Symmetry-Driven Total Synthesis of Myrioneurinol

Abstract

We report a total synthesis of the Myrioneuron alkaloid myrioneurinol enabled by the recognition of hidden symmetry within its polycyclic structure. Our approach traces myrioneurinol’s complex framework back to a symmetrical diketone precursor, a double reductive amination of which forges its central piperidine unit. By employing an inexpensive chiral amine in this key desymmetrizing event, four stereocenters of the natural product including the core quaternary stereocenter are set in an absolute sense, providing the first asymmetric entry to this target. Other noteworthy strategic maneuvers include utilizing a bicyclic alkene as a latent cis-1,3-bis(hydroxymethyl) synthon and a topologically controlled alkene hydrogenation. Overall, our synthesis proceeds in 18 steps and ~1% yield from commercial materials.

Graphical Abstract

The Myrioneuron alkaloids are a relatively small class of polycyclic natural products isolated from various plants of the Myrioneuron genus (e.g., 1–6, Figure 1A). Despite their limited number, these alkaloids display a range of bioactivities, including antimalarial, antiviral, and cytotoxic properties.¹ Biosynthetically, they are thought to originate from the combination of various lysine-derived units, resulting in a diverse array of structures containing a core decahydroquinoline framework fused to oxazine, diazine, or cyclohexane rings arranged in interlocked chair-like conformers.¹ Their challenging structures coupled with their biological properties have attracted the interest of synthetic chemists. Several simpler bi- and tricyclic members (e.g., 1) were prepared by Bodo and coworkers² and others,³ while efforts toward more complex members have proven rarer. Recent elegant syntheses of myrifabrals A and B (5–6) by She⁴ and Stoltz,⁵ and that of myrioneurinol (2) by Weinreb have begun to address this challenge.⁶

Figure 1.

(A) Representative Myrioneuron alkaloids. (B) Prior racemic total syntheses of myrioneurinol (2). (C) Our symmetry-driven approach to myrioneurinol.

In continuation of our interest in alkaloid total synthesis,⁷ we were drawn to myrioneurinol (2) as arguably one of the most structurally complex non-dimeric Myrioneuron alkaloids. This compound was isolated by Bodo and coworkers in 2007 and shown to display moderate antimalarial activity as well as weak inhibition of KB cell proliferation.⁸ As noted above, this target was prepared in racemic form by the Weinreb group in 2014 utilizing a strategy built around two key Michael reactions (a spirocyclization and malonate nitrosoalkene addition) and a late-stage intramolecular Sakurai reaction to provide (±)-2 in 27 steps (Figure 1B).⁶ As we were finalizing our manuscript, Ma and coworkers reported a more step economical approach to (±)-2 via an intramolecular [2+2] cycloaddition/retro-Mannich/Mannich approach in 14 steps from (6-iodohex-1-ynyl)trimethylsilane.^9,10 Herein, we report a distinct total synthesis of myrioneurinol enabled by the recognition of hidden symmetry within its complex framework, leveraging this strategy to provide the first asymmetric entry to this target.

Key challenges in accessing 2 revolve around construction of its tetracyclic array of fused piperidine, oxazine, and cyclohexane rings with stereocontrol at its five stereogenic centers, including one core quaternary center. We planned to employ a desymmetrization approach to construct much of the complexity of 2 from a symmetrical precursor. Desymmetrization-based strategies can greatly expedite the synthesis of complex targets by simultaneously setting many stereocenters from prochiral or meso-substrates.¹¹ Such processes can either exploit local symmetry to allow for the use of simplified synthetic fragments or break symmetry within a fully symmetrical precursor, and have been utilized in alkaloid total synthesis.¹² These strategies become even more enabling when the synthetic target does not possess any inherent symmetrical elements and ‘hidden symmetry’ is unveiled by judicious retrosynthetic disconnection back to a symmetrical precursor.

Our approach to myrioneurinol (2), which does not possess an obvious plane of symmetry, is outlined retrosynthetically in Figure 1C. The oxazine ring of 2, incorporating one of the two cis-hydroxymethyl units, might arise from oxidative cleavage and subsequent cyclization of a bicyclic alkene 7. Disconnecting both the C₆–C₁₇ bond of the D-ring and the C₂–N/C₁₀–N bonds of the piperidine A-ring of 7 leads back to a relatively simple symmetrical diketo aldehyde 8. Bicycle 8 presents an opportunity for a key desymmetrizing double reductive amination onto the C-2 aldehyde and one of the two prochiral ketones to forge the piperidine A ring, leaving the remaining ketone available for a subsequent carbocyclic ring construction. It was unclear at the outset of our studies, however, whether such a process would occur with the desired diastereoselectivity in this complex bicyclic setting,¹³ or how attainable an enantioselective version of the crucial desymmetrization might be.¹⁴ Nonetheless, given that diketone 8 should arise via sequential alkylation of known chlorodiketone 9,¹⁵ we reasoned that these questions could be probed without significant initial investment.

