The Baeyer-Villiger reaction of 23-oxosapogenins

The Baeyer-Villiger reactions of 25 R - and 25 S -23-oxosapogenins with MCPBA produced a mixture of bisnorcholanic 22 → 16 lactones and pregnan-16,20-diol carbonates. The postulated mechanism explains the reaction times in terms of the transition states for the peroxyacid attack. The explanation of the regioselectivity of the studied reaction relays on the conformational preference of the Criegee´s intermediate. Reaction barriers calculated using the semiempirical PM3 method justify the observed reaction times.


Introduction
For a long time steroid sapogenins have been used as starting materials for the synthesis of different steroidal drugs, 1 and more recently on the synthesis of ecdysone, 2 spirostanic and furostanic analogues of brassinosteroids with modified side chains. 3All synthetic procedures relay on the degradation or modification of the spirostanic side chain.The reactivity of the spirostanic side chain was extensively studied during the mid part of the last century. 4More recently, new transformations have been reported, 5,6 some of these reactions have been reported to produce the cleavage of the spiroketal side chain to the C-22→16 lactone moiety characteristic of bisnorcholanic lactones. 6isnorcholanic lactones have called the attention of scientists due to their biological properties. 7Early reports show that such compounds can be produced in low yield by treatment of the corresponding steroid sapogenin with CrO 3 , HNO 3 or as a byproduct of reaction with NaNO 2 and BF 3 ⋅Et 2 O in acetic acid. 8The conversion of solasodine and tomatidine into the corresponding .The 1 H NMR spectra of the 23-oxosapogenins (Table 1) show the downfield shift of the signals of the diasterotopic H-26 ax. and H-26 eq. in both 25R and 25S series, which together with the deshielding of H-16 and H-20 have been outlined as the main effects of the presence of a carbonyl function at C-23. 11 Downfield shift of H-26 have been attributed to magnetic anisotropy of the carbonyl function meanwhile deshielding of H-20 have been rationalized as a result of the combination of a δ-syn interaction between H-20 and the carbonylic oxygen and magnetic anisotropy of the C=O bond.(all compared with the corresponding steroid sapogenin, see Table 2).
The 1 H NMR spectra of bisnorcholanic lactones 3a-c (Figure 3) are characterized by the absence of the signals corresponding to the protons 23 to 27 and the downfield shift of the signals corresponding to H-16 (now part of the lactone moiety) and H-21 (due to magnetic anisotropy of the carboxy group).Shielding of H-20 may be attributed to a combination of magnetic anisotropy of the carboxyl and the absence of the deshielding interactions present in the 23oxosapogenins between H-20 and the carbonylic oxygen.(Compare with the corresponding 23oxosapogenin, see Table 1).
The  Treatment of the 23-oxosapogenins with MCPBA afforded a mixture of the bisnorcholanic lactone and the cyclic carbonate after long reaction times (4 weeks for the 25R-ketosapogenins 2a and 2c and 9 weeks for 25S-ketosapogenin 2b), (see Scheme 2).The carbonates may be unequivocally recognized in the 1 H NMR spectra by the new downfield signal corresponding to H-20 (now carbynolic) which shows correlation with the deshielded H-21 protons, (Figure 4).Additionally, downfield shift of H-18 and upfield shift of H-16 characterize these structures.(Compare with bisnorcholanic lactones, see Table 1    The occurrence of the observed products may be explained by a reaction path in which, after addition of the peroxyacid to C-23, migration of C-22 leads to the ortoester II that may rearrange to the lactones 3a-c or be attacked by another molecule of peroxyacid at C-22 to produce III which on migration of C-20 leads to the cyclic carbonates 4a-c, (see Scheme 4).The 20S configuration of the carbonates 4a-c stems from the known fact that in BV reactions, migration occurs with retention of the configuration of the migrating group.Marker and Shabica 13 found that treatment of 23-oxosarsasapogenin acetate with von Baeyer persulfuric reagent (potassium persulfate, potassium sulfate and concentrated sulfuric acid) in acetic acid for 16 days led to a mixture of a bisnorcholanic lactone and a pregnan-3,16,20 triol, (Scheme 5).The proposed mechanism postulates that in such acidic conditions, the spiroketal side chain is opened to the α-diketone IV which produces BV reactions to a mixture of the anhydride V and the ketoester VI.Saponification followed by acidification leads to a mixture of the pregnan-3,16,20 triol and the bisnorcholanic lactone.
A similar mechanism may be postulated for the studied BV reaction with MCPBA, the occurrence of observed carbonates can be explained only if an additional BV step from VI to VII is accepted.In our case, the following facts allow us to discard Marker´s open chain mechanism: • The strong acidic media derived from the von Baeyer reagent used by Marker may produce both, the opening of the spiroketal side chain and acid catalysis resulting in shorter reaction times.By contrast, the reaction conditions used in our case (MCPBA in CH 2 Cl 2 ), are not acid enough to produce opening of the spiroketal side chain or acid catalysis.• If the observed BV reaction with MCPBA acid follows the open chain mechanism, steric hindrance to the attack of the peroxyacid would be minimized and no differences between the re-activity of 25R and 25S compounds should be encountered.The significant difference between the reaction times on the 25R and 25S series, accounts for a closed chain mechanism in which the axial position of the 27-methyl group hinders the nucleophilic attack of the peroxyacid to the carbonylic C-23.

