|Home | About | Journals | Submit | Contact Us | Français|
A two-pot tactic is presented to reach the oxidized 2,2-dimethyl-substituted pentalene and indene ring systems.
The illudin family of natural products is a large group of sesquiterpenes of which many members display antibacterial and antitumor properties. It appears to us that the constituents in the family may be biosynthetically derived by postmetabolic selective oxidations of allylic and primary functionality in illudin D (1) followed by the necessary eliminations of hydroxy residues and/or epimerization of the α-methyl ketone functionality (Figure 1). However, de novo conventional diastereoselective synthetic strategies applicable to the oxidized gem-dimethyl residue found in compounds 3–9 are not easily imagined. The structure that is most often identified with the family is the diene illudin M (2) and all but one,1c of the past syntheses and strategies were aimed at this compound or its derivatives where the gem-dimethyl functionality remains unoxidized.1
Motivated by some recent reports of unusual divergent antibiotic activity against methicillin-resistant Staphylococcus aureus (MRSA),2 we became interested in pursuing a new strategy for construction of the oxidized illudins 4–9, along with illudinic acid (3). To this end, we chose to examine a slightly modified [3+2]-cycloaddition procedure developed by Nakamura that employed 6,6-dimethyl-1-methylene-4,8-dioxoaspiro[2.5]octane (10) as a masked and stabilized dipolarophile (Scheme 1).3 This species, which displays both dipolar and biradical characteristics, participates in both concerted and stepwise reactions.4 While the nature of this intermediate and its preferred manifold of reactivity are debatable,5 compound 10 offers quick access to complex adducts derived from an assortment of unsaturated esters and enones. Surprisingly, however, 1,1-dialkoxy-2-methylenecyclopropanes (DMCP) have rarely been utilized with unsaturated systems displaying α- and β-substitution, which could be imagined as useful in natural product synthesis, particularly for some of the oxidized members of the illudin family.
We therefore elected to examine the combination of DMCP 10 with a wide scope of five-, six-, and seven-membered cycloalkenones including enones displaying α-and β-substitution on the unsaturated portion so that we might better understand its predilections (Table 1). Our plan was to cause the elimination of the substituent and thereby afford the desired pentalene, indene or azulene system. If the reaction proved successful, we further conjectured that it could be applied with the appropriated α-or β-substituted cyclohexenone to reach some of the oxidized members of the illudins.
First, we examined the combination of DMCP 10 with both cyclohexenone (11a) and cyclopentenone (11f; Table 1, entries 1 and 8). The resulting adducts were quite sensitive to moisture. So, for the ensuing study we found it very convenient to carry out the reactions in a deuterated solvent within a sealed NMR tube so that we could compare the integration of 1H NMR signals of the product against a known standard for an estimation of the percent yield. Both of these combinations, which lead to adducts 12a and 12f, had been previously examined by Nakamura; the latter was even the focus of a procedure in Organic Synthesis.3 We were therefore gratified to find that our yields for adducts 12a and 12f were within 5% of those previously reported.3,5f
Next, we screened an assortment of α- and β-substituted cyclic enone 11j–q that we imagined might undergo subsequent elimination. However, no product was observed over a range of conditions; warming with the DMCP 10 with the respective cyclic enones 11j–q simply returned partially degraded starting materials.
On the other hand, the five- and six-membered cycloalkenones 11b–e and 11g–h underwent reaction with the DMCP 10 to afford the respective ketene acetal intermediates 12b–e and 12g–h, respectively. The yields of the α-substituted cycloalkenones ranged from 42–61%. The only β-substituted system to undergo a [3+2]-cycloaddition amongst those examined was the β-selenide 11b. However, the yield from its union with 10 was significantly lower than the successful examples of the corresponding α-substituted cyclohexenones. In addition to the ketene acetal 12, the exocyclic methylene adduct 13 was also observed. The relative ratio of this side product, however, was not affected by our choice of solvent used in the reaction. Nakamura had previously reported that very electron-deficient systems such as nitroalkenes often preferred the reaction manifold leading to product 13, and he attributed it to the intervention of a single-electron-transfer cycloaddition pathway.5d We also observed adducts corresponding to the known class of ortho-ester 14 present in appreciable quantities (5% yield) for nearly all of the cycloalkenone reactions (11c–d and 11g–h). The appreciable quantity of 14 across the series may reflect undesirable steric effects for cyclic enone displaying α-substituents. We also observed that cycloheptenone (11i) provided the ketene acetal 12i, for the first time, albeit in much lower yield than the corresponding unsubstituted five- or six-membered cyclic enones that afforded 12f and 12a respectively. However, our success for seven-membered enones was limited to this particular example, as neither the α-substituted cycloheptenone 11p nor 11q participated in the [3+2]-cycloaddition.
Next, we decided to tackle the problem of ketene acetal stability. Protonation of the similar 5,5- and 5,6-fused ketene acetals reportedly affords a 1:1 thermodynamic ratio of the corresponding diastereomeric esters. Thus, this issue needed to be addressed before prompting elimination of the α-substituent. Initially, the crude ketene acetal was subjected to a modified Simmons–Smith reaction.6 Analysis of the crude 1H NMR suggested a modest yield of the corresponding cyclopropane. However, before we could further optimize these conditions, we discovered a much simpler methylation method. Introduction of methyl iodide upon conclusion of the cycloaddition, as determined by TLC or 1H NMR, and resumption of heating (24 h) afforded the respective homologated products 15a–i, which could be isolated and further purified by chromatography (Table 2). We found that the ketene acetal undergoes kinetic methylation in good yield with 2:1 diastereoselectivity in all cases (30–67% yield from the respective 11). We speculate that the X (Cl, Br, SePh) and methyl (Me) residues are positioned on the convex, β-face of the fused ring system.
To complete the model for the corresponding oxidized 2,2-dimethyl-substituted pentalene and indene ring systems, we examined modes of elimination for the bromide and the phenyl selenide substituents (Scheme 2). Both base-induced α-elimination of the bromide, and oxidation-induced syn elimination of the α-phenyl selenide, afforded the enone functionality found in the pentalene 16 and the indene 17 (70–75% yield from diastereomeric mixture, 30–40% yield from the respective cyclic enone 11). Further reduction of the carbonyl functionality in 16 and 17 (in toluene, 6 equiv DIBAL-H in hexanes), followed by MnO2 oxidation of the crude allylic alcohol yielded 18 and 19 (83–85% yield) and completed the model.
In conclusion, it appears that the Nakamura procedure is not effective for the construction of α-substituted [5.3.0]-bicycles that might be elaborated into the corresponding azulenes. However, we have found a straightforward tactic for the construction of complex 2,2-dimethyl-substituted pentalene and indene ring systems.7 The process involves the sequential [3+2]-cycloaddition reaction of DMCP 10 with an α-bromide or α-phenyl selenide in a five- or six-membered cycloalkenone, followed by methylation and cleavage of the intermediate ketene acetal with methyl iodide, and elimination of the appropriated α-substituent. The sequence can be carried out without chromatography and results in 30–40% yield. Further bisreduction and monooxidation of the allylic alcohol provides access to oxidized 2,2-dimethyl-substituted pentalene and indene ring systems. We postulate that such a sequence may be well suited for construction of the 2,2-disubstituted hydrindane skeletons found in the illudins 3–9 from the chiral cyclohexadienone 20 (Scheme 3).
T. R. R. Pettus is deeply grateful that this work was supported with renewed funds from the National Institute of General Medical Sciences (64831-06).