|Home | About | Journals | Submit | Contact Us | Français|
An account of the total synthesis of the tetracyclic alkaloid (−)-acutumine is presented. A first-generation approach to the spirocyclic subunit was unsuccessful due to incorrect regioselectivity in a radical cyclization. However, this work spawned a second-generation strategy in which the spirocycle was fashioned via a radical–polar crossover reaction. This process merged an intramolecular radical conjugate addition with an enolate hydroxylation, and created two stereocenters with excellent diastereoselectivity. The reaction was promoted by irradiation with a sunlamp, and a ditin reagent was required for aryl radical formation. These facts suggest that the substrate may function as a sensitizer, thereby facilitating homolytic cleavage of the ditin reagent. The propellane motif of the target was then installed via annulation of a pyrrolidine ring onto the spirocycle. The sequence of reactions used included a phenolic oxidation, an asymmetric ketone allylation mediated by Nakamura’s chiral allylzinc reagent, an anionic oxy-Cope rearrangement, a one-pot ozonolysis–reductive amination, and a Lewis acid promoted cyclization of an amine onto an α,β-unsaturated dimethyl ketal. Further studies of the asymmetric ketone allylation demonstrated the ability of the Nakamura reagent to function well in a mismatched situation. A TiCl4-catalyzed regioselective methyl enol etherification of a 1,3-diketone completed the synthesis.
Acutumine (1, Figure 1) is a tetracyclic alkaloid that was originally isolated by Goto and Sudzuki in 1929 from Sinomenium acutum.1 Later, Tomita and co-workers obtained 1 from Menispermum dauricum and determined its structure via X-ray crystallography.2 Other members of the acutumine family include acutumidine2 (2), dechloroacutumine3 (3), and the epimeric alcohols dauricumine4 (4), dauricumidine4 (5), and dechlorodauricumine5 (6). Recently, a diethylamino congener of 4 known as hypserpanine (7) was isolated from Hypserpa nitida.6 Since the plants from which the acutumine alkaloids are obtained have been used in traditional Chinese medicine to treat fever and pain,6,7 it is not surprising that the pure natural products possess interesting bioactivity. For example, acutumine is endowed with both selective T-cell cytotoxicity7 and antiamnesic properties.8
The striking molecular architecture of 1 includes a propellane-type system9 and a spirocycle. The cyclopentane ring, which is common to both of the aforementioned features, possesses a neopentylic secondary chloride and three contiguous quaternary stereocenters including two all-carbon quaternary centers. Shortly after the structure of acutumine was established, Barton and co-workers postulated that it could be derived biosynthetically from a benzylisoquinoline alkaloid via an oxidative phenolic coupling followed by oxidation and rearrangement steps.10 Recently, synthetic investigations by Wipf and co-workers have suggested that modifications to the Barton proposal may be in order.11 Sugimoto and co-workers have determined that 1 is produced in nature from two units of tyrosine,12 and that dechlorodauricumine (6) is the biosynthetic precursor to 1–5.13
In spite of the fact that the structure of acutumine was published in 1967, it did not attract the attention of the synthetic community until recently. In 2005, our laboratory developed a route to the propellane core of acutumine in which the pyrrolidine ring was annulated onto an aromatic precursor.14 Shortly thereafter, Sorensen and Moreau disclosed a concise sequence featuring a β-elimination–Michael addition cascade for construction of the propellane subunit.15 Then, we fashioned the spirocycle of 1 via a radical–polar crossover reaction,16 and subsequently elaborated this advanced intermediate into the natural product.17 Herein, we provide a full account of the development and execution of our synthetic strategy, which culminated in the first total synthesis of acutumine.
Our initial retrosynthetic analysis of acutumine is depicted in Scheme 1. Manipulation of the oxygenated functional groups contained in the cyclopentenone ring of 1 leads to tetracyclic intermediate 8. The most challenging step in the conversion of 8 into 1 was anticipated to be the regioselective methyl enol etherification of a 1,3-diketone. Simplification of 8 according to our previously developed pyrrolidine annulation strategy14 reveals tricycle 9. Key reactions in this section of the synthesis would include a phenolic oxidation, an anionic oxy-Cope rearrangement, and an acid-promoted cyclization of an amine onto an α,β-unsaturated ketal. We hoped to fashion the spirocycle of 9 via 5-exo-trig radical cyclization of aryl iodide 10 with trapping of the cyclic radical intermediate by TEMPO. Cyclization substrate 10 can be dissected into two monocyclic coupling partners: Weinreb amide 11, and enantiopure vinyl iodide 12. Amide 11 could be obtained by homologation of a benzaldehyde derivative employed in our isohasubanan alkaloid synthesis.18 Vinyl iodide 12 is a single step removed from a known alcohol19 which has been constructed by a route featuring enzymatic hydrolysis of cis-3,5-diacetoxycyclopentene.20
Although some precedent existed for the radical cyclization with TEMPO trapping,21 we had three major concerns regarding this proposed transformation. First, we were unsure if the allylic chloride would survive the radical reaction conditions. However, we felt that installation of the chloride would be quite challenging after formation of the neighboring quaternary spirocyclic carbon, so we elected to introduce this substituent at an early stage of the synthesis. Second, we feared that the hindered nature of the trisubstituted alkene radical acceptor in 10 might force the cyclization to proceed via a 6-endo rather than the typically favored 5-exo pathway. Finally, although examination of molecular models suggested that TEMPO trapping of the radical intermediate on its less-hindered face would deliver the desired adduct 9, we were hesitant to predict the degree of stereoselectivity that could be achieved in this reaction.
Our attempt at constructing 1 according to the plan described above commenced with the preparation of two coupling partners, Weinreb amide 11 and vinyl iodide 12, as depicted in Scheme 2. We previously reported the synthesis of pentasubstituted benzaldehyde 13 in three steps from 4-benzyloxy-2,3-dimethoxybenzaldehyde,18 itself accessible in two steps from commercially available 2,3-dimethoxyphenol.22 Wittig homologation of 13 provided aldehyde 14, which underwent oxidation and amidation to afford 11. Then, silylation of enantiopure alcohol 15, which was obtained from a straightforward eight-step sequence beginning with cis-3,5-diacetoxycyclopentene,19,20 afforded bis-TBS-protected vinyl iodide 12.
Coupling of the vinyllithium reagent derived from 12 with Weinreb amide 11 proceeded in low yield (9%). Although not investigated in detail, it is likely that competitive deiodination of 11 was a prominent side reaction. Formation of the corresponding vinylmagnesium reagent by treating 12 with PhMgBr, PhCH2MgBr, or allylMgBr was a very sluggish process. Fortunately, the protocol of Knochel and co-workers, which utilizes the complex i-PrMgCl·LiCl in conjunction with 15-crown-5,23 generated the desired Grignard reagent from 12 in a reasonable timeframe (ca. 1 h). Addition of this species to 11 provided enone 16 in moderate yield (Scheme 3). Then, a diastereoselective reduction of 16 was accomplished by enlisting the Corey–Bakshi–Shibata (CBS) catalyst.24 The configuration of the newly formed stereocenter of alcohol 17 was assigned by Mosher ester analysis.25 The minor diastereomer was partially separable from 17, and it could be completely removed after the following step. Next, SN2 chlorination of the allylic alcohol was conducted under Corey’s conditions,26 affording chloride 10 in 43% yield. Presumably, elimination of the chloride to give the corresponding dienyl benzene species is at least partially responsible for the low yield of this transformation. Although its isolation was not pursued, the elimination product was detected by mass spectrometry. We elected not to optimize the chlorination, as sufficient quantities of 10 were obtained to permit examination of the radical cyclization. When exposed to Et3B, air, and Bu3SnH27 at a low temperature, 10 was transformed into a single tricyclic product. Unfortunately, spectroscopic studies identified the adduct as compound 18, product of a 6-endo radical cyclization instead of the desired 5-exo process. Apparently, steric hindrance associated with the 5-exo pathway enabled the typically less favored 6-endo cyclization to proceed. Since the undesired regioisomer was formed in the cyclization, the feasibility of trapping with TEMPO was not investigated. Nevertheless, we were heartened by the fact that the allylic chloride emerged unscathed from the radical cyclization. This observation gave us confidence regarding the viability of a radical-based approach to the acutumine spirocycle, as long as the required 5-exo pathway could be enforced by appropriate changes to the substrate.
