INFORMATION TO USERS
This reproduction was made from a copy of a document sent to us for microfilming. While the most advanced technology has been used to photograph and reproduce this document, the quality of the reproduction is heavily dependent upon the quality of the material submitted.
The following explanation of techniques is provided to help clarify markings or notations which may appear on this reproduction.
1. The sign or “target” for pages apparently lacking from the document photographed is “Missing Page(s)”. If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure complete continuity.
2. When an image on the film is obliterated with a round black mark, it is an indication of either blurred copy because of movement during exposure, duplicate copy, or copyrighted materials that should not have been filmed. For blurred pages, a good image of the page can be found in the adjacent frame. If copyrighted materials were deleted, a target note will appear listing the pages in the adjacent frame.
3. When a map, drawing or chart, etc., is part of the material being photographed, a definite method of “sectioning” the material has been followed. It is customary to begin filming at the upper left hand corner of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again—beginning below the first row and continuing on until .complete.
4. For illustrations that cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into your xerographic copy. These prints are available upon request from the Dissertations Customer Services Department.
5. Some pages in any document may have indistinct print. In all cases the best available copy has been filmed.
University Microfilms international 300 N. Zeeb Road Ann Arbor, Ml 48106 8526189
Hong, Won-Pyo
TOTAL SYNTHESIS OF ( + /-)-LYTHRANCEPINE-ll
The Ohio State University Ph.D. 1985
University Microfilms International 300 N. Z eeb Road, Ann Arbor, Ml 48106 TOTAL SYNTHESIS OF <*>-LYTHRANCEPINE-II
DISSERTATION
Presented in Partial Fulfillment of the
Requirements for the Degree Doctor of Philosophy
in the Graduate School of
The Ohio State University
By
Won-Pyo Hong, B.S., M.S
The Ohio State University
1985
Reading Committee: Approved by:
Dr. David J. Hart
Dr. Gideon Fraenkel
Dr. John S. Swenton Advisor
Department of Chemistry To My Parents
Soon-Kyung Hong and Hee-Kyung Kim Hong ACKNOWLEDGEMENTS
I wish to express my gratitude to Professor David J.
Hart for his invaluable advice, support, friendship, and contagious enthusiasm for chemistry throughout this endeavor. His role as an advisor and his interest in my career goals are deeply appreciated. I also wish to express my gratitude to Professor Eun Lee in Seoul
National University for his initial enthusiasm for organic chemistry. I should say the works in this thesis could not have been achieved without numerous help from others. I would like to thank all the members of the Hart group, past and present, for their numerous discussions and encouragement, especially Drs. Y.-M. Tsai, J.-K.
Choi, and Mr. D.-C. Ha. I am also grateful to Mr. R.
Weisenberger for mass spectra, Drs. C. Cottrell and B.
Chenera for 500 MHz *H- and l^C-NMR spectra, Mr. C.
Engelman for l^C-NMR spectra, Mr. D. Burnett for 200 MHz
3-H-NMR spectra, Mr. J. R. Wermer for FT-IR spectra, and
Dr. J.-J. Lee for helpful discussions of COSY spectroscopy.
iii VITA
July 11, 1953 ...... Born, Pusan, Korea
1977...... B.S. (Chemistry), Seoul National University, Seoul, Korea
1979...... M.S. (Organic Chemistry), Seoul National University, Seoul, Korea
19S0 - 1982 ...... Teaching Associate, Department of Chemistry The Ohio State University, Columbus, Ohio
1982 - present ...... Research Associate, Department of Chemistry The Ohio State University, Columbus, Ohio
PUBLICATIONS
1. Lee, E.; Hong, W.-P. "Studies on Biogenetic-type Synetheis of Natural Products (I). Synthesis and Reactions of Methyl 3-Hydroxymethylorsellinate" K9E§§Q Chenu Soc^, 1979, 23, 30.
2. Lee, E.; Hong, W.-P.; Ko, S. Y. "Fluoride-assisted Acetylation of Alcohols and Phenols" Bui1^ Korean Chem^ Soc^, 1980, 1, 144.
3. Hart, D. J.; Hong, W.-P. "Lythraceae Alkaloids: Total Synthesis of (+)-Lythrancepine-II" Qrg*. C h e m 1985, 50, 0000.
iv CONTRIBUTED PAPER
1. Hart, D. J.; Hong, W.-P. "Total Synthesis of <+)- Lythrancepine-II" The 17th Central Regional Meeting of the American Chemical Society, Akron, 1985.
FIELD OF STUDY
Major Field: Organic Chemistry
v TABLE OF CONTENTS
ACKNOWLEDGEMENTS ...... iii
VITA ...... iv
TABLE ...... viii
LIST OF FIGURES ...... ix
LIST OF SCHEMES ...... x
I. ISOLATION AND STRUCTURE DETERMINATION OF LYTHRACEAE ALKALOIDS ...... 1
A. Introduction ...... 1
B. Classification and Isolation of Lythraceae Alkaloids ...... 3
C. Type D Lythraceae Alkaloids: Structure Determination of the Lythrancine and Lythrancepine Groups ...... 5
II. MODEL STUDIES FOR LYTHRANCEPINE ALKALOID SYNTHESIS ...... 10
A. Stereochemistry at C-l, C-3, AND C-5 (N-Acyliminium Ion Cyclizations) ...... 12
B. Introduction of the C-9 Sidechain (Eschenmoser Sulfide Contraction) ...... 25
III. AN APPROACH TO THE SYNTHESIS OF LYTHRANCEPINE ALKALOID VIA BIARYL INTERMEDIATES ...... 38
IV. TOTAL SYNTHESES OF <+)-LYTHRANCEPINE-II AND (■♦•) - LYTHRANCEPINE-III ...... 50
vi V. EXPERIMENTAL
BIBLIOGRAPHY LIST OF TABLE
Table
1. Reduction of ketone 55
viii LIST OF FIGURES
3
6
21
26
38
37
44
79
61
90
91
92
93
94
ix LIST OF SCHEMES
Schemes page
I ...... 8
II...... 11
III...... 13
IV...... 13
V ...... 14
VI ...... 15
VII...... 17
VIII...... 18
IX...... 19
X ...... 20
XI ...... 21
XII...... 23
XIII...... 25
XIV...... 27
XV ...... 28
XVI...... 30
XVII...... 31
XVIII...... 33
x 34
36
38
40
42
43
45
46
47
48
50
53
54
56
57
59
62
63
65
68
71
72
74 xi X XX
88 IIAXXXX
98 ...... IAXXXX
S8 ...... AXXXX
£8 ...... AIXXXX
8 ...... IZXXXX CHAPTER I: ISOLATION AND STRUCTURE DETERMINATION OF LYTHRACEAE ALKALOIDS A. Introduction The objective of this research was the total synthesis of lythrancepine-II (4), a structurally interesting quinolizidine metacyclophane Lythraceae alkaloid which has yet to undergo pharmacological evaluation. To provide the reader with some perspective, a brief introduction describing the classification and isolation of Lythraceae alkaloids will be presented here. An overview of the structure determination of quinolizidine metacyclophane Lythraceae alkaloids will also be presented. Brief references to the Lythraceae alkaloids have appeared in Volumes X, XII, and XIV of "The Alkaloids"1 and a short review on the alkaloids from Lythrum ancega was published in Japan.2 Two major reviews on the Lythraceae alkaloids have appeared, one covering mainly structure elucidation^ and the other covering a considerable amount of synthesis as well as structure elucidation.4 1 2 Plants of the Lythraceous family are moderately well distributed in different regions of the world, from the tropics to the temperate zones, and are especially abundant in Latin America. The family consists of 22 genera composed of about 500 species, including several economically important species of the genera Heimia, ^2 9 d£ordia, Lafoensia, Lythrum, Cuphea, Ammannia, Cuphea, and Lawsonia. One reason for the early investigation of the Lythraceous plants was the pronounced but mild psychoaomimetic effect displayed by the Heimia salicifolia. This plant has been used as a medicine against bronchitis, dysentery, indigestion, and syphilis, and is used by women after delivery to close the womb and to cure inflammation of the womb. Lythrum salicaria L. has been used in case of diarrhoea and dysentery as a Chinese drug, 'qian qu cai'. At present forty-three alkaloids have been identified in Lythraceous plants. These alkaloids were detected only in the aerial parts of the plants. The structures of most of these bases have been established using chemical and spectroscopic data and/or X-ray analysis. 3 >OH HO, OMe OMe Type A: Decinine Type B: Vertaline Type C: Lythranidine H HO OAc Me OH Me< MeO OH OMe 4 5 Type D: Lythrancepine-II Type E: Abresoline Figure 1 B. Classification and Isolation of Lythraceae Alkaloids Lythraceae alkaloids have been classified according to five structural types (type A: lythrine and vertine) and Heimia salicifglia (type A: lythrine) by Douglas and his coworkers.9 in 1967, Fujita and his coworkers isolated three piperidine metacyclophane alkaloids from Lythrum ancegs Makino (type C: lythranine, lythranidine, and lythramine).10 The fourth structural variant of the Lythraceae alkaloids, quinolizidine metacyclophanes, were isolated by Fujita and his coworkera^ from Lythrum anceps in 1971 (type D: lythrancines I-VII and lythrancepines I-III) and by Ferris*2 from Lythrum lanceolatum in 1973 (type D: lythrumine and monoacetyllythrumine). Finally, Rother and SchwartinglS isolated several simple phenylquinol- izidine alkaloids (type E: demethyllasubine-I and demethyllasubine-II) from young seedlings of Heimia salicifglia plants in 1974. This report was followed by the isolation of other simple phenylquinolizidine alkaloids from Heimia salicifolia (type E: 5 demethoxyabresoline and 10-epidemethoxyabresoline) by Rother and Schwarting14 and fro* aubcoatato (type E: laaubine-I and II, subcosine-I and II) by Fujita and his coworkera.6 Theae alkaloids (type E) were absent in extracts of plants obtained at later atagea of growth, suggesting this type of alkaloid is an intermediate in the biosynthesis of macrocylic Lythraceae alkaloids. Several of the type A, B, C and E alkaloids have been synthesized. The synthesis of the type D alkaloids, however, has never been achieved.15 This thesis will describe the first total synthesis of the type D alkaloids, lythrancepine-II (4) and lythracepine-III (14). Therefore, the structure determination of the lythrancine and lythrancepine groups of Lythraceae alkaloids will be reviewed in following section. C. Type D Lythraceae Alkaloids: Structure Determination of the Lythrancine and Lythrancepine Groups Ten alkaloids containing a cis-quinolizidine ring system and a biphenyl group, i.e. lythrancine-I (6), -II (7), -III (8), -IV (9), -V (10), -VI (11), -VII (12), and lythrancepine-I (13), -II (4), and -III (14), belong to the type D family of Lythraceae alkaloids. All of these 6 6 Rl=r2=r3=H 10 Rl=R2=Ac 13 Rl=R2=H 7 r1=r2=H , r 3=Ac 11 R*=Ac, R2=H 4 Rl=H, r 2=Ac S Rl=R3=Ac, R2=H 12 Rl=H, R2=Ac 14 R1=R2=Ac 9 r !=r 2=r 3=a c Figure 2 alkaloids were isolated from Lythrum anceps. These compounds are shown in Figure 2.18 All of these type D Lythraceae alkaloids have methoxyl groups at C-17 and C- 21 and share a common skeleton. Lythrancine-type alkaloids (6-12) have hydroxyl or acetoxyl groups at C-3, C-4, and C-ll. Lythrancines 7-9 and lythrancines 10-12 are epimeric at C-3. Lythrancepines 13, 4, and 14 are C-4 deoxy derivatives of lythrancines 6, 7, and 8 (9), respectively. Because the structural relationship between the lythrancine and lythrancepine groups is so close, their structure determinations are interrelated. 