The Cyclic Nitronate Route to Pharmaceutical Molecules: Synthesis of GSK’s Potent PDE4 Inhibitor as a Case Study

An efficient asymmetric synthesis of GlaxoSmithKline’s potent PDE4 inhibitor was accomplished in eight steps from a catechol-derived nitroalkene. The key intermediate (3-acyloxymethyl-substituted 1,2-oxazine) was prepared in a straightforward manner by tandem acylation/(3,3)-sigmatropic rearrangement of the corresponding 1,2-oxazine-N-oxide. The latter was assembled by a (4 + 2)-cycloaddition between the suitably substituted nitroalkene and vinyl ether. Facile acetal epimerization at the C-6 position in 1,2-oxazine ring was observed in the course of reduction with NaBH3CN in AcOH. Density functional theory (DFT) calculations suggest that the epimerization may proceed through an unusual tricyclic oxazolo(1,2)oxazinium cation formed via double anchimeric assistance from a distant acyloxy group and the nitrogen atom of the 1,2-oxazine ring.

Our group has a long-term interest in developing another approach towards the modification of cyclic nitronates, which utilizes C-H functionalization of the position next to the nitronate group (α-C-atom, Scheme 1c) [17]. Some time ago, we demonstrated that upon silylation, cyclic nitronates 1 and 2 are transformed into N-siloxyenamines 4, in which the double bond is shifted to the exocyclic α-position [18]. Enamines 4 exhibit umpolung reactivity and react with nucleophiles in the presence of Lewis acids (LA) to give α-substituted cyclic oxime ethers 5 (1,2-oxazines or isoxazolines) via S N ' substitution of TMSO-group (Scheme 1c). Using this approach, nucleophilic halogenation [19], oxygenation [20][21][22], azidination [23] of cyclic nitronates were performed (route 1). Although we succeeded in using this methodology in the total synthesis of some pharmaceutical molecules, in Scheme 1. Approaches towards modification of cyclic nitronates. (a) Synthesis of cyclic nitronates from nitroalkenes; (b) 1,3-Dipolar addition with cyclic nitronates and its application in total synthesis; (c) C-H functionalization of cyclic nitronates via N-siloxyenamines (route 1); (d) C-H functionalization of cyclic nitronates via [3,3]-rearrangement of N-acyloxyenamines (route 2). of nitroalkene 8 with the vinyl ether 9 bearing Whitesell's chiral auxiliary group ((+)-trans-2-phenyl-1-cyclohexanol ether).  6 6 separation from C-6 epimer Oxygenation of the methyl group in the N-oxide 10 proved to be challenging, and several methods were tested (Scheme 3). N-Siloxy,N-oxyenamine 16 was generated from the N-oxide 10 under mild conditions and then subjected to the LA-assisted nucleophilic addition of the bromide anion [24,25]. However, the desired 3-bromomethyl-1,2-oxazine 11, which served as a precursor to nitrate 12, was formed in moderate yields (best results are shown in Scheme 3a). Another issue was the epimerization of the sensitive C-6 acetal stereocenter leading to a mixture of 4,6-trans/4,6-cisdiastereomers 11 and 11′, which had to be separated by column chromatography (Scheme 3a,b). Unfortunately, the epimer 11′ could not be used in the synthesis of CMPO as it produced the undesired 3,4-cis-stereoisomer upon the reduction of the C=N bond in 5,6-dihydro-4H-1,2-oxazine ring on the later stages of the synthesis (Scheme 3c) [25].
The reason for the epimerization may lie in the mechanism of the LA-promoted reaction of Noxyenamines with nucleophiles, which involves heterolytic cleavage of the N-O bond (Scheme 3b). Experimental [20,22] and computational data [22] suggest that the SN' substitution of the TMSOgroup may proceed through an epimerizable N-vinyl,N-oxynitrenium cation C1. In our later study, we were able to optimize the epimerization ratio to 6: 1 by using Cr(NO3)3 as both a mild Lewis acid and the source of the nitrate anion [20]. However, the yield of nitroxy-derivative 12 was still not very high (ca. 40% from nitronate 10, Scheme 3a). Thus, further optimization of the C-H functionalization stage was reasonable. Oxygenation of the methyl group in the N-oxide 10 proved to be challenging, and several methods were tested (Scheme 3). N-Siloxy,N-oxyenamine 16 was generated from the N-oxide 10 under mild conditions and then subjected to the LA-assisted nucleophilic addition of the bromide anion [24,25]. However, the desired 3-bromomethyl-1,2-oxazine 11, which served as a precursor to nitrate 12, was formed in moderate yields (best results are shown in Scheme 3a). Another issue was the epimerization of the sensitive C-6 acetal stereocenter leading to a mixture of 4,6-trans/4,6-cis-diastereomers 11 and 11 , which had to be separated by column chromatography (Scheme 3a,b). Unfortunately, the epimer 11 could not be used in the synthesis of CMPO as it produced the undesired 3,4-cis-stereoisomer upon the reduction of the C=N bond in 5,6-dihydro-4H-1,2-oxazine ring on the later stages of the synthesis (Scheme 3c) [25].
The reason for the epimerization may lie in the mechanism of the LA-promoted reaction of N-oxyenamines with nucleophiles, which involves heterolytic cleavage of the N-O bond (Scheme 3b). Experimental [20,22] and computational data [22] suggest that the S N ' substitution of the TMSO-group may proceed through an epimerizable N-vinyl,N-oxynitrenium cation C1. In our later study, we were able to optimize the epimerization ratio to 6: 1 by using Cr(NO 3 ) 3 as both a mild Lewis acid and the source of the nitrate anion [20]. However, the yield of nitroxy-derivative 12 was still not very high (ca. 40% from nitronate 10, Scheme 3a). Thus, further optimization of the C-H functionalization stage was reasonable.

