This preview has intentionally blurred parts. Sign up to view the full documentView Full Document
- Download Document
- Word Count: 9904
Unformatted Document Excerpt
Am. 8266 J. Chem. Soc. 1996, 118, 8266-8277 Tandem [4 + 2]/[3 + 2] Cycloadditions of Nitroalkenes. 9. Synthesis of (-)-Rosmarinecine Scott E. Denmark,* Atli Thorarensen, and Donald S. Middleton Contribution from the Roger Adams Laboratory, Department of Chemistry, UniVersity of Illinois, Urbana, Illinois 61801 ReceiVed May 15, 1996X Abstract: (-)-Rosmarinecine (2) is the necine base portion of the pyrrolizidine alkaloid (-)-rosmarinine. (-)Rosmarinecine is a representative of the group of pyrrolizidines that show a cis relationship between adjacent stereocenters C(1), C(7), and C(7a), in addition to a highly oxygenated skeleton. (-)-Rosmarinecine (2) has been synthesized in eight steps and 14.8% overall yield, as an illustration of a general approach for the construction of pyrrolizidines having this stereochemical feature. The key step in the asymmetric synthesis is a Lewis acid-promoted, tandem inter-[4 + 2]/intra-[3 + 2] cycloaddition between a fumaroyloxy nitroalkene and a chiral vinyl ether. Introduction and Background Pyrrolizidine alkaloids have long attracted the interest of synthetic organic chemists due to their diverse biological properties and their structural complexity.1 Rosmarinine (1) was isolated from S. rosmarinifolius Linn. (family Compositae) and upon hydrolysis afforded the necine base (-)-rosmarinecine (2).2 Warren and co-workers3 determined the structure of (-)rosmarinecine by synthesis from retronecine (9), the structure of which had already been determined by Adams and Hamlin.4 In addition, an X-ray crystal structure of rosmarinine (1) has been reported for the purpose of examining the folding of the macrolactone and the necine nucleus.5 The X-ray crystal structure serves to establish independently the stereostructure of (-)-rosmarinecine (2) with respect to the configurations at C(1), C(6), C(7), and C(7a). The necine portion of 1 has been found in other pyrrolizidine alkaloids with different acid chains attached to the hydroxyl groups: e.g., angularine (3), 12-O-acetylrosmarinine (4), neorosmarinine (5), and petitianine (6) (Figure 1).12,13 Rosmarinecine has since been isolated from various plants in the Compositae family, including S. pleitocephalus,6 S. X Abstract published in AdVance ACS Abstracts, August 1, 1996. (1) (a) Leonard, N. J. In The Alkaloids; Manske, R. H. F., Holmes, H. L., Eds.; Academic Press: New York, 1950; Vol. 1, Chapter 4. (b) Leonard, N. J. In The Alkaloids; Manske, R. H. F., Ed.; Academic Press: New York, 1960; Vol. 6, Chapter 3. (c) Warren, F. L. Fortschr. Chem. Org. Naturst. 1955, 12, 198. (d) Warren, F. L. In The Alkaloids; Manske, R. H. F., Ed.; Academic Press: New York, 1970; Vol. 12, Chapter 4. (e) Natural Occuring Pyrrolizidine Alkaloids; Rizk, A-F. M., Ed.; CRC: Boston, 1991. (f) Bull, L. B.; Culvenor, C. C. J.; Dick, A. T. The Pyrrolizidine Alkaloids; NorthHolland Publishing Co. Amsterdam, The Netherlands, 1968. (g) Pyrrolizidine Alkaloids; Enviromental Health Organization: Geneva, Switzerland, 1988. (h) Mattocks, A. R. Chemistry and Toxicology of Pyrrolizidine Alkaloids; Academic Press: New York, 1986. (i) Robins, D. J. Nat. Prod. Rep. 1995, 12, 413. (j) Robins, D. J. Nat. Prod. Rep. 1994, 11, 613. (k) Robins, D. J. Nat. Prod. Rep. 1993, 10, 487. (l) Robins, D. J. Nat. Prod. Rep. 1992, 9, 313. (m) Robins, D. J. Nat. Prod. Rep. 1991, 8, 213. (n) Robins, D. J. Nat. Prod. Rep. 1990, 7, 377. (o) Robins, D. J. Nat. Prod. Rep. 1989, 6, 577. (p) Robins, D. J. Nat. Prod. Rep. 1989, 6, 221. (q) Robins, D. J. Fortschr. Chem. Org. Naturst. 1982, 41, 116. (2) (a) Waal, H. L. Nature 1940, 146, 777. (b) Waal, H. L. Onderst. J. Vet. Sci. An. Ind. 1940, 15, 241. (c) Waal, H. L. Onderst. J. Vet. Sci. An. Ind. 1941, 16, 149. (3) Dry, L. J.; Koekemoer, M. J.; Warren, F. L. J. Chem. Soc. 1955, 59. (4) Adams, R.; Hamlin, K. E., Jr. J. Am. Chem. Soc. 1942, 64, 2597. (5) Freer, A. A.; Kelly, H. A.; Robins, D. J. Acta Cryst. C 1986, 42, 1348. (6) Kunec, E. K.; Robins, D. J. J. Chem. Soc., Chem. Commun. 1986, 250. Figure 1. Pyrrolizidine alkaloids that contain rosmarinecine. triangularis,7 S. taiwanensis Hayata,8 S. pterophorus,9 S. hygrophilus,10 S. adnatus D.C.,11 S. angulatus L.,12 S. hadiensis and S. syringifolius.13 In addition, rosmarinine, among other alkaloids, has been isolated from butterflies in Southern Australia of the Danaus plexippus L. and Danaus chrysippus L species.9 It was demonstrated that the butterflies mainly obtained rosmarinine by consumption of S.pterophorus. The butterflies can store the alkaloids for extended period of time, and it is speculated that the purpose is to make the insects distasteful to predators. Finally, the biosynthesis of (-)rosmarinecine is now well understood.14,15 (7) Roitman, J. N. Aust. J. Chem. 1983, 36, 1203. (8) Lu, S.-T.; Lin, C.-N.; Wu, T.-S.; Shieh, D.-C. J. Chin. Chem. Soc. 1972, 19, 127. (9) Edgar, J. A.; Cockrum, P. A.; Frahn, J. L. Experientia 1976, 15, 1535. (10) Richardson, M. F.; Warren, F. L. J. Chem. Soc. 1943, 452. (11) Hughes, C. A.; Gordon-Gray, C. G.; Schlosser, F. D.; Warren, F. L. J. Chem. Soc. 1965, 2370. (12) Porter, L. A.; Geissman, T. A. J. Org. Chem, 1962, 27, 4132. (13) (a) Were, O.; Benn, M.; Munavu, R. M. J. Nat. Prod. 1991, 54, 491. (b) Were, O.; Benn, M.; Munavu, R. M. Phytochemistry 1993, 32, 1595. (14) (a) Kelly, H. A.; Robins, D. J. J. Chem. Soc., Perkin Trans. 1 1987, 177. (b) Kelly, H. A.; Robins, D. J. J. Chem. Soc., Perkin Trans. 1 1987, 2195. (c) Kelly, H. A.; Robins, D. J. J. Chem. Soc., Chem. Commun. 1988, 329. (d) Kunec, E. K.; Robins, D. J. J. Chem. Soc., Perkin Trans. 1 1989, 1437. (e) Denholm, A. A.; Kelly, H. A.; Robins, D. J. J. Chem. Soc., Perkin Trans. 1 1991, 2003 S0002-7863(96)01630-7 CCC: $12.00 1996 American Chemical Society Synthesis of (-)-Rosmarinecine Scheme 1 J. Am. Chem. Soc., Vol. 118, No. 35, 1996 8267 Figure 2. Structure of several cis-substituted necines. The simpler necines such as isoretronecanol have served as common targets to demonstrate new synthetic methodology, as witnessed by the myriad of synthetic approaches on record.1,16 (-)-Rosmarinecine, on the other hand, is structurally one of the more complex necines bases, with four stereogenic centers on the eight atoms that form the 1-azabicyclo[3.3.0]octane ring system. It is, therefore, not very surprising that, prior to our effort, there was only one synthesis of this natural product, reported by Tatsuta et al., who prepared (-)-rosmarinecine in 17 steps from D-glucosamine.17 Synthesis Design The asymmetric, tandem [4 + 2]/[3 + 2] nitroalkene cycloaddition reaction has been explored extensively in recent years.18,19 This transformation is ideally suited for the synthesis of necine bases since the core pyrrolizidine skeleton arises from hydrogenation of the bicyclic nitroso acetals, which are the end products of the tandem cycloaddition. As an illustration of the power of the asymmetric tandem [4 + 2]/[3 + 2] nitroalkene cycloaddition reaction, we recently disclosed the synthesis of (-)-hastanecine.20 