Chemoenzymatic Total Synthesis of (+)-10-Keto-Oxycodone from Phenethyl Acetate †

The total synthesis of (+)-10-keto-oxycodone was attained from phenethyl acetate in a stereoselective manner. Absolute stereochemistry was established via enzymatic dihydroxylation of phenethyl acetate with the recombinant strain JM109 (pDTG601A) that furnished the corresponding cis-cyclohexadienediol whose configuration corresponds to the absolute stereochemistry of the ring C of (+)-10-keto-oxycodone. Intramolecular Heck reaction was utilized to establish the quaternary carbon at C-13, along with the dibenzodihydrofuran functionality. The C-14 hydroxyl and C-10 ketone were installed via SmI2-mediated radical cyclization, and oxidation of a benzylic alcohol (obtained from an intermediate nitrate azide), respectively. The synthesis of (+)-10-keto-oxycodone was completed in a total of 14 operations (21 steps) and an overall yield of ~2%. Experimental and spectral data are provided for key intermediates and new compounds.


Introduction
Interest in the preparation of 10-keto opiates and related derivatives such as the 10-hydroxy-morphinans was aimed at pursuing κ-selective analgesics [1]. Of the three types of opioid receptors (µ, δ, κ), the κ-opioid receptor has been especially interesting because its activation produces analgesia with minimal physical dependence [2]. A highly selective κ-opioid agonist may provide a useful analgesic free from abusive potential [3]. Among the efforts directed toward the preparation of the afore-mentioned 10-keto opiates are those aimed at the preparation of 10-keto, 10α-, 10β-hydroxy-TRK-820 [4], 10-keto-naltrexone, 10-keto-oxymorphone, 10-keto-oxycodone ( Figure 1) [5], 10-keto-naloxone, and 10-keto-naloxone 3-methyl ether [6] from the corresponding opiates. These preparations are all conducted from natural morphinans by semi-syntheses. Furthermore, the above 10-keto opiates were also identified as purported intermediates in the decomposition of the parent opiates upon storage [6]. As part of our continuing program in the preparation of morphine alkaloids and related compounds [7,8], we recently had the opportunity to access a suitable intermediate that could potentially be converted to ent-10-keto-oxycodone.
The published methods for the preparation of 10-keto-oxycodone and 10-hydroxy-oxycodone involve the direct oxidation of oxycodone with chromium [5,9,10], or selenium dioxide [11] reagents. Recently, the use of ceric ammonium nitrate (CAN) for such an oxidation led exclusively to the 10-hydroxy product, further oxidized to the C-10-ketone with the use of stronger oxidants such as Dess-Martin periodinane [9,10]. Oxycodone, on the other hand, is prepared in two-steps from thebaine by the oxidation of the diene moiety with a peroxyacid to an enone and subsequent hydrogenation [12][13][14][15][16]. The amount of available thebaine, itself a minor constituent of opium, limits the production of oxycodone. Recently, thebaine and oripavine have become available from genetically modified poppies that produce much higher percentages of these alkaloids [17][18][19][20], now supplied by Tasmanian Alkaloids, Inc. [21]. Since the milestone synthesis of morphine by Gates [22] in 1952, there have been more than 30 total syntheses of morphine and related alkaloids and the academic pursuit continues unabated [23][24][25][26][27]. None of the reported syntheses are practical, however, for an industrial scale production. Even the most efficient academic synthesis, published by Rice [28], may not be amenable for scale-up in the industrial preparation of morphinans. Although the development of a truly practical total synthesis of any morphinan or an opiate-derived agent on a commercial scale seems to be not feasible in the near future, we have attempted to develop methods for the synthesis of oxycodone and derivatives from readily available starting materials. A de novo preparation of oxycodone and related medicinal opiate-derived agents for medicinal use could alleviate the negative impacts of any future unfavorable events that may limit the supply of natural sources-such as climate or political changes in the opium-producing areas.
Molecules 2019, 24, x 2 of 12 poppies that produce much higher percentages of these alkaloids [17][18][19][20], now supplied by Tasmanian Alkaloids, Inc. [21]. Since the milestone synthesis of morphine by Gates [22] in 1952, there have been more than 30 total syntheses of morphine and related alkaloids and the academic pursuit continues unabated [23][24][25][26][27]. None of the reported syntheses are practical, however, for an industrial scale production. Even the most efficient academic synthesis, published by Rice [28], may not be amenable for scale-up in the industrial preparation of morphinans. Although the development of a truly practical total synthesis of any morphinan or an opiate-derived agent on a commercial scale seems to be not feasible in the near future, we have attempted to develop methods for the synthesis of oxycodone and derivatives from readily available starting materials. A de novo preparation of oxycodone and related medicinal opiate-derived agents for medicinal use could alleviate the negative impacts of any future unfavorable events that may limit the supply of natural sources-such as climate or political changes in the opium-producing areas.