7 The structures of these alkaloids were initially established using a combination of spectral techniques and chemical correlations.The first confirmation that these assignments were correct came in 1974 when Sim and his coworkers reported X-ray structure of lythrancine-IV (9) and the 3-0-p-bromobenzenesulfonate of lythrancine-II <7>.20 Rather than presenting the complex arguments originally used to establish the structures of lythrancepines I-III, this X-ray data will be used as the starting point to convince the reader of the correctness of structures 13, 4, and 14. First, acetylation of lythrancine-II (7) with acetic anhydride in pyridine at room temperature (2 h) furnished lythrancine-III (6), whereas acetylation at 110°C for 3 h gave lythrancine-IV (9) as shown in Scheme I. Hydrolysis of lythrancine-II (7) with 1 X methanolic potassium hydroxide <25°C, 2 h) yielded lythrancine-I (6). As expected, lythrancine-I (6) gave lythrancine-III (8) on acetylation with acetic anhydride in pyridine at room temperature. The structural relationship between the lythrancine and lythrancepine groups was established by lithium aluminum hydride reduction of the CC4)-tosylate of lythrancine-III (8) and acetylation of the product to give lythrancepine-III (14) and its C-3 epimer. Hydrolysis of lythrancepine-III (14) with 1 X methanolic potassium hydroxide (25°C) gave the 8 lythrancepine-I (13) and Identical treatment of lythrancepine-II (4) also gave 13. Acetylation of lythrancepine-II <4> with acetic anhydride in pyridine (25^0, 5 h) gave lythrancepine-III (14) and identical treatment of lythrancepine-I (13) also gave 14. Thus, from the relationships established by these acetylation and hydrolysis studies, the structures and absolute stereochemistry of 6-9, 13, 4, and 14 are on firm ground. Scheme I Lythrancine-I KOH Lythrancine-II 6 7 Ac20, A c 20, Py, r1 Py, 110°C Lythrancine-III Lythrancine-IV 8 9 l.TsCl, Py 2 .LAH 3.Ac20, Py, rt Ac 20, Py, rt Lythrancepine-III Lythrancepine-II 14 4 Lythrancepine-I 13 The structures and stereochemistry of lythrancine-V, -VI, and -VII were elucidated by analysis of NMR and mass spectra and by comparison with those of lythrancine-IV <9>.19 Although all structures of type D Lythraceae alkaloids were established as described above, no X-ray analysis of a lythrancepine group alkaloid has been carried out. Thus, the following syntheses of 4 and 13 provide additional proof of the structures of the lythrancepine group of Lythraceae alkaloids. CHAPTER II: MODEL STUDIES FOR LYTHRANCEPINE ALKALOID SYNTHESIS Several total syntheses of lactonic Lythraceae alkaloids (type A and B) have been reported. Most of these syntheses use Mannlch reactions of pelletlerine (15) with substituted benzaldehydes,21 biaryls,22 or diaryl ether&23 to assemble the quinolizidine moiety. For example, Hanaoka and his coworkers21 condensed pelletlerine (15) with G-bromoveratraldehyde (16) and obtained a 5:2 ratio of quinolizidine derivatives 17 and 18, respectively (Equation 1). None of the '‘pelletlerine" syntheses of type A and B Lythraceae alkaloids are highly stereoselective. H H NaOH (eq.1) OMe OMe 15 16 17 18 As mentioned before, no syntheses of the 10 11 quinolizidine metacyclophane alkaloids (type D) have been reported. Recently, Hart and Kanai developed an efficient and highly stereoselective synthesis of the lactonic alkaloid vertaline (2).24 Their approach to this type B Lythraceae alkaloids differs from all previous approaches most notably by passing through C-9 functionalized intermediates. Thus, it was hoped that this route could be adapted to the synthesis of type D Lythraceae alkaloids. The general plan for adapting this chemistry to type D Lythraceae alkaloid synthesis is outlined antithetically in Scheme II. Methods for the preparation Scheme II H R^O 20 19 Ar 23 22 21 12 of compounds such aa 21 from 22 and 23 had been developed during the previoualy mentioned ayntheeia of vertaline <2). Thus, the four major problema which had to be addreaaed were: 1) Adjustment of stereochemistry at C-3 (21— *>20), 2) Attachment of the C-9 aidechain with control of stereochemistry. 3) Establiahing the stereochemistry at C-ll, and 4) Closure of the macrocylic ring. Before attempting a total synthesis of the lythrancepine alkaloids, some model atudiee addressing the firat two problema were performed. This chapter will focua on these studies, but first, an overview of the N-acyliminium ion chemistry leading to the synthesis of compounds of type 21 will be presented. A. Stereochemistry at C-l, C-3, and C-5 Ion Cyclizations) Although acyclic N-acyliminium ions have been used in synthesis for several decades, the chemistry of cyclic N-acyliminium ions received little attention until Speckamp and his coworkers began their studies.25 jn 1971, Speckamp developed a facile entry to cyclic N- acyliminium ion precursors 26 via pH controlled sodium 13 borohydride reduction of cyclic imides 24 in ethanol (Scheme III).26 He proceeded by studying a series of Scheme III T* NaBH^ T EtOH X N ^ O HCl OX ^ N ^ O H HC1 I I R R 24 25 26 intramolecular cyclization reactions of N-acyliminium ions with built-in olefins.27 a relavant example is shown in Scheme IV. Treatment of ethoxy lactam 27 with formic Scheme IV HCOOH 27 28 HCO- H" 30 29 acid gave 29 (90 *) and 30 (10 X> in nearly quantitative 14 yield with high stereoselectivity. A reasonable explanation of the stereochemical course of this reaction involves initial ionization of 27 to afford cyclic N- acyliminium ion 28 followed by anti-periplanar addition of an electrophile-nucleophile pair across the olefin via a transition state in which the 2-aza-l,5-hexadiene unit resembles a chair-like conformation. In large part due to the pioneering work of Speckamp, this N-acyliminium ion initiated olefin cyclization has been documented as a potent tool in alkaloid synthesis.28 Although Speckamp delineated several stereochemical features of these reactions, the effect of asymmetric centers on their stereochemical course was not explored until the current decade. In 1980, Hart found that asymmetric centers alpha to the nitrogen can excert profound influence over the stereochemistry of N- acyliminium ion cyclization.29 t 0 evaluate the effect of this chiral center on the stereochemical course of N- SchemeV OH H HOCO OH a,b 31 32 33 a. glutarimide, DEAD, Ph3P b. DIBAL-H c. HC00H 15 acyliminium ion cyclizationa In a conformationally nonblaaed aystem, carbinolamide 32 was prepared as outlined in Scheme V. Treatment of 32 with formic acid gave cyclization product 33 (64 X) and only small amounts <2-5 X) of substances stereoiaomeric to 33 at C-3. A possible explanation of this observation is shown in Scheme VI. The N-acyliminium ion derived from 32 cyclizes via a chair-like conformation in which the incipient C-l substituent occupies an axial site, in contrast to the equatorial orientation of substituents usually observed in olefin cyclizations and other reactions whose transition-state geometries resemble chair cyclohexane.30 This unusual observation can be attributed to the strong unfavorable a <1'3) strain between the amide carbonyl and the phenyl group in the transition-state leading to the C-l isomer 36.31 Scheme VI 34 HOCO 36 0 33 16 In summary, two factors seem to be responsible for stereoselectivity in this N-acyliminium ion cyclization. First, the cyclization of carbinolamide 32 most likely proceeds through an N-acyliminium ion which can adopt chair conformations like 34 and 35. Second, the development of a severe A <3-'3) interaction in 34 may force the C-l substituent to occupy an axial position as shown in conformation 35. It is emphasized that these two factors are believed to be responsible for the stereochemical course of all N-acyliminium cyclizations performed during the lythrancepine alkaloid syntheses described herein. Before presenting the lythrancepine model studies, one more piece of background information will be presented. During the course of executing the aforementioned total synthesis of the vertaline (2), a need arose to convert benzaldehyde 16 to homoallylic amine 37 (Equation 2).24 After several conventional procedures for accomplishing this task met with failure. CHO 2 Br (eq.2) OMe OMe OMe 16 37 17 a convenient route was found by adding an organometalllc reagent to an N-silyl Imlne which prepared by a modification of a procedure developed by Wannagat and R o c h o w . 3 2 Thus, treatment of aldehyde 16 with a slight excess of lithium hexamethyldisilazide in tetrahydrofuran gave the corresponding N-silyl imine. Addition of a slight excess of allylmagnesium bromide gave an excellent yield <97 X) of amine 37 after an aqueous workup and purification. This was the first report of an organometallic adding to an N-silyl imine. Although this method is limited to nonenolizable carbonyl compounds,33 it is extremely valuable in the early stages of the synthesis of lythrancepine alkaloids. Scheme VII CHO H HOCO., OMe 38 OMe 39 Problem H RO H RO Problem 2 C02Et OMe OMe 41 40 18 Keeping in mind the general plan outlined in Scheme II, it waa decided to examine the first two problems described above as outlined in Scheme VII. Treatment of commercially available p-anisaldehyde (38) with 1.2 equivalents of lithium hexamethyldisilazide in tetrahydrofuran followed by addition of 1.22 equivalents of ethereal allylmagnesium bromide at 0°C gave an amine 43 in 87 * yield after an aqueous workup and distillation as shown in Scheme VIII. As mentioned above, this reaction Scheme VIII NSiMe b,c OMe 42 a. LiHMDS b. CH2=CHCH2MgBr c. H20 sequence presumably proceeds via initial formation of N- silyl imine 42 followed by addition of the organometallic reagent. To maximize the yield, the silyl imine intermediate 42 was not purified or fully characterized. The formation of 42, however, was established by crude proton NMR which showed the silyl imine proton as a singlet at 8 8.90 as well as no aldehyde proton due to 36 at 8 9.87. 19 Amine 43 was converted to amide 44 using the excellent method of Weinreb.34 Thus, sequential treatment of amine 43 in dichloromethane with trimethylaluminum35 and methyl 5,5-dimethoxypentanoate (22) gave amide 44 in 98 x yield as shown in Scheme IX. The methyl 5,5- dimethoxypentanoate (22) was prepared from dimethyl Scheme IX 1. AlMe OMe 43 44 Me2OC NaOMe co 2 MeO.C 45 46 47 H C (OMe)3 p-TsOH OMe NaCl OMe wet DMF 0 OMe Me02C 22 48 20 malonate (45) and acrolein (46) using the known procedure of Stevens3& in an overall 22 X yield. Treatment of amide 44 with formic acid in dichloromethane (25°C, 3 h) gave a mixture of four products. The major product was quinolizinone 39 (66 X") with the expected stereochemistry. In addition, isomeric formate 50 (3.3 X) and an equal mixture of olefins 49 (14 X) were obtained (Scheme X). Several aspects of this Scheme X MeO OMe HOCO HC00H OMe OMe 39 HOCO-[3R] 44 50 HOCO-[3S1 cyclization reaction are notable. The conversion of 44 to 49, 39, and 50 presumably proceeds via initial acetal hydrolysis, C-N bond formation to give a carbinol amide such as 51, dehydration to give an N-acyliminium ion 52, and cyclization (Scheme XI). No products derived from 0- alkylation of the amide 44 (e.g. amine 43), were obtained. It is also noted that the subsequent iminium ion cyclization (52— ►49, 39 and 50) proceeds with very clean stereochemistry at C-l and C-5. All of the 2 1 quinolizinone products (49, 39, and 50) have the same stereorelationship predicted by the transition-state model proposed for such iminium ion cyclizations.29 Scheme XI MeO .OMe 49 + 39 4- 50 OMe 44 The stereochemical assignments for 39 and 50 were made primarily on the basis of proton NMR analysis (Figure 3) OMe HOCO OMe 39 HOCO. OMe HOC OMe 50 Figure 3 2 2 Both 39 and 50 have sharp doublets at 6 6.24 and 5 5.85, respectively, due to H-l. The coupling constants for H-l 4.5 Hz). The large coupling constants suggest that both H-3 and H-5 are axially disposed in 39. The appearance of a quintet (J=3.5 Hz) at S 5.13 in 50 confirmed the equatorial nature of H-3 in this minor