Results
We speculated that the pericyclic (3,3)-rearrangement of N-acyloxyenamine intermediate I1 generated by the acylation of nitronate 10 may proceed without any epimerization of the C-6 stereogenic center. To test this idea, cyclic nitronate 10 was treated with pivaloyl chloride/Et3N (1.5/2.0 equiv.) under conditions previously optimized for model 1,2-oxazine-N-oxides (MeCN, −30 °C, 2 h) [26]. The desired pivalate 17 was formed in a 61% yield together with some amount of unreacted N-oxide 10. After a short optimization of conditions, we found that the use of a bigger access of the PivCl/Et3N system (2.0/2.5 equiv.) and prolonged reaction time (18 h) at lower temperature resulted in an increase in the yield up to 76% (Scheme 4). Gratifyingly, no noticeable epimerization at the C-6 position was observed under these conditions.

Results
We speculated that the pericyclic (3,3)-rearrangement of N-acyloxyenamine intermediate I1 generated by the acylation of nitronate 10 may proceed without any epimerization of the C-6 stereogenic center. To test this idea, cyclic nitronate 10 was treated with pivaloyl chloride/Et 3 N (1.5/2.0 equiv.) under conditions previously optimized for model 1,2-oxazine-N-oxides (MeCN, −30 • C, 2 h) [26]. The desired pivalate 17 was formed in a 61% yield together with some amount of unreacted N-oxide 10. After a short optimization of conditions, we found that the use of a bigger access of the PivCl/Et 3 N system (2.0/2.5 equiv.) and prolonged reaction time (18 h) at lower temperature resulted in an increase in the yield up to 76% (Scheme 4). Gratifyingly, no noticeable epimerization at the C-6 position was observed under these conditions. We further investigated whether pivalate 17 could be used in the synthesis of CMPO. Hydrogenolysis of the pivalate group in 17 (to give alcohol 13) prior the reduction is challenging since 5,6-dihydro-4H-1,2-oxazines are known to undergo fragmentation via a retro-[4+2]-cycloaddition process under the action of bases [31]. Therefore, pivalate 17 was subjected to the hydride reduction with NaBH 3 CN in acetic acid (Scheme 5). Surprisingly, the reaction produced two separable isomeric products 18 and 18 in 3: 1 ratio (62% combined yield, 91% based on converted 17). From the coupling constants in 1 H-NMR spectra, it was deduced that both isomers had trans-arrangement of the substituents at the C-3 and C-4 atoms, while the configuration of the C-6 stereocenter was different. Thus, the C-6 acetal moiety underwent epimerization in the course of the reduction (see Discussion section). The amount of 4,6-cis-isomer 18 increased with time, demonstrating that the isomerization took place in the reduced product 18 and not in the starting compound 17. This is also confirmed by the fact that stereisomers 18 and 18 had same configuration of the newly formed C-3 stereocenter. If the epimerization preceded the reduction, the C-6 epimerized 5,6-dihydro-4H-1,2-oxazine 17 would also give the reduced product with the 3,4-cis-disposition of substituents (see Scheme 3c). We further investigated whether pivalate 17 could be used in the synthesis of CMPO. Hydrogenolysis of the pivalate