Results and Discussion
Scheme 1 outlines our retrosynthesis of (+)-10-keto-oxycodone. Disconnection of ring D leads to highly functionalized lactone 2. In the forward sense, Fukuyama's lactamization approach to oxycodone [29] can be utilized to construct the C-9 stereogenic center. The amino group required for the cyclization reaction is derived from hydroxyazidation of alkene lactone 3 [30,31], which is envisioned to be prepared from the keto acetal 4 following a deprotection of acetal and a SmI2mediated pinacol-type coupling reaction. The key intermediate 4 can be obtained from alkene 5 via dihydroxylation followed by selective mesylation of the less hindered hydroxyl group and the elimination of the mesylate to reveal the ketone functionality in 4. Alkene 5 is available in two steps from alcohol 7 via a sequence of steps that involves a Mitsunobu coupling with an iodo phenol acetal to furnish aryl ether 6, followed by an intramolecular Heck reaction. The absolute stereochemistry in 7 is incorporated via microbial dihydroxylation of phenethyl acetate 8 in the whole-cell fermentation with toluene dioxygenase, overexpressed in E. coli JM109(pDTG601A) [32]. The enzymatically derived arene cis-dihydrodiols such as 7 have found widespread use in enantioselective synthesis of natural products [33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51].

Results and Discussion
Scheme 1 outlines our retrosynthesis of (+)-10-keto-oxycodone. Disconnection of ring D leads to highly functionalized lactone 2. In the forward sense, Fukuyama's lactamization approach to oxycodone [29] can be utilized to construct the C-9 stereogenic center. The amino group required for the cyclization reaction is derived from hydroxyazidation of alkene lactone 3 [30,31], which is envisioned to be prepared from the keto acetal 4 following a deprotection of acetal and a SmI 2 -mediated pinacol-type coupling reaction. The key intermediate 4 can be obtained from alkene 5 via dihydroxylation followed by selective mesylation of the less hindered hydroxyl group and the elimination of the mesylate to reveal the ketone functionality in 4. Alkene 5 is available in two steps from alcohol 7 via a sequence of steps that involves a Mitsunobu coupling with an iodo phenol acetal to furnish aryl ether 6, followed by an intramolecular Heck reaction. The absolute stereochemistry in 7 is incorporated via microbial dihydroxylation of phenethyl acetate 8 in the whole-cell fermentation with toluene dioxygenase, overexpressed in E. coli JM109(pDTG601A) [32]. The enzymatically derived arene cis-dihydrodiols such as 7 have found widespread use in enantioselective synthesis of natural products [33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51].
As previously described in our earlier publications on the preparation of ent-oxycodone [30,31], the synthesis began with the microbial dihydroxylation of phenethyl acetate 8 (Scheme 2) via a whole cell fermentation with E. coli JM109 (pDTG601A) to afford the known intermediate cyclohexadiene diol 7 (obtained in 5gL −1 yield) [52,53], which was subjected to a selective reduction of the more accessible alkene to afford the known diol 9 (85% yield) [54,55]. The distal, less hindered, hydroxyl in diol 9 was protected with tert-butyl dimethylsilyl chloride and the free allylic alcohol was then subjected to a Mitsunobu reaction with iodo phenol 10 [56], derived from isovanillin, to furnish the coupled product ether 6 (45% yield over two steps). A subsequent intramolecular Heck reaction of 6 resulted in the formation of olefin 5 (87% yield) whose osmium tetroxide-catalyzed dihydroxylation led to diol 11 (81% yield). This compound possesses the features of the three (ACE) rings of 10-ketooxycodone. The diol functionality was converted to ketone 4 via mesylation of the less hindered hydroxyl group followed by elimination with excess 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU). The mechanism may also be viewed formally as a 1,2-hydride shift. [57] of the resulting mesylate (63% yield over two steps). With the attainment of 4, deprotection of the acetal with aqueous TFA followed by a pinacol-type coupling of the intermediate keto aldehyde using SmI2 afforded diol 12, tentatively assigned as the cis isomer (65% yield over two steps) [58]. Mesylation of the less hindered hydroxyl group of diol 12 followed by treatment with excess DBU resulted in the formation of styrene alcohol 13 (56% yield over two steps). Hydrolysis of the acetate followed by oxidation with tetrapropylammonium perruthenate (TPAP) in the presence of N-methylmorpholine N-oxide (NMO) resulted in the formation of the key intermediate lactone 3 (91% over two steps). Scheme 1. Retrosynthetic analysis of the route to ent-10-keto-oxycodone.