product. Adjustment of the stereocenter at C-3 in the major formate 39 was performed in a straightforward manner as shown in Scheme XII. Treatment of the formate 39 alone with sodium hydroxide in aqueous methanol gave alcohol 53 in 94 X yield. However, after confirming the iminium ion cyclization stereochemistry, the mixture of 49, 39, and 50 was hydrolyzed directly with sodium hydroxide for practical purposes. Thus, treatment of amide 44 on a moderate scale <21.3 g; 66.2 mmol) with formic acid <2 h, 25°C) followed by hydrolysis gave an inseparable mixture of alcohols 53 and 54 in 67 X overall yield from amide 44 as well as a mixture of olefins 49 in 11 X yield. It was possible to crystallize about 80 x of alcohols 53 and 54 directly from the crude hydrolysis products using ethyl 23 Scheme XII HOCO HO CrO NaOH 3 OMe OMe OMe 39 HOCO-[3R] 53 HO-[3R] 55 50 HOCO-[3S] 54 HO-[3S] LiEt3BH or K-sec-BUjBH HO AcO- OMe OMe OMe 56 57 54 HO-[3Sj 53 HO-[3R] acetate-hexane. The mother liquor was separated easily by column chromatography over silica gel. Chromatographic separation of 49, 39, and 50 prior to hydrolysis was complicated by their similar mobilities on silica gel. The mixture of alcohols 53 and 54 was treated with Jones reagent in acetone to give ketone 55 in 74 X yield as well as recovered alcohols 53 and 54 in 19 X yield. Oxidation of alcohols 53 and 54 with pyridinium chlorochromate in dichloromethane3& did not improve the yield of 55. Two bulky reducing agents, lithium triethylborohydride sec-butylborohydride (K-Selectride),3^b were chosen for converting rigid ketone 55 to the desired axial alcohol 54. The resulting inseparable alcohols 54 and 53 were treated with acetic anhydride, triethylamine, and a catalytic amount of 4-dimethylaminopyridine40 to give a separable mixture of acetates 56 and 57. The results are shown in Table I. The table shows that the lithium Table I: Reduction of ketone 55* Reducing agents Yield of alcohols Yield of acetates Ratio 53 and 54 56 and 57 56:57 K-sec-Bu3 BH 98 * 83 X 6.6:1 LiEt3BH 99 * 87 X 9.5:1 • Both reactions were performed in tetrahydrofuran at - 78° C triethylborohydride is slightly superior to potassium tri-sec-butylborohydride in terms of stereoselectivity. Thus, lithium triethylborohydride was used in all subsequent reactions involving reduction of a C-3 k e t o n e . 41 As expected, the C-3 proton in axial acetate 56 appeared as a quintet Having established the appropriate stereochemistry at C-l, C-3, and C-5, we turned to introduction of the C- 9 sidechain as described in the next section. B. Introduction of the C-9 Sidechain (Eschenmoser Sulfide Contraction) This section will describe introduction of the C-9 sidechain via an Eschenmoser sulfide contraction^^ Qnd the establishment of the C-9 stereochemistry <40— ^41 in Scheme V I I ) . Lactam 5 6 , with appropriate stereochemistry at carbons 1, 3, and 5, was converted to thiolactam 58 using the excellent method developed by Lawesson and his coworkers43 (Scheme X I I I ) . A mixture of 0.5 equivalents Scheme XIII AcO AcO, Lawesson1s reagent 59 OMe OMe 56 58 of 2,4-bis<4-methoxyphenyl)-1,3-dithia-2,4-diphosphetane- 2,4-disulfide CLawesson's reagent (59)] and lactam 56 in toluene gave thiolactam 58 in a nearly quantitative yield. Lawesson's reagent (59) was prepared from anisole 2 6 OMe 155°C |) + p4s 10 (eq.3) 6 s 59 and phosphorus pentasulfide in 72 * yield (Equation 3). It was found that H-l in thiolactam 58 was more than 1 ppm downfield from H-l in lactam 56 by proton NMR analysis. Thus, the H-l proton in lactam 56 appeared at 8 6.08 but that in thiolactam 58 was at 8 7.35, even lower than the aromatic protons. The differences between 58 and 56 may arise from the position of the sulfur in the third row of the periodic table. The higher atomic number of 0 sulfur causes an increased covalent radius (0: 0.74 A; S: O 1.04 A), larger polarizability, and a decreased tendency to form double bonds with second row elements. In terms of the valence bond theory, this leads to a stronger contribution due to polar resonance structures (60) in thioamides than in amides (Figure 4 ).44 © s 60 Figure 4 This electronic effect is probably responsible for the downfield shift of H-l in thiolactam 58. 27 The next step involved introduction of the sidechain at the C-9 position. The Eschenmoser sulfide contraction,42 in which a thioamide is converted into a vinylogous urethane, seemed to be a viable method. Surprisingly, initial attempts to introduce a C-9 carbethoxymethylidene group from thiolactam 58 met with failure. For example, sequential treatment of thiolactam 56 with ethyl iodoacetate and triphenylphosphine- triethylamine at room temperature gave lactam 56 in 74 X yield and only 12 X of vinylogous urethane 61 (Scheme XIV). Scheme XIV u H C 02Et AcO CN OMe OMe 56 62 28 Variations of solvents the use of reagents such as ethyl bromoacetate and iodoacetonitrile did not improve the yield of desired products. For example, identical treatment of thiolactam 58 with iodoacetonitrile gave lactam 56 in 50 H yield and enamino nitrile 62 in 23 ft yield. Before discussing the solutions to the problem of lactam formation, the Scheme XV AcO AcO C0,Et OMe OMe AcO OMe AcO AcO COjEt Ph tP C 02E1 OMe OMe 66 61 probable mechanism of the sulfide contraction should be explained (Scheme XV). The first step of this sequence involves thiolactam S-alkylation. This step may be reversible. For example, Ireland and Brown showed that heating thioiminium salt 66 with ester 67 in acetonitrile give a 1:1 mixture of salts 66 and 66 (Equation 4 ).45 The second step of the reaction involves C 02Et I S 'CO,i-Pr /^ N (C H 3)2 + ,CH2CCV - Pr L © + ICH2C0 2Et .© 32 ^ N ( C H 3), (eq.4) ' I0 66 67 68 63 sequential deprotonation, episulfide formation, and removal of sulfur by the thiophilic triphenylphosphine. To identify the problem, a series of experiments were performed. First, it was shown that thiolactam alkylation was not the problem. Thus, thiolactam 58 was treated with excess methyl iodide in a mixture of diethyl ether and dichloromethane at room temperature to give salt 69 in 80 X yield as a pale yellow solid. This material was easily characterized by iH-NMR and IR spectroscopy. This hygroscopic salt 69 was then treated with sodium borohydride in methanol to give amine 71 in 92 X yield. Identical treatment of thiolactam 58 with iodoacetonitrile in diethyl ether gave salt 70 in 85 X 30 yield. Reduction with sodium borohydride gave amine 71 in 98 * yield (Scheme XVI). These experiments showed that thioiminium salt formation was not a problem and showed that such ions were susceptable to nucleophilic attack.46 Scheme XVI AcO AcO CH_ I © CH OMe OMe 58 69 ICH2CN NaBH, AcO AcO, NaBH, CH2CN OMe OMe 70 71 Since it appeared that the deprotonation or sulfur scavenging steps may be the problem, proton NMR studies were performed at room temperature. Thus, a solution of pure thioiminium ion 70 in chloroform was placed in an NMR tube and 1.2 equivalents of triphenylphosphine was 31 added. No change in the NMR spectrum of 70 was observed. However, addition of 1.2 equivalents of DABCO to the mixture caused an immediate change.47 The NMR spectrum clearly showed disappearance of doublets at 6 4.35 (J=18.9 Hz) and 5 5.10 Scheme XVII AcO AcO :B OMe OMe 70 72 h 2o AcO AcO N OMe OMe 56 62 32 doublet, at 5 4.50 (H-5) in 70 and the appearance of multiplete at 6 5.35, 4.90, and a ainglet at 6 3.20 suggested that salt 70 had probably been converted to enamine 72: S 3.20 (SCH2 CN), 3.90 3), and 5.35 should account for charge neutralization by deprotonation of thioiminium ion 70 to ketene N,S-acetal 72. Therefore, it appears that base kinetically deprotonates 70 at C-8 rather than the carbon bonded directly to sulfur. Aqueous workup then converts 72 to lactam 56, perhaps via salt 70 (Scheme XVII). Several examples of enamine formation in sulfide contractions have been reported. For example, Ireland and Brown found that attempted sulfide contraction of salt 73 led only to thiophene 74 presumably via enamine formation from the salt 73 (Equation 5 ).45 After completion of these (eq.5) 73 74 studies, Rapoport and his coworkers also found that sulfide contraction of thiolactam 75 produced a-alkylated thiolactam 80 and lactam 78 in 2-10 x and 5-18 X yields, respectively, as side products depending on reaction 33 conditions (Scheme XVIII).48 Scheme XVIII t-Bu02C"'\N/■ O ^n 0 I Bzl t-Bu02C"'^N^^/- O - -co’Me 2 78 I Bzl 81 r \ t t-BuOjC'^MAs^^co^Me ' L Bzl t-B u 0 2C'"\N K ^S____ ^t-BuOjC'^N^S^^CQ^e 77 Bzl Bzl J 75 76 r^ C Q ,M e t-Bu02C"'C§As/ \ c . ^ I 2 I— Bzl t-Bu02C"'SNAs 79 I Bzl 80 Since it was clear that competitive deprotonation at C-8 was causing the problems, conditions were sought to deprotonate the methylene sidechain. It was reasoned that the pKa's of DABCO-HI and the C-8 and SCH2 CN positions of salt 70 might be close enough that deprotonation of 70 adjacent to sulfur might occasionally take place under forcing conditions. It was anticipated that such a deprotonation might then lead to episulfide formation and 34 irreversible removal of sulfur by the thiophilic triphenylphosphine. Thus, salt 70 was warmed under reflux with triphenylphosphine and DABCO. A dramatic improvement in the yield of enamino nitrile 62 was observed. Overall, treatment of thiolactam 58 with iodoacetonitrile gave 62 in 70 X yield. Similarly, treatment of thiolactam 62 wi^.i ethyl bromoacetate in a mixture of tetrahydrofuran and diethyl ether for 2 days followed by the addition of triphenylphosphine and DABCO under reflux for 5 min gave vinylogous urethane 61 in 65 X yield. With the sidechain carbons introduced, attention was turned to fixing the C-9 stereochemistry. Reduction of vinylogous urethane 61 with sodium cyanoborohydride49 at pH 4 gave amino ester 82 (77 X) with the desired stereochemistry at C-9. None of the C-9 diastereomer was obtained (Scheme XIX). This step should not pass without Scheme XIX AcO. AcO. NaBH3CN C 02Et pH 4 C 02Et OMe OMe 61 82 comment. Since the C-9 isomer of amino ester 82 was not isolated in this model study and some other relevant experiments were performed later, an explanation of thiB 35 stereoselective reduction will be deferred until chapter IV (vide infra). However, evidence for the assignment of structure 82 is presented here. Proton NMR analysis of amino ester 82 suggests that it adopts a cis-fused chair- chair conformation. For example, 200 MHz NMR analysis of amino ester 82 showed a 9-line triplet of triplets (J=11.2, 11.2, 5.0, and 5.0 Hz) at 5 4.75-5.20 for H-3. This indicates that the A-ring of 82 is in a conformation in which this hydrogen is axially disposed. An analysis of H-9 suggested that it was equatorially disposed. Thus, independent irradiation of the C-10 protons <5 2.28 and 6 2.75) caused H-9 (6 3.20) to collapse to a broad doublet Since a large coupling to the axial hydrogen on C-8 was not observed, this suggested that H-9 was equatorially disposed. These NMR experiments are consistent with structure 82 in the aforementioned cis-fused chair-chair conformation. Some other experiments were performed to ensure that the stereochemical assignment at C-9 and the identification of H-9 in the NMR were correct. The amino ester 82 was converted to C-10 deuterated hydroxy ester 83 upon treatment with excess methyl alcohol-d and sodium methoxide under reflux in 81 x yield. The resulting hydroxy ester 83 was converted to acetate 84 (63 X) as shown in Scheme XX. 200 MHz proton NMR analysis of 36 Scheme XX AcO, HO AcO MeOD Py D C0>Me DMAP CO,Me OMe OMe OMe 82 83 84 acetate 84 clearly showed a doublet of doublets at 6 4.02 and a broad doublet at 6 3.16 which is equatorially disposed. Finally, the structure of amino ester 82 was proven by X-ray analysis of a crystal grown from methanol (Figure 5).5* 37 It. is int.ereBt.ing that only upon formation of amino ester 82 does the quinolizidine skeleton release the severe 1,3-diaxial interaction between the C-l aryl group and C-3 substituents. All quinolizidine structures up to vinylogous urethane 61 and enamino nitrile 62 adopt conformations corresponding to 50 in Figure 3. After release of the 1,3-diaxial interaction, amino ester 82, for the first time, affords a quinolizidine with the cis- fused conformation 82a shown in Figure 6. ^ COzEt AcO C 0 2Et AcO OMe 82 Figure 6 In summary, the model studies established methods for controlling all the stereochemistry from C-l to C-9 and provided a method for introducing a carbon-carbon bond at the lactam carbonyl group (C-9). The next two chapters will present synthetic approaches to <+>- lythrancepine-II itself. CHAPTER III. AN APPROACH TO THE SYNTHESIS OF LYTHRANCEPINE ALKALOID VIA BIARYL INTERMEDIATES The first approach to <+)-lythrancepine-II started with a biaryl system. The synthetic plan is shown in Scheme XXI. Scheme XXI CHO CHO CHO OMe NH OMe OMe OMe OMe OMe OMe CHO 85 86 87 PhCH.O HOCO OMe OMe OMe OMe OMe OMe OMe CHO 91 89 3S CO.Et C02Et OAc OAc OAc MeO MeO . MeO 92 93 HO OAc •SPh OH Me eO MeO 4 96 95 We began with the known biaryl 86.