This constituted the first example of the use of nitroalkene cycloaddition chemistry in the synthesis of natural products. Herein, we disclose a general approach to necines which bear a cis relationship between any or all of the stereocenters at C(1), C(7), and C(7a), as can be seen in examples 2 and 7-10 (Figure 2).21 The retrosynthetic analysis for (-)-rosmarinecine is summarized in Scheme 1. It was proposed that rosmarinecine could be obtained from the corresponding R-hydroxy lactam 15 by exhaustive reduction. The R-hydroxy lactam would arise from hydrogenolytic cleavage of nitroso acetal 14, which itself originates from a tandem inter-[4 + 2]/intra-[3 + 2] cycloaddition of nitroalkene 12 and a chiral vinyl ether 11. The geometrical constraints imposed by the tether require that the intramolecular [3 + 2] cycloaddition must occur in an endo (15) (a) Robins, D. J. Chem. Soc. ReV. 1989, 18, 375. (b) Robins, D. J. The Alkaloids; Cordell, G. A., Ed.; Academic Press: New York, 1995; Vol. 46, Chapter 1. (16) (a) Ikeda, M.; Sato, T.; Ishibashi, H. Heterocycles 1988, 27, 1465. (b) Dai, W.-M.; Nagao, Y. Heterocycles 1990, 30, 1231. (c) Wrobel, J. T. The Alkaloids: Brossi, A., Ed.; Academic Press: New York, 1985; Vol. 26, Chapter 7. (17) Tatsuta, K.; Takahashi, H.; Amemiya, Y.; Kinoshita, M. J. Am. Chem. Soc. 1983, 105, 4096. (18) Denmark, S. E.; Thorarensen, A. Chem. ReV. 1996, 96, 137. (19) (a) Denmark, S. E.; Moon, Y.-C.; Senanayake, C. B. W. J. Am. Chem. Soc. 1990, 112, 311. (b) Denmark, S. E.; Senanayake, C. B. W.; Ho, G. D. Tetrahedron 1990, 46, 4857. (c) Denmark, S. E.; Senanayake, C. B. W. J. Org. Chem. 1993, 58, 1853. (d) Denmark, S. E.; Schnute, M. E.; Senanayake, C. B. W. J. Org. Chem. 1993, 58, 1859. (e) Denmark, S. E.; Schnute, M. E.; Marcin, L. R.; Thorarensen, A. J. Org. Chem. 1995, 60, 3205. (20) Denmark, S. E.; Thorarensen, A. J. Org. Chem. 1994, 59, 5672. (21) Preliminary communication: Denmark, S. E.; Thorarensen, A.; Middleton, D. S. J. Org. Chem. 1995, 60, 3574. Scheme 2 mode on the same face of the nitronate dipole 13 to which the tether is attached. This would give rise to a cis relationship between the stereocenters at C(1) and C(7). Furthermore, the stereocenter at C(7a) would be cis to those at C(1) and C(7) and would establish the correct relative configuration between C(1), C(7), and C(7a) for (-)-rosmarinecine. The configuration at C(6) is established as a consequence of the dipolarophile geometry, which in this case must be cis. Finally, the absolute stereochemical outcome is established by the appropriate combination of chiral auxiliary and Lewis acid for the intermolecular [4 +2] process to engender the correct configuration at C(4) in the intermediate nitronate 13 (C(1) in (-)-rosmarinecine). This center, in turn, will dictate the creation of all the other required stereocenters in both a relative and an absolute sense. Thus, the synthesis precursors simplify to the maleatederived -acyloxy nitroalkene 12 and a nonracemic chiral vinyl ether 11. This initial and highly direct approach had to be abandoned due to the instability of the nitroalkene 12, which rapidly isomerized to the fumarate isomer 16. After many unsuccessful attempts to employ 12, it was recognized that 16 could also serve as a suitable precursor for (-)-rosmarinecine (Scheme 2). This more tractable nitroalkene would lead to an R-hydroxy lactam 17, by analogy to the maleate series, in a tandem [4 + 2]/[3 + 2] cycloaddition, followed by hydrogenolysis. Since the two-dimensional stereochemistry of the fumarate will translate into the incorrect configuration at C(6) in 17 for (-)rosmarinecine, an inversion at this center is necessitated. While a strategic inelegance, the inversion of the hydroxyl at C(1) should elongate the synthesis by only one or two steps, and therefore, this would still constitute an efficient synthesis. Results and Discussion Use of Achiral Dienophiles. To evaluate the feasibility of the approach outlined in Scheme 2, the nitroalkene 21 was prepared by analogy with 2-(benzoyloxy)-1-nitroethene that was used in the synthesis of (-)-hastanecine.20,22 The commercially available ethyl ester 18 was easily converted to the acid chloride (22) Kraus, G. A.; Thurston, J.; Thomas, P. J.; Jacobson, R. A.; Su, Y. Tetrahedron Lett. 1988, 29, 1879. 8268 J. Am. Chem. Soc., Vol. 118, No. 35, 1996 Scheme 3 Scheme 5 Denmark et al. Scheme 4 19 (93%), which was subsequently treated with 2023 to afford the nitroalkene 21 as a yellow solid in 90% yield (Scheme 3). For orienting experiments, butyl vinyl ether was selected as the test dienophile, in order to simplify product analysis by 1H NMR spectroscopy. Addition of methylaluminum bis(2,6-ditert-butyl-4-methylphenoxide) (MAD)24 to a solution of nitroalkene 21 and butyl vinyl ether at -78 C afforded the nitroso acetal 22 in 59% yield as a 1.9:1 mixture of diastereomers, with the endo diastereomer (R-anomer) predominating. Changing the Lewis acid to methylaluminum bis(2,6-diphenylphenoxide) (MAPh) gave 22 in 64% yield as a 1:1 mixture of diastereomers. If TiCl2(O-iPr)2 was used as the Lewis acid, a single isomer of 22 was obtained. However, the yield was poor (28%), and the reaction was found to give a number of byproducts that made purification difficult. With ready access to multigram quantities of 22, an extensive study was then undertaken to establish the optimal conditions for hydrogenolysis of the nitroso acetal 22 to R-hydroxy lactam 17 (Scheme 4). This study focused on four different catalysts (PtO2, Raney nickel, Pd/C, and Pd(OH)2) in six different solvents (methanol, 2-propanol, ethyl acetate, THF, toluene, and chloroform) at three different pressures (1 atm, 100 psi, and 200 psi) for periods of time between 14 and 40 h. Disappointingly, none of these conditions afforded the desired product 17. Either decomposition to a number of unidentified byproducts occurred or starting material was recovered. The inability to effect the desired hydrogenolytic cleavage of 22 to 17 stands in sharp contrast to the facile hydrolysis of nitroso acetals such as 23 to the R-hydroxy lactam 24 (Scheme 4). The key structural difference between 22 and 23 is clearly the lactone function, and several reasons for the failure of 22 (23) Babievskii, K. K.; Belikov, V. M.; Tikhohova, N. A. IzV. Akad. Nauk SSSR, Ser. Kim. 1970, 5, 1161. (24) (a) Maruoka, K.; Itoh, T.; Yamamoto, H. J. Am. Chem. Soc. 1985, 107, 4573. (b) Nonoshita, K.; Banno, H.; Maruoka, K.; Yamamoto, H. J. Am. Chem. Soc. 1990, 112, 316. (c) Maruoka, K.; Ooi, T.; Yamamoto, H. J. Am. Chem. Soc. 1990, 112, 9011. (d) Maruoka, K.; Itoh, T.; Shirasaka, T.; Yamamoto, H. J. Am. Chem. Soc. 1988, 110, 310. to be converted to 17 can be formulated. First, it is possible that the lactone undergoes competitive reduction under the hydrogenolysis conditions. Second, the shorter C-O bonds and sp2-hybridized carbonyl group in the -lactone could be creating strain in the tricyclic nitroso acetal. This strain may disfavor recyclization after N-O bond cleavage, thus allowing unwanted side reactions to occur preferentially. Third, unmasking the aldehyde at C(6) may lead to -elimination of the lactone