Our synthesis began with the preparation of decagram quantities of 9 from commercial pentachlorocyclopropane 10 through a known¹⁵ one-pot process involving initial HCl elimination to the cyclopropene, Diels–Alder reaction with cyclopentadiene and cyclopropane ring-opening to give 11, and subsequent basic hydrolysis (Scheme 1). Although initial attempts at conversion of 9 to allyldiketone 13, including dechlorination/monoallylation or Claisen rearrangement-based approaches, were plagued by poor selectivity and low yields, we ultimately found that a decarboxylative Tsuji–Trost allylation provided a scalable means to access 13.¹⁶ Thus, crude 9 was converted to O-Alloc derivative 12 in 80% yield (over 2 steps), followed by decarboxylative C-allylation with catalytic Pd(PPh₃)₄ to give an intermediate 2-chloro-2-allyldiketone that could be dechlorinated by addition of Zn and AcOH to the same reaction vessel. With robust access to multigram quantities of 13, a mild Michael addition to acrolein delivered key tricarbonyl precursor 14 in 75% yield, albeit in moderate dr (1.4:1) favoring our desired diastereomer (despite significant attempts at improvement, see the Supporting Information (SI) for full details).

Scheme 1.

Synthesis of tetracyclic alkene 19 via desymmetrizing double reductive amination and its attempted stereoselective hydrogenation.

With an inseparable epimeric mixture of diketo aldehyde 14 in hand, we were poised to investigate the key desymmetrizing double reductive amination. Initially, we elected to pursue the synthesis of (±)-15 using N-benzylamine, and preliminary screens identified NaBH₃CN and AcOH as the optimal reductant and acid promoter, respectively. Through further trials (see selected results in the inset table in Scheme 1), we identified high temperature (100 °C) as the most impactful parameter for the success of this transformation; reaction monitoring by ¹H NMR showed the first reductive amination of the aldehyde to occur essentially instantaneously at room temperature, with the second reductive cyclization onto the hindered ketone requiring high temperature to proceed efficiently. Ultimately, we could access tricyclic amine 15 in 62% yield as a single diastereomer on gram-scale. This transformation constructs the piperidine A-ring with the correct relative configuration for myrioneurinol (initially confirmed by NOE studies) and provides an opportunity to remove the minor aldehyde diastereomer, which likewise delivers a separable piperidine product as a single diastereomer (not shown).

Next, chemoselective dihydroxylation of the strained bicyclic alkene followed by acetonide protection produced a masked diol 16 as a single diastereomer, setting the stage for a later oxidative cleavage. Attempted allylation of ketone 16 proved surprisingly difficult with many common allylation protocols, including Grignard, organoboron, and organolanthanide nucleophiles, failing to engage this hindered carbonyl. Ultimately, this step was only successfully achieved through the use of excess (10 equiv) allyllithium, cleanly providing a single diastereomer of the allyl adduct (not shown). This crude diene was subjected to ring-closing metathesis (RCM) with Hoveyda-–Grubbs Second Generation (HG-II) catalyst,¹⁷ delivering pentacyclic alkene 17 in 67% yield over the 4 steps from 15 with only a single chromatographic purification being required. With the final carbocyclic ring secured, the superfluous alkene could be hydrogenated (H₂, Pd/C, AcOH, MeOH) with concomitant hydrogenolysis of the N-Bn group. Reprotection of the crude secondary amine as a more tractable N-tosylamide gave a compound (18) whose relative configuration could be confirmed through X-ray crystallographic analysis.

At this stage, the final C-10 stereocenter of 2 could be addressed. Overall, this required a deoxygenation of the tertiary alcohol 18 with retention of configuration. While Barton–McCombie conditions failed due to our inability to derivatize this hindered hydroxyl, a simple workaround involved elimination to the alkene 19 with SOCl₂/pyridine (92%) followed by alkene hydrogenation (85%) to give a single diastereomer of saturated product 20, at this stage of unknown configuration. Noting the prior hydrogenation in the sequence, we were able to execute a shorter sequence involving earlier elimination of 17 to its corresponding diene (not shown, 70%), which could similarly undergo hydrogenation/debenzylation and N-tosyl protection (59%, 2 steps) to deliver the same reduction product (20). Unfortunately, X-ray crystallography ultimately demonstrated that the obtained diastereomer bore a C-10 configuration opposite to that of myrioneurinol. Further attempts to invert the stereoselectivity through a variety of hydrogen atom-transfer (HAT), homogeneous, and heterogeneous catalytic hydrogenations of 19 yielded solely this undesired diastereomer (20). We attribute this outcome to steric hindrance from the nearby bridged system (see inset A, Scheme 1).