^2 A mixture of aniaaldehyde (38), iodine monochloride, and acetic acid under gentle reflux gave 3-iodoanisaldehyde (85) in 71 X yield. A mixture of 85 and copper p o w d e r ^ 3 Was heated at 210-220°C to give biaryl 86 in 78 X yield. Monoacetalization of dialdehyde 86 gave a atatiatical mixture of producta aa ahown in Scheme XXII. Therefore, a mixture of dialdehyde 86, 1 equivalent of neopentylglycol, and a catalytic amount of p- tolueneaulfonic acid monohydrate in benzene gave an easily separable mixture of monoacetal 87 (50 X) , diacetal 97 (16 *>, and starting dialdehyde 86 (21 X). Obviously diacetal 97 could be recycled by hydrolysis to give monoacetal 87. This was performed by stirring a mixture of diacetal 97 and 1 N aqueous sulfuric acid in acetone to give monoacetal 87 (42 X), dialdehyde 86 (29 X> , and starting diacetal 97 (25 X) using TLC to monitor the reaction progress. Further experiments were performed to verify the statistical formation of acetals in this biaryl system. A mixture of dialdehyde 86, 1 equivalent of 1,3-propanedithiol, and a catalytic amount of Scheme XXII CHO 41 phosphorous oxychloride In chloroform also gave an easily separable mixture of monothioacetal 98 <35 x>, dithioacetal 99 <25 X) , and 25 X of starting dlaldehyde 86 Conversion of monoacetal 87 to amide 89 was stralghtfoward. Thus, sequential treatment of nonoaldehyde 87 with 1.3 equivalents of lithium hexamethyldisilazide followed by the addition of 1.3 equivalents ethereal allylmagnesium bromide gave amine 88 <85 X). Treatment of 88 with trimethylaluminum54 and methyl 5,5-dimethoxylpentanoate <22)36 in sequence gave amide 89 in 94 x yield as shown in Scheme XXI. 34 Treatment of amide 89 with formic acid in dichloromethane for 2 h gave five products: 86 <4 x), 100 <36 X), 101 <9 X), 90 <43 X), and 102 <7 X). Products 86 and 100 are unlike those formed in the model study Therefore, several points will be described here. First, as shown in Scheme XXIII, hydrolysis of the cyclic acetal took place during the cyclization. This is not surprising as it is known that electron donating aryl groups accelerate acetal hydrolysis.55 in addition, acetal 42 Scheme XXIII HOCO OMe HCOOH OMe CHO OMe CHO 89 86 90 HOCO-[3R] 102 HOCO-[3S] H OMe + OMe CHO 101 migration <89— *-100) waa obaerved. Apparently, hydrolyaia of dimethylacetal 89 givea an aldehyde which undergoea carbinolamide formation and cyclization or ketalization with neopentylglycol to give 100. Thia acetal ia aomewhat resistant to hydrolyaia due to the gem-dimethyl effect,55 but will aerve aa an intermediate in the bicycloannelation. Thia waa ahown by treating pure iaolated amide 100 with formic acid for 24 h to give olefin 101 (15 X), formate 90 (53 X), and isomeric formate 102 (2 X ). The formation of dialdehyde 86 can be explained as follows. In 1981, during the course of a synthesis of gephyrotoxin 2 2 3 AB,S6 Hart and Tsai discovered that an N-acyl-2-aza-Cope rearrangements? was underlying an N-acyliminium ion cyclization. They found that treatment of hydroxy lactam 103 with formic acid- dichloromethane (25°C, 10 h) gave a 60 X yield of formate 104, and trace amounts of isomeric formate 105, and olefins 106 as expected (Scheme XXIV). But it was quite a Sc he me XXIV HOCO, HHHOCO. H + 104 105 106 HCOOH-CH2Cl2 OH CF3COOH-CH2Cl2-Et3SiH surprise when none of the cyclized product was detected upon treatment of hydroxy lactam 103 with a mixture of trifluoroacetic acid-dichloromethane-triethylsilane (25°C, 5 min). Instead, a separable 3:5 mixture of lactams 107 and 100 was obtained in a 73 X yield. Also, treatment of 103 with formic acid-dichloromethane- triethylsilane (25°C, 10 h) gave cyclized formate 104 (29 X) in addition to a 3:5 mixture of lactams 107 and 108 respectively (42 X). The observations outlined above suggest that treatment of carbinolamide 103 with acid gives N- acyliminium ion 109 which rearranges to ion 110, presumably via a 2-aza-Cope process as shown in Figure 7. Reduction products 107 and 108 apparently result from a trapping of N-acyliminium ions 109 and 110 by triethylsilane. In addition, it was found that the ratio Figure 7 of lactams 107 and 108 did not vary with the reaction media. This suggested that an equilibrium mixture of 109 45 and 110 was rapidly established, followed by slower secondary processes such as reduction and/or cyclization. In trifluoroacetic acid, reduction was faster than cyclization. In the presumably more nucleophilic formic acid media, cyclization and reduction took place at comparable rates. This was the first example of a hetero- Cope rearrangement in an N-acyliminium ion system.58 Thereafter, Hart and Yang studied the 2-aza-Cope rearrangements in detail and applied it to the synthesis of pyrrolizidine alkaloids.59*60 it was also found that there appears to be little difference in the behavior of the five- and six-membered ring iminium ions. Therefore, it is presumed that dialdehyde 86 is the result of 2-aza- Cope rearrangement of the initially formed N-acyliminium Scheme XXV 112 CHO 111 86 46 ion 111 t.o an ion 112, followed by hydrolysis (Scheme XXV). This was the first isolation of the aldehyde portion resulting from hydrolysis of an iminium ion formed in an N-acyl-2-aza-Cope rearrangement. Continuing with the synthesis, formate 90 was hydrolyzed to alcohol 113 with sodium hydroxide in aqueous methanol (96 *) and resulting alcohol 113 was treated with neopentylglycol and a catalytic amount of g- toluenesulfonic acid monohydrate to give acetal 114 (100 *> Scheme XXVI HOCO Cl ho o h OMe NaOH OMe H CHO CHO 90 113 DMSO, (C 0C 1)., Et3N 1 I 114 + 113 + OMe + OMe OMe OMe HO 116 115 47 Oxidation of alcohol 114 using a Swern oxidation, however, aroused some troubles.61 Sequential treatment of alcohol 114 with oxalyl chloride and dimethyl sulfoxide in dichloromethane followed by the addition of triethylamine gave ketone 115 in variable yields (20-96 *). When yields were low, starting alcohol 114, hydrolyzed alcohol 113, and/or hydrolyzed ketone 116 were obtained (Scheme XXVI). Because alkyl methylthiomethyl ether 118 (the common side-product of Moffatt type oxidations resulting from a Pummerer-type rearrangement) was not isolated, recovery of alcohol 114 results from only partial conversion of alcohol to its alkoxysulfonium salt 117 before addition of triethylamine (Scheme XXVII). Scheme XXVII o o r § ©>» Me,SO ♦ CtCCCI [Me2SCICI J r ’r'c h o h 114 ,CH c s TE— (M e^O CH Rtfci) ♦ HCI I I H CH2 117 RfRzCH0®* CH^CHZ 115 I R'r'c HOH ♦ rV c HOCH.SCH j 114 118 48 Acetal hydrolysis should be resulted from protonation by hydrogen chloride produced in this Swern oxidation. Although hydrolysis of acetal 113 and 116 and recovery of alcohol 114 may be explained as above, it is still uncertain that what factors may influence the yield. Ketone 115 was converted to an inseparable mixture of alcohols 119 and 114 with lithium triethylborohydride Scheme XXVIII HO OMe OMe OMe OMe OMe 114 119 115 I NaH, /PhCH2Br PhCHzO. OMe HCl OMe OMe CHO 120 91 (72 X) and resulting mixture of alcohols was treated with benzylbromide and sodium hydride to give mainly axial benzyl ether 91 (44 X) as well as recovered alcohols (15 X). No attempt was made to isolate the benzyl ether isomeric to 91. Hydrolysis of the acetal was also a problem during the aqueous workup of this reduction. Finally, the acetal 91 was hydrolyzed by hydrochloric acid in aqueous acetone to give aldehyde 120 in 87 X yield (Scheme XXVIII). The next steps in this approach would have been adjustment of the aldehyde oxidation state, introduction of the C-10 carbethoxymethylidene group, and conversion of enamino ester 93 to /?-amino ester 94. This was to be followed by the synthesis of the 0-ketosu If oxide 95, activation of hydroxyl group, high dilution cyclization of the intermediate tosylate, and reduction of the phenyl sulfinyl group. The synthesis of lythrancepine-II (4) was then to be completed by reducing 9G, acetylation, and removing the blocking group. However, due to difficulties associated with the acetal protecting group and solubility problems encountered with several of the intermediates, an alternative route was pursued. This route, which revolves around construction of the biaryl via an intramolecular Ullmann reaction, will be the subject of the next chapter. CHAPTER IV. TOTAL SYNTHESES OF <£)-LYTHRANCEPINE-II AND <♦)-LYTHRANCEPINE-III The next approach to lythrancepine-II la outlined antithetically in Scheme XXIX. In the final atage of thia Scheme XXIX H HO OAc > OAc OMe ieO 121 OMe MeO OMe MeO O OMe OMe 22 123 122 synthesis, we envisioned the 13-membered ring closure via an intramolecular biaryl coupling. The C-9 aubstituent in 121 would be introduced by the Eachenmoaer aulfide 50 51 contraction. The synthesis of lactam 122 was anticipated to be quite stralghtfoward according to the procedures established In the model study. This synthetic strategy has several advantages. First of all. It avoids the problem of acetal hydrolysis which occurred in the scheme using biaryl systems (chapter III>. Second, if introduction of the entire C-9 aidechain succeeds, the scheme is convergent. Lastly, the basic strategy allows for functional group differentiation in all rings and should allow the synthesis of all other members of type D Lythraceae alkaloids, even the lythrum alkaloids (see ref. IS, i and ii). Before describing the synthesis, it is noted that all the conformations of intermediates are identical to the corresponding structures prepared in the model study up to amino ester 82. To avoid repetition, the details of the chemistry will only be mentioned on occasion. The synthesis began with m-iodoanisaldehyde (85) which had already been prepared for the synthesis of the biaryl 8 6 . It is noted that the proton NMR spectrum of 85 clearly showed a doublet at 5 6.93 for ArHs with J=9 Hz <2££li2 coupling), a doublet of doublets at 6 7.84 for ArH& with J=9, 2 Hz (ortho and meta couplings), and