Emil Fischer.” For the record, Fischer’s death was actually self-inflicted; but it probably didn’t help that he was suffering from an excruciatingly painful case of intestinal carcinoma that was likely caused by exposure to a molecule that he discovered, phenylhydrazine (see structure above). More anecdotal evidence along these lines lies in the mushroom-derived toxin, gyromitrin, a carcinogen present in several members of the distinctive fungal genus Gyromitra. Gyromitrin is basically a pro-drug delivery system for methylhydrazine. In the body, methylhydrazine reacts with pyridoxal 5-phosphate, the active form of vitamin B6, to form a hydrazone, resulting in reduced production of GABA which leads to neurological symptoms. Gyromitrin is also metabolized to reactive nitrosamide intermediates that decompose to methyl radicals causing liver necrosis. Because of examples like these, I’m not a great supporter of ‘neoglycorandomization’ as an approach to drug discovery in spite of its conciseness. The neoglycoside molecules that comprise a ‘glycodiverse’ library all contain R₁R₂N-OMe functionality that seems like it might not be an ideal starting point for drug discovery. Personally, I would rather invest time and resources towards the additional effort required to synthesize a smaller, more focused set of well-designed screening targets. Moreover, the bigger problem with digitoxin and related cardiotonic steroids as chemotherapeutic lead compounds is their notoriously low cytotoxic selectivity for human cancer cells versus human non-malignant cells, which is typically not higher than ten-fold. To this point, Thorson, in his recent MedChem manuscript, notes that the aminosugars described in his study present opportunities for conjugation to cancer-targeting antibodies as a strategy to improve their therapeutic window of efficacy and enable their application in the chemotherapeutic treatment of cancer.

New Cyclopentane-Based Sialyltranferase Inhibitors as Lead Compounds for the Chemotherapeutic Treatment of Cancer

            Sialic acid-containing glycoconjugate antigens play a critical role in a number of physiological and pathological biochemical processes, including cell-cell adhesion, immune defense and, importantly, tumor cell metastasis. Sialyltransferase enzymes catalyze the transfer of sialic acids to terminal non-reducing positions on growing oligosaccharide chains of glycoconjugates. Sialyltransferases of all origins and subtypes share the same donor substrate, cytidine monophosphate N-acetylneuraminic acid (CMP-Neu5Ac). The enzyme-catalyzed transfer reaction is thought to proceed via an S_N1-like mechanism (outlined below) wherein partial dissociation of the cytosine monophosphate leads to formation of a trigonal planar oxocarbenium species in the transition state. Overexpression these enzymes and the consequent overpresentation of sialylated antigens on cell surfaces are correlated with poor prognosis in several different types of carcinomas. As such, the discovery of cell-permeable inhibitors of sialyltransferase is considered a promising strategy for antitumor drug development.

            Soyasaponin I is a glycosylated pentacyclic triterpenoid natural product derived from soybean that displays significant inhibition (K_i = 210 nM) of a particular sialyltransferase subtype. Related derivatives of the bile acid steroid lithocholic acid were later developed as potent inhibitors of a sialyltransferase and one of those (Lith-O-Asp, structure shown below) could effectively attenuate the total sialylation on cancer cell surfaces and suppress tumor cell metastasis in in vivo animal models of cancer.

            However, the most potent sialyltransferase inhibitors developed to date are structures that mimic the aforementioned three-dimensional structure of the transition state of the enzymatic process (for a classic example of transition state analogue design, see here). A potent transition-state analogue related to the CMP-Neu5Ac glycosyl donor was first described by Richard Schmidt’s group in 2002. More recently, Xin-Shan Ye’s laboratory in Beijing, China has reported a series of highly substituted cyclopentane-containing compounds (highlighted example shown below) that were designed based on similar principles of enzyme-binding. I’ve advocated elsewhere that the cyclopentane ring is an excellent scaffold for drug discovery. Ye’s new cyclopentanoid phosponates, whose overall conformation (likely an interconverting half-chair and envelope) effectively mimics the somewhat planar character of the CMP-Neu5Ac-derived oxocarbenium ion in the enzyme transition state, provide additional support for this arguably underutilized MedChem concept. The most potent cyclopentane derivative was synthesized by a 20-step sequence of reactions and displays outstanding inhibitory activity against recombinant human ST6Gal-I. Detailed structure-activity relationships across the series are also reported. This study further illustrates the utility of the cyclopentane motif as a modular scaffold for medicinal chemistry development programs.