group in 17 (to give alcohol 13) prior the reduction is challenging since 5,6-dihydro-4H-1,2-oxazines are known to undergo fragmentation via a retro-[4+2]cycloaddition process under the action of bases [31]. Therefore, pivalate 17 was subjected to the hydride reduction with NaBH3CN in acetic acid (Scheme 5). Surprisingly, the reaction produced two separable isomeric products 18 and 18′ in 3: 1 ratio (62% combined yield, 91% based on converted 17). From the coupling constants in 1 H-NMR spectra, it was deduced that both isomers had transarrangement of the substituents at the C-3 and C-4 atoms, while the configuration of the C-6 stereocenter was different. Thus, the C-6 acetal moiety underwent epimerization in the course of the reduction (see Discussion section). The amount of 4,6-cis-isomer 18′ increased with time, demonstrating that the isomerization took place in the reduced product 18 and not in the starting compound 17. This is also confirmed by the fact that stereisomers 18 and 18′ had same configuration of the newly formed C-3 stereocenter. If the epimerization preceded the reduction, the C-6 epimerized 5,6-dihydro-4H-1,2-oxazine 17′ would also give the reduced product with the 3,4-cis-disposition of substituents (see Scheme 3c).  We further investigated whether pivalate 17 could be used in the synthesis of CMPO. Hydrogenolysis of the pivalate group in 17 (to give alcohol 13) prior the reduction is challenging since 5,6-dihydro-4H-1,2-oxazines are known to undergo fragmentation via a retro-[4+2]cycloaddition process under the action of bases [31]. Therefore, pivalate 17 was subjected to the hydride reduction with NaBH3CN in acetic acid (Scheme 5). Surprisingly, the reaction produced two separable isomeric products 18 and 18′ in 3: 1 ratio (62% combined yield, 91% based on converted 17). From the coupling constants in 1 H-NMR spectra, it was deduced that both isomers had transarrangement of the substituents at the C-3 and C-4 atoms, while the configuration of the C-6 stereocenter was different. Thus, the C-6 acetal moiety underwent epimerization in the course of the reduction (see Discussion section). The amount of 4,6-cis-isomer 18′ increased with time, demonstrating that the isomerization took place in the reduced product 18 and not in the starting compound 17. This is also confirmed by the fact that stereisomers 18 and 18′ had same configuration of the newly formed C-3 stereocenter. If the epimerization preceded the reduction, the C-6 epimerized 5,6-dihydro-4H-1,2-oxazine 17′ would also give the reduced product with the 3,4-cis-disposition of substituents (see Scheme 3c). Finally, deprotection of the N-Boc moiety with TFA and treatment of the resulting prolinol trifluoroacetate with Im2CO/Et3N afforded the desired PDE4 inhibitor CMPO (Scheme 6). Thus, the asymmetric synthesis of PDE4 inhibitor CMPO was completed in seven steps from a known nitroalkene 8 in 8% overall yield. Chiral HPLC analysis revealed the enantiomeric purity of the product > 97% ee. The racemic sample of CMPO for HPLC analysis was prepared according to the same synthetic sequence starting from the racemic trans-2-phenylcyclohexanol.