oxygen, also relieving ring strain. A potential solution to this problem is to relieve the strain (and leaving group ability) engendered by the carbonyl group in the five-membered ring by changing the hybridization of C(3) in 22 from sp2 to sp3. This could be accomplished by selectively reducing the carbonyl function. Gratifyingly, treatment of the nitroso acetal 22 with 1 equiv of diisobutylaluminum hydride (DIBAL-H) at -78 C gave the lactol 25 in 55% yield (Scheme 5). Treatment of this product with Raney nickel at 150 psi of hydrogen for 36 h gave the R-hydroxy lactam 26 in 57% yield as a highly crystalline, white solid after chromatography. Having thus established the feasibility for the creation of a key intermediate in our planned synthesis of (-)-rosmarinecine, we turned to the use of chiral vinyl ethers. Optimization of Reaction Parameters for the [4 + 2]/[3 + 2] Cycloaddition. The tandem [4 + 2]/[3 + 2] cycloaddition is the centerpiece in the planned synthesis of (-)-rosmarinecine. In that reaction, all of the stereogenic centers are installed in the correct relative and absolute sense, except that at C(6). To assure both a high yielding and a highly selective sequence of cycloadditions, we examined a number of permutations of Lewis acids and chiral vinyl ethers. To maximize efficiency, the entire synthesis was initially performed employing racemic auxiliaries. Initial experiments employed MAPh and TiCl2(O-iPr)2 as the Lewis acids. MAD was not considered due to its generally low selectivity as the Lewis acid in tandem [4 + 2]/[3 + 2] cycloadditions.18 The use of nitroalkene 21 and vinyl ether 27 (3 equiv) in reactions promoted by MAPh (3 equiv) led to a 93% yield of nitroso acetal 30 (Table 1). The use of less than 3 equiv of either the Lewis acid or the vinyl ether resulted in low yields (Table 1, entries 1-3). Although a high yield of the tandem cycloaddition product could be obtained, the nitroso acetal 30 was produced as a 6.3:1 exo/endo mixture of diastereomers, (entry 1, Table 2). The use of TiCl2(O-iPr)2 as the promoter improved the selectivity to 1:10 exo/endo, but this reaction was rather low yielding. More importantly, the product isolated from this cycloaddition was contaminated with unseparable impurities. We therefore turned our attention to the vinyl ether (-)-28 derived from trans-2(1-methyl-1-phenylethyl)cyclohexanol with MAPh as the promoter.25,26 The nitroso acetal 31 was isolated in excellent yield (25) (a) Comins, D. L.; Salvador, J. M. J. Org. Chem. 1993, 58, 4656. (b) Comins, D. L.; Salvador, J. M. Tetrahedron Lett. 1993, 34, 801. (c) Comins, D. L.; LaMunyon, D. H. Tetrahedron Lett. 1994, 35, 7343 (26) Denmark, S. E.; Thorarensen, A. J. Org. Chem. 1996, in press. Synthesis of (-)-Rosmarinecine Table 1. Cycloaddition with Vinyl Ether 27 Promoted by MAPh J. Am. Chem. Soc., Vol. 118, No. 35, 1996 8269 Table 3. Selected 1H NMR Data of the Major Diasteromeric Nitroso Acetals 30-32 entry 1 2 3 4 a MAPh, equiv 4.0 1.1 2.0 3.0 27, equiv 1.5 3 3 3 yield,a % 69 37 64 93 entry 1 2 3 4 5 a nitroso acetal 30a 30b 31a 32a 32b HC(6) ppm (J, Hz) a a 4.72 (t, 6.9) 4.64 (t, 7.1) 4.95 (dd, 2.2, 6.9) HC(2) ppm (J, Hz) 5.25 (d, 3.7) 5.28 (d, 3.6) 5.20 (d, 3.6) 5.30 (d, 3.7) 5.30 (d, 3.9) Yield after chromatographic purification. Table 2. Cycloaddition with Vinyl Ethers 27-29 Obscured by ethyl ester signal. entry 1 2 3 4 5 6 R Et Et i-Pr i-Pr Et Et auxiliary 27 27 28 28 29 29 promoter MAPha TiCl2(Oi-Pr)2 MAPha TiCl2(Oi-Pr)2 MAPha TiCl2(Oi-Pr)2 diastereoselectivity, exo/endo 6.3:1 1:10 12.2:1d 6.6:4.1:1e 41:1 >10:1b nitroso acetal 30 30 31 31 32 32 yield, % 93 57c 95 16 95 37c a All cycloadditions promoted by MAPh utilized 3 equiv of MAPh and 3 equiv of vinyl ether. b Minor could not be detected due to impurities. c Contaminated by an unknown impurity. d Minor assumed to be the other exo diastereomer. e The major diastereomer was the same as that from entry 3. (95%) and with an improved exo/exo selectivity of 12.2:1. To determine if the two diastereomeric nitroso acetals 31 belonged to the same or opposite enantiomeric families, the mixture was converted to an intermediate in the synthesis of (-)-rosmarinecine.27 The enantiomeric purity for this intermediate (12.9:1 enantiomeric ratio by chiral HPLC analysis) was shown to correspond well with the observed diasteromeric ratio of 12.2: 1. Thus, the two diasteromeric nitroso acetals 31 do, in fact, belong to the opposite enantiomeric families. Unfortunately, they could not be separated by chromatography or recrystallization; consequently, it was not possible to enrich the mixture. The third vinyl ether to be examined was derived from (R)2,2-diphenylcyclopentanol 29.19e Cycloaddition of 21 and 29 promoted by MAPh afforded the nitroso acetal 32 in both excellent yield (95%) and high diastereoselectivty (Table 2, entry 5). In repeated experiments, the diastereoselectivty was found to be variable (41:1 to 20:1), probably due to unintentional enrichment during chromatographic purification. The minor diasteromer 32b was confirmed to be an endo-derived cycloadduct, since it was formed preferentially from the cycloaddition with TiCl2(Oi-Pr)2 as the promoter. Under these conditions, mainly 32b was produced but in rather low yield (Table 2, entry 6). We have demonstrated that cycloadditions promoted by (27) The enantiomeric ratio was determined on the 4-nitrobenzoate 50 by a sequence analogous to Scheme 11. TiCl2(Oi-Pr)2 are generally endo selective, even with 29 as the vinyl ether.19e Configurational Assignment. In nitroso acetals of the general structure 23, it had been observed that the relative configuration of the anomeric center, C(6), with respect to the ring junction carbon, C(4a), could be determined by analysis of the splitting pattern of HC(6).28 In the cycloaddition promoted by MAPh, the relationship should be trans if the exo product were generated kinetically, and the anomeric proton should be observed as a triplet. In contrast, with TiCl2(Oi-Pr)2 as the promoter, a cis relationship is dictated by the endo approach of the vinyl ether in the [4 + 2] cycloaddition. In this scenario, the anomeric proton should be observed as a doublet of doublets. It is therefore interesting that the coupling pattern of HC(6) in the nitroso acetals 30-32 displayed the same behavior. The anomeric proton appeared as a triplet in the nitroso acetal obtained from a MAPh-promoted reaction, supporting a trans relationship to C(4a) (Table 3, entries 3 and 4). On the other hand, the anomeric proton appeared as a doublet of doublets when the cycloaddition was promoted by a titanium Lewis acid, confirming a cis relationship between C(6) and C(4a) (entry 5). Interestingly, the chemical shift of the anomeric proton is also very sensitive to the auxiliary employed, while all the other protons on the tricyclic ring system are only slightly affected. The coupling pattern of the anomeric proton provided a means to identify the different diastereomeric nitroso acetals that were obtained in each cycloaddition. Armed with the knowledge that 29 behaved in a predictable and selective fashion with MAPh, it was easy to choose the appropriate configuration