a doublet at 6 8.27 for ArH2 with J*2 Hz (meta coupling). All compounds in this synthesis exhibited this aromatic NMR pattern as long as they maintained the 3-iodo-4- 52 methoxyphenyl moiety. Sequential treatment of m- iodoaniealdehyde (85) with lithium hexamethyldieilazide and allylmagneaium bromide gave homoallylic amine 123 (76 X) along with 9 X of reduced amine 43. The crude proton NMR of the ailyl imine intermediate analogoua to 42 ahowed no reduction of the iodophenyl moiety by lithium hexamethyldieilazide. Thua, a poaaible explanation for the formation of amine 43 may be metal-halogen exchange between the aryl iodide and Grignard reagent followed by protonation during the aqueoua workup. In fact, the yield of amine 43 waa 9 X only if a alight exceaa (1.05 equivalenta) of allylmagneaium bromide waa uaed. The uae of larger amounta of the Grignard reagent increaaed the yield of 43. Sequential treatment of amine 123 with trimethylaluminum and eater 22 gave amide 124 in 96 X yield.54 Treatment of amide 124 with formic acid in dichloromethane <2 h, 25QC) gave formate 126 (48 X) , iaomeric formate 127 (5 X), inaeparable alcohola 126 and 129 (22 X), olefina 125 (9 X), and m-iodoaniaaldehyde (85) (11 X). Formate 126 could be cryatallized directly from the mixture of crude cyclized producta in 38 X yield (Scheme XXX). The olefin mixture 125 waa cryatallized from ethyl acetate to give a aingle iaomer by 13C-NMR apectroacopy. Thia iaomer waa tentatively aaaigned at the ^2»3-iaomer based on 1 H-NMR and mass spectral data. 3-H NMR of olefin mixture 125 showed approximately a 3:1 ratio of the A2'3-isomer and A^'^-isomer, respectively Scheme XXX CHO 1.LiHMDS 2.CH2=CHCH2MgBr 3.H20 1.AlMe CHO OMe HCOOH — Y o OMeOh OMe 124 R 126 R=H0C0-[3R] + 127 R=H0C0-[3S] 128 R=H0-[3R] 129 R=H0-[3S] OMe Both formates (126 and 127) and both alcohols (126 and 129) are useful as synthetic intermediates. 54 Treatment of the mixture of formates 126 and 127 with Scheme XXXI HOCO HO NaOH OMe OMe 126 HOCO- [ 3R ] 128 HO- [ 3R ] 127 HOCO- [ 3 S ] 129 HO- [ 3 S ] DMSO, Et.N (C0C1) 3 AcO HO AC 2 O LiEt3BH OMe OMe OMe 130 131 129 HO-[3S] + 128 HO-[3R] AcO. AcO Lawesson1s reagent 59 OMe OMe 132 133 sodium hydroxide in aqueoua methanol gave alcohols 1 2 8 and 1 3 0 in 95 X yield. These alcohols, together with the alcohols produced during the cyclization, were converted to ketone 1 2 9 in 99 X yield using a Swern oxidation.61 It is noted that the Swern oxidation gave ketone 1 3 0 almost quantitatively, unlike ketoacetal 1 1 5 (Chapter III). Ketone 1 3 0 was reduced at -78°C with lithium triethylborohydride to give a mixture of alcohols 1 2 9 and 1 2 8 (100 X) and the resulting inseparable mixture of 1 2 9 and 1 2 8 was treated with acetic anhydride, triethylamine, and a catalytic amount of 4-dimethylaminopyridine to afford a separable mixture of acetate 1 3 2 (84 X) along with 7.5 X of isomeric acetate 1 3 1 (Scheme XXXI). Having established the appropriate stereochemistry at C-l, C-3, and C-5, we turned to introduction of the C-9 sidechain. Treatment of lactam 1 3 2 with Lawesson's r e a g e n t 4 3 gave easily crystal1izable thiolactam 1 3 3 in 98 X yield. Treatment of thiolactam 1 3 3 with ethyl iodoacetate in tetrahydrofuran (25°C, 2 days) gave a quantitative yield of crystalline salt 1 3 4 . It is noted that ethyl iodoacetate is much more efficient than ethyl bromoacetate in thia reaction. Also using excess electrophile shortened the reaction time considerably. Treatment of salt 1 3 4 with triphenylphosphine and DABC0 in chloroform under reflux gave vinylogous urethane 1 3 5 (76 X) . I t should be mentioned that formation of 1 3 5 from 56 the salt 134 required a longer reaction time than formation of 61 from 64. TLC analysis of the reaction mixture after 1 0 min showed considerable amounts of lactam 132. The crude proton NMR, however, showed no peaks due to lactam 132, suggesting the existence of an equilibrium between iminium ion and enamine as mentioned before. The formation of lactam presumably occurs only on the TLC plate or upon aqueous workup of the reaction media. Reduction of vinylogous urethane 135 at pH 4 with Scheme XXXII H H AcO AcO ICH2C02Et OMe OMe 133 134 DABCO, Ph3P AcO AcO. NaBH3CN pH 4 OMe OMe 136 CH2C02Et-[9R] 137 CH2C02Et-[ 9S] 135 57 sodium cyanoborohydride49 gave amino ester 137 <94 X) and 5 X of isomeric amino ester 136 <25°C, 30 min) (Scheme XXXII). Sequential treatment of thiolactam 133 with 5 equivalents of iodoacetonitrile in a mixture of tetrahydrofuran and diethyl ether <25°C, 3 days) followed by the addition of triphenylphosphine and DABCO under reflux <20 min) gave enamino nitrile 138 in 79 X yield. Reduction of enamino nitrile 138 at pH 4 with sodium cyanoborohydride gave 84 X of amino nitrile 140 and 8 X SchemeXXXIII H H AcO AcO 1. ICH2CN CN 2.DABCO, Ph3P OMe OMe 133 138 H AcO 58 of isomeric amino nitrile 139 (25°C, 10 min) (Scheme XXX1I1). Thus, as expected, the chemistry developed in Chapter II translated well to this system. Attention was next turned to introduction of the entire C-9 sidechain. This required the preparation of «- bromoketone 1 5 0 . This was accomplished in 8 steps from p- hydroxybenzaldehyde ( 1 4 1 ) as outlined in Scheme XXXIV. Treatment of 1 4 1 with malonic acid and pyridine gave cinnamic acid 1 4 2 (44 X) .62 Reduction of 1 4 2 with 5 X palladium on chacoal gave hydrocinnamic acid 1 4 3 (89 X ). Acid 1 4 3 was treated with potassium iodide and iodine in ammonium hydroxide to give monoiodide 1 4 4 in 46 X yield (25°C, 2 h) along with a trace of diiodide.63 Treatment of hydroxyhydrocinnamic acid 1 4 4 with excess diazomethane (25°C, 2 days) gave ester 1 4 5 in 90 X y i e l d . 63 Methyl ester 1 4 5 was hydrolyzed with sodium hydroxide in aqueous methanol (25°C, 3 h) to give acid 1 4 6 in 99 X yield. Treatment of acid 1 4 6 with thionyl chloride under reflux (30 min) gave acid chloride 1 4 7 (100 X) and slow addition of acid chloride 1 4 7 to excess diazomethane at 0°C gave diazoketone 1 4 8 (99 X ).64 Hydrogen chloride was passed through a solution of diazoketone 1 4 9 in diethyl ether to give chloromethylketone 1 4 9 (76 X ) and identical treatment of 1 4 8 with hydrogen bromide gave bromomethyl- ketone 1 5 0 in 72 X y i e l d . 65 iodomethylketone 1 5 1 was prepared by treating chloromethylketone 1 4 9 with sodium 59 Scheme XXXIV COjH CHO Pd/C Py OH 141 142 KI c o 2h NaOH CH„N OMe OMe OH 146 145 144 | SOC1, HC1 OMe OMe OMe 147 148 149 HBr Nal HI OMe OMe OMe 150 152 151 60 iodide in acetone in 89 X y i e l d . ^ Initially, iodomethylketone 151 was used as the electrophile in the preparation of the required thioiminium salt. However, iodomethylketone 151 was slowly reduced to methylketone 152, causing complications. Bromomethylketone 150 did not reduce to methylketone 152 under the same conditions and was quite suitable for use. Sequential treatment of thiolactam 133 with 2 equivalents of a-bromoketone 150 in chloroform (25°C, 3 days) followed by the addition triphenylphosphine and DABCO under reflux (30 min) gave a single product in 6 8 X yield (Scheme XXXV). However, spectroscopic data suggested this material was not the desired enamino ketone 154. For example, IR spectrum showed no carbonyl absorption bands due to a vinylogous amide. The proton NMR spectrum was so complicated that it was very difficult to characterize the exact structure. However, critical information was provided by the mass spectrum and 13C-NMR spectrum. First of all, the parent ion did not match that required by structure 154. Instead of m/e 729 required by 154, the parent ion appeared at m/e 743, consistant with the molecular formula C2 9 H3 1 NO4 I2 S. The 13C-NMR spectra (broad band and off resonance) clearly showed the presence of 6 methylenes and 1 0 quaternary carbons, inconsistant with the 7 methylenes and 9 quaternary required by 154. Finally, the UV spectrum gave a Amax at 227 nm 61 inconsistent with the Xmax 327 nm <7r-ir« band) calculated by 1 5 4 using Woodward's rules. It waa concluded from these spectral data that the product obtained from the attempted sulfide constraction was, in fact, thiophene 1 5 5 . This structure agrees with all spectral data including the parent peak of 743 In a similar fashion, this reaction with bromoacetone& 7 gave thiophene 1 5 8 in 65 X yield. Scheme XXXV DABCO, Ph3P DABCO 63 The disappointing results with 150 forced us to adopt a procedure which involved stepwise introduction of the C-9 sidechain. Since this required the use of carbanion chemistry (vide infra), adjustment of the C-3 blocking group was required. Thus, a mixture of alcohols 129 and 128 obtained from lithium triethylborohydride reduction of 130, was treated with sodium hydride and benzyl bromide in DMF to afford a separable mixture of benzyl ether 122 (66 X) and 159 (10 X) as shown in Scheme XXXVI. Treatment of lactam 122 with Lawesson's reagent^3 Scheme XXXVI HO PhCH20 NaH OMe OMe 128 ♦ 129 159 + PhCH20. Lawesson's reagent 59 OMe OMe 160 122 64 gave thiolactam 160 (10 min, 98 X) . At this point, it is worth mentioning additional data which is consistent with the downfield shift of H-l in thiolactams as described in chapter II. Lawesson and his coworkers®® have intensively studied the 13C-NMR of thiolactams. A linear relation between the chemical shifts of the thiocarbonyl group of the thioamides and carbonyl group of the corresponding amides has.been found by a least square linear regression analysis (Equation 6 ). According to this equation, 8 (C=S> = 1.60 x <5 (C=0) - 72.3 (eq.6) the calculated chemical shift of C-9 of thiolactam 160 is 5 199.1 (chemical shift of C-9 of lactam 122 is 8 169.6) which is quite close to the observed value of 8 201.0. It was suggested that the shifts may partly result from different inductive effects of the thioamide group operating through bonds and from the considerably higher dipole moment of thioamides operating through space relative to amides. It is noted that the a-carbons attached to the nitrogen are also shifted downfield in 160 (chemical shifts at C-l: 8 48.3 for lactam 122, 8 57.3 for thiolactam 160: chemical shift at C-5: 5 47.1 for 122, 8 50.4 for 160). Treatment of thiolactam 160 with 10 equivalents of ethyl iodoacetate in chloroform (25°C, 24 h) followed by 65 the addition of DABCO and triphenylphosphine under reflux, gave enamino ester 161 in 92 * yield. Reduction of 161 at pH 4 with sodium cyanoborohydride (25°C, 10 nin) gave amino esters 162 (68 X) and 163 (10 X ) (Scheme XXXVII). Reduction of 161 with Adams' catalyst (Pt02> in acetic acid gave 51 * of amino eater 162 and 27 X of iodide-reduced amino ester analogous to 162 (58 psi, 90 min) . At this point, a plausible reason for the good stereoselectivity observed in the iminium ion reduction Scheme XXXVII 1 .ICHoC0oEt C 0 2Et 2.DABCO, Ph^P OMe OMe 160 161 pH 4 PhC H ,0 CO, Et C 02Et OMe OMe 162 163 (161— ►162 and 163) ahould be presented. Initially, it was felt that 162 would be the thermodynamically most stable reduction product and that an equilibrium might be established between 162 and 163 via an elimination- addition process after reduction took place. Although this is a resonable suggestion, it is not the case as subjecting 162 and 163 to the reduction reaction conditions did not cause their interconversion. Therefore, the partitioning of 161 between 162 and 163 is kinetically controlled. In rationalizing the observed results, it is assumed that iminium ion 164, derived from protonation of 161, is the species undergoing