Since this bicyclic substructure was ultimately not required for myrioneurinol (2), we sought to alleviate this steric issue by cleaving the bridging system prior to alkene hydrogenation (Scheme 2). The first step in such a plan, acetonide deprotection of 19, proved to be unexpectedly challenging as standard acidic conditions led to decomposition or an undetermined rearranged product. After screening a variety of conditions, a mild cerium trichloride/oxalic acid system was discovered to provide a tractable solution delivering diol 21 in 65% yield along with some recovered 19 (18%).¹⁸ Oxidative diol cleavage with PhI(OAc)₂¹⁹ followed by reduction of the resulting dialdehyde in the same pot yielded a bridged-cleaved primary diol 22 that was advanced to bis-MOM ether 23 in 86% yield (two steps).

Scheme 2.

Completion of (±)-myrioneurinol (2) via a topologically controlled HAT hydrogenation.

With the topology of the alkene altered in 23 and its precursor diol 22, we attempted its hydrogenation. Unfortunately, preliminary screens again delivered unsatisfactory selectivity; the best result was obtained with bis-MOM ether 23 which yielded a 1.1:1 mixture of diastereomers under standard hydrogenation conditions (H₂, Pd/C; entry 1, inset). Pleasingly, it was found that HAT reduction of 23 was able to favor the desired isomer, with optimized conditions using Baran’s Fe-catalyzed system providing the desired saturated 24 as the major product (dr = 12:1) in 64% yield.²⁰ Tricycle 24 is a known intermediate in the prior Weinreb synthesis.⁶ Thus, subjection of 24 to a slightly modified version of their two-step sequence, involving tosyl deprotection with Li/NH₃ followed by local desymmetrization of the two primary MOM-ethers by acid-mediated oxazine formation and deprotection, respectively, gave racemic myrioneurinol [(±)-2] in 35% yield over the two steps. Overall, our total synthesis proceeds in 18 steps and ~1% yield from commercial materials.

A key advantage of our desymmetrization-based strategy is that it is readily adaptable to an asymmetric synthesis of myrioneurinol. Thus, simply substituting benzylamine for inexpensive (R)-α-methylbenzylamine (25) in the double reductive amination led to a diastereoselective desymmetrization proceeding with reasonable selectivity (dr = 4:1) for one of the two diastereotopic ketones (formally enantiotopic once the α-methylbenzyl unit is removed), allowing for the isolation of pure major isomer 26 in 34% yield by column chromatography (Scheme 3).^21,22 This transformation sets the absolute configuration of four of the five stereocenters of myrioneurinol (2), including the quaternary center, in a single step. 26 could be submitted to the same sequence of reactions (via intermediates 27–28) as our racemic benzyl series to arrive at (−)-18 (>99% ee by HPLC), whose absolute configuration was be determined by single-crystal X-ray analysis. The synthesis of (−)-18 thus constitutes a formal asymmetric synthesis of (−)-ent-myrioneurinol [(−)-2]; given that (S)-α-methylbenzylamine is equally available, access to the natural (+)-enantiomer via such a process should be trivial.

Scheme 3.

Formal synthesis of (−)-myrioneurinol [(−)-2] via asymmetric desymmetrization of 14.

In summary, we have developed an 18-step total synthesis of myrioneurinol (2), including the first asymmetric approach via a formal synthesis of (−)-2. Our synthesis exploits hidden symmetry to construct its polycyclic framework, centering on a key desymmetrizing double reductive amination of a bicyclic diketo aldehyde to assemble its core ring system in a stereocontrolled manner. Utilization of an inexpensive, enantiopure chiral amine in this process provides a convenient asymmetric entry to the myrioneurinol scaffold. Other noteworthy features of our synthesis include masking the cis-bis(hydroxymethyl) unit of 2 as a bicyclic olefin, and a diastereoselective alkene hydrogenation that relied on careful control of substrate topology. Future studies from our group will look to expand this symmetry-driven approach to other alkaloid targets both within the Myrioneuron class and beyond.