SIMONSEN LECTURE* Synthetic Approaches to Vitamin D and its Relatives By B. Lythgoe DEPARTMENT OF ORGANIC CHEMISTRY, THE UNIVERSITY, LEEDS LS2 9JT 1 Introduction The natural vitamin D3, required for normal bone calcification, is formed in man by the action of ultraviolet light on 7-dehydrocholesterol present in the skin (Scheme 1). Reversible photochemical cleavage of the 9,lO-bond first gives precalciferola (1); a reversible thermal hydride shift then gives the vitamin (2). A third 9,10-seco-sterol, tachysterols (3), is formed by the photochemical cis ---t trans isomerization (also reversible) of the central double bond in precalciferols. The technical preparation of vitamin D3 uses the route of Scheme 1;and of course, if combined with a synthesis of 7-dehydrocholesterol, this route also affords a total synthesis of the vitamin.However, a more direct synthetic route can be imagined, in which a mono- cyclic fragment representing ring A and an appropriate hydrindane derivative representing rings c and D are united so as to generate the conjugated triene system of the vitamin. Experiments on this, the A --t CD approach, were started in Manchester (Burkhardtl) and in Gottingen (Dimroth2) shortly after the gross structure of vitamin D2 was elucidated, and before the stereochemical details became clear; these initiatives did not, however, reach a successful conclusion. I was at that time a research worker in Manchester and, no doubt, a latent interest in the problem was then implanted which eventually led me into active participation.The present lecture describes experiments on the total synthesis of the vitamins D2 and D3 and some important relatives. 2 Early Photochemical Routes to Vitamins D It may be useful to summarize here developments in the field up to about 1960. The hydrindanes (4)3 and (6),4 obtained by the oxidation of vitamin D2 (Scheme 2), and the corresponding compounds from vitamin D3, offer readily available starting materials for experiments on partial synthesis. The pioneer workerslt2 envisaged a synthesis of vitamin D2 from the epimeric dienones (S), obtained by interaction of the ap-unsaturated aldehyde (6) with 4-hydroxycyclohexanone. *First delivered at the Scientific Societies’ Lecture Theatre, London W1, on 18th Jan 1979.J. B. Aldersley and G. N. Burkhardt, J. Chem. SOC.,1938, 545; J. B. Aldersley, G. N. Burkhardt, A. E. Gillam, and N. C. Hindley, ibid., 1940, 10.* K. Dimroth, Ber., 1938, 71, 1333, 1346. a A. Windaus and W Grundmann, Annalen, 1936,524,295. I. M. Heilbron, R. N. Jones, K. M. Samant, and F. S. Spring, J. Chem. SOC.,1936, 905. Synthetic Approaches to Vitamin D and its Relatives 17 (1) 7-Dehydrocholesterol PrecaIciferol, 17 I aOHHO'l (2) (3) Vitamin D, Tachysterol, Scheme 1 Note: All structures in this and subsequent Schemes, in which stereochemical detail is shown by heavy and/or dotted lines, represent absolute configurations R I I HO'l Vitamin D, 0 Scheme 2 That, however, required the regiospecific conversion of the keto-group into an exocyclic methylidene group, a reaction which only became possible with the Lythgoe advent of Wittig’s olefin synthesis.This reaction was then employed indepen- dently in Leeds and in Braunschweig to effect partial syntheses of vitamins D (Scheme 3). We in Leedss isomerized the c27 epimer mixture (8) to the 52-hv+ Vitamin D, 17 CHI= PPh, -Vitamin D:, Scheme 3 epimers (9) before replacing the keto-group by a methylidene group and separat- ing the epimers to give vitamin Dz. Inhoffen’s group6 separated the 3p-epimer (10) of the cZ6 dienones and converted it in a Wittig reaction into 5E-vitamin D3 (1 l), which was then isomerized photochemically to give vitamin