of 2,2-diphenylcyclopentanol for the synthesis of natural (-)rosmarinecine. Unmasking of the Nitroso Acetal. From our experience with lactone 22, it was anticipated that reduction of the lactone 32 would be required to allow the ring closure to occur.29 Diisobutylaluminum hydride, which had been successfully employed previously, afforded the lactol 33 in, at best, 38% yield (Table 4).30 The transformation was complicated by the competitive reduction of the ethyl ester, affording a mixture of aldehyde 34a and alcohol 34b. Thus, a survey was undertaken (28) Schnute, M. E. Ph.D. Thesis, University of Illinois at UrbanaChampaign, Urbana, IL, 1995; pp 21-23. (29) This was verified by the failure of 32 to undergo the desired hydrogenolysis. (30) (a) Hijfte, L. V.; Little, R. D.; Petersen, J. L.; Moeller, K. D. J. Org. Chem. 1987, 52, 4647. (b) Ley, S. V.; Santafianos, D.; Blaney, W. M.; Simmonds, M. S. J. Tetrahedron Lett. 1987, 28, 221. (c) Nakata, T.; Saito, K.; Oishi, T. Tetrahedron Lett. 1986, 27, 6345. (d) Corey, E. J.; Albonico, S. M.; Koelliker, U.; Schaaf, T. K.; Varma, R. K. J. Am. Chem. Soc. 1971, 93, 1490. (e) Yoon, N. M.; Gyoung, Y. S. J. Org. Chem. 1985, 50, 2443. Review: (f) Winterfeldt, E. Synthesis 1975, 617. 8270 J. Am. Chem. Soc., Vol. 118, No. 35, 1996 Table 4. Survey of Reducing Agents for the Reduction of 32 Denmark et al. Scheme 6 Scheme 7 mass recovery, % entry 1a 2a 3b 4b 5b 6a 7d reagent DIBAL-H Red-AlEtOH LiAlH(Ot-Bu)3 Red-Al Red-AlN-methyl piperazine L-Selectride LS-Selectride lactol 33 38 24 trace 0 0 55 0 overreduction 34 18 26 trace 29 85c 0 0 starting material 32 33 22 82 58 0 0 quantitative a Isolated yield. b Ratios determined by 1H NMR. c Numerous side products also observed. d No reduction product could be detected by 1H NMR. with reagents capable of effecting the lactone-to-lactol conversion:31 sodium bis(2-methoxyethoxy)aluminum hydride (RedAl),32 Red-AlEtOH,33 Red-Alamine,34 and lithium tris(tertbutoxy)aluminum hydride (LiAlH(Ot-Bu)3).35 Red-Al reduction of the lactone provided none of the desired product; instead, both overreduction products 34a and 34b (29%) were obtained. The ethanol-modified reagent afforded some of the desired product 33 (24%), but the mass recovery was very poor. The use of the N-methylpiperazine-modified Red-Al was ineffective. More promising was the use of lithium tris(sec-butyl)borohydride (L-Selectride), which afforded the desired lactol 33 in 55% yield. These results were still rather unsatisfactory for such a simple functional group transformation, and two separate strategies were undertaken to differentiate these groups: (1) hydrolysis of the ester to a carboxylic acid and (2) increasing the size of the ester to disfavor sterically reduction. The simple hydrolysis of the ethyl ester 32 to the carboxylic acid 35 was more difficult than expected. The nitroso acetal 32 was found to be extremely base sensitive, and, in most instances, the free auxiliary was the only recovered material. Also unsuccessful were various reagents that have been (31) (a) Walker, E. R. H. Chem. Soc. ReV. 1976, 5, 23. (b) Brown, H. C.; Krishnamurthy, S. Tetrahedron 1979, 35, 567. (c) Barrett, A. G. M. In ComprehensiVe Chemistry, Reductions; Fleming, I., Ed.; Pergamon Press: Oxford 1991; Vol. 8, Chapter 1.10. (32) (a) Doria, C.; Gaio, P.; Gandolfi, C. Tetrahedron Lett. 1972, 4307. (b) Klotzer, W.; Teitel, S.; Brossi, A. HelV. Chim. Acta 1971, 54, 2057. (c) Disselnkotter, H.; Lieb, F.; Oediger, H.; Wendisch, D. Liebigs Ann. Chem. 1982, 150. (d) Hohlbrugger, R.; Klotzer, W. Chem. Ber. 1979, 112, 849. (33) (a) Hung, H.-K.; Lam, H.-Y.; Niemczura, W.; Wang, M.-C.; Wong, C.-M. Can. J. Chem. 1978, 56, 638. (b) Tokoroyama, T.; Matsuo, K.; Kotsuki, H.; Kanazawa, R. Tetrahedron 1980, 36, 3377. (34) (a) Bartsch, H.; Schwarz, O. J. Heterocycl. Chem. 1982, 19, 1189. (b) Kanazawa, R.; Tokoroyama, T. Synthesis 1976, 526. (35) (a) Girotra, N. N.; Wendler, N. L. Tetrahedron Lett. 1975, 227. (b) Nicolaou, K. C.; Seitz, S. P.; Pavia, M. R. J. Am. Chem. Soc. 1981, 103, 1222. (c) Tsuda, Y.; Yoshimoto, K.; Yamashita, T.; Kaneda, M. Chem. Pharm. Bull. 1981, 29, 3238. (d) Cassidy, F.; Moore, R. W.; Wootton, G.; Baggaley, K. H.; Geen, G. R.: Jennings, L. J. A.; Tyrrell, A. W. R. Tertahedron Lett. 1981, 22, 253. (e) Stevens, R. V.; Vinogradoff, A. P. J. Org. Chem. 1985, 50, 4056. (f) Skattebl, L.; Stenstrm, Y. Acta Chem. Scand. B 1985, 39, 291. employed to cleave ethyl or methyl esters and ethers, such as potassium trimethylsilanolate,36 iodotrimethylsilane,37 lithium ethane thiolate,38 dimethyl sulfide aluminum tribromide,39 and sodium hydroxide in methanol.40 The only reaction that showed even moderate success was the transesterification of 32 to a methyl ester (36) in methanol promoted by CeCl3 or CoCl2 (Scheme 6).41 Increasing the bulk of the ester group proved to be a successful strategy. Our initial approach was to use the corresponding tert-butyl ester. From 18, the tert-butyl ester was prepared in 90% yield by esterification with isobutylene promoted by sulfuric acid (Scheme 7).42,43 Selective hydrolysis of the ethyl ester with 1 equiv of LiOH in THF afforded the acid 37 in 93% yield.44 The formation of the acid chloride from 37, and coupling with the potassium salt 20, afforded the nitroalkene 38 in 48% yield. (36) (a) Laganis, E. D.; Chenard, B. L. Tetrahedron Lett. 1984, 25, 5831. (b) Baldwin, J. E.; Adlington, R. M.; Mitchell, M. B. J. Chem. Soc., Chem. Commun. 1993, 1332. (37) (a) Olah, G. A.; Narang, S. C.; Balaram Gupta, B. G.; Malhotra, R. J. Org. Chem. 1979, 44, 1247. (b) Lott, R. S.; Chauhan, V. S.; Stammer, C. H. J. Chem. Soc., Chem. Commun. 1979, 495. (c) Ho, T.-L.; Olah, G. A. Angew. Chem., Int. Ed. Engl. 1976, 15, 774. (d) Jung, M. E.; Lyster, M. A. J. Am. Chem. Soc. 1977, 99, 968. (e) Morita, T.; Okamoto, Y.; Sakurai, H. J. Chem. Soc., Chem. Commun. 1978, 874. (38) (a) Vaughan, W. R.; Baumann, J. B. J. Org. Chem. 1962, 27, 739. (b) Feutrill, G. I.; Mirrington, R. N. Aust. J. Chem. 1972, 25, 1731. (c) Kornblum, N.; Scott, A. J. Am. Chem. Soc. 1974, 96, 590. (d) Bartlett, P. A.; Johnson, W. S. Tetrahedron Lett. 1970, 4459. For a review of dealkylation methods, see: (e) McMurray, J. Org. React. 1976, 24, 187. (39) (a) Node, M.; Nishide, K.; Sai, M.; Fuji, K.; Fujita, E. J. Org. Chem. 1981, 46, 1991. (b) Node, M.; Nishide, K.; Sai, M.; Fujita, E. Tetrahedron Lett. 1978, 5211. (c) Williard, P. G.; Fryhle, C. B. Tetrahedron Lett. 1980, 21, 3731. (40) (a) Liu, H.-W.; Auchus, R.; Walsh, C. T. J. Am. Chem. Soc. 1984, 106, 5335. (b) Gawronski, J. K.; Reddy, S. M.; Walborsky, H. M. J. Am. Chem. Soc. 1987, 109, 6726. (41) (a) Luche, J.-L.; Gemal, A. L. J. Chem. Soc., Chem. Commun. 1978, 976. (b) Gemal, A. L.; Luche, J.-L. J. Org. Chem. 1979, 44, 4187. For a review of transesterification, see: (c) Otera, J. Chem. ReV. 1993, 93, 1449. (42) Nelsen, S. F.; Gillespie, J. P. J. Org. Chem. 1975, 40, 2391. (43) An alternative method for the preparation of this compound has been reported: Neises, B.; Steglich, W. Org. Synth. 