We reasoned that use of an α,β-unsaturated ketone instead of a simple alkene as radical acceptor would cause the electron-rich aryl radical to cyclize onto the electron-deficient β-carbon of the enone. This radical conjugate addition28 could allow us to surmount the steric obstacles associated with the desired 5-exo cyclization. A modified retrosynthesis of 1 based on this concept is portrayed in Scheme 4. We did not envision our new strategy requiring changes to the final stages of the synthesis; accordingly, tetracycle 8 was still projected as a key precursor to 1. However, 8 would now be derived from spirocyclic α-hydroxy ketone 19. Removal of the hydroxyl moiety from 19 and disconnection of the aryl–alkyl C–C bond from the spirocyclic carbon reveals enone 20, substrate for the radical cyclization. Instead of relying on TEMPO trapping to install the hydroxyl group, we planned to take advantage of the α-keto radical intermediate that would form upon 5-exo-trig cyclization of 20. These species are known to react with reagents such as Et3B,29 Et2Zn,30 and Et3Al31 to generate enolates, which can then participate as nucleophiles in polar reactions. The three-step sequence of a radical process, conversion of a radical intermediate to a polar intermediate, and a polar process has been termed a radical–polar crossover reaction by Murphy.32 Thus, the transformation of 20 into 19 would constitute a radical–polar crossover reaction which merges an intramolecular radical conjugate addition with an enolate hydroxylation.33 A report by Kunz and Rück provided encouraging precedent, but successful reactions in their study were limited to intermolecular conjugate additions of methyl radicals generated by homolysis of Me2AlCl to α,β-unsaturated N-acyl oxazolidinones. Moreover, the diastereoselectivity of the reaction was low.34 Clearly, a new protocol would be required to enable the stereoselective formation of 19 from 20.
In the proposed cyclization–hydroxylation of 20, two new stereocenters would be formed. We hypothesized that the bulky silyloxy group would direct the aryl radical to the opposite face of the enone, thereby setting the spirocyclic carbon in the proper configuration. Then, examination of molecular models suggested that the aryl group would shield the re face of the putative enolate intermediate, whereas the neighboring chlorine atom would protrude away from the si face (Figure 2). The pseudoequatorial orientation of the silyl ether would minimize its impact on the steric environment of the enolate. As a result, the hydroxyl group would likely be delivered to the si face. Consequently, it appeared that, out of four possibilities, the desired isomer of 19 would predominate in the planned radical–polar crossover reaction of 20.
We envisioned preparing substrate 20 via the same Weinreb amide–Grignard reagent coupling employed in the previous route (Scheme 3). In fact, the identical Weinreb amide (11) would be used in this case. However, construction of the enone moiety required a vinyl iodide in which the two hydroxyl groups were differentiated. Therefore, vinyl iodide 21 would be employed in place of 12.
The synthesis of 21, which largely parallels the known route19,20 used to prepare 12, is shown in Scheme 5. Protection of enantiopure alcohol 22 (available in three steps from cis-3,5-diacetoxycyclopentene)20 as a TES ether afforded 23, which was subjected to pivaloate cleavage with DIBAL-H, providing alcohol 24. Oxidation followed by iodination19 delivered α-iodo enone 26. Then, Luche reduction35 and masking of the resulting alcohol as a TBS ether gave 21, in which the two hydroxyl groups are differentiated.
Coupling of the Grignard reagent derived from 21 with Weinreb amide 11provided enone 28 (Scheme 6). As before, Knochel’s procedure23 was essential to obtaining reproducible yields. The CBS reduction of 28 proceeded under carefully optimized conditions (28/CBS cat./BH3·THF = 1.0:0.2:1.2, −10 °C, ≤4 h) to give allylic alcohol 29 in 89% yield and 9:1 dr. The configuration of the hydroxyl-bearing carbon was determined by Mosher ester analysis, and was consistent with the CBS reduction of 16 (Scheme 3).25 The next transformation, SN2 chlorination of 29, afforded low and variable yields when the Corey protocol26 was employed, presumably due to elimination of HCl from the product 30. Fortunately, consistent results with minimal elimination could be achieved by enlisting MsCl and Et3N.36 Then, the TES group of 30 was selectively removed with HF·pyridine, and oxidation of the resulting alcohol 31 afforded enone 20.
Later, we attempted to streamline the route to 20 by examining the asymmetric Nozaki–Hiyama–Kishi coupling37 of vinyl iodide 21 with aldehyde 14 (Scheme 7). Allylic alcohol 29 could be obtained directly from this reaction. Unfortunately, the yield and dr remained low despite attempts at optimization. Since this more direct pathway was less efficient, we continued to use the route depicted in Scheme 6 to construct enone 20.
At this point, we commenced our investigation of the radical–polar crossover reaction, which is summarized in Table 1. Since tin radicals are commonly employed to generate aryl radicals,27 we included hexabutylditin, which functions as a source of tin radicals under nonreducing conditions, in the reaction mixture. We found that the desired cascade process occurred upon photolysis with a simple sunlamp at 0 °C.38 Efforts to discover the optimal promoter for the radical–polar crossover step revealed that Et3Al31 was more effective than Et3B29 or Et2Zn30 (cf. entry 6 vs entries 1 and 3 or entry 7 vs entries 2 and 4). As for the hydroxylating agent, 3-phenyl-2-(phenylsulfonyl)oxaziridine39 emerged as superior to O2,40 DMDO,41 t-BuOOH,42 or (Me3SiO)2.43 After varying other parameters, we established the optimal conditions listed in entry 8. The α-hydroxy ketone 19 was obtained in 62% yield, along with iodide 32 (7%) and reduced compound 33 (3%). Thus, the yields of the cyclization and hydroxylation steps were 72% and 86%, respectively. Importantly, no diastereomers of 19, 32, or 33 were detected in the reaction mixture, and consistent results were obtained when the reaction was conducted on a preparative scale (59% yield of 19 on >100 mg scale). Ketone 33 is likely produced by reduction of the α-keto radical intermediate or protonation of the enolate intermediate. However, the origin of iodide 32 is less certain. It is possible that I• or I2, both of which would be present if the aryl iodide is cleaved directly via photolysis (vide infra), could react with the α-keto radical intermediate to create this byproduct. Alternatively, the α-keto radical or the enolate may react with either Bu3SnI (presumably generated by iodine atom abstraction from 20 by Bu3Sn•) or an electrophilic species formed upon in situ oxidation of Bu3SnI44 to form 32.
The configurations of the two stereocenters formed in the radical–polar crossover reaction were assigned with the aid of NOE experiments conducted on a PMB ether derivative of 19. The diagnostic correlations are illustrated in Figure 3. These assignments were ultimately confirmed by the conversion of 19 into (−)-1 and are consistent with the reaction pathway proposed above (see Figure 2).
In order to improve the overall efficiency of the process, we explored the conversion of iodide 32 into α-hydroxy ketone 19. In principle, this transformation could be accomplished via a radical–polar crossover reaction consisting of α-keto radical formation, enolate generation, and hydroxylation. In practice, the use of Et2Zn/O2 in conjunction with the oxaziridine provided 19 in 62% yield (Scheme 8). This raised the overall yield of 19 from 20 to 66%. Interestingly, the use of Et3Al in place of Et2Zn afforded lower yields (40%).
We also examined the possibility of using SmI2/HMPA45 instead of hexabutylditin to generate the aryl radical in the radical–polar crossover reaction. α-Hydroxy ketone 19 was produced in low yields (ca. 15%) from these reactions, along with 32, 33, and several other uncharacterized byproducts. This negative outcome is likely a result of the number of functional groups (aryl iodide, enone, allylic chloride) in 20 which are capable of reacting with SmI2.
