Article pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. 2019, 141, 10883−10904
Chemoenzymatic Total Synthesis of ent-Oxycodone: Second‑, Third‑, and Fourth-Generation Strategies Mariia Makarova, Mary Ann A. Endoma-Arias, Helen E. Dela Paz, Razvan Simionescu, and Tomas Hudlicky* Department of Chemistry and Centre for Biotechnology, Brock University, 1812 Sir Isaac Brock Way, St. Catharines, Ontario L2S 3A1, Canada
Downloaded via KEAN UNIV on July 17, 2019 at 08:16:24 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: Four distinct approaches to ent-oxycodone were designed and accomplished. All rely on the same starting material, the diene diol derived from phenethyl acetate by the whole-cell fermentation with E. coli JM109 (pDTG601A), a strain that overexpresses toluene dioxygenase. The key step in the first-generation approach involves the construction of the C-9/C-14 bond by a SmI2-mediated cyclization of a keto aldehyde. The second-generation design relies on the use of the Henry reaction to accomplish this task. In both of these syntheses, Parker’s cyclization was employed to construct the D-ring. The thirdgeneration synthesis provides an improvement over the second in that the nitrogen atom at C-9 is introduced by azidation of the C-9/C-10 olefin, followed by reduction and lactam formation between the C-9 amine and the Fukuyama-type lactone. Finally, the fourth generation takes advantage of the keto−nitrone reductive coupling to generate the C-9/C-14 linkage. The four generations of the total syntheses of ent-oxycodone were accomplished in 13, 18, 16, and 11 operations (19, 23, 24, and 18 steps), respectively. Experimental and spectral data are provided for all new compounds.
■
INTRODUCTION Morphine (1) and its congeners (Figure 1) have proven to be challenging targets for synthetic organic chemists. Since the first total synthesis of morphine by Gates in 1952,1 more than 30 total or formal syntheses have been reported.2 The most recent is Metz’s synthesis of (−)-codeine in 2018.3 Most synthetic strategies have targeted codeine (2), which is easier to handle than the more sensitive morphine and whose O-demethylation to morphine has been well established by Rice.4a In 2009, Stork reported a successful synthesis of the racemic 6-methoxycodeine with a subsequent one-step transformation to thebaine. 5 Five years later, Opatz accomplished an enantioselective synthesis of (−)-dihydrocodeine in 12 linear steps in 31% yield, which also constituted a formal synthesis of (−)-thebaine.6 In 2018, he achieved the de novo total synthesis of thebaine (4).7a The attainment of dihydrocodeinone,7b−d dihydrothebainone,7e−g codeinone,7h−k or other advanced intermediates7l,m may be considered as formal syntheses of thebaine as well. Our own group published a model study toward the synthesis of thebaine by a cycloaddition strategy based on 1,2-pyridazines and considerations of latent symmetry issues in design.8 The pentacyclic skeleton of morphine alkaloids possesses fully dissonant connectivity,9 presenting a serious challenge in © 2019 American Chemical Society
design to any practitioner. The likelihood of a truly practical chemical synthesis of any morphine alkaloid that is competitive in cost with the compounds available from natural sources is extremely low; even the most efficient synthesis accomplished on a kilo scale by Rice4b would not meet such a requirement. There are, however, at least two reasons to continue the quest for an efficient synthesis. The first is purely educational and aesthetic, as any total synthesis effort offers the ultimate in training of students. The second reason may present itself in the effective combination of molecular biology and chemistry by approaching the preparation by means of well-understood biosynthetic pathways overexpressed in robust bacterial strains. Such an approach would yield key intermediates by fermentation and allow access not only to the natural alkaloids but also to those medicinal agents derived by semisynthesis. A commercial preparation of both analgesics and antagonists is now best accomplished from either oripavine (3) or thebaine (4) because these compounds are now available in large amounts from selectively cloned plants developed by Tasmanian Alkaloids.10 It was established that Received: May 10, 2019 Published: June 11, 2019 10883
DOI: 10.1021/jacs.9b05033 J. Am. Chem. Soc. 2019, 141, 10883−10904
Article
Journal of the American Chemical Society
Figure 1. Natural morphine alkaloids and some opiate-derived agents.