reduction. Proton NMR spectra of iminium salts 64, 69, 70, 134, 153, and 156 suggest the C-l and C-3 substituents in 164 are axially disposed. Thus, in discussing the addition of hydride to 164, two formations, 164a and 164b, must be considered as shown in Scheme XXXVIII. It has been suggested that stereoelectronic features govern the course of addition of nucleophiles to iminium ions. Bohlmann and his coworkers&9 have shown that the 67 borohydride reduction of iminium salts of type 167 gave products 168 in preference to its isomer 169 (Equation 7). This results, and others,70 suggest that a stereoelectronic requirement in iminium ion reductions is maintenance of maximum orbital overlap between the developing nitrogen lone pair and incoming nucleophile. Applying this principle to the case in hand suggests that there are four possible transition states in the reduction of 164 where maximum orbital overlap can be maintained with respect to the attacking hydride reagent and the developing lone pair on nitrogen, resulting in products where the sp3-hybridized orbitals generated are anti- coplanar. Two of these, 165 (path a; si-face attack) and 166 (path c; re-face attack), require boat-like transition states and are probably kinetically disfavored.71 Before discussing the remaining two possibilities, path b (re-face attack) and path d (si- face attack), the relative stabilities of conformations 164a and 164b will be discussed. Although the conformation of the B-rings in 164a and 164b are drawn as half-chairs, there may be some differences in the case of 164a. Because of the adjacent rigid chair-like A-ring, C- 6 cannot move as much as expected when C-7 is flip- floping to interconvert both possible half-chair conformations in the B-ring. Thus, in 164a, the B-ring should have an envelope-like conformation72a j.n order to Scheme XXXVIII PhCH2o k ^ I ^COjEt / W T ° PHCH.01JH ? A ^ 162a O * ,, ' P ath a P ath b \ P ath a - M s 164a P ath b 165 Me P ath d ,© P a th c 166 164 b P ath c P ath d OMe PhCH2cj| PhCH ,0 PhCH CO,Et C02Et 163 b place C -6 at an equatorial position of rigid A-ring, In 164b, however, the B-ring can adopt a true half-chair conformation. Hence 164b seems to be more favorable than conformation 164a. However path d clearly suffers from an unfavorable 1,3-diaxial interaction developing between the C-l phenyl group and the incoming nucleophile (hydride ion). In addition, the most obvious changes that occur on introduction of a heteroatom into a six-membered ring have to do with bond lengths and bond angles. O Carbon-nitrogen bond lengths <1.47 A) are shorter than 0 the carbon-carbon bond length of 1.54 A. The C-N-C bond angle is somewhat smaller than the C-C-C angle in cyclohexane. So, nitrogen containing heterocyclic rings (e.g. piperidine) are somewhat more puckered than cyclohexane itself. Applying this concept to conformations 164a and 164b, and considering the exocyclic ^N=C double bond to A-ring, this ring might also contain a little bit of envelope character to compensate for increasing C(l)-N-C(5) angle resulting from the exocyclic double bond. As a result, the phenyl substituent and C(2)-C<3) bond should rotate little bit clockwise. Keeping in mind the shorter bond lengths of C <1)-N and ^N=C,7 2b this result suggests that path d suffers from a stronger interaction between the incoming nucleophile and the phenyl group than expected and path b might not suffer from an interaction between the 70 nucleophile and C In summary, rather than the relative stability of conformations 164a and 164b, the sterically more favorable transition state which satisfies the stereoelectronic control requirements leads to the product. As mentioned before, assuming that 162 and 163 will prefer to adopt chair-chair conformations, an evaluation of all possible conformers indicates that 162c should be the thermodynamically most possible reduction product.7 3 In fact, proton NMR analyses of 162 and 163 proved that 162 and 163 adopt the conformations of 162c and 163a respectively.7 4 To complete the introduction of the C-9 sidechain, some /3-keto sulfone chemistry was examined.75 it was hoped that methyl phenyl sulfone could be used as a one carbon synthon to link 162 and a 3-iodo-4-methoxybenzyl group. Therefore, the model study outlined in Scheme XXXIX was performed. Sequential treatment of methyl phenyl sulfone 1707& with sodium hydride in DMSO <65°C, 30 min) and a solution of ethyl caproate (171) in tetrahydrofuran gave /3-keto sulfone 172 in 84 X yield.7 7 £-Keto sulfone 172 was treated with sodium hydride in DMSO (25°C, 50 min) 71 followed by benzyl chloride 173?B (25°C, 2 h) to give a- aubstituted-0-keto aulfone 174 in 43 X yield.79 Reduction of aulfone 174 waa performed with aluminum amalgam in 10 X aqueoua tetrahydrofuran (reflux, 3 h) to give ketone 175 in 71 x yield.Thia model atudy aeemed to give a resonable way of introducing the C-9 aidechain. Becauae we were aware that the conditione ueed in the reductive cleavage of aulfone might alao reduce an aryl iodide, a Scheme XXXIX PhS02 CH3 + CH3(CH2)4C02Et 170 171 NaH OMe 173 o CH3(CH2)4C C H S 0 2Ph O CH, Al-Hg CH3(CH2)4CCH2CH. OMe OBzl OMe 175 174 model atudy uaing 3-iodo-4-methoxybenzyl chloride (177) in place of 173 waa performed (Scheme XXXX). 3-Iodo-4- 72 methoxy-benzaldehyde (85) was reduced with sodium borohydride to give alcohol 176 in 96 * yield (25°C, 10 min). Treatment of alcohol 176 with thionyl chloride and pyridine (25°C to reflux, 3 h) gave benzyl chloride 177 in 70 X yield. /3-Ketosulfone 172 was treated with sodium hydride in DMSO (25°C, 1 h) followed by benzyl chloride 177 <25°C, 30 h> to give a-substituted-/?-keto sulfone 178 in 63 X yield along with 24 X of recovered /3-keto sulfone Scheme XXXX CHO H ri NaBH4 soci2 OMe OMe OMe 177 85 176 j NaH, 172 CH2 O 179 OMe 178 172. Sulfone 178 could also be prepared by sequential treatment of /3-keto sulfone 172 with potassium t-butoxide in t-butanol <60°C, 30 min) and a solution of benzyl 73 chloride 177 in tetrahydrofuran (60°C, 36 h) in 75 X yield.®! The reductive cleavage of aulfone 178 waa firat performed with aluminum-amalgam. However, treatment of aulfone 178 with aluminum-amalgam in 10 X aqueous tetrahydrofuran under reflux <1 h) gave 8 8 X of ketone 179 in which the aryl iodide had also been reduced. Several other methods for desulfonylation were examined. For example, zinc-acetic acid (25°C to reflux, 24 h)®2 gave no desulfonylation product but the aryl iodide waa reduced. Sodium amalgam (3 *) in the presence of diaodium hydrogen phosphate in methanol (25°C, 2 h ) ® 3 gave doubly reduced product 179. W-2 Raney-nickel in ethanol®** (reflux, 24 h) gave only aryl iodide reduction. These disappointing results forced us to look for different coupling process which would circumvent aryl iodide cleavage during removal of the activating substituent. The next possibility was the Horner-Wadsworth- Emmons reaction.®5 Avoiding model studies, sequential treatment of 2 . 2 equivalents of dimethyl methyl- phosphonate with 2 equivalents of butyllithium (0°C, 15 nin) followed by 1 equivalent of ester 162 (0°C, 75 min) gave /3-ketophosphonate 180 in 99 X yield after an aqueous workup (Scheme XXXXI).®® /?-Ketophosphonate 180 was then treated with sodium hydride in DME (25°C, 1 h) followed by 3-iodo-4-methoxybenzaldehyde (85) (25°C, 6 h> to give a ,/?-unsaturated ketone 181 in 81 X yield. 74 Scheme XXXXI CH,P(0)(OMe) P(OMe) OMe OMe 162 180 1 .NaH, 2.85 OMe OMe 181 The next step in the ayntheaia required reduction of the unsaturated ketone moiety in 161 without touching the aryl iodides. As a matter of course, appropriate model studies were performed. Treatment of aldehyde 65 with 1- triphenylphosphoranylidene-2 -propanone (182)87 in dichloromethane (25°C, 48 h) gave a,/?-unsaturated ketone 183 in 8 6 % yield (Scheme XXXXII). The next objectives of this model study were the selective 1,2- or 1,4-reduction of the enone system. First, attempts were made to obtain 75 Scheme XXXXII CHO S o + Ph.P-%>A v OMe 182 85 OMe 183 Et_SiH 10 eq CF-COOH CF-COOH OH OH NaOH HN=NH OMe OMe OMe 185 187 186 the 1,4-reduction product 184. However, moat popular 1,4- reduction methods such as catalytic hydrogenations and dissolving metal redutions were not used since both might be expected to also reduce the aryl iodide moiety. Reduction of 183 with sodium borohydride in pyridine8 8 or copper(I) bromide with sodium bis-(2 -methoxyethoxy>- aluminum hydride (RED-AL)®® met with failure, giving poor regioeelectivity. It had been reported that treating an a,/3-unsaturated ketone with triethylsilane (1 equivalent) and trifluoroacetic acid (10 equivalents) would afford the saturated carbonyl compound.91 However, treatment of ketone 183 with 1 equivalent of triethylsilane and 10 equivalents of trifluoroacetic acid in chloroform gave a mixture of saturated ketone 184 and starting ketone 183 (reflux, 40 h ) . With 8 equivalents of triethylsilane and 11 equivalents of trifluoroacetic acid, ketone 183 was converted to trifluoroacetate 185 in 79 X yield (reflux, 2 h). This trifluoroacetate 185 was easily converted to saturated alcohol 187 upon treatment with 3 N aqueous sodium hydroxide in methanol in 96 X yield (25°C, 5 min). Direct conversion of ketone 183 to saturated alcohol 187 was also performed with 3 equivalents of triethylsilane and 16 equivalents of trifluoroacetic acid (reflux, 2 h) followed by basic workup with 3 N aqueous sodium hydroxide in 95 X y i e l d . ^2 Although this reduction was potentially suitable for use in the lythrancepine-II synthesis, the presence of a benzylic amine in the real system caused some worries. Thus, selective 1,2-reduction of unsaturated ketone 183 was also attempted. Treatment of ketone 183 with 1.2 equivalents of lithium triethylborohydride^® in 77 tetrahydrofuran gave unsaturated alcohol 186 quantitatively (0°C, 10 min). The unaaturated alcohol 186 was reduced to saturated alcohol 187 with diimide, generated from p-toluenesulfonylhydrazide and sodium acetate, in 87 X yield (reflux, 5 h).