As we pondered the role of the ditin reagent in the radical–polar crossover reaction, we were intrigued by the fact that the process utilized photochemical initiation. Hexabutylditin does not absorb light (λmax = 236 nm); accordingly, photolytic reactions which employ this reagent typically require a sensitizer.46 However, the conversion of 20 into 19 proceeds in the absence of typical sensitizing agents. Accordingly, we considered two possible roles for the ditin species. First, the enone moiety of 20 could act as a sensitizer and mediate the homolytic cleavage of hexabutylditin into two tributyltin radicals. Then, abstraction of the aryl iodide by Bu3Sn• would initiate the radical–polar crossover reaction. Alternatively, the aryl iodide might be cleaved directly by photolysis. In this scenario, the ditin reagent would function as an iodine trap, thereby preventing iodine radicals from reacting with subsequent intermediates.47 If hexabutylditin is involved in trapping iodine but not in aryl radical formation, then omitting it from the reaction mixture should result in atom transfer cyclization47 due to the presence of untrapped I•. In fact, the formation of α-iodo ketone 32 as a byproduct lends some credence to the idea that the aryl radical is formed by direct photolysis of 20. However, reactions run in the absence of hexabutylditin did not proceed, and starting material was recovered. Although further study is required to more fully elucidate the mechanism of the radical–polar crossover reaction, this observation indicates that the ditin reagent is instrumental in generating the aryl radical intermediate from 20 and is not merely functioning as an iodine trap. Thus, the enone moiety of 20 may be mediating homolytic cleavage of hexabutylditin by functioning as a sensitizer, as posited above. Another possibility is that coordination of the iodine atom in 20 to the ditin reagent weakens the C–I and/or Sn–Sn bonds, thereby facilitating the photolysis.48
Before we could fashion the propellane system of acutumine, some routine functional group manipulations of spirocycle 19 were necessary. Although the carbonyl carbon of this intermediate is maintained at the same oxidation state in 1, the incompatibility of a ketone at this position with an upcoming allylation reaction required us to perform a reduction–protection sequence. Accordingly, treatment of 19 with L-Selectride provided diol 34 in 9:1 dr (Scheme 9). The cis relative stereochemistry of the diol moiety was assigned based on the coupling constant (6.6 Hz) measured for the two α-hydroxy hydrogen atoms.49 This configuration is consistent with attack of the bulky reducing agent on the less hindered si face of the ketone, or the face opposite the neighboring OTBS, hydroxyl, and chloro substituents. A stereoselective reduction of 19 was not essential, since the alcohol would be oxidized in the final stages of the synthesis. However, the isolation and characterization of intermediates was simplified by working with diastereomerically pure compounds, so the L-Selectride protocol was beneficial. Then, the more accessible secondary alcohol of 34 was selectively silylated. Subsequent cleavage of the benzyl ether of 35 via hydrogenolysis afforded phenol 36.
Using a sequence of reactions devised with a model compound,14 phenol 36 was transformed into amine 42 as outlined in Scheme 10. Phenolic oxidation of 36by PhI(OAc)2 in MeOH delivered masked o-benzoquinone 37. Masked o-benzoquinones have great utility in organic synthesis due to the density of functionality which is contained in a single six-membered ring.50 We planned on utilizing each of the functional groups (enone, dimethyl ketal, methyl enol ether) created in the phenolic oxidation to complete the synthesis of 1. Thus, this reaction was one of the cornerstones of our strategy. Then, the neopentylic alcohol of 37 was benzylated in preparation for the crucial asymmetric ketone allylation. Examination of molecular models of 38 revealed that the re face of the ketone was slightly less congested than the si face. However, the difference in steric hindrance between the two diastereotopic faces appeared to be minor. Consequently, we postulated that substrate-directed stereocontrol would be minimal and that catalyst or reagent control would be required to accomplish the allylation. During the course of our synthesis of isohasubanan alkaloids, we discovered that Nakamura’s chiral allylzinc reagent (S,S)-3951 was uniquely effective in enantioselective allylations of achiral ketones related to 38.18a We were pleased to discover that allylation of 38 with (S,S)-39 proceeded in good yield (79%) and dr (93:7) to afford homoallylic alcohol 40. Although an excess of 39 (1.6 equiv) was required to ensure a reasonable reaction rate with the bulky ketone, the pure bisoxazoline ligand could be re-isolated from the reaction mixture.52 The configuration of the newly formed stereocenter in alcohol 40 was tentatively assigned by considering the six-membered cyclic transition state for the asymmetric ketone allylation proposed by Nakamura and co-workers51 (Figure 4). The dimethyl ketal of 38 likely occupies an equatorial position on the ring, which causes the sp2 alkene carbon to reside in an axial position. The ability of the phenyl groups on the bisoxazoline ligand to avoid steric interactions with the substrate in the orientation shown leads to excellent facial selectivity. The stereochemical assignment based on this reasoning was eventually confirmed by conversion of 40 into (−)-1.
To ascertain the degree of substrate-directed stereocontrol and determine the prospects for reagent control with 39 in a mismatched case, we performed two additional allylations of 38. These experiments are described in Table 2 along with the allylation discussed above. The use of allylmagnesium bromide in place of (S,S)-39 afforded a 70:30 mixture of 40 and its diastereomer 40’ (entry 2), thereby confirming our hypothesis regarding moderate levels of stereocontrol by the substrate. Interestingly, allylation with (R,R)-39, the enantiomeric Nakamura reagent, delivered 40’ as the major product in good yield and dr (entry 3). Thus, (R,R)-39 is able to overcome the mismatched stereochemistry of 38, albeit with slighty reduced yield and dr. The performance of the Nakamura reagent in this work and in our isohasubanan alkaloid synthesis18a demonstrates that its scope is not limited to the alkynyl ketones for which it was designed.51 Studies to determine the types of ketones amenable to asymmetric allylation by 39 are in progress and will be the subject of a future report.
Exposure of 40 to KOt-Bu and 18-crown-6 triggered an anionic oxy-Cope rearrangement that produced ketone 41 in excellent yield (92%, Scheme 10). Notably, this reaction was capable of forming an extremely congested C–C bond at 0 °C. The facile nature of this process is likely a consequence of conjugation of the trisubstituted double bond with the methyl enol ether. Such conjugation has previously been noted to accelerate anionic oxy-Cope rearrangements.53 The allylation–anionic oxy-Cope sequence accomplishes a formal conjugate addition of an allyl group to enone 38. The direct 1,4-addition was not attempted due to discouraging results with model compounds.54
We approached the next step, oxidative cleavage of 41 followed by reductive amination of the crude aldehyde, with some trepidation due to problems encountered in the model studies. Specifically, ozonolysis of a compound analogous to 41was plagued by competitive oxidation of the methyl enol ether. Dihydroxylation was possible, but oxidative cleavage of the resulting diol was unsuccessful.14 These challenges persisted, as treatment of 41 with OsO4/NaIO4 returned negligible quantities (ca. 10%) of the aldehyde. Similarly poor results were obtained from ozonolysis reactions conducted by bubbling O3 through the reaction mixture. Numerous byproducts were generated, and it became apparent that a method of controlling the stoichiometry of O3 was required. We were intrigued by a report from Wender and co-workers describing the use of standard solutions of O3.55 In the model system, we found EtOAc to be the most effective solvent for the ozonolysis.14 Accordingly, we determined the concentration of a saturated solution of O3 in EtOAc to be 0.007 M by titration with styrene.55 Addition of 1.5 equiv of O3 from this solution to a solution of 41 in EtOAc provided ca. 30% of the aldehyde, along with 30% of recovered 41. Greater quantities of O3 gave lower yields due to increased byproduct formation. Fortunately, a significant improvement was achieved by including pyridine in the reaction mixture. This additive has been shown to modulate the reactivity of O3.56 The best results (54% yield of amine 42, 27% recovery of 41) were obtained by adding 1.5 equiv of O3 via standard solution to a solution of 41, pyridine, and Et3N in EtOAc, then conducting the reductive amination in the same pot. Amine 42 and recovered alkene 41 were readily separable on SiO2.