As previously described in our earlier publications on the preparation of ent-oxycodone [30,31], the synthesis began with the microbial dihydroxylation of phenethyl acetate 8 (Scheme 2) via a whole cell fermentation with E. coli JM109 (pDTG601A) to afford the known intermediate cyclohexadiene diol 7 (obtained in 5gL −1 yield) [52,53], which was subjected to a selective reduction of the more accessible alkene to afford the known diol 9 (85% yield) [54,55]. The distal, less hindered, hydroxyl in diol 9 was protected with tert-butyl dimethylsilyl chloride and the free allylic alcohol was then subjected to a Mitsunobu reaction with iodo phenol 10 [56], derived from isovanillin, to furnish the coupled product ether 6 (45% yield over two steps). A subsequent intramolecular Heck reaction of 6 resulted in the formation of olefin 5 (87% yield) whose osmium tetroxide-catalyzed dihydroxylation led to diol 11 (81% yield). This compound possesses the features of the three (ACE) rings of 10-keto-oxycodone. The diol functionality was converted to ketone 4 via mesylation of the less hindered hydroxyl group followed by elimination with excess 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU). The mechanism may also be viewed formally as a 1,2-hydride shift. [57] of the resulting mesylate (63% yield over two steps). With the attainment of 4, deprotection of the acetal with aqueous TFA followed by a pinacol-type coupling of the intermediate keto aldehyde using SmI 2 afforded diol 12, tentatively assigned as the cis isomer (65% yield over two steps) [58]. Mesylation of the less hindered hydroxyl group of diol 12 followed by treatment with excess DBU resulted in the formation of styrene alcohol 13 (56% yield over two steps). Hydrolysis of the acetate followed by oxidation with tetrapropylammonium perruthenate (TPAP) in the presence of N-methylmorpholine N-oxide (NMO) resulted in the formation of the key intermediate lactone 3 (91% over two steps). Based on our earlier work on a similar CAN-mediated functionalization of styrene to deliver entoxycodone [31], it was envisioned that the lactone 3 could be readily converted to a hydroxy lactam en route to ent-10-keto-oxycodone (Scheme 3). It was previously reported [31] that hydrogenation of azido nitrate 14 intermediate in methanol as solvent resulted in the formation benzylic methoxy ether lactam 15. Therefore, it seemed logical that switching to non-nucleophilic solvent (e.g., EtOAc, THF) would afford the benzylic alcohol instead. To our surprise and disappointment, the reduction of azide functionality did not occur and it appears that the benzylic alcohol derived by hydrogenation of the nitrate was slowly oxidized to the ketone 16 on exposure to air during workup and further manipulation [59]. To achieve the desired reduction of the azide to amine and subsequent lactamization, the solvent had to be changed to methanol. The desired lactam formation did occur, but the product was the hemiketal 17. This route was not followed through as this would necessitate deprotecting the hemiketal to the ketone prior to reduction with borane reagent. The hemiketal, when subjected to reduction with sodium borohydride, resulted in the formation of an ether, tentatively assigned as the C-10 epimeric benzylic methoxy ether (epi-15). Based on our earlier work on a similar CAN-mediated functionalization of styrene to deliver ent-oxycodone [31], it was envisioned that the lactone 3 could be readily converted to a hydroxy lactam en route to ent-10-keto-oxycodone (Scheme 3). It was previously reported [31] that hydrogenation of azido nitrate 14 intermediate in methanol as solvent resulted in the formation benzylic methoxy ether lactam 15. Therefore, it seemed logical that switching to non-nucleophilic solvent (e.g., EtOAc, THF) would afford the benzylic alcohol instead. To our surprise and disappointment, the reduction of azide functionality did not occur and it appears that the benzylic alcohol derived by hydrogenation of the nitrate was slowly oxidized to the ketone 16 on exposure to air during workup and further manipulation [59]. To achieve the desired reduction of the azide to amine and subsequent lactamization, the solvent had to be changed to methanol. The desired lactam formation did occur, but the product was the hemiketal 17. This route was not followed through as this would necessitate deprotecting the hemiketal to the ketone prior to reduction with borane reagent. The hemiketal, when subjected to reduction with sodium borohydride, resulted in the formation of an ether, tentatively assigned as the C-10 epimeric benzylic methoxy ether (epi-15). Based on our earlier work on a similar CAN-mediated functionalization of styrene to deliver entoxycodone [31], it was envisioned that the lactone 3 could be readily converted to a hydroxy lactam en route to ent-10-keto-oxycodone (Scheme 3). It was previously reported [31] that hydrogenation of azido nitrate 14 intermediate in methanol as solvent resulted in the formation benzylic methoxy ether lactam 15. Therefore, it seemed logical that switching to non-nucleophilic solvent (e.g., EtOAc, THF) would afford the benzylic alcohol instead. To our surprise and disappointment, the reduction of azide functionality did not occur and it appears that the benzylic alcohol derived by hydrogenation of the nitrate was slowly oxidized to the ketone 16 on exposure to air during workup and further manipulation [59]. To achieve the desired reduction of the azide to amine and subsequent lactamization, the solvent had to be changed to methanol. The desired lactam formation did occur, but the product was the hemiketal 17. This route was not followed through as this would necessitate deprotecting the hemiketal to the ketone prior to reduction with borane reagent. The hemiketal, when subjected to reduction with sodium borohydride, resulted in the formation of an ether, tentatively assigned as the C-10 epimeric benzylic methoxy ether (epi-15).