1985, 63, 183. (44) Patel, D. V.; VanMiddlesworth, F.; Donaubauer, J.; Gannett, P.; Sih, C. J. J. Am. Chem. Soc. 1986, 108, 4603. Synthesis of (-)-Rosmarinecine Table 5. Reductions of 39 with Various Hydride Reagents J. Am. Chem. Soc., Vol. 118, No. 35, 1996 8271 Scheme 8 mass recovery, %a entry 1 2 3 a reagent DIBAL-H LiAlH(Ot-Bu)3 L-Selectride lactol 40 31 65 86 side product 34 41 14 0 21 0 starting material 39 33 0 0 Isolated yield after chromatography. The nitroalkene 38 was then combined in a cycloaddition with vinyl ether 29 promoted by MAPh. The desired nitroso acetal 39 was isolated in 93% yield as a 19:1 mixture of exo/ endo diastereomers. The nitroso acetal could be enriched by a simple recrystallization from ethanol to a 31:1 exo/endo diastereomeric mixture. Surprisingly, DIBAL-H was still a rather nonselective reducing agent, affording only 31% yield of the desired lactol 40 with considerable reduction of the tertbutyl ester to 34 (Table 5). Lithium tris(tert-butoxy)aluminum hydride (2 equiv) afforded the desired lactol 40 in 65% yield, along with the unwanted diol 41 in 21% yield. However, when L-Selectride was employed as the reducing agent, the lactol 40 was isolated cleanly in 86% yield, with no overreduction detected. Subjecting of lactol 40 to the standard hydrogenolysis conditions, employing Raney nickel as the catalyst, afforded none of the desired product 26 after workup. The isolated product was very polar, as judged by TLC analysis, and seemed to include the tert-butyl ester by 1H NMR spectroscopy. It appeared that the initial stages of the hydrogenolysis had been successful, but the final lactamization required to give the desired product did not occur. All attempts to cyclize the suspected intermediate failed. Apparently, the tert-butyl group was too bulky to allow the pyrrolidine to attack the carbonyl group in the strained bicyclic framework. We thus elected to test the corresponding isopropyl ester in the hope that it would serve effectively in both stages of the unmasking processes. The requisite nitroalkene 44 was easily prepared from maleic anhydride (Scheme 8). Reaction of maleic anhydride with potassium isopropoxide, according to the procedure of Veibel and Pedersen, afforded the potassium maleate salt 42 in excellent yield (91%).45 Treatment of 42 with (COCl)2/DMF as before resulted in the formation of the desired acid chloride 43 (66%).46 The attendant isomerization of the double bond was assured by the 15.4 Hz vicinal coupling constant, which is in good agreement with that observed in the nitroalkenes derived from 18. The acid chloride 43 was coupled with 20 to afford the target nitroalkene 44 in 69% yield. (45) Veibel, S.; Pedersen, C. Acta Chem. Scand. 1955, 9, 1674. (46) For alternative methods for the preparation of the acid chloride, see: (a) Oiwa, M.; Matsumoto, A. Jpn. Kokai Tokkyo Koho JP 62 45,561, 1987; Chem. Abstr. 1987, 107, 176634b. Several methods exist for the preparation of the monoisopropyl fumaric acid, as an example: (b) Zecher, W.; Merten, R. Eur. Pat. Appl. EP 69926 A1, 1983. (c) Gordinskii, B. Y.; Shimanskii, V. M.; Vishnyakova, R. S. Zh. Prikl. Kim. 1967, 40, 1881. (d) Ushakov, S. N.; Nikolaev, A. F.; Toroptseva, A. M.; Trizno, M. S. Zh. Prikl. Kim. 1959, 32, 667. Scheme 9 The nitroalkene 44 was then transformed into the corresponding nitroso acetal 45 by a tandem cycloaddition with 29 in excellent yield. The reduction of the lactone to the lactol 46 was consistently achieved in greater than 90% yield with L-Selectride, with no evidence of overreduction. Moreover, subjecting the lactol 46 to the hydrogenolytic conditions with Raney nickel gratifyingly afforded the desired product 26 in 59% yield. To optimize the yield of 26, a survey of hydrogen pressure was undertaken. In this system, higher pressures afforded lower yields, though little difference was observed between 14.6 and 160 psi. In some instances, a very small amount of overreduction was detected wherein the lactol had been reduced to a diol (47). The optimum conditions employed 160 psi of hydrogen pressure for 48 h. In addition, following removal of the catalyst, the filtrate was heated at 60 C for 1 h to ensure complete ring closure to the lactam. Correction of the Configuration at C(6). The lactam 26 possessed all of the correct structural features of (-)-rosmarinecine. However, to convert the lactam to (-)-rosmarinecine, inversion of the configuration at C(1) (which becomes C(6) of (-)-rosmarinecine) was required (Scheme 9). In addition, it remained to reveal the free amino triol by reduction of the lactam and the lactol. The free hydroxyl at C(1) makes the Mitsunobu reaction the perfect choice to correct the configuration of C(1).47 In addition, the resulting ester could be deprotected under the same reducing conditions in which the amino triol would be revealed. Initial attempts to selectively invert the stereocenter at C(1) in the presence of the free lactol in 26 were foiled. The only isolated product (54%) contained one benzoate group; unfortunately, it was as a result of an inversion of the lactol, not the C(1) hydroxyl. The selective protection of the lactol as a methyl (47) Review of the Mitsunobu reaction: Hughes, D. L. Org. React. 1992, 42, 335. 8272 J. Am. Chem. Soc., Vol. 118, No. 35, 1996 Scheme 10 Denmark et al. Scheme 11 Figure 3. Hydrolysis of 50 as a function of TFA concentration. Scheme 12 Scheme 13 acetal was easily accomplished (MeOH/pTsOH/HC(OMe)3/ reflux) to afford 48 in 74-84% yield as a single anomer, as judged by 1H NMR spectrocopy (Scheme 10). If the reaction was carried out for a shorter time, or at room temperature, a mixture of anomeric methyl acetals was obtained.48 The assignment of configuration of the acetal is made on the assumption of thermodynamic control; the most stable methyl acetal should be of the R configuration due to steric interaction on the concave side of the tricyclic structure. In addition, the acetal proton appeared as a singlet, indicating a trans relationship between HC(7) and HC(7a). The minor anomer appeared at 5.14 ppm as a doublet (J ) 5.6 Hz), indicating a cis relationship between HC(7) and HC(7a). The configuration at C(1) was now easily corrected under standard Mitsunobu reaction conditions utilizing benzoic acid as the nucleophile. The product 49 was isolated in 73% yield (Scheme 11). Recent reports have demonstrated that more acidic nucleophiles, such as 4-nitrobenzoic acid, afforded higher yield in reactions with sterically hindered alcohols.49 This simple modification dramatically improved the yield in the inversion with 48, affording the inverted lactam 50 in 94% yield. The only complication in this reaction was the similar polarity of the substitution product and the byproduct triphenylphosphine oxide, requiring multiple purifications by silica gel chromatography to obtain 50 cleanly. As confirmation that the methyl acetal 48 had, indeed, undergone invertive substitution and was not simply acylated in the Mitsunobu reaction, the R-hydroxy lactam 48 was treated with 4-nitrobenzoyl chloride to obtain the 4-nitrobenzoate 51 in 79% yield (Scheme 11). Comparison of the 1H NMR spectra of the two esters confirmed that the Mitsunobu reaction had occurred with inversion. The 1H NMR spectrum of lactam 50 has a singlet