The cyclization of amine 42 to give the tetracyclic framework of acutumine was projected to occur via acid-promoted ionization of the dimethyl ketal and subsequent attack of the amine at the β-carbon of the resulting α,β-unsaturated oxonium ion intermediate.57 In our synthesis of a model system representing the propellane core of 1, we discovered that TMSOTf could mediate this reaction, affording moderate yields (50%) of the desired product.14 Unfortunately, application of these conditions to the highly functionalized substrate 42 induced decomposition, and the desired tetracycle was only produced in trace amounts (<10%). Since we possessed limited quantities of 42, we returned to the more readily available model compound 43 in order to develop alternative conditions for the cyclization (Table 3). In our prior investigation, we found that enol 44, not the corresponding ketone, was produced by cyclization of 43. We believe that the combination of a stabilizing intramolecular hydrogen bond in the enol and destabilizing steric interactions in the ketone are responsible for this phenomenon.14
Our study of the cyclization of 43 is summarized in Table 3. A brief survey of Brønsted acids (entries 1–4) showed that reasonable yields of 44 could be obtained with TFA at 0 °C (entry 2). Similar conditions led to an undesired rearrangement in our attempted synthesis of hasubanan alkaloids, producing unnatural compounds which we refer to as isohasubanan alkaloids.18 These contrasting results reflect subtle differences in the structures of the acutumine and hasubanan alkaloids. Surpisingly, the use of HCl afforded low yields of hemiaminal 45 (entry 3). It is unclear why 45 was only detected under these conditions and not in any other reactions. Switching to the Lewis acid BCl3 yielded promising results, and detailed optimization established the superiority of the conditions listed in entry 11. Interestingly, use of the solvent hexafluoroisopropanol (HFIP) without any added Lewis or Brønsted acid58 resulted in modest yields of 44 (entries 13 and 14). However, substitution of HFIP for CH2Cl2 as solvent in the optimized BCl3 conditions did not improve the yield further (cf. entry 11 and entry 15). In all cases, 4 Å MS were employed to suppress cleavage of the acid-sensitive methyl enol ether of 43.
TFA and BCl3 emerged from this study as the reagents of choice for the key cyclization reaction. Accordingly, both were evaluated in the cyclization of amine 42. We recognized that 42 was likely to be more acid-sensitive than the simpler compound 43; consequently, we employed slightly milder conditions than were optimal in the model study. When 42 was exposed to TFA at −10 °C, the desired tetracycle 46 was detected by 1H NMR and MS. Unfortunately, it was produced in low yield (<40%) and alongside an inseparable byproduct which was not identified. Fortunately, BCl3 (1.5 equiv) at −40 °C afforded better results, as pure 46 could be isolated in 45% yield (Scheme 11). The modest outcome of this cyclization is likely a consequence of the congested nature of the C–N bond which is formed combined with the acid sensitivity of substrate 42.
The elaboration of 46 into 1 commenced with TBAF-mediated cleavage of the two TBS ethers. The resulting diol was surprisingly unstable, so it was immediately converted into the isolable 1,3-diketone 47 by treatment with TPAP and NMO. Then, hydrogenolysis of the benzyl ether proceeded in virtually quantitative yield without competitive reduction of the tetrasubstituted alkene or the alkyl chloride. This facile reaction delivered alcohol 48, the immediate precursor to acutumine.
A multitude of protocols exist for the conversion of 1,3-diketones into β-keto enol ethers, encompassing basic,59 Lewis acidic,60 and neutral (CH2N2)61 conditions. Nonetheless, we were unsure of the prospects for a regioselective transformation of 48 into 1. The outcomes of O-methylations of asymmetric 1,3-diketones are difficult to predict, and equimolar mixtures of the two regioisomeric products are often produced.61a After considering the various procedures available, we were attracted to the method of Porta and co-workers, which employs catalytic quantities of TiCl4 in MeOH. This protocol was modestly selective (ca. 3:1 ratio of products) for generation of the less hindered enol ether from an asymmetric cyclic 1,3-diketone.60a We were also mindful of the report by Danishefsky and co-workers of a regioselective methyl enol etherification with CH2N2,61b although the complete lack of selectivity observed by Pettus and Wang with this reagent61a provides a sobering contrast. Unfortunately, our experience resembled that of Pettus and Wang, as exposure of 48 to CH2N2 provided 40% of 1 along with 35% of its enol ether regioisomer 49. However, switching to TiCl4/MeOH yielded a more favorable 3.7:1 ratio in favor of 1. The overall yield of this reaction (66%) was lower, but the isolated yield of 1 (52%) was greater. Accordingly, we employed the Porta methodology in our efforts. The synthetic sample of 1 produced by this route was identical to an authentic natural sample of acutumine as evidenced by spectroscopic and chromatographic techniques.
We have achieved the enantioselective total synthesis of (−)-acutumine (1). A first-generation approach to the spirocyclic subunit based on a 5-exo aryl radical cyclization with TEMPO trapping was unsuccessful, as the 6-endo pathway prevailed. However, this result paved the way for a successful second-generation strategy featuring a novel radical–polar crossover reaction for construction of the spirocycle. This cascade process was comprised of a 5-exo intramolecular conjugate addition of an aryl radical onto an enone acceptor, conversion of the resulting α-keto radical into an enolate, and hydroxylation of the enolate. Both the radical cyclization and enolate hydroxylation steps were highly stereoselective, and a single product was generated which possessed the requisite stereochemistry for conversion into 1. The reaction was conducted under irradiation by a sunlamp, yet the ditin reagent which was required for aryl radical formation does not possess a chromophore. Thus, it is possible that the enone moiety of the substrate may function as a sensitizing agent, or complexation of the aryl iodide with the ditin reagent may facilitate photolysis.
The tetracyclic framework of the natural product was then fashioned by annulation of a pyrrolidine ring onto the spirocycle. The annulation sequence consisted of a phenolic oxidation, a diastereoselective ketone allylation utilizing Nakamura’s chiral allylzinc reagent, an anionic oxy-Cope rearrangement, a one-pot ozonolysis–reductive amination, and a Lewis acid-promoted cyclization of an amine onto an α,β-unsaturated dimethyl ketal. Further studies of the asymmetric ketone allylation revealed the ability of the Nakamura reagent to control the stereochemistry of the reaction in a mismatched case. The successful cyclization conditions (BCl3, CH2Cl2, −40 °C) were discovered by extensive optimization of a model reaction. Finally, elaboration of the tetracycle into 1 required four steps, including the TiCl4-mediated regioselective methyl enol etherification of a 1,3-diketone. We are hopeful that the methods which were developed and refined in the course of this venture will find application in the synthesis of other complex molecules.
A solution of 20 (107 mg, 0.166 mmol) in anhydrous THF (1.5 mL) at 0 °C was treated with hexabutylditin (89 µL, 102 mg, 0.168 mmol) and triethylaluminum (1.0 M solution in THF, 490 µL, 0.49 mmol). The resultant mixture was irradiated at 0 °C with a 660 W sunlamp for 6 h (frequent addition of ice to the cooling bath was necessary to maintain this temperature). Then, 3-phenyl-2-(phenylsulfonyl)-oxaziridine39 (ca. 0.5 M solution in THF, 1.53 mL, 0.765 mmol) was added to the mixture, and it was stirred at 0 °C (without irradiation) for 5 h, then at rt for 2 h. The resultant mixture was extracted with EtOAc (3 × 5 mL), dried (MgSO4), and concentrated in vacuo. Flash chromatography (SiO2, 1.5 × 14 cm, 10–15% EtOAc in hexanes gradient elution) afforded 19 (52.7 mg, 0.0989 mmol, 59%) as a colorless oil: [α]25D +26 (c 0.23, CH2Cl2); 1H NMR (CDCl3, 300 MHz) δ 7.53-7.32 (m, 6H), 5.06 (s, 2H), 4.82 (t, J = 7.2 Hz, 1H), 4.45 (t, J = 6.6 Hz, 1H), 4.25 (s, 1H), 3.90 (s, 3H), 3.84 (s, 3H), 2.99 (d, J = 6.9 Hz, 1H), 2.91 (d, J = 6.6 Hz, 1H), 2.62 (d, J = 6.9 Hz, 1H), 2.02 (d, J = 7.2 Hz, 1H), 1.51 (br s, 1H), 0.93 (s, 9H), 0.13 (s, 3H), 0.11 (s, 3H); 13C NMR (CDCl3, 75 MHz) δ 207.5, 149.7, 149.6, 146.4, 142.5, 134.8, 127.8 (2C), 127.3, 126.4 (2C), 121.8, 103.4, 92.2, 72.1, 66.2, 64.2, 63.8, 63.1, 49.7, 42.7, 38.8, 22.2 (3C), 16.1, −4.9, −5.0; DEPT NMR (CDCl3, 75 MHz) C: 207.5, 149.7, 149.6, 146.4, 142.5, 134.8, 121.8, 64.2, 16.1 CH: 127.8, 127.3, 126.4, 103.4, 92.2, 66.2, 49.7 CH2: 72.1, 42.7, 38.8 CH3: 63.8, 63.1, 22.2, −4.9, −5.0; 2D 1H–1H COSY NMR (CDCl3, 500 MHz) 4.82/2.62 (s), 4.82/2.11 (s), 4.45/2.99 (s), 4.45/2.91 (s); 2D 1H–13C HMQC NMR (CDCl3, 500 MHz) 7.53-7.32/127.8, 7.53-7.32/127.3, 7.53-7.32/126.4, 7.53-7.32/103.4, 5.06/72.1, 4.82/66.2, 4.45/49.7, 4.25/92.2, 3.90 and 3.84/63.8 and 63.1, 2.99/38.8, 2.91/38.8, 2.62/42.7, 2.11/42.7, 0.88/22.2, 0.13 and 0.11/−4.9 and −5.0; IR (film) νmax 3012, 2955, 2878, 2857, 1728, 1471, 1356, 1251, 1134, 1087 cm−1; HRMS (ESI) m/z 533.21177 (MH+, C28H37O6ClSiH+ requires 533.21207).