201419 and by Chen at the end of 2018.20 Our own firstgeneration effort was published in early 2019.21 Also, in early 2019, Opatz accomplished a total synthesis of (−)-oxycodone by means of electrochemical methods.22 In the present Article, we report the details of several generations of chemoenzymatic total synthesis of ent-oxycodone from phenethyl acetate.
the main factors controlling the yield of morphine alkaloids are the genotype and the nitrogen nutrition of the crop along with the phosphorus nutrition, soil water status, weeds, pests, and disease. Researchers from Tasmanian Alkaloids successfully attempted to arrest the biosynthetic pathway of morphine at thebaine by means of mutagenesis. The most valuable selection was named Norman. The latex of this plant contained only two major alkaloids, thebaine and oripavine, with no morphine or codeine. The development of the Norman poppy paralleled the production of a slow-release formulation of oxycodone in USA.11 Thus a two-step sequence involving the oxidation of thebaine with peroxy acid to 14-hydroxycodeinone, followed by hydrogenation of the enone moiety, produces oxycodone (5), first developed by Freund and Speyer in 1916 in Germany in an attempt to improve the analgesic properties of morphine.12 This is now the method of choice for large-scale manufacturing13 of this important analgesic.14 Similarly, oripavine is converted to oxymorphone (6), which is essential for the preparation of the important antagonists such as naloxone (7), naltrexone (8), and the mixed agonist−antagonist nalbuphine (9) (Figure 1). Its preparation is more arduous because it requires several steps for the N-demethylation/alkylation sequence.15 For more than 25 years, our own group has pursued a chemoenzymatic strategy toward the synthesis of morphine alkaloids by utilizing cis-dihydrodiols derived by the toluenedioxygenase-mediated dihydroxylation of aromatic substrates. A summary of our efforts was published in 2015.16 Access to cis-dihydrodiols is accomplished by the fermentation of aromatic substrates with a recombinant strain of E. coli JM109 (pDTG601A), developed by Gibson and overexpressing toluene dioxygenase.17 The applications of these metabolites in the enantioselective synthesis of natural products have been extensively reviewed.18 It seemed prudent to us to pursue a direct synthesis of oxycodone from commercially available materials rather than a synthesis of any of the natural morphinans, given that only two such accomplishments have already been reported by Fukuyama in
■
RESULTS AND DISCUSSION The first de novo total synthesis of oxycodone, reported by Fukuyama and coworkers,19 was accomplished in 24 steps from 2-bromoisovanillin (10) with an overall yield of 0.016% (Scheme 1). The key steps included direct intramolecular arylation to form the tricyclic core, an oxidative dearomatization to install the C-14 hydroxyl group, an intramolecular Michael addition, and a Hofmann rearrangement/lactamization cascade to introduce the amino group and to construct the D-ring. Evans’ oxazolidinone chiral auxiliary established the chirality and absolute stereochemistry, as shown in Scheme 1. Four years later, Chen achieved the second synthesis of oxycodone in 16 steps starting from MOMprotected 2-bromoisovanillin (15).20 This synthesis shares some of the key features with the one reported by Fukuyama.19 The C-14 hydroxyl group was installed early in the synthesis by means of oxidative dearomatization, followed by Rovis desymmetrization of peroxyquinol 16 to introduce the chirality (Scheme 1). Heck cyclization allowed the construction of a B-ring, followed by the Stork−Ueno reaction to form the C-13 quaternary center. Amidation and the introduction of the olefin provided the allylic alcohol 19, which successfully underwent Parker-type cyclization to complete the synthesis of (−)-oxycodone. We published our first-generation synthesis of oxycodone in early 2019 (Scheme 2).21 The synthesis, starting with the enzymatic dihydroxylation of phenethyl acetate to diene-diol 20, was completed in 13 operations (19 steps) and in an overall yield of 1.5%. The crucial steps involved a Mitsunobu reaction to couple the A- and C-rings, a Heck cyclization to 10884
DOI: 10.1021/jacs.9b05033 J. Am. Chem. Soc. 2019, 141, 10883−10904
Article