Neil deGrasse Tyson on Baseball

Dionicio Siegel’s ‘Nonbiomimetic’ Polyene Cyclization Process Enables the Total Synthesis of Celastroid Pentacyclic Triterpenoids

            The natural product celastrol exhibits diverse biological properties including anti-inflammatory and anti-cancer, as well as suppression of phenotypes associated with neurodegenerative disorders. For example, the unique quinone methide triterpenoid acts as a downregulator of mediators of anti-inflammatory responses such as interleukin-1a, TNF-a and nuclear factor kB (NF-kB). It also inhibits human prostate tumor growth and human glioma xenografts in mice. Recently, celastrol was identified as an inhibitor of heat shock protein 90 (Hsp90), which is over-expressed in cancer cells and plays a role in activation of certain pro-oncogenic signaling molecules. Celastrol is a non-ATP-competitive inhibitor of Hsp90 and acts via a mechanism that likely involves (in some as yet undefined fashion) conjugate addition of cysteine residues within biological nucleophile[s] to the electrophilic quinone methide substructure. In 2011, Richard Silverman’s laboratory at Northwestern University demonstrated that a range of soft nucleophiles add to the pharmacophore of celastrol in a highly stereospecific fashion. It is interesting to note that a related triterpene derivative, bardoxolone methyl (or CDDO, structure shown above), also contains a reactive cyanoenone Michael acceptor motif embedded in the western A-ring of its pentacyclic architecture. Its ability to generate reversible Michael adducts with biological sulfur nucleophiles is also speculated to be relevant to the molecular mechanism of action of CDDO. Bardoxolone methyl is currently in phase III clinical trials for the treatment of severe chronic kidney disease in type 2 diabetes mellitus patients.
            The total synthesis of celastrol was recently reported by Dionicio Siegel’s laboratory at the University of Texas at Austin. The ultimate success of Siegel’s synthetic approach to celastroid pentacyclic triterpenoids hinged on a polyene cyclization reaction that is described by the authors as ‘nonbiomimetic.’ Siegel and co-workers note that “biological polyene cyclization leading to celastrol is exceedingly difficult to reproduce in the laboratory due to a set of complex and energetically unfavorable methyl and hydride shifts.” The UT Austin team references synthetic work targeting alnusenone and friedelin, conducted in the late 1960’s and 70’s by Bob Ireland’s group. At this juncture, it may be instructive for readers to revisit the enzyme-catalyzed p-cation polyene cyclization of oxidosqualene into the ornate pentacyclic carbon skeleton of b-amyrin (detailed mechanism shown below) in order to understand why Siegel chooses to characterize his own cyclization reaction as ‘nonbiomimetic.’

            Tsutomu Hoshino’s laboratory at Niigata University in Japan has shown that oxidosqualene (and related unnatural polyene substrates) must be correctly ‘folded’ into a chair-chair-chair-boat-boat conformation in order to facilitate polycyclization leading to the intact b-amyrin carbocyclic scaffold. Hoshino’s group conducted studies involving incubation of synthetic derivatives of oxidosqualene in the presence of b-amyrin synthase derived from the African plant, Euphorbia tirucalli. They demonstrated that the methyl group at carbon position 30 of oxidosqalene plays an important role in binding to a hydrophobic recognition site of the enzyme, leading to appropriate construction of the ordered architectural conformation of the polyene substrate. Hoshino's studies suggest that a correct folding conformation of the polyene strongly influences the success of the polycyclization cascade. Oxidosqualene analogues that possess an intact Me-30 are more efficiently converted into pentacyclic terpenoid systems as compared to those lacking a terminal (Z)-Me group. Substrates that do not contain Me-30 generate far more abortive cyclization products. It is clear from the enzymatic mechanism depicted above, as well as the Hoshino group’s recent biosynthetic studies (outlined below), that the chemical polyene cyclization process developed by Siegel and co-workers at UT Austin is indeed ‘nonbiomimetic.’

            The UT Austin total synthesis of celastrol is initiated by execution of a two-directional approach starting from 2,3-dimethylbutadiene. In relatively short order, an advanced polyene-aldehyde intermediate (structure depicted below) is obtained by an expedient sequence of reactions featuring a tin-lithium exchange/alkylation to install the aromatic moiety. Stork-enamine Robinson annulation with methyl vinyl ketone (MVK) was successfully conducted on multigram scale to generate a critical cyclohexenone intermediate and subsequent lithium aluminum hydride reduction furnished the polyene cyclization substrate as an inconsequential diastereomeric mixture. Remarkably, exposure of a dilute solution of the cyclohexenol intermediate to the Lewis acid ferric chloride at low temperature promoted the stereocontrolled formation of the desired pentacycle with useful efficiency, in light of the complexity of the overall transformation. Siegel’s ‘nonbiomimetic’ polycyclization reaction was demonstrated on 1-gram scale. The Jones reagent was then used to oxidize the benzylic position located in the B-ring and subsequent selenoxide elimination installed the requisite enone. Demethylation afforded the catechol natural product wilforol A, which served as an intermediate that was suitable for eventual conversion into celastrol and its corresponding methyl ester, pristimerin. The total synthesis of racemic celastrol (obtained as a red-orange solid) was achieved in a total of 31 linear operations, starting from 2,3-dimethylbutadiene. The work is extremely important because it provides synthetic access to a medicinally relevant quinone methide triterpenoid that could not be easily obtained by a more cost-effective semisynthetic approach. For example, it is difficult to envision a method by which one might oxidatively convert a readily available pentacyclic triterpenoid such as oleanolic acid (the starting material for bardoxolone methyl) or b-amyrin into a sensitive quinone methide derivative. Hopefully, synthetic access to meaningful quantities of celastrol will facilitate studies to elucidate the precise biochemical mode of action leading to the diverse biological activies exhibited by this unique natural product.