The iodide 32 (6.9 mg, 0.011 mmol, 6.4%) and reduced compound 33 (3.0 mg, 0.0058 mmol, 3.5%) were also obtained. For 32: 1H NMR (CDCl3, 300 MHz) δ 7.49-7.34 (m, 6H), 5.13 (s, 2H), 4.98-4.92 (m, 1H), 4.72-4.63 (m, 1H), 4.59 (s, 1H), 3.91 (s, 3H), 3.87 (s, 3H), 3.00 (d, J = 6.6 Hz, 1H), 2.93 (d, J = 6.6 Hz, 1H), 2.65 (d, J = 6.6 Hz, 1H), 2.02 (d, J = 6.9 Hz, 1H), 0.89 (s, 9H), 0.14 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 202.1, 146.3, 146.2, 144.4, 139.1, 139.0, 120.8 (2C), 120.1 (2C), 119.5, 113.6, 91.4, 71.2, 67.1, 63.9, 63.5, 63.0, 53.2, 47.4, 41.1, 37.3, 22.0 (3C), 14.4, −4.4, −4.5; HRMS (ESI) m/z 643.11245 (MH+, C28H36O5ClISiH+ requires 643.11380). For 33: 1H NMR (CDCl3, 300 MHz) δ 7.46-7.31 (m, 6H), 5.10 (s, 2H), 4.93-4.88 (m, 1H), 4.65 (t, J = 6.3 Hz, 1H), 3.88 (s, 3H), 3.84 (s, 3H), 2.92 (d, J = 6.6 Hz, 1H), 2.85 (d, J = 6.6 Hz, 1H), 2.65 (d, J = 6.9 Hz, 1H), 2.49 (s, 1H), 2.05 (s, 1H), 2.00 (d, J = 6.9 Hz, 1H), 0.92 (s, 9H), 0.11 (s, 6H); 13C NMR (CDCl3, 125 MHz) δ 208.2, 145.4, 144.1, 143.8, 142.6, 139.8, 120.1 (2C), 118.8, 118.3 (2C), 114.4, 96.5, 73.0, 69.6, 65.5, 63.9, 62.7, 51.2, 40.6, 40.0, 39.2, 25.8 (3C), 17.6, −4.3, −4.4; HRMS (ESI) m/z 534.23974 (M(NH4)+, C28H37O5ClSi(NH4)+ requires 534.24370).
A solution of 19 (150 mg, 0.281 mmol) in anhydrous THF (2 mL) at 0 °C under Ar was treated with L-Selectride (1.0 M solution in THF, 280 µL, 0.28 mmol). The resultant mixture was stirred at 0 °C for 1.5 h, then treated with sat aq NH4Cl (1 mL) and warmed to rt. The mixture was extracted with EtOAc (3 × 3 mL), dried (MgSO4), and concentrated in vacuo. Flash chromatography (SiO2, 2.5 × 11 cm, 20% EtOAc–hexanes elution) afforded 34 (132 mg, 0.247 mmol, 88%) as a pale yellow solid in 9:1 dr. A diastereomerically pure sample could be obtained after further purification: [α]25D +22.7 (c 1.39, CH2Cl2); 1H NMR (CDCl3, 300 MHz) δ 7.42-7.12 (m, 5H), 6.75 (s, 1H), 5.07 (s, 2H), 4.87 (dd, J = 11.1, 5.7 Hz, 1H), 4.27 (d, J = 6.6 Hz, 1H), 4.08 (br s, 1H), 3.84 (s, 3H), 3.80 (s, 3H), 3.64 (br s, 1H), 3.56-3.38 (m, 2H), 3.06 (t, J =12.0 Hz, 1H), 2.89 (dd, J =12.6, 7.2 Hz, 1H), 2.03-1.98 (m, 1H), 1.70-1.61 (m, 1H), 0.88 (s, 9H), 0.21 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 151.6, 146.4, 146.3, 144.3, 139.1, 139.0, 120.8 (2C), 120.1, 119.5 (2C), 113.6, 73.5, 71.2, 67.2, 64.0, 63.5, 63.0, 53.2, 47.4, 41.1, 37.3, 23.9, 22.0 (3C), −4.4, −4.5; IR (film) νmax 3548, 2911, 1626, 1450, 1219, 1091, 933 cm−1; HRMS (ESI) m/z 557.20989 (MNa+, C28H39ClO6SiNa+ requires 557.20966).
The cis relative stereochemistry of 34 was assigned based on the 6.6 Hz coupling constant of the two α-hydroxy hydrogens. This value is similar to coupling constants reported by Hartung and Paquette49a for related cis compounds (4.2–5.8 Hz) and differs markedly from the value reported by Christol and Vanel49b for a related trans compound (10 Hz). Additionally, molecular models of 20 demonstrate that approach of the reducing agent to the top (re) face of the carbonyl, which would afford the trans isomer, is hindered by the neighboring chloride substituent.
A solution of 34 (140 mg, 0.262 mmol) in anhydrous CH2Cl2 (5.0 mL) under Ar was treated with Et3N (450 µL), then cooled to 0 °C. TBS-Cl (59 mg, 0.39 mmol, 1.5 equiv) was added portionwise to the mixture, and it was stirred at 0 °C for 2 h, then at rt for 1 h. The resultant mixture was diluted with EtOAc (5 mL), treated with sat aq NH4Cl (3 mL), and the layers were separated. The aqueous layer was extracted with EtOAc (3 × 5 mL), dried (MgSO4), and concentrated in vacuo. Flash chromatography (SiO2, 2.5 × 10 cm, 7.5% EtOAc–hexanes elution) afforded 35 (148 mg, 0.228 mmol, 87%) as a light yellow oil: [α]25D +17 (c 1.2, CH2Cl2); 1H NMR (CDCl3, 300 MHz) δ 7.49-7.27 (m, 5H), 6.54 (s, 1H), 5.07 (s, 2H), 4.67 (dd, J = 12.6, 7.2 Hz, 1H), 4.16 (d, J = 6.9 Hz, 1H), 3.86 (s, 3H), 3.80 (s, 3H), 3.71 (br s, 1H), 3.62-3.44 (m, 2H), 3.05 (t, J = 12.0 Hz, 1H), 2.70 (dd, J = 12.6, 7.5 Hz, 1H), 2.01-1.94 (m, 1H), 1.67-1.63 (m, 1H), 1.17 (s, 9H), 0.99 (s, 9H), 0.20 (s, 6H), 0.14 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 161.2, 160.4, 155.1, 151.8, 145.9, 145.6, 137.6 (2C), 136.9, 136.0 (2C), 133.7, 79.9, 79.5, 75.2, 72.6, 72.2, 70.3, 53.7, 51.8, 38.1, 34.1 (3C), 34.0 (3C), 33.2, 26.3, 17.1,−4.4 (2C), −4.5 (2C); IR (film) νmax 3577, 2897, 1610, 1442, 989 cm−1; HRMS (ESI) m/z 649.31418 (MH+, C34H53ClO6Si2H+ requires 649.31420).