Discussion
Epimerization of the acetal moiety in the course of the hydride reduction of 5,6-dihydro-4-H-1,2oxazine 17 is of special note. In the previously reported hydride reduction of 1,2-oxazine 13 possessing a free hydroxymethyl group, no epimerization at the C-6 atom was observed (cf. data in Scheme 2; Scheme 5) [24]. We hypothesized that such a difference in the behavior of 3hydroxymethyl-and 3-acyloxymethyl-substituted 1,2-oxazines 13 and 17 may be attributed to an anchimeric assistance from the carbonyl group, which stabilizes the intermediate cation C2 [33] by forming a bridged system with an eight-membered ring (cation C3). In carbohydrates, a similar anchimeric assistance of the acyloxy group from the distant 1,4-position has been proposed, yet it was not confirmed unambiguously by experimental data [34][35][36][37]. In our case, density functional theory (DFT) calculations at the MN15/Def2TZVP level of theory (see Supplementary material for On the next step, careful saponification of the pivalate moiety in pyrrolidine 19 with KOH in aqueous methanol gave Boc-prolinol 20 in 84% yield (Scheme 6). It is noteworthy that hydrolysis of the Boc-group was also observed to some extent under these conditions. For this reason, the reaction mixture was treated with Boc 2 O after neutralization to convert the unprotected prolinol into the N-Boc-derivative 20, which was isolated by column chromatography.
Finally, deprotection of the N-Boc moiety with TFA and treatment of the resulting prolinol trifluoroacetate with Im 2 CO/Et 3 N afforded the desired PDE4 inhibitor CMPO (Scheme 6). Thus, the asymmetric synthesis of PDE4 inhibitor CMPO was completed in seven steps from a known nitroalkene 8 in 8% overall yield. Chiral HPLC analysis revealed the enantiomeric purity of the product >97% ee. The racemic sample of CMPO for HPLC analysis was prepared according to the same synthetic sequence starting from the racemic trans-2-phenylcyclohexanol.

Discussion
Epimerization of the acetal moiety in the course of the hydride reduction of 5,6-dihydro-4-H-1,2-oxazine 17 is of special note. In the previously reported hydride reduction of 1,2-oxazine 13 possessing a free hydroxymethyl group, no epimerization at the C-6 atom was observed (cf. data in Scheme 2; Scheme 5) [24]. We hypothesized that such a difference in the behavior of 3-hydroxymethyl-and 3-acyloxymethyl-substituted 1,2-oxazines 13 and 17 may be attributed to an anchimeric assistance from the carbonyl group, which stabilizes the intermediate cation C2 [33] by forming a bridged system with an eight-membered ring (cation C3). In carbohydrates, a similar anchimeric assistance of the acyloxy group from the distant 1,4-position has been proposed, yet it was not confirmed unambiguously by experimental data [34][35][36][37]. In our case, density functional theory (DFT) calculations at the MN15/Def2TZVP level of theory (see Supplementary material for details) revealed that the bridged cation C3 is much less stable compared to the initial monocyclic cation C2. Interestingly, the formation of a third ring between the nitrogen atom and the acyloxy group may lead to a tricyclic cation C4, which is predicted to be much more stable than the mono-or bicyclic structures C2 and C3 (Scheme 7). The formation of such a stable tricyclic cation as an intermediate or a resting state may account for the observed facile epimerization of pivalate 18. The higher thermodynamic stability of the 4,6-cis-isomer 18 over the 4,6-trans-isomer 18, as shown by DFT calculations, is likely to be the driving force for the epimerization at the C-6 atom.
Molecules 2020, 25, x 7 of 13 details) revealed that the bridged cation C3 is much less stable compared to the initial monocyclic cation C2. Interestingly, the formation of a third ring between the nitrogen atom and the acyloxy group may lead to a tricyclic cation C4, which is predicted to be much more stable than the mono-or bicyclic structures C2 and C3 (Scheme 7). The formation of such a stable tricyclic cation as an intermediate or a resting state may account for the observed facile epimerization of pivalate 18. The higher thermodynamic stability of the 4,6-cis-isomer 18′ over the 4,6-trans-isomer 18, as shown by DFT calculations, is likely to be the driving force for the epimerization at the C-6 atom. Another remarkable observation was that 1,2-oxazine 18 (as well as its precursor 17) did not undergo epimerization in acetic acid (rt, 2 h). Isomerization to the cis-isomer 18′ was observed only in the presence of NaBH3CN. Hence, the fragmentation of the acetal moiety is most likely promoted by some Lewis acidic boron species generated from NaBH3CN in acidic medium. Indeed, slow epimerization of 1,2-oxazine 18 was observed upon treatment of 18 with B(OBu)3 or BF3·Et2O in acetic acid.
It is noteworthy that an anchimeric-assisted epimerization in 1,2-oxazine series has not been reported previously. Moreover, to our knowledge, this is the first reported example of a remote neighboring group participation from the 1,4-position in a six-membered ring confirmed by DFT calculations [38][39][40]. The formation of an eight-membered ring in this case may be driven by an unusual secondary anchimeric interaction involving the nitrogen atom of the 1,2-oxazine ring leading to an unusual tricyclic oxazolo(1,2)oxazinium cation C4.