^3 other methods for the generation of diimide such as hydrazine hydrate with sodium metaperiodate and copper(II) sulfate^ or potassium azodicarboxylate with acetic acid$5 could not be adopted for the reduction of unaaturated alcohol 186. The former method gave alcohol 187 along with material in which the iodide hod also been reduced in about a 3:5 ratio. The latter method gave only a low yield of alcohol 187. Moreover, it was suspected that acidic conditions might hurt the amino moiety of actual compound 181. These results led us to use lithium triethylborohydride rather than triethylsilane. It was hoped that metal hydride reduction of ketone 181 might show some diastereoselectivity due to chelation between the metal (Li+), nitrogen lone pair of quinolizidine skeleton, and carbonyl group. Treatment of a ,/3- unsaturated ketone 181 with 2 equivalents of lithium triethylborohydride in a mixture of tetrahydrofuran and dichloromethane gave an easily separable mixture of allylic alcohols 188 (71 x') and 189 (25 X , 0°C, 30 min) as shown in Scheme XXXXIII. As expected, it was difficult to characterize the stereochemistry at C-ll in 188 and 78 Scheme XXXXIII OMe OMe OMe 181 OH 189 OMe + OMe OH HN=NH OMe 188 OMe OH OMe 190 189. An extremely tentative assignment, however, was made on the basis of TLC mobilities. The Rf values for 188, 181, and 189 (silica gel; ethyl acetate-hexane, 1 :1 ) were 0.57, 0.50, and 0.14, respectively. The unusually low polarity of alcohol 188 may be explained by intramolecular hydrogen bonding^ between the C-ll hydroxy group and nitrogen lone pair as shown in Figure 8 . Both alcohols 188 and 189 could adopt conformations H OMe OH OMe 188 P hC H p OMe OH OMe 189 OMe Figure 8 80 in which the hydroxyl and amino groups are intramolecularly hydrogen bonded. However, in 188, the C- 11 alkenyl sidechain is pseudo-equatorially disposed whereas in 189, the C-ll alkenyl sidechain is paeudo- axially disposed and experiences more severe steric interaction with the C-l aryl group. As a result, it was anticipated that 188 should have more intramolecular hydrogen bonding than 189 and therefore be less polar than 189. It is on this basis that stereochemistry was originally assigned. This reduction might also have been expected to give 188 as the major product based on a chelation controlled reduction. Perhaps the lithium coordinates with both the nitrogen lone-pair and a non bonded carbonyl oxygen lone-pair. This locks 181 into a conformation where the incoming triethylborohydride should potentially attack the re-face rather than the si- face which is relatively hindered by the H-5 axial proton. Although TLC analysis and spectroscopic arguments allowed us to make a tentative C-ll stereochemical assignment at this point, proof of stereochemistry waa obtained at the next stage of the synthesis. The major alcohol 188 was reduced with diimide generated from p-toluenesulfonylhydrazide and sodium acetate in aqueous DME to afford saturated alcohol 190 in 92 X yield (reflux, 4 h) . The IR spectrum of 190 like, 188, showed no sharp free hydroxyl stretching band and 8 1 only showed a broad hydrogen bonded hydroxyl stretching band around 3100 cm~l. The structure of alcohol 190 was firmly extablished by X-ray crystallographySl also showed hydrogen bonding between the C-ll hydroxyl group and nitrogen lone-pair (Figure 9). Figure 9 Alcohol 190 was acetylated with acetyl anhydride, triethylamine, and a catalytic amount of 4- dimethylaminopyridine (25°C, 3 days) to give acetate 121 in 85 % yield. With the entire C-9 sidechain in place, we 82 were set for the critical biaryl construction. Synthetic procedures for the preparation of biaryls by the classical Ullmann reaction have, in recent years, been supplanted by the use of zerovalent nickel to effect the reductive coupling of aryl halides under homogeneous conditions. The original method, discovered by Semmelhack and his coworkers, used the isolable albeit air-sensitive bis<1 ,5 -cyclooctadiene)nickel<0 ).^7 Their studies were extended to intramolecular biaryl coupling especially using tetrakis(triphenylphosphine)nickel(0) (191) under high dilution conditions.36 In addition, it had been reported that this method had been used to prepare the biphenyl unit in a model study for the synthesis of the alkaloid protostephanine.59 This result suggested that a tertiary amine might not interfere with such a coupling. Fresh 191 was prepared using a classical method from anhydrous bis(2,4-pentanedionato)nickel, triphenyl phosphine, and triethylaluminum and stored in a dry box (Equation 8 ).100 This reddish brown solid 191 could be Ni(CHjCOCH COCH3)2 + 2Et3AI + 4Ph3P (Ph3p)«Ni + 2(C5H70 2 )AIEt2 + 2 C2H, (eq. 8) 191 S3 stored for a long time in a fairly well equipped dry box but became inactive upon exposure to oxygen. Treatment of 121 with 1.5 equivalents of 191 in DMF (S mM concentration of 121) at 55°C for 48 h gave the long- awaited biaryl 192 in 20-25 X yield along with 12 X of reduction product 193 (Scheme XXXXIV). The structure 192 Schem e XXXXIV H OMe OH OMe 190 PhCHp OMe OAc OMe 121 191 PhCH20 OAc OMe OAc Mei eO OMe 192 193 84 was apparent from spectral data. In particular, the aromatic region was identical to the patterns reported for type D Lythraceae alkaloids. Although this reaction gave poor mass balance, no intensive attempt was made to optimize the yield of biaryl 192. All that remained to complete the synthesis was removal of the benzyl group from the C-3 oxygen. Cleavage of this benzyl ether by hydrogenolysis was anticipated to be simple. However, initial attempts to cleave benzyl ether 192 with 10 X palladium on charcoal in 95 x aqueous ethanol gave no reaction. A variety of conditions were tried with no success. The reason why 192 was resistant to hydrogenolysis in the presence of palladium on charcoal is still uncertain. This surprising result induced us to perform some model studies (Scheme XXXXV). First, treatment of benzyl ether 162 with 10 X palladium on charcoal in ethyl acetate <75 psi, 24 h) gave no hydrogenolysis product, only benzyl ether 194 was obtained confirming that palladium was not a good choice of catalyst. Treatment of benzyl ether 162 with W-4 Ra-Ni in ethanol1 0 1 '1 0 2 gave 61 X of piperidine 195 <1 atm, 48 h), the product of benzyl ether and benzylic C-N bond cleavage.1 0 3 We next moved to non-catalytic methods using a slightly different model system. Some methods such as dissolving metals (Na/NHg), DDQ, lithium aluminum 85 Scheme XXXXV Ph CHjO C 0 2Et PhCH 20 OMe Pd/C 194 CO,Et H H W-4 OMe Ra-Ni 162 COzEt OMe 195 hydride, electrolytic reduction or oxidation, free radical bromination-hydrolyeis, or oxidation-hydrolysis were felt to be unusable for cleavage of benzyl ether 192 due to the presence of sensitive functional groups in 192.104 a new model compound was prepared as shown in Scheme XXXXVI. Alcohols 5 3 and 5 4 (mainly 5 4 ) were converted to benzyl ether 196 with benzyl bromide and sodium hydride in DMF <25°C, 6 h) in 48 % yield after separation by chromatography. Treatment of lactam 196 with lithium aluminum hydride gave amine 197 in 94 x 8 6 Scheme XX XXVI HO NaH O Me OMe 53 HO-[3R] 196 54 HO-[3S] LAH HO BBr OMe OMe 198 197 yield (reflux, 1 h). A variety of benzyl ether cleavage conditions such as catalytic transfer hydrogenations <10 X palladium on charcoal with 1,4-cyclohexadiene in ethanol at 25°Cl05 or io X palladium on charcoal with cyclohexene in ethanol under reflux^O^ ) , boron trifluoride etherate with ethanethiol in dichloromethane (25°C, 24 h),107 triraethylsilyl chloride with sodium iodide in acetonitrile <50°C, 12 h>,108 and trimethylsilyl iodide in chloroform <25°C, 24 h>109 met with failure giving sluggish reactions at best. At long last, cleavage of benzyl ether 197 was accomplished using 3 equivalents of boron tribromide in dichloromethane <0°C, 15 min) giving 198 in 65 * yield.We were still concerned that treatment of benzyl ether 192 with boron tribromide might destroy the acetoxy group because it had been reported that boron trihalides will cleave carboxylic esters to the corresponding alkyl halides.m Nonetheless, treatment of acetate 121 with 2 equivalents of boron tribromide in dichloromethane at 0°C for 3 min followed by an aqueous workup showed that cleavage of the benzyl ether might be faster than reaction of the acetate. This result encouraged us to attempt the cleavage of benzyl ether 192 using similar conditions. Thus, sequential treatment of benzyl ether 192 with 4 equivalents of boron tribromide in dichloromethane at 0°C for 3 min followed by aqueous workup gave <+)- lythrancepine-II, in 54 yield (Scheme XXXXVII). Unlike the reaction conditions for benzyl ethers 197 and 1 2 1 , excess boron tribromide was needed in this case, presumably because oxygens could coordinate with boron tribromide. Even though the yield of this reaction was not extremely high, no side products were isolated and TLC analysis of the reaction mixture showed only a single spot due to <+)-4. The synthetic (+)-lythrancepine-II (4) was identical to an authentic sample of < + )- Scheme XXX XVII PhCH,0. HO OAc BBr. OAc eO leO 192 Py AcO OAc Mei eO 14 lythrancepine-II kindly provided by Professor Fujitall2 (TLC, FT-IR, mass spectrum, 500 MHz ^H-NMR). Further proof of structure was obtained by acetylation of ( + )-4 with acetic anhydride in pyridine <25°C, & h, 64 X) to give FT-IR, mass spectrum, 500 MHz iH-NMR). A J-Resolved 2D ^H-NMR spectrum11^ was taken for the <+)-l4 and consequently, most of the chemical shifts and coupling constants of (0-4 and <0-14 were Identified. These spectra as well as those of natural <+)-lythrancepine-II and -III are shown in Figures 10 and 11 respectively. In conclusion, we have completed the first total synthesis of a member of the largest structural family of Lythraceae alkaloids. The synthesis required 18 steps from anisaldehyde and proceeds in approximately 1 * overall yield. The synthetic scheme allowed us to synthesize all three members of lythrancepine alkaloids as shown in Figure 1 (Chapter I). Furthermore, the basic strategy should allow the synthesis of all other members of this family of Lythraceae alkaloids. h 2o r ,• » ...-..-p., r~i T- T T • • r * • a . :i l.n r.n l.C» u.o 8.0 6.0 I «-M Figure 10a. 1H NMR Spectrum of (+)-Lythrancepine-II (4) (CDCl3, 500 MHz) HO, OAc Mei eO HzO t 1 -’T-’ T* •*~1 ’ 1 T’ 8.0 6.0 5 .0 4.U 3. U 2 .0 1. 0 0.0 PPM CO Figure 10b. H NMR Spectrum of (+)-Lythrancepine-II (4) (CDCl^, 500 MHz) MtATI LTIHRPfltEiMNi Figure 11a. NMR Spectrum of (+)-Lythrancepine-III (+)-Lythrancepine-III of Spectrum NMR 11a. Figure 1 7.0 I \ AcO eO OAc r- *- r *t- T T 't ifti tl;.!; ?.o (14) HzO CC^ 50 MHz) 500(CDCl^, l.U o.n >r T PPH Figure lib. NMR Spectrum of (+)-Lythrancepine-III (+)-Lythrancepine-III of Spectrum NMR lib. Figure oJ uJ ii Jk lu Jo Jo k r W u .0 7 AcO 6.0 eO l j OAc A .0 5 . -j--.-. 4.0 HHK .0 3 (14) ■ I" ■'“ 2.0 CC^ 500MHz) (CDCl^, .^iniu<)|f PPM V. EXPERIMENTAL All melting points were taken with a Thomas-Hoover capillary melting point apparatus and are uncorrected. nuclear magnetic resonance spectra were recorded on Varian Associates EM-360 <60 MHz), Brucker NR-80 (80 MHz), Varian Associates EM-390 <90 MHz), Bruker WP-200 (200 MHz), or Bruker AM-500 <500 MHz) spectrometers and are reported in parts per million from internal tetramethylsilane on the 6 scale. Data are reported as follows: chemical shift [multiplicity CHCHC0Me=ArHs or ArHs', CHCHC0Me=ArH& or ArHg'. 13c nuclear magnetic resonance spectra were recorded on a Bruker WP-80 <20.11 MHz), Bruker WP-200 <50.28 MHz), Nicolet NT-300 <75.42 MHz), or Bruker AM-500 <125.69 MHz) spectrometers and are reported in parts per million from internal tetramethylsilane. NMR data are reported as 95 follows: chemical shift, [multiplicity (s=singlet, d=doublet, t=triplet, q=quartet), interpretation]. Host 13c NMR spectra were recorded as Broad-Band or DEPT (Distortionless Enhancement by Polarization Transfer) spectra. Thiophene 155 is the only compound whose multiplicities were determined via off-resonance techniques. Infrared spectra were taken with Perkin-Elmer 457, Perkin-Elmer 263B, or Mattson Cygnus 25 FT-IR instruments. Mass spectra were recorded on AEI-MS9, Kratos DS-55, or Kratos MS-30 instruments at an ionization energy of 70 eV. Samples on which exact masses were measured exhibited no significant peaks at m/e values greater than those of the parent. The parent ions of some amines and a few other compounds were too small for exact mass measurements to be obtained and the parent ions of some compounds which have molecular weights of greater than 700 were beyond calibration to get exact masses. In these cases, the fragmentation patterns were in accord with the assigned structures. Laser desorption Fourier transform mass spectrometry was performed for 161 using a Nicolet FT/MS-1000 and Tachisto 215 G TEA CO2 laser doped with KBr to supply K+ without calibration. Combustion analyses were performed by Micro-Analysis, Inc., Wilmington, DE. Solvents and reagents were dried and purified prior to use when deemed necessary: acetic anhydride (distilled 97 from P2055; tetrahydrofuran and diethyl ether (distilled from sodium metal); dichloromethane and chloroform (passed through activity I alumina or distilled from CaH2 >; benzene, dimethoxyethane, dimethylformamide, toluene, and triethylamine (distilled from CaH2 >; methanol (distilled from magnesium methoxide), dimethyl sulfoxide (distilled from CaH2 at reduced pressure). All reaction temperatures refer to those of the reaction mixture unless indicated otherwise. Reactions requiring an inert atmosphere were run under a blanket of argon or nitrogen. Formic acid (98 X) was used in all cyclizations. 