D3.Inhoffen’s group further effected a synthesis, starting from Hagemann’s ester [(61) p. 4621, of the $-unsaturated aldehyde (7) which had been used to prepare the dienone (10). They thus completed7 the first total synthesis of a vitamin D by the A -CD approach. However, the magnitude of the overall yield, ca. %, made it clear I. T. Harrison and B. Lythgoe, Proc. Chem. Soc., 1957, 261 ;J. Chem. SOC.,1958, 837. H. H. Inhoffen, K. Irmscher, H. Hirschfeld, U. Stache, and A. Kreutzer, Chern.Ber., 1958, 91,2309, H. H. Inhoffen, H. Burkhardt, and G. Quinkert, Chem. Ber., 1959,92,1564, and refs. there cited. 45 1 Synthetic Approaches to Vitamin D and its Relatives that further studies were needed in both stages of the work. For example, in the sequence (7) -(10) -(1 1) -(2) the configuration at C-3 was established non- stereoselectively, and the mode of establishment of the 52-geometry was not efficient.3 Synthesisof Tachysterols The next objective to be attained was a partial synthesis of tachysterols (3). The presence in this compound of the central trans-disubstituted double bond made it probable that the synthesis could be effected by a Wittig reaction between appropriate CSand CU fragments. These fragments correspond to the allylic alcohols (16) and (19). Scheme 4 outlines a route8 to the monocyclic alcohol (16). Reduction of the HO C0,Me Reagents: i, Li-NH,; ii, H,O+; iii, resolve; iv, NaBH,; H,O+; v, NaOMe-MeOH; vi, LiAlH,(OEt) Scheme 4 Ring A fragment for tachysterol, aromatic acid (12) with lithium and ammonia, followed by acid hydrolysis, gave the keto-acid rac-(l3). The required enantiomer (13) was obtained by resolution with quinine.’ Reduction of the keto-group gave the y-lactone (14), which on treatment with methanolic sodium methoxide yielded the conjugated ester (15).Reduction then gave the alcohol (16). The bicyclic alcohol (19) can, on paper, be derived simply from cholesterol by cleavages at the 9,lO-and 6,7-positions, but to realise this plan in practice was less easy. Our first attemptQ used as an intermediate duoannelic acid (17), which can be obtained from cholesterol in low yield by direct oxidation, or in better yield by an indirect method. Its conversion into 8-hydroxymethyl-des-~~-cholest-8-ene (19), using standard methods of alicyclic chemistry, is shown in Scheme 5.Tt was found that treatment of the conjugated ester (18) with alkali converted it partly * J. Dixon,B. Lythgoe, I. A. Siddiqui, and J. Tideswell, J. Chem. Soc.(C), 1971, 1301. * R. S. Davidson, W. H. H. Gunther, S. M. Waddington-Feather, and B. Lythgoe, J. Chem. SOC.,1964, 4907. Lythgoe into the dg(ll)-isomer, illustrating the fact that, owing to conformational factors, the 8,g-double bond occupies a position of high energy. . .. ...&H17 1. 11. 111, Ho&H17 iv,v, iii-___) 0 IH IA C02H C02Me C02Me 4,vi, i I kH20H C02Me Reagents: i, CH,N,; ii, CF,CO,H; iii, NaOMe-MeOH; iv, PhS0,Cl-pyridine; v, Me,NCHO-H,O; vi, KOH-EtOH; vii, LiAlH,(OEt) Scheme 5 Later, a more efficient degradative route was found,1° using as an intermediate the ‘Westphalen’ methoxy-ketone (20) (Scheme 6).Acetoxylation at C-7 opened the way to an oxidative cleavage of the 6,7-bond; the 9,lO-double bond was then ozonized, giving the bicyclic keto-benzoate (21). Normal manipulation of the functional groups in this compound gave the &unsaturated aldehyde (22), which can be obtained from, or converted into, the allylic alcohol (19). The latter was so obtained from cholesterol in an overall yield of 22.5%, and became relatively easily available. In principle, either of the two alcohols (16) and (19) can be used to make the phosphorane component for the Wittig reaction, leaving the other to be