Materials and Methods
All reactions were carried out in oven-dried (150 °C) glassware. NMR spectra (Bruker AM 300 spectrometer, Karlsruhe, Germany) were recorded at room temperature (if not stated otherwise) with residual solvents peaks as an internal standard. Peak multiplicities are indicated by s (singlet), d (doublet), t (triplet), dd (doublet of doublets), q (quartet), quint (quintet), ddd (doublet of doublets of doublets), tt (triplet of triplets), tdd (triplets of doublets of doublets), m (multiplet), br (broad). The numeration of atoms used in the assignment of NMR spectra is given in Figure 1.
Another remarkable observation was that 1,2-oxazine 18 (as well as its precursor 17) did not undergo epimerization in acetic acid (rt, 2 h). Isomerization to the cis-isomer 18 was observed only in the presence of NaBH 3 CN. Hence, the fragmentation of the acetal moiety is most likely promoted by some Lewis acidic boron species generated from NaBH 3 CN in acidic medium. Indeed, slow epimerization of 1,2-oxazine 18 was observed upon treatment of 18 with B(OBu) 3 or BF 3 ·Et 2 O in acetic acid.
It is noteworthy that an anchimeric-assisted epimerization in 1,2-oxazine series has not been reported previously. Moreover, to our knowledge, this is the first reported example of a remote neighboring group participation from the 1,4-position in a six-membered ring confirmed by DFT calculations [38][39][40]. The formation of an eight-membered ring in this case may be driven by an unusual secondary anchimeric interaction involving the nitrogen atom of the 1,2-oxazine ring leading to an unusual tricyclic oxazolo(1,2)oxazinium cation C4.

Materials and Methods
All reactions were carried out in oven-dried (150 • C) glassware. NMR spectra (Bruker AM 300 spectrometer, Karlsruhe, Germany) were recorded at room temperature (if not stated otherwise) with residual solvents peaks as an internal standard. Peak multiplicities are indicated by s (singlet), d (doublet), t (triplet), dd (doublet of doublets), q (quartet), quint (quintet), ddd (doublet of doublets of doublets), tt (triplet of triplets), tdd (triplets of doublets of doublets), m (multiplet), br (broad). The numeration of atoms used in the assignment of NMR spectra is given in Figure 1.
Oxygenation of nitronate 10. Enantipure or racemic nitronate 10 (347 mg, 0.72 mmol) was dissolved in dry acetonitrile (1.7 mL) in a Schlenk tube under argon atmosphere, and then Et3N (251 μL, 1.8 mmol) was added. The solution was cooled to −40 °C and pivaloyl chloride (174 μL, 1.41 mmol) was added. The reaction mixture was stirred at ca. −40 °C for 2 h and then kept in a freezer (ca. −25 °C) overnight. The mixture was diluted with EtOAc (5 mL) and transferred into a separating funnel containing EtOAc (20 mL) and 0.25 M aq. NaHSO4 solution (20 mL). The aqueous layer was extracted with EtOAc (20 mL), the combined organic layers were washed with water (30 mL) and brine (30 mL), dried over anhydrous Na2SO4 and concentrated under reduced pressure. The residue was subjected to a column chromatography on silica gel (Hexane/EtOAc = 10/1) to give 311 mg (76%) of enantiopure or racemic pivalate 17.

Conclusions
In conclusion, we were able to solve the problem of site-selective C-H oxygenation of the cyclic nitronate intermediate in the asymmetric synthesis of a potent PDE4 inhibitor CMPO by using tandem acylation/(3,3)-sigmatropic rearrangement. In comparison with the previous synthesis, this method afforded the required 3-oxymethyl-substituted 1,2-oxazine intermediate in a much higher yield (76% vs. 27%). This key intermediate could be readily converted into the target (−)-CMPO by the reductive contraction of the 1,2-oxazine ring followed by deprotection and carbamylation with Im 2 CO. A rapid epimerization of the C-6 acetal moiety was observed upon the reduction of the 5,6-dihydro-4H-1,2-oxazine ring with NaBH 3 CN in acetic acid. DFT calculations suggest that the epimerization is favored by an unprecedented double anchimeric assistance from a remote acyloxy group and the nitrogen atom of the 1,2-oxazine ring.