at 5.40 ppm, indicating a trans relationship between and HC(7a) HC(1), while the lactam 51 displayed a doublet at 6.06 ppm (J ) 8.5 Hz), indicating a cis relationship between HC(7a) and HC(1). (48) The production of a single anomer or a high anomeric ratio was important to achieve good yield in the subsequent reaction, as the minor component could never be isolated after workup (vide infra). (49) Dodge, J. A.; Trujillo, J. I.; Presnell, M. J. Org. Chem. 1994, 59, 234. Synthesis of Rosmarinecine. The final production of amino triol 2 by deprotection of the 4-nitrobenzoate and reduction of the lactam lactol required that the methyl acetal first be cleaved. Simple hydrolysis with 1 N HCl provided the desired product in very low yield, and iodotrimethylsilane37 or boron trifluoride etherate/tetrabutyl ammonium iodide50 afforded no reaction at all. More promising results were obtained with the use of trifluoroacetic acid (TFA), which provided the desired product 51 in variable yield.51 Since the reaction appeared to be very sluggish, the TFA concentration was optimized in a series of 1H NMR experiments. The approximate half-life of the hydrolysis was determined at three different acid concentrations: 80%, 90%, and 95% TFA. All the reactions were performed at room temperature in TFA-d, diluted with D2O to the appropriate concentration. As can be seen in Figure 312, all of the reactions displayed pseudo-first-order kinetics. The 80% TFA reaction was very slow, with a t1/2 of 14.5 h. Under these reaction conditions, only a trace amount of byproducts was generated, even after a long reaction time. The use of 95% TFA increased the rate dramatically (t1/2 ) 3.1 h) and afforded a more tolerable reaction time. Unfortunately, under these strongly acidic conditions, a large number of byproducts were observed in the 1H NMR spectrum. The 90% TFA solution provided an excellent balance between rate and selectivity, with (50) Mandal, A. K.; Soni, N. R.; Ratnam, K. R. Synthesis 1985, 274. (51) Nakajima, N.; Uoto, K.; Yonemitsu, O.; Hata, T. Chem. Pharm. Bull. 1991, 39, 64. Synthesis of (-)-Rosmarinecine J. Am. Chem. Soc., Vol. 118, No. 35, 1996 8273 Figure 4. (a) 13C NMR and (b) 1H NMR spectra of i, ii, and rosmarinecine 2 after silica gel chromatography. Scheme 14 a t1/2 of 4.8 h and a byproduct profile similar to that obtained with the 80% TFA reaction. The isolation, purification, and identification of 52 proved to be challenging. The 1H NMR spectra of the lactol 52 usually contained broad peaks with little fine structure, regardless of the solvent employed. Therefore, to ensure that our assignment of the lactol structure was correct, 52 was methylated with iodomethane and silver oxide (Scheme 12). The methyl acetal 50 was isolated in 53% yield, along with 28% of recovered 52. The product isolated was the same methyl acetal 50 used in the hydrolysis by comparision of the 1H and 13C NMR spectra; thus, the assignment of the lactol structure as 52 was secured. The final transformation involved an exhaustive reduction of the lactam, ester, and latent lactol carbonyl groups. In our (-)-hastanecine synthesis, a similar reduction with lithium aluminum hydride revealed the desired amino diol. To complete the synthesis of (-)-rosmarinecine, the 4-nitrobenzoate 52 was treated with lithium aluminum hydride to afford compound i in 66-84% yield after silica gel chromatography (Scheme 13). The 1H and 13C NMR spectra of i differed significantly from 8274 J. Am. Chem. Soc., Vol. 118, No. 35, 1996 Table 6. Selected 1H NMR Data for Nitroso Acetals 45a,b Denmark et al. satisfying 64% yield with a small amount of an overreduced lactol. The chiral auxiliary (R)-2,2-diphenylcyclopentanol was recovered in 98% yield after the hydrogenation. Protection of the lactol (+)-26 as its methyl acetal (+)-49 (73%), followed by inversion at C(2) by a Mitsunobu reaction utilizing 4-nitrobenzoic acid as the nucleophile, afforded the 4-nitrobenzoate (+)-50 (94%), which was determined to be 97.3% by chiral HPLC. The deprotection of (+)-50 was accomplished with 90% trifluoroacetic acid, to afford the lactol (+)-52 in 87% yield. Reduction of lactol (+)-52 with Red-Al in THF at reflux afforded, after extensive chromatographic purification on both silica gel and basic alumina, a white solid, which was recrystallized from acetone/pentane to afford (-)-rosmarinecine (2) as an analytically pure, white solid in 66% yield. The physical properties matched those of the reported values: mp 169-170 C, [R]21D -117.6 (EtOH, c ) 0.96), lit.52 mp 170-172 C, [R]23D -119.1 (EtOH, c ) 0.94); lit.10 mp 171-172 C, [R]25D -118.5 ; lit.12 mp 171-172 C, [R]25D -116.5 (EtOH, c ) 0.01); lit.17 [R]21D -121 (EtOH, c ) 0.01). Importantly, the 1H NMR and 13C NMR data of the synthetic product are nearly identical to those of the natural compound (mp 168-170 C, [R]21D -119.8 (EtOH, c ) 1.01)) obtained by hydrolysis of rosmarinine (see supporting information). Conclusion. The synthesis of (-)-rosmarinecine was accomplished in eight steps and 14.8% overall yield from the fumaroyl acid chloride 43. All the stereocenters were installed in a single transformation with high selectivity, which demonstrates the utility of the tandem [4 + 2]/[3 + 2] cycloaddition strategy and serves to illustrates the predictive power of the various stereochemical features of the reaction sequence. Since (-)-rosmarinecine is among the more complex necine bases containing an all-cis relationship, this synthesis demonstates the generality of the tandem [4 + 2]/[3 + 2] cycloaddition chemistry for the synthesis of this class of compounds. Experimental Section General Experimental. See supporting information for details. (E)-4-Chloro-4-oxo-2-butenoic Acid 1-Methylethyl Ester (43). In a three-neck, 500 mL, round-bottom flask fitted with a Schlenk filtration tube was placed the potassium isopropoxy maleate 42 (15.0 g, 76.5 mmol) in CH2Cl2 (120 mL). To the solution was added oxalyl chloride (16.7 mL, 191 mmol, 2.49 equiv), followed by dimethylformamide (0.90 mL, 11.6 mmol, 0.15 equiv), whereupon massive evolution of gas was observed. The resulting solution was stirred at room temperature in an aluminum foil-covered flask for 22 h. Pentane (210 mL) was added, and the precipitated salts were filtered off and washed with pentane (30 mL). The filtrate was concentrated in vacuo, and the black residue was filtered again and fractionally distilled to afford 8.90 g (66%) of 43 as a clear, colorless oil: bp 44-45 C (0.8 Torr); 1H NMR (400 MHz, CDCl3) 6.97 (d, J ) 15.4 Hz, 1H), 6.92 (d, J ) 15.4 Hz, 1H), 5.12 (septet, J ) 6.3 Hz, 1H, HC(5)), 1.30 (d, J ) 6.1 Hz, 6H, HC(6)); 13C NMR (100 MHz, CDCl ) 165.42 (C(1)), 163.22 (C(4)), 138.49 3 (C(3)), 136.44 (C(2)), 69.92 (C(5)), 21.62 (C(6)); IR (neat) 2985 (m), 2940 (w), 1766 (s), 1726 (s), 1376 (m), 1352 (m), 1337 (m), 1303 (s), 1278 (s), 1262 (s), 1239 (m), 1184 (s), 1146 (m), 1104 (s), 1024 (s), 984 (s), 972 (s), 907 (m); MS (70 eV) mz (relative intensity) 176 (M+, 1), 141 (27), 135 (14), 119 (17), 118 (10), 117 (54), 99 (40), 91 (19), 89 (62), 85 (10), 83 (16), 82 (26), 59 (31), 55 (29), 54 (36), 53 (16), 43 (100), 42 (86). Anal. Calcd for