Scheme 3.
Initial studies on hydroxy lactam formation. Scheme 3. Initial studies on hydroxy lactam formation.
To address the above issue of the premature oxidation of C-10, the intermediate azido nitrate 14 was subjected to nucleophilic substitution with acetate to afford an intermediate acetate azide 2 (Scheme 4), with the acetate serving as a "protecting group" against premature oxidation of C-10. An overall inversion of stereochemistry of benzylic (C-10) carbon was tentatively assigned as in 2, in contrast with the formation of 15 (retention). It is believed that 15 was formed from an intermediate benzylic alcohol upon hydrogenolysis of the nitrate ester followed by lactamization and then nucleophilic displacement of OH by methoxy group. The stereochemical properties of the intermediate benzylic alcohol lactam favored a beta face attack by methanol as nucleophile. The facile displacement of OH in the benzylic position was facilitated by the highly activating methoxy group at the para position. For the formation of epi-15, on the other hand, the hydride served as the nucleophile and thus in the product the methoxy group ended up on the alpha face. The assignment of 15 and epi-15 as beta and alpha epimers was based on the coupling constants (J 9-10 ) between the H-10 (benzylic C-H) and H-9 on similar systems [4,60]. For compounds similar to 15, J values are approximately 5.0 Hz and for epi-15, J values are around 0 Hz. The acetate azide 2 was hydrogenated to the amine, which underwent a spontaneous lactam formation to afford the acetate lactam 18 (63% yield over three steps). Reduction of the lactam 18 with borane followed by reductive amination with formaldehyde converted the lactam to the N-methyl amine 19. Because of the sensitivity (to basic and acidic reagents needed for the deprotection of C-6 hydroxyl) of the benzylic alcohol functionality in 19, it was first oxidized with manganese dioxide to the corresponding ketone to increase stability. It has been suggested by one of the referees that the benzylic alcohol 19 may be subject to a Grob-type fragmentation under basic conditions. Deprotection of the TBS group using TBAF and oxidation of the C-6 hydroxyl afforded ent-10-keto-oxycodone (61% yield over four steps). To address the above issue of the premature oxidation of C-10, the intermediate azido nitrate 14 was subjected to nucleophilic substitution with acetate to afford an intermediate acetate azide 2 (Scheme 4), with the acetate serving as a "protecting group" against premature oxidation of C-10. An overall inversion of stereochemistry of benzylic (C-10) carbon was tentatively assigned as in 2, in contrast with the formation of 15 (retention). It is believed that 15 was formed from an intermediate benzylic alcohol upon hydrogenolysis of the nitrate ester followed by lactamization and then nucleophilic displacement of OH by methoxy group. The stereochemical properties of the intermediate benzylic alcohol lactam favored a beta face attack by methanol as nucleophile. The facile displacement of OH in the benzylic position was facilitated by the highly activating methoxy group at the para position. For the formation of epi-15, on the other hand, the hydride served as the nucleophile and thus in the product the methoxy group ended up on the alpha face. The assignment of 15 and epi-15 as beta and alpha epimers was based on the coupling constants (J9-10) between the H-10 (benzylic C-H) and H-9 on similar systems [4,60]. For compounds similar to 15, J values are approximately 5.0 Hz and for epi-15, J values are around 0 Hz. The acetate azide 2 was hydrogenated to the amine, which underwent a spontaneous lactam formation to afford the acetate lactam 18 (63% yield over three steps). Reduction of the lactam 18 with borane followed by reductive amination with formaldehyde converted the lactam to the N-methyl amine 19. Because of the sensitivity (to basic and acidic reagents needed for the deprotection of C-6 hydroxyl) of the benzylic alcohol functionality in 19, it was first oxidized with manganese dioxide to the corresponding ketone to increase stability. It has been suggested by one of the referees that the benzylic alcohol 19 may be subject to a Grob-type fragmentation under basic conditions. Deprotection of the TBS group using TBAF and oxidation of the C-6 hydroxyl afforded ent-10-keto-oxycodone (61% yield over four steps).