Journal of the American Chemical Society Scheme 1. Synopses of Fukuyama’s and Chen’s Syntheses of Oxycodone
numbering) could be achieved via the dihydroxylation/ elimination sequence of intermediate 33. Heck cyclization would allow the construction of a dihydrofuran ring and a C13 quaternary center. The coupling of 2-bromoisovanillin (10) (to become the A-ring) and cis-dihydrodiol (20) (to become the C-ring) could be performed using the Mitsunobu reaction. cis-Dihydrodiol (20) is produced from phenethyl acetate (35) by enzymatic dihydroxylation, as previously reported.21 The first Henry reaction would allow the introduction of nitrogen functionality, whereas the second would produce the B-ring with two stereocenters at C-14 and C-9 (Scheme 3). We were hopeful for a cis stereochemical relationship between the resulting hydroxyl group at C-14 and the nitro group at C-9 to facilitate the closure of the D-ring. Thus we envisioned two different pathways for the transformation of amino alcohol 30 to ent-oxycodone (5) via Parker-type23 and Fukuyama-type chemistry,19 as shown in Scheme 4. The details of the second-generation synthesis of ent-oxycodone follow. Synthesis of Amino Alcohol 30. Our synthesis began with the enzymatic dihydroxylation of phenethyl acetate (35) by whole-cell fermentation with E. coli JM109 (pDTG601A) to produce cis-dihydrodiol 20,24 which was immediately subjected to the selective reduction of the less substituted
construct the tricyclic core, the introduction of the keto group with the subsequent SmI2-mediated closure of the Bring, the simultaneous formation of the C-14 hydroxyl group, and Parker’s hydroamination to furnish the D-ring. Most recently, Opatz prepared oxycodone in 11 steps from methyl gallate (25) (Scheme 2).22 The key transformations employed a Bischler−Napieralski reaction, followed by a Noyori asymmetric transfer hydrogenation to obtain the core of (R)-reticuline (Scheme 2). Anodic coupling allowed the cyclization of the B-ring and furnished tetracyclic intermediate 27. The dihydrofuran ring was constructed by means of conjugate nucleophilic substitution by the phenol liberated from acetate 27. The C-14 hydroxyl group was introduced by the [4 + 2] cycloaddition of singlet oxygen to the diene in 28. Second-Generation Approach. Following the completion of our first approach, we embarked on subsequent generations of design to shorten and improve the synthesis. We envisioned a second-generation approach where the C-9 and C-14 stereogenic centers are created by means of nitroalkyl anionic addition to a C-14 ketone (morphine numbering). We assumed that ent-oxycodone (5) could be simplified to aminoalcohol 30, which could be obtained from ketoaldehyde 32 by means of a double Henry reaction (Scheme 3). The formation of the ketone at C-14 (morphine 10885
DOI: 10.1021/jacs.9b05033 J. Am. Chem. Soc. 2019, 141, 10883−10904
Article
Journal of the American Chemical Society Scheme 2. Synopses of Hudlicky’s and Opatz’s Syntheses of Oxycodone
Scheme 3. Retrosynthetic Analysis for the Second-Generation Synthesis
double bond with diimide (Scheme 3).25 The protection of the distal, more accessible hydroxyl group as its silyl ether was followed by the Mitsunobu coupling of the allylic alcohol
37 to 2-bromoisovanillin (10)26 to afford intermediate 34 (Scheme 5). A reagent system27 consisting of N,N,N′,N′tetramethylazodicarboxamide (TMAD) and n-Bu3P was 10886
DOI: 10.1021/jacs.9b05033 J. Am. Chem. Soc. 2019, 141, 10883−10904
Article
Journal of the American Chemical Society
tion sequence. (These are described in detail in the Supporting Information, Part 2.) In the final and most successful route, olefin 33 was dihydroxylated and converted to monomesylate 38, whose subsequent DBU-mediated elimination, via the formation of an intermediate lactol 39, furnished ketone 32 in high yield (80%). Ketoaldehyde 32 was subjected to the first Henry reaction, according to a procedure reported by Magnus,28 to provide the conjugated nitroalkene, which was selectively reduced to nitroalkane 31 in 88% yield with Hantzsch ester29 (Scheme 5). The use of NaCNBH3 in AcOH gave only 25% of the desired product, and magnesium in MeOH reduced the keto group along with the olefin. The key nitroalkane 31 was subjected to the DBU-mediated cyclization (the use of Et3N or N-ethyl-N,N-diisopropylamine (DIPEA) showed