C7H9ClO3 (176.59): C, 47.60; H, 5.13. Found: C, 47.45; H, 5.17. (E)-[(2-(Nitroethenyl)oxy]-4-oxo-2-butenoic Acid 1-Methylethyl Ester (44). To a slurry of potassium nitroacetaldehyde 20 (2.20 g, 17.3 mmol) in nitromethane (32 mL) at -6 C was added rapidly a solution of 43 (3.09 g, 17.3 mmol, 1.0 equiv) in CH2Cl2 (12 mL). The mixture was allowed to stir for 45 min at -6 C. During this time, the color changed to light brown, with precipitation of a white solid. Dichloromethane (250 mL) was then added, and the resulting suspension was filtered through Whatman No.1 filter paper. The filtrate was nitroso acetal (-)-45a (+)-45b HC(6) ppm (J, Hz) 4.63 (t, 7,2) 4.95 (dd, 2.2, 6.8) HC(7b) ppm (J, Hz) 4.39 (dd, 6.6, 8.6) 4.26 (dd, 6.4, 8.6) those reported for natural (-)-rosmarinecine.52 In only one instance under these reaction conditions was (-)-rosmarinecine isolated from 52 (45%). Employing Red-Al instead as the reducing agent with 52 afforded an 82% yield of yet another amino triol ii, which differed from both rosmarinecine and i. Still more surprising was the fact that LiAlH4 reduction of the benzoate lactol 53 (prepared by TFA hydrolysis of 49) produced 2 in moderate yield (69%) after silica gel chromatography. The dissimilarity of the NMR spectra of i, ii, and 2 is easily seen, as depicted in Figure 4. This unanticipated outcome caused considerable consternation. The gross similarity of the NMR spectra of i, ii, and 2 suggested that epimeric forms of the desired product were being generated.1,53 Some of these possibilities could be eliminated, and we eventually realized that the anomalies in the spectra were caused by variable concentrations of hydrates or Ncomplexes of rosmarinecine itself. This was proven by additional purification of each of the three substances i, ii, and 2 on basic alumina to afford the same compound, which in each instance possessed the same 1H and 13C NMR spectra as those reported for natural (-)-rosmarinecine.52 Total Synthesis of (-)-Rosmarinecine: Summary. The synthesis of enantiomerically pure (-)-rosmarinecine employing the optically active vinyl ether 29 is detailed in Scheme 14, where the yields refer to homogenous, analytically pure material. The requisite nitroalkene 44 was subjected to the tandem [4 + 2]/[3 + 2] cycloaddition sequence, utilizing MAPh as the Lewis acid promoter and (-)-29. Reaction of the nitroalkene with 3 equiv of the chiral vinyl ether in the presence of 3 equiv of MAPh afforded the nitroso acetal 45 in 94% yield. The nitroso acetal 45 was isolated as a 25:1 exo/endo mixture of diastereomers that could not be enriched by recrystallization. The diastereomers 45a,b were, therefore, separated by silica gel chromatography. The chromatographic fractions containing the desired product (-)-45a were concentrated in two separate portions. The composition of the major portion (74%) was determined by 1H NMR spectroscopy to be approximatly 95:1 exo/endo mixture of diastereomers, and the minor portion consisted of a 6.3:1 exo/endo mixture. The stereochemical assignment of the diastereomers was based, 1H NMR: the anomeric proton in the exo diastereomer (-)-45a appears as a triplet, while the same proton in the endo diastereomer (+)45b appears as a doublet of doublets (Table 6). The lactone was then reduced with L-Selectride to afford the lactol (-)-46 in excellent yield (91%). The lactol (-)-46 was then submitted to the standard hydrogenation conditions to afford the tricyclic R-hydroxy lactam-lactol (+)-26 in a (52) Denholm, A. A. Ph.D. Thesis, University of Glasgow, 1990. We gratefully acknowledge professor David J. Robins for the copies of this experimental procedure. (53) We are not aware of a report in the open literature discussing this phenomena, but considering the enormous numbers of papers in this field, a report could have been unintentionally missed. Synthesis of (-)-Rosmarinecine concentrated by rotary evaporation, and the remaining nitromethane was removed under high vacuum with a water bath at room temperature to give 3.77 g of the crude nitroalkene. The crude product was recrystallized from hexane/Et2O 10:1 (118 mL) with hot gravity filtration to afford 2.80 g (69%) of 44 as yellow, crystalline solid: mp 67-68 C (hexane/Et2O); 1H NMR (400 MHz, CDCl3) 8.90 (d, J ) 11.0 Hz, 1H, HC(2)), 7.37 (d, J ) 11.2 Hz, 1H, HC(1)), 7.07 (d, J ) 15.6 Hz, 1H, HC(5)), 6.92 (d, J ) 15.9 Hz, 1H, HC(6)), 5.14 (septet, J ) 6.3 Hz, 1H, HC(9)), 1.32 (d, J ) 6.4 Hz, 6H, HC(10)); 13C NMR (100 MHz, CDCl3) 163.23 (C(7)), 160.07 (C(4)), 147.40 (C(2)), 138.87 (C(5)), 130.30 (C(1)), 129.16 (C(6)), 69.80 (C(9)), 21.59 (C(10)); IR (KBr) 3131 (m), 3081 (m), 2981 (m), 1744 (s), 1711 (s), 1671 (m), 1648 (w), 1532 (m), 1373 (m), 1362 (m), 1340 (w), 1315 (m), 1283 (w), 1189 (m), 1147 (w), 1109 (w), 1007 (w), 945 (w); MS(CI, CH4) m/z (relative intensity) 230 (M+ + H, 2), 159 (12), 142 (9), 141 (100), 117 (6), 99 (9). Anal. Calcd for C9H11NO6 (229.18): C, 47.16; H, 4.83; N, 6.11. Found: C, 47.17; H, 4.84; N, 6.12. (2R,2aS,4aR,6S,7bR)- and (2S,2aR4aS,6S,7bS)-6-[(1R)-(2-Diphenylcyclopentyl)oxy]-3-oxooctahydro-1,4,7-trioxa-7a-azabicyclopent[cd]indene-2-carboxylic Acid 1-Methylethyl Ester (45). To a solution of 2,6-diphenylphenol (8.202 g, 33.30 mmol, 6.0 equiv) in CH2Cl2 (61 mL) was added trimethylaluminum (2.0 M in toluene, 8.3 mL, 17 mmol, 3.0 equiv). Gas evolution was observed, and the resulting light yellow solution was stirred at room temperature for 40 min. To a solution of 44 (1.272 g, 5.550 mmol) in CH2Cl2 (22 mL) at -75 C was added a solution of (R)-2,2-diphenylcyclopentoxyethene ((-)-29)19e (4.402 g, 16.65 mmol, 3.0 equiv) in CH2Cl2 (13 mL). MAPh was added slowly to the resulting mixture over 36 min, keeping the internal temperature below -70 C. During the addition, the solution changed color from light yellow to deep, dark brown. The brown solution was stirred for additional 1.5 h and then was quenched with MeOH (15.6 mL), poured into CH2Cl2 (1 L), and washed with water (2 0.6 L). The aqueous phases were back-extracted with CH2Cl2 (2 0.5 L), and the combined organic extracts were dried (Na2SO4), filtered through Celite, and concentrated in vacuo. The crude product was purified by silica gel (330 g) column chromatography, eluting with hexane/EtOAc (1:0 (0.9 L), 30:1 (0.9 L), 8:1 (0.9 L), 4:1 (1.8 L), 2:1 (0.9 L), 1:1 (0.9 L), 0:1(0.9 L)). The first fractions contained 2,6diphenylphenol, which was recrystallized from hexane (300 mL) to afford 7.00 g (85%) of recovered phenol. The chromatographic fractions containing the desired product were concentrated in two separate portions. The major portion afforded 2.051 g (74.7%) of a white foam, the composition of which was determined by 1H NMR to be a 95:1 exo/endo mixture of diastereomers ((-)-45a/(+)-45b). The minor portion afforded 0.574 g (20.9%) as a white foam, which consisted of a 6.3:1 exo/endo mixture. An analytically pure sample of the endo diastereomer (+)-45b was obtained by radial chromatography (1 mm silica plate), eluting with hexane/EtOAc (4:1, 2:1, 1:1). For (-)-45a: mp 81-85 C; 1H NMR (400 MHz, CDCl3) 7.257.09 (m, 10H), 5.25 (d, J ) 3.6 Hz, 1H, HC(2)), 5.09 (septet, J ) 6.3 Hz, 1H, HC(9)), 