Scheme 4. Completion of the synthesis of ent-10-keto-oxydocone via lactam 18.
In conclusion, a relatively short chemoenzymatic total synthesis of ent-10-keto-oxycodone has been accomplished in 14 operations from phenethyl acetate. To the best of our knowledge this accomplishment constitutes the first synthesis of ent-10-keto-oxycodone.

General Methods
Inoculum was obtained from viable cells stored −78 °C in cryovials. They were grown in suitable media as previously described [28]. The substrate was fed in 5 mL increments over the course of ~3 h with metabolites being harvested in the usual manner. All non-aqueous reactions were conducted in an argon atmosphere using standard Schlenk techniques for the exclusion of moisture and air. In conclusion, a relatively short chemoenzymatic total synthesis of ent-10-keto-oxycodone has been accomplished in 14 operations from phenethyl acetate. To the best of our knowledge this accomplishment constitutes the first synthesis of ent-10-keto-oxycodone.

General Methods
Inoculum was obtained from viable cells stored −78 • C in cryovials. They were grown in suitable media as previously described [28]. The substrate was fed in 5 mL increments over the course of~3 h with metabolites being harvested in the usual manner. All non-aqueous reactions were conducted in an argon atmosphere using standard Schlenk techniques for the exclusion of moisture and air. Methylene chloride was distilled from calcium hydride, THF and toluene were dried over sodium/benzophenone. Analytical thin layer chromatography was performed on Silicycle 60 A • 250 mm TLC plates with F-254 indicator. Flash column chromatography was performed using Kieselgel 60 (230-400 mesh). Melting points were recorded on a Hoover Unimelt apparatus (Hoover, VA, USA) and are uncorrected. IR spectra were obtained on a Perkin-Elmer One FT-IR spectrometer (PerkinElmer, Waltham, MA, USA) Optical rotation was measured on a Perkin-Elmer 341 polarimeter (PerkinElmer, Waltham, MA, USA) at a wavelength of 589 nm. 1 H and 13 C spectra were recorded on a 300 MHz and 400 MHz Bruker spectrometer (Bruker, Billerica, MA, USA). All chemical shifts are referenced to TMS or residual nondeuterated solvent. Data of proton spectra are reported as follows: chemical shift in ppm (multiplicity: singlet (s), doublet (d), triplet (t), quartet(q) and multiplet (m)), coupling constants [Hz], integration). Carbon spectra were recorded with complete proton decoupling and the chemical shifts are reported in ppm (δ) relative to solvent resonance as internal standard. Mass spectra and high resolution mass spectra were performed by the Analytical Division at Brock University. The 1 H and 13 C NMR spectra of some compounds are in the Supplementary Materials.

ent-10-keto-oxycodone (ent-(1))
To the crude N-methyl amine alcohol 19 (obtained from the previous step) dissolved in DCM (1.0 mL) was added MnO 2 (191 mg, 2.2 mmol). The suspension was stirred vigorously at room temperature for 1 day. The reaction mixture was filtered through a plug of Celite and the residue was washed with DCM (5 mL). The filtrate was concentrated by rotary evaporation and the residue was used as crude in the next step. To a stirred solution of N-methyl ketone (obtained from the previous step) in THF (1 mL) was added TBAF (250 µL, 0.25 mmol). The reaction mixture was stirred at room temperature for 5 h after which it was diluted with EtOAc/H 2 O (5 mL/5 mL). The layers were separated and the aqueous layer was further extracted with DCM (2 × 5 mL). The organic layers were combined, dried over MgSO 4 , filtered and concentrated to afford a residue that was used as crude in the next step. To the crude alcohol dissolved in DCM (3 mL) was added Dess-Martin periodinane (93.3 mg, 0.22 mmol). The reaction mixture was stirred at rt for 2 h. It was diluted with satd. Na 2 S 2 O 3 (1 mL), followed by satd. NaHCO 3 solution (1 mL). The layers were separated and the aqueous phase was further extracted with DCM (2 × 10 mL). The organic layers were combined, dried over MgSO 4 , filtered and concentrated to afford a residue that was chromatographed on silica gel using DCM/MeOH as eluent (9:1) to afford the product ent-(1) as a solid (24 mg, 0.07 mmol, 60.7% yield over four steps).