con-verted into the aldehyde component.We chose the alcohol (16) for conversion into the phosphorane (23) since this avoided the need for protecting the hydroxy- group. Union of the two components as shown in Scheme 7 gave tachysterols, isolated in good yield as the crystalline 4-methyl-3,5-dinitrobenzoate.11 When Wittig reactions with an allylic phosphorane are used to construct internal disubstituted olefins, a mixture of cis-and trans-isomers is normally obtained. Here, however, the reaction was trans-stereospecific; no precalciferol3 lo P. J. Kocienski, B. Lythgoe, and D. A. Roberts,J. Chem. Soc., Perkin Trans.I, 1980, 897. R. S. Davidson, S. M. Waddington-Feather, D. H. Williams, and B. Lythgoe, J. Chem. SOC.,1967, 2534.453 Synthetic Approaches to Vitamin D and its Relatives Cholesterol M eO M eO OAc 0 0 (20) ii, iii. ii, iv 17 fH CH ,OCOPh Me0 OCOPh‘1”)$58H17-pSH”vii, viii, ix Ts0~c8H’7~ HO I CH 20C0Ph eHO CHO (22) Reagents: i, trimethylsilyl enol ether treated Pb(OAc),-Et,NHF; ii, LiAlH4; iii, Pb(OAc),; iv, PhCOCI-pyridine; v, 03,Me2S;vi, LiAIH(OBut,); vii, MeC,H4S02CI-pyridine; viii, KOH-H20; ix, (COCI),-Me2S0, NEt,; x, KF-NaOAc-Me2S0 Scheme 6 Route to 8-formyl-des-AB-cholest-8-ene was detected. Two factors seem responsible.12 Firstly, in such reactions the initial betaine formation is readily reversible. Secondly, owing to the cyclic disposition of the outer double bonds, the product is branched at both positions adjacent to the new (central) double bond.This sets up adverse steric interactions in the transition state leading to the cis-isomer, so that the reaction is steered towards formation of the trans-isomer. 4 Precalciferol3 The dominating feature of precalciferola (1) is the cis-geometry of the central l2 P. J. Kocienski, B. Lythgoe, and I. Waterhouse, J. Chem. SOC.,Perkin Trans. I, 1980, 1045. Lythgoe CH20H (19) CHzOH I iii, iv QOH nOH (3) Tachysterol, Reagents: i, Mn0,-Et,O; ii, Bui,AlH ;iii, PhCH,P(OPh),Br-PPh, ;iv, 2BuLi Scheme 7 Synthesis of tachysterol, double bond. We proposed to secure this feature by using as an intermediate the acetylenic compound (33), constructed by a union at the 7,8-position. It was important for success that the double bond positions in the enynene (33) should be unambiguous. This was achieved for the ring A part by using as a starting material the optically active enyne (24), prepared13 as shown in Scheme 8 by a one-carbon homologation of the allylic alcohol (16).HO CH,OH HO CHO HO CH :CHCI Reagents: i, Mn0,-Me,CO; ii, CICH: PPh,; iii, NaNH,-NH,; ivy Me,SiCl-pyridine- (Me,Si),NH; v, BuLi Scheme 8 Ring A fragment for precalciferol, l9 T. M. Dawson, J. Dixon, P. S. Littlewood, and B. Lythgoe, J. Chem. SOC.(C), 1971,2352. Synthetic Approaches to Vitamin D and its Relatives The double bond in the ring c part of (33) occupies the high energy 8,g-position. To secure this we used as the hydrindane component 9a-chloro-des-m- cholestan-8-one (30), prepared initiallyl4 from des-~~-cholestan-8/%01 (26) by way of des-~~-cholest-8-ene(27), the 8/3,9ar-diol(28), and the 8&9@epoxide (29), as shown in Scheme 9.I+ -FEHOH H vii vi t---HO" Reagents: i, LiAlH,; ii, 00,-pyridine; iii, heat benzoate to 360°C; iv, ClC,H,CO,H; v, HCI0,-H,O-Me,CO ;vi, KOH-MeOH on the 9 a-monotosylate; vii, HCI-dioxan; viii, Na,Cr,O,-H,SO,-H,O-Et,O Scheme 9 cpfragment for precalcfero13 The two components were then unitedl5 '(see Scheme 10) by reaction of the protected lithium derivative (25) of the enyne (24) with the chloro-ketone (30) to give, after de-protection, the vicinal chlorohydrin (3 1). The elimination of the l4 P. S. Littlewood, B. Lythgoe, and A. K. Saksena, J.Chem. SOC.