4.79 (ddd, J ) 2.2, 3.6, 6.4 Hz, 1H, HC(4a)), 4.72 (d, J ) 4.1 Hz, 1H, HC(1)), 4.63 (t, J ) 7.2 Hz, 1H, HC(6)), 4.39 (dd, J ) 6.6, 8.6 Hz, 1H, HC(7b)), 3.90 (dd, J ) 3.4, 8.3 Hz, 1H, HC(2a)), 2.64-2.56 (m, 1H, HC(3)), 2.30-2.26 (m, 2H, HC(5 and 3)), 2.142.06 (m, 1H, HC(5)), 1.89-1.85 (m, 2H, HC(5 and 4)), 1.54 (ddd, J ) 2.0, 7.2, 13.5 Hz, 1H, HC(5)), 1.45-1.38 (m, 1H, HC(4)), 1.30 (d, J ) 6.1 Hz, 3H, HC(10)), 1.29 (d, J ) 6.4 Hz, 3H, HC(10)); 13C NMR (100 MHz, CDCl3) 173.81 (C(3)), 166.97 (C(8)), 146.13 (C), 145.25 (C), 128.31 (CH), 128.16 (CH), 127.65 (CH), 126.82 (CH), 125.92 (CH), 125.59 (CH), 99.55 (C(6)), 86.05 (C(1)), 84.84 (C(2)), 74.04 (C(7b)), 73.98 (C(4a)), 70.60 (C(9)), 60.09 (C(2)), 49.13 (C(2a)), 35.01 (C(3)), 31.72 (C(5)), 28.30 (C(5)), 21.59 (C(10)), 21.55 (C(10)), 20.44 (C(4)); IR (KBr) 3056 (w), 3023 (w), 2979 (m), 2876 (w), 1783 (s), 1740 (s), 1494 (m), 1447 (m), 1385 (m), 1375 (m), 1356 (m), 1287 (s), 1237 (s), 1205 (m), 1176 (s), 1147 (m), 1106 (s), 1067 (s), 1049 (s), 1032 (m), 1012 (m), 932 (m), 914 (s); MS (FAB) m/z (relative intensity) 532 (M+ + K, 2), 494 (M+ + H, 4), 309 (15), 274 (17), 223 (100), 221 (100), 155 (60), 154 (17), 153 (21), 152 (10), 149 (10), 143 (12), 137 (11), 135 (48), 120 (16), 118 (99), 117 (37), 102 (49); [R]21D -80.1 (CHCl3, c ) 1.00); TLC Rf ) 0.45 (hexane/EtOAc 1:1). Anal. J. Am. Chem. Soc., Vol. 118, No. 35, 1996 8275 Calcd for C28H31NO7 (493.55): C, 68.13; H, 6.33; N, 2.83. Found: C, 68.28; H, 6.42; N, 3.01. For (+)-45b: mp 80-84 C (hexane/EtOAc); 1H NMR (400 MHz, CDCl3) 7.27-7.09 (m, 10H), 5.28 (d, J ) 3.9 Hz, 1H, HC(2)), 5.10 (septet, J ) 6.3 Hz, 1H, HC(9)), 4.95 (dd, J ) 2.2, 6.8 Hz, 1H, HC(6)), 4.79 (ddd, J ) 3.0, 6.0, 9.3 Hz, 1H, HC(4a)), 4.72 (d, J ) 5.4 Hz, 1H, HC(1)), 4.26 (dd, J ) 6.4, 8.6 Hz, 1H, HC(7b)), 3.93 (dd, J ) 3.7, 8.6 Hz, 1H, HC(2a)), 2.63-2.55 (m, 1H, HC(3)), 2.26-2.13 (m, 2H, HC(3 and 5)), 1.94-1.86 (m, 1H, HC(5)), 1.84-1.72 (m, 2H, HC(5 and 4)), 1.50 (ddd, J ) 3.1, 6.9, 16.0 Hz, 1H, HC(5)), 1.371.25 (m, 1H, HC(4)), 1.31 (d, J ) 6.1, 3H, HC(10)), 1.30 (d, J ) 6.1 Hz, 3H, HC(10)); 13C NMR (100 MHz, CDCl3) 173.72 (C(3)), 166.88 (C(8)), 146.04 (C), 145.25 (C), 128.42 (CH), 128.12 (CH), 127.61 (CH), 126.88 (CH), 125.90 (CH), 125.54 (CH), 101.07 (C(6)), 87.40 (C(1)), 84.48 (C(2)), 74.54 (C(4a)), 72.56 (C(7b)), 70.54 (C(9)), 60.25 (C(2)), 49.46 (C(2a)), 34.78 (C(3)), 32.33 (C(5)), 29.85 (C(5)), 21.60 (C(10)), 21.56 (C(10)), 20.39 (C(4)); IR (KBr) 3056 (w), 3023 (w), 2979 (m), 2878 (w), 1781 (s), 1741 (s), 1494 (w), 1447 (w), 1386 (w), 1375 (m), 1359 (m), 1346 (w), 1297 (s), 1288 (m), 1238 (s), 1209 (m), 1180 (s), 1154 (s), 1105 (s), 1082 (m), 1072 (m), 1046 (m), 1012 (m), 919 (m); MS (FAB) m/z (relative intensity) 494 (M+ + H, 4) 307 (31), 289 (14), 258 (10), 237 (12), 221 (43), 155 (27), 154 (100), 139 (13), 138 (29), 137 (55), 136 (62), 123 (10), 119 (10), 116 (20), 106 (17); [R]21D +12.9 (CHCl3, c ) 0.54); TLC Rf ) 0.45 (hexane/EtOAc 1:1). Anal. Calcd for C28H31NO7 (493.55): C, 68.13; H, 6.33; N, 2.83. Found: C, 68.12; H, 6.32; N, 2.99. (2R,2aS,4aR,6S,7bR)-6-[(1R)-(2-Diphenylcyclopentyl)oxy]-3-hydroxyoctahydro-1,4,7-trioxa-7a-azabicyclopent[cd]indene-2-carboxylic Acid 1-Methylethyl Ester ((-)-(46)). To a solution of (-)-45a (1.965 g, 3.981 mmol) in THF (71 mL) at -74 C was added dropwise lithium tri(sec-butyl)borohydride (0.95 M in THF, 4.20 mL, 3.98 mmol, 1.0 equiv) over 5 min. The resulting solution was stirred for 1 h and then quenched with a solution of aqueous phosphate buffer (pH 6.9)/ glycerol (1:1, 70 mL). The mixture was immediately poured into CH2Cl2 (0.5 L) and then washed with water (150 mL) and brine (2 150 mL) and back-extracted with CH2Cl2 (2 150 mL). The combined organic extracts were dried (Na2SO4), filtered, and concentrated in vacuo. The crude product was purified by silica gel column chromatography, eluting with hexane/EtOAc (8:1, 4:1, 2:1, 1:1, 0:1) to afford 1.810 g (91.8%) of (-)-46 as a white foam: mp 82-88 C; 1H NMR (400 MHz, CDCl3) 7.25-7.09 (m, 10H), 5.49 (s, 0.88H, HC(3R)), 5.41 (dd, J ) 6.8, 12.7 Hz, 0.12H, HC(3)), 5.21 (d, J ) 3.4 Hz, 0.12H, HC(2)), 5.11 (septet, J ) 6.3 Hz, 1H, HC(9)), 4.96 (d, J ) 6.6 Hz, 0.88H, HC(2R)), 4.74-4.73 (m, 1.12H, HC(1 and 6)), 4.63 (t, J ) 7.1 Hz, 0.88H, HC(6R)), 4.52-4.49 (m, 0.88H, HC(4aR)), 4.46 (d, J ) 13.0 Hz, 0.12H, OH()), 4.33 (dd, J ) 6.0, 8.9 Hz, 1.12H, HC(7b and 4a)), 3.56 (ddd, J ) 3.4, 6.9, 8.5 Hz, 0.12H, HC(2a)), 3.28 (ddd, J ) 1.5, 5.3, 9.0 Hz, 0.88H, HC(2aR)), 3.04 (d, J ) 2.6 Hz, 0.88H, OH(R)), 2.64-2.56 (m, 1H, HC(3)), 2.30-2.20 (m, 2H, HC(3 and 5)), 2.16-2.10 (m, 1H, HC(5)), 1.85-1.80 (m, 2H, HC(4 and 5)), 1.69 (ddd, J ) 3.4, 7.4, 15.3 Hz, 1H, HC(5)), 1.46-1.36 (m, 1H, HC(4)), 1.28 (d, J ) 6.2 Hz, 5.3H, HC(10R)), 1.27 (d, J ) 6.3 Hz, 0.7H, HC(10)); 13C NMR (100 MHz, CDCl3) 164.46 (C(8)), 168.19 (C(8R)), 146.43 (C(R)), 146.22 (C()), 145.39 (C(R)), 145.22 (C()), 128.42 (CH(R)), 128.28 (CH()), 128.15 (CH()), 128.07 (CH(R)), 127.63 (CH()), 127.52 (CH(R)), 126.91 (CH(R)), 126.80 (CH()), 125.89 (CH()), 125.80 (CH(R)), 125.57 (CH()), 125.48 (CH(R)), 102.00 (C(3R)), 100.53 (C(6)), 99.38 (C(6R)), 98.34 (C(3)), 86.20 (C(1)), 85.51 (C(1R)), 84.24 (C(2R)), 83.85 (C(2)), 78.22 (C(7b)), 77.53 (C(7bR)), 73.16 (C(4aR)), 73.03 (C(4a)), 69.97 (C(9)), 69.80 (C(9R)), 60.05 (C(2)), 58.57 (C(2aR)), 53.63 (C(2a)), 35.10 (C(3R)), 35.05 (C(3)), 31.74 (C(5R)), 31.67 (C(5)), 29.48 (C(5)), 28.05 (C(5R)), 21.60 (C(10)), 20.41 (C(4)); IR (KBr) 3441 (w, br), 3022 (w), 2978 (m), 2946 (m), 2874 (m), 1736 (s), 1494 (m), 1446 (m), 1384 (m), 1376 (m), 1322 (m), 1286 (m), 1267 (m), 1241 (m), 1218 (m), 1198 (m), 1147 (m), 1106 (s), 1073 (s), 1064 (s), 1036 (s), 1019 (m), 974 (m), 949 (m), 926 (m); MS (FAB) m/z (relative intensity) 496 (M+ + H, 10), 309 (13), 276 (11), 222 (19), 221 (100), 186 (11), 155 (57), 154 (14), 153 (18), 152 (27), 144 (14), 143 (15), 137 (10), 134 (43), 120 (15), 118 (88), 117 (45), 102 (42); [R]21D -42.1 (CHCl3, 8276 J. Am. Chem. Soc., Vol. 118, No. 35, 1996 c ) 0.90); TLC Rf ) 0.40 (hexane/EtOAc 1:1). Anal. Calcd for C28H33NO7 (495.57): C, 67.86; H, 6.71; N, 2.82. Found: C, 67.80; H, 6.86; N, 3.18. (1R,5aS,7S,7aR,7bR)-1,7-Dihydroxy-6-oxaoctahydro-2H-cyclopenta[gh]pyrrolizin-2-one ((+)-(26)) and (1R,6S,6aR,7S)-1,6-Dihydroxy7-(hydroxymethyl)hexahydro-1H-pyrrolizin-2-one (47). To a solution of (-)-46 (1.783 g, 3.598 mmol) in methanol (100 mL) was added a catalytic amount of methanol-washed (6 80 mL) W-2 Raney nickel. The suspension was stirred at room temperature in a 300 mL flask inside a steel autoclave for 48 h under 160 psi atmosphere of H2. The catalyst was filtered off through Celite and washed with methanol (200 mL), and the solution was stirred at 65 C for 1 h and then concentrated in vacuo. The residue was separated by silica gel column chromatography, eluting with CHCl3/MeOH (1:0, 99:1, 19:1, 9:1, 6:1, 4:1, 2:1) into three fractions. The first fraction contained a mixture of (+)-26 and (R)-2,2-diphenylcyclopentanol. The second portion contained 0.406 g of pure (+)-26, and the third fraction contained 0.040 g of a mixture of (+)-26 and 47. The first fraction was repurified by silica gel column chromatography, eluting wit...