A solution of 35 (148 mg, 0.228 mmol) in anhydrous MeOH (5.0 mL) was treated with 10% Pd/C (40 mg, 0.27 wt equiv). The resultant mixture was stirred at rt under H2 (1 atm) for 4 h, then filtered through a plug of Celite (washed with CH2Cl2), dried (MgSO4), and concentrated in vacuo. Flash chromatography (SiO2, 1.5 × 8 cm, 10% EtOAc–hexanes elution) afforded 36 (123 mg, 0.220 mmol, 96%) as a pale yellow oil: [α]25D +27 (c 1.7, CHCl3); 1H NMR (CDCl3, 300 MHz) δ 6.82 (s, 1H), 4.43 (dd, J = 12.6, 7.2 Hz, 1H), 4.21 (d, J = 6.9 Hz, 1H), 3.91 (s, 3H), 3.88-3.84 (br s, 1H), 3.84 (s, 3H), 3.63 (br s, 1H), 3.54-3.31 (m, 2H), 2.87 (t, J = 12.0 Hz, 1H), 2.66 (dd, J = 12.6, 7.2 Hz, 1H), 2.00-1.90 (m, 1H), 1.74-1.62 (m, 1H), 1.16 (s, 9H), 1.10 (s, 9H), 0.19 (s, 6H), 0.16 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 148.1, 142.8, 139.1, 133.2, 128.6, 128.1, 82.7, 75.8, 75.3, 69.5, 69.0, 66.5, 61.1, 40.8, 36.7, 25.8, 22.1 (6C), 13.8, −4.4 (2C), −4.5 (2C); IR (film) νmax 3212, 1258, 1122, 1077 cm−1; HRMS (ESI) m/z 559.26731 (MH+, C27H47ClO6Si2H+ requires 559.26725).
A solution of 36 (98.0 mg, 0.175 mmol) in anhydrous CH3OH (3.0 mL) was added to a mixture of KHCO3 (30 mg, 0.35 mmol, 2.0 equiv), PhI(OAc)2 (62 mg, 0.19 mmol, 1.1 equiv), and anhydrous CH3OH (3.0 mL) at −10 °C under Ar. The resulting yellow–orange mixture was stirred for 10 min, diluted with CH2Cl2 (5 mL), and washed with brine (10 mL). The layers were separated, and the organic layer was dried (MgSO4) and concentrated in vacuo. Flash chromatography (SiO2, 2.5 × 10 cm, 10% EtOAc–hexanes elution) afforded 37 (69.0 mg, 0.117 mmol, 67%) as a yellow oil: [α]25D −15 (c 1.2, CHCl3); 1H NMR (CDCl3, 300 MHz) δ 6.25 (s, 1H), 4.55 (dd, J = 12.6, 7.5 Hz, 1H), 4.13 (d, J = 6.9 Hz, 1H), 3.98 (s, 3H), 3.61 (br s, 1H), 3.49-3.24 (m, 2H), 3.37 (s, 3H), 3.32 (s, 3H), 2.79 (t, J = 11.8 Hz, 1H), 2.58 (dd, J = 12.6, 7.5 Hz, 1H), 1.93-1.81 (m, 1H), 1.58-1.49 (m, 1H), 0.91 (s, 9H), 0.86 (s, 9H), 0.10 (s, 6H) 0.09 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 191.1, 142.8, 139.1, 133.2, 121.4, 117.7, 82.8, 69.9, 69.5, 69.0, 66.5, 61.1, 56.5, 56.4, 40.9, 37.7, 23.8, 22.1 (6C), 13.8, −4.4 (2C), −4.6 (2C); IR (film) νmax 3337, 2450, 1755, 1233, 956 cm−1; HRMS (ESI) m/z 606.30440 (M(NH4)+, C28H49ClO7Si2(NH4)+ requires 606.30436).
A solution of 37 (110 mg, 0.187 mmol) in anhydrous DMF (1.5 mL) at rt under Ar was treated with NaH (60% dispersion in mineral oil, 7.6 mg, 4.6 mg NaH, 0.19 mmol), tetrabutylammonium iodide (70 mg, 0.19 mmol), and benzyl bromide (23 µL, 32.9 mg, 0.192 mmol). The resultant brown solution was stirred at 60 °C for 5 h, cooled to rt, diluted with CH2Cl2 (2 mL), and washed with brine (2 mL). The layers were separated, and the aqueous layer was extracted with CH2Cl2 (5 × 3 mL). The combined organic layers were dried (MgSO4) and concentrated in vacuo. Flash chromatography (SiO2, 1.5 × 12 cm, 1% Et3N in 5% EtOAc–hexanes elution) afforded 38 (111 mg, 0.163 mmol, 88%) as a brown oil: [α]25D −21 (c 1.1, CHCl3); 1H NMR (CDCl3, 300 MHz) δ 7.42-7.18 (m, 5H), 6.39 (s, 1H), 5.20 (s, 2H), 4.99 (dd, J = 12.6, 7.5 Hz, 1H), 4.76-4.72 (m, 1H), 3.92 (s, 3H), 3.63-3.46 (m, 2H), 3.56 (s, 3H), 3.51 (s, 3H), 3.14 (t, J = 11.8 Hz, 1H), 2.77 (dd, J = 12.6, 7.5 Hz, 1H), 1.83-1.71 (m, 1H), 1.49-1.38 (m, 1H), 0.98 (s, 9H), 0.97 (s, 9H); 0.089 (s, 6H), 0.086 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 192.7, 150.3, 145.1, 141.3, 135.4, 126.6 (2C), 126.0, 125.4 (2C), 123.1, 107.4, 72.1, 71.5, 71.0, 69.5, 66.0, 59.2, 59.0, 55.1, 43.4, 39.6, 27.8, 24.1 (3C), 24.0 (3C), 16.3 (2C), −4.4 (2C), −4.5 (2C); IR (film) νmax 3284, 2566, 1727 cm−1; HRMS (ESI) m/z 679.32488 (MH+, C35H55ClO7Si2H+ requires 679.32476).
A solution of bis((S)-4-phenyl-4,5-dihydrooxazol-2-yl)methane62 (76 mg, 0.24 mmol) and 2,2´-dipyridyl (2 crystals), in anhydrous THF (200 µL) at 0 °C under Ar was treated dropwise with n-BuLi (1.6 M in hexanes, 250 µL, 0.40 mmol) until the mixture turned a reddish-brown color. The solution was warmed to rt and stirred for 1 h, then treated dropwise with allylzinc bromide (1.0 M in THF, 240 µL, 0.24 mmol) and cooled to −78 °C. A solution of ketone 38 (95 mg, 0.14 mmol) in anhydrous THF (220 µL) was added dropwise, and the resultant mixture was stirred at −78 °C under Ar for 1 h. The reaction was quenched by the addition of MeOH–H2O (1:1, 1 mL), and the mixture was extracted with Et2O (3 × 1 mL). The combined organic layers were washed with NaOH (0.5 M, 1 mL), dried (MgSO4), and concentrated in vacuo. Flash chromatography (SiO2, 2:23:75 Et3N/EtOAc/hexanes elution) afforded 40 (79 mg, 0.11 mmol, 93:7 dr, 79%) as a colorless oil: [α]25D −36 (c 1.2, CHCl3); 1H NMR (CDCl3, 300 MHz) δ 7.47-7.31 (m, 5H), 6.37-6.27 (m, 1H), 6.21 (s, 1H), 5.24 (dd, J = 12.3, 7.2 Hz, 1H), 5.09 (s, 2H), 4.93-4.79 (m, 2H), 4.65 (d, J = 6.9 Hz, 1H), 3.82 (s, 3H), 3.47-3.27 (m, 3H), 3.41 (s, 3H), 3.37 (s, 3H), 3.09 (t, J = 12.0 Hz, 1H), 2.74 (dd, J = 12.3, 7.2 Hz, 1H), 1.85-1.77 (m, 1H), 1.73-1.65 (m, 1H), 1.57-1.53 (m, 1H), 1.50-1.37 (m, 1H), 0.91 (s, 9H), 0.85 (s, 9H), 0.11 (s, 6H), 0.08 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 150.3, 145.1, 141.3, 140.9, 135.4, 126.7 (2C), 126.1, 125.5 (2C), 123.3, 123.2, 107.4, 73.8, 72.1, 71.5, 71.0, 69.1, 65.4, 59.1, 59.0, 55.5, 48.0, 43.1, 39.6, 27.9, 24.1 (3C), 24.0 (3C), 16.4 (2C), −4.4 (2C), −4.5 (2C); IR (film) νmax 3087, 2991, 2836, 1629, 1467, 933 cm−1; HRMS (ESI) m/z 721.37162 (MH+, C38H61ClO7Si2H+ requires 721.37171).