Theoretical calculations 14
Semiempirical PM3 calculations of the fragments corresponding to 23-oxosapogenins of both 25R and 25S series indicate that neither charges nor LUMO energies can explain the different reactivity observed for the two C-25 epimers.Both reactivity indexes are almost the same for both epimers, therefore the cause of these differences should be other than electronic effects (see Table 3).Figure 6 shows the PM3 optimized geometries of the fragments corresponding to the side chains of both 25R-and 25S-23-oxosapogenins.These optimized geometries are consistent with the previously described NMR data for steroid 25R-and 25S-sapogenins and the coupling pattern of both axial and equatorial H-26 with H-25. 11rom these results it can be assumed that in the 25S epimer, the β-side of ring F is sterically hindered by two methyl groups, as a consequence, axial approach to C=O is expected to be highly disfavored.In contrast, equatorial approach to C=O (from the α-side) seems to be less hindered in both 25R and 25S series as well as axial approach to C=O in the 25R epimer (see Figure 6).PM3 calculations of the barriers of all possible reaction paths derived from the approach of the peroxyacid to C=O from the axial or equatorial directions in both 25R and 25S series, were performed.The results of the reaction barriers calculations as well as the molecular graphics of transition states are shown in Figure 7.The 25R epimer appears to be slightly more reactive to equatorial attack but also axial attack is possible.By contrast, according to the results, the axial approach to the 25S epimer is undoubtedly disfavoured because of the steric hindrance previously mentioned, leaving only one possibility, the equatorial attack, for the production of the Criegee´s intermediate.In addition the barrier for equatorial attack in the 25R is 0.6 kcal/mole lower than this of 25S indicating its higher reactivity.These results plenty justify the experimental evidence that the reaction time in the 25S epimer is twice longer than this of the 25R epimer.

Equatorial approach 25R 25S
Barrier The regioselective production of bisnorcholanic lactones 3a-c, may be explained in terms of the preference for antiperiplanar migration (App-migration) in the migration step. 15This assumption arises from the primary stereoelectronic effect in which orientation of the peroxide bond (O1-O2) should be App to that of the migrating group, in such a way that the σ bonding orbital of the migrating group and the σ * antibonding orbital of the peroxide bond overlap. 16adkiewicz-Poutsma and coworkers have shown that App-migrations are strongly favored with barriers that can be 58.0 kcal/mol lower in energy than the gauche migration barriers.Additionally they stated that no transition state could be located for gauche migration. 17n this context, the observed regioselectivity can be a consequence of the conformational preference of the Criegee´s peroxyester intermediate.PM3 calculations showed that those conformers of the Criegee´s intermediate with App arrangement of C22-C23-O1-O2 favorable to the migration of C-22, and hence to produce lactones 3a-c, are preferred over those with App arrangement of C24-C23-O1-O2 favorable to migration of C-24 which would produce the regioisomeric 23-oxo-24-oxa moiety, see Scheme 6 and Figure 8.

Conclusions
The Baeyer-Villiger reactions of (25R)-and (25S)-23-oxosapogenins produce the cleavage of the spiroketal side chain resulting in a mixture of a bisnorcholanic lactone and a cyclic carbonate of 16,20 diols.This reaction opens a synthetic alternative to bisnorcholanic lactones from the readily available steroid sapogenins.Synthetic applications of these reactions and studies on the effects of acid catalysis are on development.
The regioselective production of the bisnorcholanic lactones 3a-c may be explained in terms of the preference of those conformers which are favorable to migration of C-22.The reaction follows a course in which the steric factors play a determining role.PM3 calculated reaction barriers accurately reproduce the experimentally observed facts, and therefore its results are confident enough to explain the behavior of this system.
13  C NMR spectra of bisnorcholanic lactones 3a-c are characterized by the absence of the signals corresponding to C-23 to C-27, the downfield shifts of C-22 (now carboxy), C-21 and C-20 and the upfield shift of C-18 and C-17.(Compare with the corresponding 23-oxosapogenin, see Table2).