61 X of Oil dispersion of sodium hydride was used without washing it. Most reactions were followed by thin layer chromatography over silica gel or alumina using EM Laboratories glass-backed 0.25-mm thick precoated silica gel 60 F-254 plates or EM Laboratories glass-backed 0.25-mm thick precoated aluminum oxide 60 F- 254 plates. Column chromatography was performed over EM Laboratories silica gel 60 (70-230 mesh) or Woelm neutral alumina. Preparative thin layer chromatography was performed over EM Laboratories glass-backed 2.0-mm thick precoated silica gel 60 F-254 plates or EM Laboratories glass-backed 0.5-mm thick precoated silica gel 60 F-254 plates. Medium pressure liquid chromatography (MPLC) was performed over EM Laboratories Lobar columns using an FMI RPSY lab pump. Rotary diak chromatography (chromatotron) waa performed over EH Laboratoriea allica gel PF-254 with CaSC>4 1/2 H2 O type 60 platea uaing an FMI RP G-150 lab pump. Ethyl acetate and n-hexane, uaed aa eluenta in column chromatography, were diatilled prior to uae. Order of experimental liata generally followed their echeme aa well aa numbera of compounda. 4-A»ino-4-(4-methoxyphenyl)-l-butene <43). To a aolution of 10.1 mL (47.7 mmol) of 1,1,1,3,3,3,-hexamethyldieilazane in 12 mL of tetrahydrofuran waa added 28.3 mL <43.9 mmol) of 1.55 M n-butyllithium in hexane with cooling in an ice bath. The aolution waa atirred for 1 0 min and 5.0 g <36.7 mmol) of p-methoxy- benzaldehyde waa added via ayringe with cooling in an ice bath. The mixture waa atirred at room temperature for 1 h followed by addition of 61.5 mL <44.9 mmol) of 0.73 M ethereal allylmagneaium bromide with cooling in an ice bath. The resulting mixture waa stirred for 20 min at room temperature, poured carefully into 100 mL of saturated aqueous ammonium chloride, and extracted with three 100-mL portions of diethyl ether. The combined organic layers were washed with 100 mL of saturated aqueous sodium chloride and 100 mL of water, dried (Na2 S0 4 >, and concentrated in vacuo. The residue was distilled to give 5.63 g (87 X) of amine 43 as a colorless oil: bp 84-85QC at 0.5 mrnHg; IR 1608 cm-1; NMR (CDClg) 6 1.50 2H, CH2 CH=>, 3.80 ArCHN), 4.95-5.25 (m, 2H, =CH2 >, 5.50-6.00 (m, 1H, =CH>, 6.85 (d, J=9 Hz, 2H, ArH), 7.28 CH2CH=CH2» 100), 121 (47), 93 (8 ); exact mass calcd for CsHiqNO (Mi‘-CH2 CH=CH2 ) ffi/e 136.0763, found m/e 136.0762. N-[1-(4-Methoxyphenyl)but-3-en-l-yl3-5,5-dimethoxy- pentanamide (44). To a solution of 5.93 g (33.5 mmol) of amine 43 in 40 mL of dichloromethane at room OMe temperature was added 27.0 mL (40.2 mmol) of OMe 1.49 M trimethyl- OMe aluminum in heptane under nitrogen. The 44 solution was stirred for 30 min followed by addition of 5.90 g (33.5 mmol) of methyl 5,5-dimethoxypentanoate (22) in 5 mL of dichloromethane. The resulting solution was warmed under reflux for 2 0 h, cooled to room temperature, poured into 50 mL of 1 N aqueous sodium hydroxide, and extracted with five 80-mL portions of dichloromethane. The organic layers were washed with three 100-mL portions of saturated aqueous sodium chloride, dried (MgS0 4 ), and concentrated in vacuo. The residual pale yellow solid was chromatographed over 150 g of silica gel (eluted with ethyl acetate-hexane, 7:3) to give 10.5 g <98 X) of amide 44 as a pale yellow oil which solidified upon standing for several weeks in the refrigerator: mp 50-51°C; IR CH2 CH2 C(OMe)2 > » 2 .1 0 -2 . 3 5 Hr, 2 H, CH2 C-), 3.30 ArOCH3 >, 4.25-4.40 =CH2 and ArCHN), 5.45-6.05 (m, 2H, =CH and NH), 6.82 J =9 Hr, 2H, ArH), 7.20 (3), 280 (8 ), 249 (7), 248 (42), 176 (5), 145 (10), 136 (100), 71 (59); exact mass calcd for C1 7 H2 4 NO3 rel-<4R,9aS)-4-<4-Methoxyphenyl)-1,6,7,8,9,9a- hexahydrophenyl-4H-quinolirin-6-one (49), rel-(2R,4R,9a5)- Formyloxy-4-<4-methoxyphenyl)octahydro-4H-quinolirin-6- 101 o n m (39), and rel-(2S,4R,9aS)-2-Formyloxy-4-(4-*ethoxy- phenyl)octahydro-4H-quinolizin-6-one (50). To a solution H H M HOCO,, HOCO OMe OMe OMe 49 39 50 of 512 mg (1.60 mmol) of amide 44 in 6 mL of dichloromethane at room temperature was added 16 mL of 98 X formic acid. The solution was stirred at room temperature for 3 h, neutralized with saturated aqueous sodium bicarbonate, extracted with three 50-mL portions of dichloromethane, washed with three 50-mL portions of saturated aqueous sodium bicarbonate, dried (MgS0 4 >, and concentrated in vacuo. The resulting mixture was chromatographed over a Lobar size B column (eluted with ethyl acetate-hexane, l:l> to give 56 mg (14 X) of a mixture of isomeric olefins 49, 319 mg (66 X) of formate 39, and 16 mg (3.3 X) of isomeric formate 50 as colorless oils. Isomeric olefins 49: IR (CH2 CI2 ) 1650 cm“l; NMR (CCI4 ) S 1.35-2.65 (m, 8 H, CH2 ), 3.30-3.80 (m with s at 3.72, 4H, OCH3 and NCH), 5.70-5.98 (m, 2H, CH=CH), 6.00- 102 6.22 (broad a, 1H, ArCHN), 6.68 (d, J=9 Hz, 2H, ArH), 7.25 (d, J=9 Hz, 2H, ArH); mass spectrum, m/e (relative intensity) 257 (M+, 100), 242 (5), 228 (4), 160 (27); exact mass calcd for C 1 6 H1 9 NO2 m/e 257.1416, found m/e 257.1423. Formate 39: IR (CH2 CI2 ) 1719, 1633 cm"!; NMR (CCI4 ) 8 1.20-2.90 (m, 10H, CH2 ), 3.15-3.55 (m, 1H, NCH), 3.75 (s, 3H, OCH3 ), 4.85-5.45 (broad m, 1H, CO2 CH), 6.23 (broad d, J=4.5 Hz, 1H, ArCHN), 6.85 (d, J=9 Hz, 2H, ArH), 7.20 (d, J=9 Hz, 2H, ArH), 8.03 (s, 1H, HCO2 ); mass spectrum, m/e (relative intensity) 303 (M+, 100), 302 (12), 288 (1), 274 (1), 258 (8 ), 256 (23), 161 (10); exact mass calcd for C 1 7 H2 1 NO4 m/e 303.1470, found m/e 303.1477. Isomeric formate 50: NMR (CCI4 ) 8 1.50-2.70 (m, 10H, CH2 ). 3.70 (s, 3H, OCH3 ), 3.55-3.85 (m, 1H, NCH), 5.13 (qu, J=3.5 Hz, 1H, CO2 CH), 5.85 (dd, 1H, ArCHN), 6.65 (d, J=9 Hz, 2H, ArH), 6.92 (d, J=9 Hz, 2H, ArH), 7.50 (s, 1H, HCO2 ); mass spectrum, m/e (relative intensity) 303 (M*, 100), 302 (13), 288 (3), 274 (8 ), 258 (15), 256 (22), 161 (16); exact mass calcd for C3.7 H2 1 NO4 m/e 303.1470, found m/e 303.1477. rel-(2R,4R,9aS)-2-Hydroxy-4-(4-methoxyphenyl)- octahydro-4H-quinolizin-6-one (53). To a solution of 299 103 mg CO.958 mmol) of formate 39 in 7 H mL of methanol at room temperature was added 0.95 mL of 3 N aqueous sodium hydroxide. The mixture was stirred at room temperature for 1 OMe h, poured into 30 mL of saturated 53 aqueous sodium bicarbonate, and extracted with three 30-mL portions of dichloromethane. The organic extract was washed with three 50-mL portions of brine, dried (MgS0 4 ), and concentrated in vacuo. The residual solid was recrystallized from ethyl acetate-hexane to give 254 mg (94 *) of alcohol 53 as a white solid: mp 131°C; IR (CHCI3 ) 3592 (sharp), 3400 (broad), 1627 cm“l; NMR (CDCI3 ) 6 1.25-2.90 (m with s at 2.05, 11H, CH2 and OH), 3.10-3.60 (broad m, 1H, 0CH), 3.70-4.20 (broad m with s at 3.80, 4H, NCH and OCH3 ), 6.18 (broad d, J=4 Hz, 1H, ArCHN), 6.85 (d, J=9 Hz, 2H, ArH), 7.00 (d, J=9 Hz, 2H, ArH); mass spectrum, m/e (relative intensity) 275 (M*, 100), 274 (51), 257 (5), 245 (4), 161 (25), 150 (43), 135 (45); exact mass calcd for C 1 6 H2 1 NO3 m/e 275.1521, found m/e 275.1530. rel-<4R,9aS>-4-(4-Methoxyphenyl)octahydro-4H- quinolizin-2,6-dione (55). To a solution of 200 mg (0.73 104 mmol) of alcohol 53 in 5 mL of H acetone, cooled in an ice bath O was added 0.33 mL (0.66 mmol) of Jones reagent <2.66 M) . The mixture was stirred for 2 0 min with cooling in an ice bath OMe followed by dilution with 20 mL 55 of dichloromethane. The organic layer was washed with four 50-mL portions of water, dried 147 mg (74 X) of ketone 55 as a colorless oil: IR CH2 >, 3.20-3.80 7.08 (d, J=9 Hz, 2H, ArH); mass spectrum, m/e (relative intensity) 273 149 (24), 134 (30); exact mass calcd for C 1 6 H1 9 NO3 m/e 273.1365, found m/e 273.1372. rel-<2S,4R,9aS)-2-Hydroxy-4-(4-methoxyphenyl)- octahydro-4H-quinolizin-6-one (54), rel-<2S,4R,9aS)-2- Acetoxy-4-<4-*ethoxyphenyl)octahydro-4H-quinolizin-6-one (56), and rel-(2R,4R,9aS)-2-Acetoxy-4-(4-*ethoxyphenyl>- 105 octahydro-4H-qu±nolizin-6-one (57). Method A. To a HO AcO OMe OMe OMe 54 56 57 solution of 1.36 g <4.97 nmol) of ketone 55 in 25 mL of tetrahydrofuran was added 6 mL <6.0 mmol) of 1 M lithium triethylborohydride in tetrahydrofuran with cooling in a dry ice bath under nitrogen. The mixture was stirred for 1 h with cooling in a dry ice bath, hydrolyzed with 1.5 mL of water at room temperature, and oxidized with 2 mL of 30 H hydrogen peroxide followed by stirring at room temperature for 3 h. The aqueous layer was saturated with potassium carbonate and the organic layer was concentrated in vacuo to give 1.36 g <99 *i) of an inseparable mixture of alcohol 54 and isomeric alcohol 53 as a white solid. To a solution of 1.36 g <4.95 mmol) of alcohols 53 and 54 in 5 mL of dichloromethane was added 0.93 mL <9.S4 mmol) of acetic anhydride, 0.69 mL <4.95 mmol) of triethylamine and a catalytic amount of 4- dimethylaminopyridine. The mixture was atirred vigorously 106 at. room temperature for 5 h, diluted with 100 mL of dichloromethane, washed with three 30-mL portions of aqueous saturated sodium bicarbonate and two 50-mL portions of 1 N aqueous hydrochloric acid, dried (MgS0 4 ), and concentrated in vacuo. The residual pale yellow solid was chromatographed over 30 g of silica gel (eluted with ethyl acetate) to give 1.23 g (79 X) of acetate 56 as a white solid and 0.11 g (8.3 X) of isomeric acetate 57 as a pale yellow oil. Method B. To a solution of 205 mg (0.75 mmol) of ketone 55 in 4 mL of tetrahydrofuran was added 0.95 mL (0.95 mmol) of 1 M potassium tri-sec-butylborohydride in tetrahydrofuran with cooling in a dry ice bath under nitrogen. The mixture was stirred for 1 h with cooling in a dry ice bath, hydrolyzed with 0.2 mL of water at room temperature, and oxidized with 0.4 mL of 30 X hydrogen peroxide followed by stirring at room temperature for 2 h. The aqueous layer was saturated with potassium carbonate and the organic layer was concentrated in vacuo to give 203 mg (98 X) of mixture of alcohol 54 and isomeric alcohol 53 as a white solid. To a solution of 200 mg (0.73 mmol) of alcohols 53 and 54 in 2 mL of dichloromethane was added 1.5 mL (15.9 nmol) of acetic anhydride, 1.0 mL (7.17 mmol) of triethylamine and a catalytic amount of 4-dimethylamino- pyridine. The mixture was stirred vigorously at room 107 temperature for 1 h, and concentrated In vacuo. The reaidue was diluted with 20 mL of dichloromethane, washed with three 20-mL portions of saturated aqueous sodium bicarbonate and two 20-mL portions of 1 N aqueous hydrochloric acid, dried (MgS0 4 >, and concentrated in vacuo. The residual pale yellow solid mixture was chromatographed over a Lobar size B column (eluted with ethyl acetate) to give 167 mg <72 ss) of acetate 56 and 25 mg <11 *) of isomeric acetate 57. Acetate 56: mp 74.5°C; IR (CH2C12) 1732, 1610 cm'l; , 2.30-2.55 , 3.85