A mixture of 18-crown-6 (34 mg, 0.13 mmol), KOt-Bu (14 mg, 0.13 mmol), and anhydrous THF (700 µL) at 0 °C under Ar was stirred for 15 min, then treated with a solution of 40 (30.0 mg, 0.0416 mmol) in anhydrous THF (150 µL, added dropwise over 3 min). The resulting mixture was stirred at 0 °C under Ar for 1 h. The reaction was quenched by the addition of H2O (1 mL), and the mixture was diluted with Et2O (2 mL). The layers were separated, and the organic layer was dried (MgSO4) and concentrated in vacuo. Flash chromatography (SiO2, 1% Et3N in 10% EtOAc–hexanes elution) afforded 41 (27.5 mg, 0.0381 mmol, 92%) as a colorless oil: [α]25D −22 (c 1.5, CHCl3); 1H NMR (CDCl3, 300 MHz) δ 7.52-7.34 (m, 5H), 5.88-5.76 (m, 1H), 5.27 (dd, J = 12.2, 7.0 Hz, 1H), 5.20 (s, 2H), 5.00-4.88 (m, 2H), 4.60 (d, J = 6.9 Hz, 1H), 3.91 (s, 3H), 3.58-3.42 (m, 2H), 3.50 (s, 3H), 3.45 (s, 3H), 3.10 (t, J = 11.8 Hz, 1H), 2.98 (d, J = 14.7 Hz, 1H), 2.72 (dd, J = 12.2, 7.0 Hz, 1H), 2.56 (d, J = 15.0 Hz, 1H), 1.89-1.81 (m, 1H), 1.78-1.69 (m, 1H), 1.67-1.62 (m, 1H), 1.58-1.51 (m, 1H). 0.97 (s, 9H), 0.94 (s, 9H), 0.10 (s, 6H), 0.08 (s, 6H); 13C NMR (CDCl3, 75 MHz) δ 192.5, 147.7, 142.5, 138.8, 132.8, 124.2 (2C), 123.5, 122.9 (2C), 120.7, 104.8, 71.1, 69.6, 68.9, 68.4, 66.6, 56.6, 56.4, 52.9, 49.8, 41.2, 40.5, 37.0, 36.3, 25.3, 21.5 (3C), 21.4 (3C), 13.9 (2C), −5.0 (2C), −5.1 (2C); IR (film) νmax 3055, 2978, 2844, 1782, 1631, 1423, 1012, 941 cm−1; HRMS (ESI) m/z 721.37180 (MH+, C38H61ClO7Si2H+ requires 721.37171).
A saturated solution of O3 in EtOAc was prepared by bubbling ozone through EtOAc at −78 °C for 10 min. The concentration was determined to be 0.007 M as measured by titration with styrene.55 Then, a solution of 41 (27 mg, 0.037 mmol), pyridine (10 µL), and Et3N (16.0 µL, 11.6 mg, 0.115 mmol, 3.1 equiv) in EtOAc (0.5 mL) was cooled to −40 °C. A portion of the previously prepared solution of O3 in EtOAc (0.007 M, 8 mL, 0.056 mmol, 1.5 equiv), which was precooled to −78 °C, was then added to this solution. The resultant mixture was stirred at −78 °C for 5 min, then diluted with anhydrous MeOH (1.0 mL) and treated with powdered 4 Å molecular sieves (30 mg) and CH3NH2 (2.0 M in MeOH, 76 µL, 0.15 mmol, 4.1 equiv). This mixture was stirred at rt under Ar for 30 min, then treated with NaBH3CN (4.8 mg, 0.076 mmol) and stirred for 16 h. It was then diluted with EtOAc (2 mL), washed with aq KOH (10 M, 1 mL), and the layers were separated. The aqueous layer was extracted with EtOAc (3 × 2 mL), and the combined organic layers were dried (MgSO4) and concentrated in vacuo. Flash chromatography (SiO2, 1% Et3N in 15–20% EtOAc–hexanes gradient elution) afforded recovered 41(7.3 mg, 27% recovery) and 42 (15 mg, 0.020 mmol, 54%, 74% based on recovered 41) as a yellow oil: [α]25D −25 (c 1.4, CHCl3); 1H NMR (CDCl3, 300 MHz) δ 7.52-7.32 (m, 5H), 5.05 (s, 2H), 4.83 (dd, J = 12.0, 6.9 Hz, 1H), 4.61 (d, J = 6.9 Hz, 1H), 4.02 (s, 3H), 3.87-3.73 (m, 2H), 3.62 (s, 3H), 3.57 (s, 3H), 3.06 (t, J = 11.8 Hz, 1H), 2.97 (s, 3H), 2.71-2.62 (m, 2H), 2.38 (dd, J = 12.2, 7.0 Hz, 1H), 2.28 (d, J = 15.0 Hz, 1H), 2.23 (br s, 1H), 2.18-2.13 (m, 1H), 2.00 (d, J = 15.0 Hz, 1H), 1.69-1.61 (m, 1H), 1.44-1.38 (m, 1H), 1.35-1.28 (m, 1H), 0.87 (s, 9H), 0.81 (s, 9H), 0.10 (s, 6H), 0.07 (s, 6H); 13C NMR (CDCl3, 125 MHz) δ 197.1, 152.4, 147.1, 137.4, 128.8 (2C), 128.1, 127.5 (2C), 109.4, 75.1, 74.1, 73.6, 73.0, 71.2, 61.2, 61.0, 54.1, 50.0, 47.0, 45.9, 45.1, 41.7, 40.9, 39.2, 29.9, 26.1 (3C), 26.0 (3C), 18.2 (2C), −4.5 (2C), −5.0 (2C); IR (film) νmax 3125, 2923, 2810, 1741, 1633, 1420, 1208, 1138, 982 cm−1; HRMS (ESI) m/z 760.38014 (MNa+, C38H64ClNO7Si2Na+ requires 760.38021).
A mixture of 42 (15 mg, 0.020 mmol), 4 Å MS (80 mg), and anhydrous CH2Cl2 (2.0 mL) was stirred at rt under Ar for 10 min, then cooled to −40 °C. Next, BCl3 (1.0 M in CH2Cl2, 30 µL, 0.030 mmol, 1.5 equiv) was added dropwise, and the resultant mixture was stirred at −40 °C under Ar for 18 h, then concentrated in vacuo. The residue was purified by flash chromatography (SiO2, 2:30:68 Et3N/EtOAc/hexanes elution), affording 46 (6.4 mg, 0.0091 mmol, 45%) as a colorless oil: [α]25D −79 (c 1.2, CHCl3); 1H NMR (pyridine-d5, 500 MHz) δ 7.44-7.41 (m, 3H), 7.36-7.33 (m, 2H), 5.25 (dd, J = 12.2, 6.8 Hz, 1H), 5.12 (s, 2H), 4.80 (d, J = 7.0 Hz, 1H), 4.14 (s, 3H), 3.84 (s, 3H), 3.50-3.44 (m, 1H), 3.41-3.36 (m, 1H), 3.20 (t, J = 12.0 Hz, 1H), 3.11 (d, J =15.5 Hz, 1H), 2.75-2.69 (m, 3H), 2.61 (d, J = 15.5 Hz, 1H), 2.52-2.47 (m, 1H), 2.46 (s, 3H), 1.71-1.68 (m, 1H), 1.55-1.50 (m, 1H), 1.46-1.40 (m, 1H), 0.96 (s, 9H), 0.92 (s, 9H), 0.26 (s, 3H), 0.22 (s, 3H), 0.13 (s, 6H); 13C NMR (pyridine-d5, 125 MHz) δ 192.0, 158.8, 142.3, 138.1, 129.9 (2C), 129.2, 128.4 (2C), 76.2, 75.8, 75.3, 74.2, 72.0, 67.4, 59.5, 59.2, 56.9, 52.3, 50.8, 46.3, 44.0, 40.5, 37.6, 35.4, 30.0 (3C), 29.8 (3C), 20.0 (2C), −4.1 (2C), −4.2 (2C); IR (film) νmax 3209, 2974, 2795, 1763, 1651, 1402, 1265, 912 cm−1; HRMS (ESI) m/z 706.37199 (MH+, C37H60ClNO6Si2H+ requires 706.37205).
A solution of 46 (8.7 mg, 0.012 mmol) in anhydrous THF (100 µL) at 0 °C under Ar was treated with TBAF (1.0 M in THF, 27 µL, 0.027 mmol, 2.2 equiv) in one portion. The resulting mixture was stirred at 0 °C for 20 min. The reaction was quenched by the addition of ice water (0.5 mL), and the mixture was extracted with cold CH2Cl2 (cooled in ice bath, 3 × 0.5 mL). The combined organic layers were dried (MgSO4) and concentrated in vacuo in an ice bath. The unstable crude diol was dissolved in acetone (0.5 mL) at 0 °C, then 4 Å MS (50 mg), NMO (4.2 mg, 0.036 mmol), and TPAP (0.4 mg, 0.001 mmol) were added in order to the solution. The resulting mixture was stirred at 0 °C under Ar for 30 min, then slowly warmed to rt over 1 h and stirred at rt for 1 additional h. It was then filtered through a plug of SiO2 (rinsed with 5 mL EtOAc), dried (MgSO4), and concentrated in vacuo. Flash chromatography (SiO2, 1% Et3N in 2% MeOH–CH2Cl2 elution) afforded 47 (3.3 mg, 0.0070 mmol, 57%) as a white solid: [α]25D −122 (c 0.7, CH2Cl2); 1H NMR (pyridine-d5, 500 MHz) δ 7.46-7.42 (m, 3H), 7.39-7.37 (m, 2H), 5.20 (dd, J = 12.0, 6.5 Hz, 1H), 5.02 (s, 2H), 4.87 (s, 1H), 4.10 (s, 3H), 3.92 (d, J = 14.0 Hz, 1H), 3.80 (s, 3H), 3.74 (d, J = 14.0 Hz, 1H), 3.16 (t, J = 12.5 Hz, 1H), 3.07 (d, J =16.0 Hz, 1H), 2.70-2.63 (m, 3H), 2.55 (d, J = 15.5 Hz, 1H), 2.45-2.41 (m, 1H), 2.40 (s, 3H), 1.65-1.62 (m, 1H); 13C NMR (pyridine-d5, 125 MHz) δ 202.8, 201.4, 193.3, 160.2, 143.6, 139.4, 131.6 (2C), 130.9, 130.3 (2C), 73.4, 71.9, 71.2, 69.1, 60.9, 60.6, 59.3, 58.3, 52.1, 47.7, 46.9, 41.9, 39.0, 36.8; IR (film) νmax 3024, 2931, 2795, 1825, 1633, 1429, 1176, 955 cm−1; HRMS (ESI) m/z 474.16785 (MH+, C25H28ClNO6H+ requires 474.16779).
To a solution of 47 (3.0 mg, 0.0063 mmol) in anhydrous MeOH (1.0 mL) under Ar was added 10% Pd/C (10 mg, 3.3 wt equiv). The resulting mixture was stirred at rt under H2 (1 atm) for 2 h, then filtered through a plug of Celite (washed with CH2Cl2), dried (MgSO4), and concentrated in vacuo. Flash chromatography (SiO2, 1.5 × 8 cm, 5% MeOH–CH2Cl2 elution) afforded 48 (2.4 mg, 0.0063 mmol, 99%) as a pale yellow oil: [α]25D −135 (c 0.6, CH2Cl2); 1H NMR (pyridine-d5, 500 MHz) δ 8.66 (br s, 1H), 5.30 (dd, J = 12.0, 6.5 Hz, 1H), 5.13 (s, 1H), 4.14 (s, 3H), 4.03 (d, J = 14.0 Hz, 1H), 3.83 (s, 3H), 3.76 (d, J = 14.0 Hz, 1H), 3.26 (t, J = 12.2 Hz, 1H), 3.17 (d, J = 15.5 Hz, 1H), 2.82-2.74 (m, 3H), 2.65 (d, J = 15.5 Hz, 1H), 2.55-2.52 (m, 1H), 2.50 (s, 3H), 1.76-1.73 (m, 1H); 13C NMR (pyridine-d5, 125 MHz) δ 203.1. 201.7, 194.1, 160.9, 140.2, 72.9, 71.7, 69.3, 60.2, 59.9, 56.8, 55.5, 53.2, 48.9, 46.1, 42.0, 38.4, 36.4; IR (film) νmax 3054, 2832, 1836, 1477, 1201, 934 cm−1; HRMS (ESI) m/z 406.10270 (MNa+, C18H22ClNO6Na+ requires 406.10279).
TiCl4 (0.04 M solution in CH2Cl2, 20 µL, 0.0008 mmol) was added to a solution of 48 (2.0 mg, 0.0052 mmol) in anhydrous MeOH (100 µL). The solution was stirred at rt for 15 min, then treated with Et3N (4 µL, 2.9 mg, 0.029 mmol) and stirred at rt for 45 min. The mixture was concentrated in vacuo, and the residue was purified by flash chromatography (SiO2, 1.5 × 6 cm, 1% Et3N in 5–15% MeOH–CH2Cl2 gradient elution), affording 1 (1.1 mg, 0.0027 mmol, 52%) and enol ether regioisomer 49 (0.3 mg, 0.00075 mmol, 14%). For 1: white film, [α]25D −171 (c 0.81, pyridine), lit2b [α]25D −206 (c 0.69, pyridine); 1H NMR (pyridine-d5, 500 MHz) δ 8.47 (br, s, 1H), 5.61 (s, 1H), 5.20 (dd, J = 11.8, 6.8 Hz, 1H), 5.03 (s, 1H), 4.04 (s, 3H), 3.80 (s, 3H), 3.73 (s, 3H), 3.16 (t, J = 12.0 Hz, 1H), 3.07 (d, J = 15.5 Hz, 1H), 2.69-2.63 (m, 3H), 2.54 (d, J = 15.5 Hz, 1H), 2.45-2.42 (m, 1H), 2.39 (s, 3H), 1.65-1.62 (m, 1H); 13C NMR (pyridine-d5, 125 MHz) δ 201.3, 192.8, 188.9, 159.7, 138.9, 105.5, 72.9, 70.7, 68.3, 60.4, 60.1, 58.8, 57.8, 53.2, 51.6, 47.2, 41.4, 38.5, 36.3; IR (film) νmax 3410, 2899, 2817, 1655, 1641, 1364, 1205, 1079, 935 cm−1; HRMS (ESI) m/z 398.13655 (MH+, C19H24ClNO6H+ requires 398.13649).
For 49: white film, [α]25D −112 (c 0.3, pyridine), 1H NMR (pyridine-d5, 500 MHz) δ 8.40 (br, s, 1H), 5.32 (s, 1H), 5.16 (dd, J = 12.0, 7.0 Hz, 1H), 4.94 (s, 1H), 4.07 (s, 3H), 3.74 (s, 3H), 3.57 (s, 3H), 3.13 (t, J = 12.5 Hz, 1H), 3.02 (d, J = 16.5 Hz, 1H), 2.67-2.60 (m, 3H), 2.48 (d, J = 15.5 Hz, 1H), 2.43-2.38 (m, 1H), 2.36 (s, 3H), 1.62-1.60 (m, 1H); HRMS (ESI) m/z 398.13664 (MH+, C19H24ClNO6H+ requires 398.13649).
We thank the National Science Foundation (CHE-716991), the National Institutes of Health (GM70483), and Brigham Young University (Graduate Research Fellowship to F.L., Undergraduate Research Award to S.S.T.; Mentoring Environment Grant to S.L.C.) for financial support.
Supporting Information Available. Experimental procedures and characterization data for new compounds not described in the Experimental Section, and NMR spectra for all new compounds. This material is available free of charge via the Internet at http://pubs.acs.org.