(C), 1971, 2955. l6 T. M. Dawson, J. Dixon, P.S. Littlewood, B. Lythgoe, and A. K. Saksena, J. Chem. SOC. (C),1971,2960. Lythgoe X @0 -ii Li I 111 iii (1) Precalciferol, Reagents: i, reaction product treated H,O+; ii, compound (3 1) treated with bis(ethy1enedia- mine)chromium(r~~-Me,NCHO;iii, 1 H,-Lindlar Pd Scheme 10 Synthesis of precalcifero13 elements of hypochlorous acid to generate the 8,g-double bond was effected by Kochi’slG method. The resulting enynene (33) was semihydrogenated over Lindlar’s catalyst to give precalciferols, isolated as the 3,5-dinitrobenzoate in ca. 21 % yield from the chloro-ketone (30). Thermal equilibration of the synthetic material provided vitamin D3; this was the first occasion on which it had been obtained without the intervention of photochemical methods.At first sight it may seem that the above result could be more simply achieved by using in place of the chloro-ketone (30) the more readily available ketonel7 (5). This should lead to the tertiary alcohol (32), from which the elimination of the elements of water could, in theory at any rate, yield the enynene (33). How- ever, conformational and hyperconjugational influences would be expected to direct this elimination towards the d8(l4)-isomer of (33). For this reason the l6 J. K. Kochi and D. Singleton, J. Am. Chem. Soc., 1968, 90, 1582. l7 H. Brockman and A. Busse, 2.Physiol. Chem.. 1938. 256.252. Synthetic Approaches to Vitamin D and its Relatives chloro-ketone (30), which permits unambiguous control of the ring c double bond position, was used in the synthesis.5 A Stereoselective Route to the Vitamins D The lack of stereoselectivity inherent in the routes of Scheme 3 led us to study another approach, in which the two component fragments were to be united to construct the 7,8-double bond, This plan allows the ring A component to contain from the outset the proper S-chirality at the hydroxy-centre, and the 52-tri- substituted double bond. It was tested in a synthesis18 of the model conjugated 2-triene (39) (Scheme 11). OH PPh,+Br’ (35) (36) liii v, vi iv c--0-8c--&OH (37) Reagents: i, LiA1H4;ii, PhCH,P(OPh),Br, PPh,; iii, 2 BuLi; iv, cydohexanone; v, MnO,; vi, CH2: PPh, Scheme 11 Reduction of the butenolide (34) gave the allylic diol(33, the primary hydroxy- group of which was selectively converted into a phosphonium bromide group.The derived phosphorane (37) reacted with cyclohexanone to give the con- jugated dienol(38), from which the triene (39) was obtained in two routine steps. The sequence (35) +. (38) provided the first demonstration that an allylic alcohol with the less stable 2-geometry can be converted into a Wittig reagent, and used to make a conjugated diene, with essentially complete retention of the l8 I. T. Harrison and B. Lythgoe, J. Chem Sac., 1958, 843. Lythgoe original ally1 geometry. The method has since found applications in general synthetic work. Although normally practicable, it is not always free from experimental difficulty.The rather weakly nucleophilic nature of triphenylphos- phine requires for the. quaternization step the allylic bromide rather than the more stable chloride, and this can occasion partial loss of geometry. Moreover, phosphonium halides are difficult to purify by conventional methods, particu- larly when other polar groups (e.g. OH) are present. These difficulties are minimized in a method described below (p. 460). At first it seemed possible to carry out a synthesis of vitamin D3 on the lines of Scheme11,starting with the appropriate hydroxy-derivative of the butenolide (34) and, at the correct step, using the c18 ketone (5) in place of cyclohexanone. However, experiments with the appropriate hydroxyderivative of (35) failed to provide a pure phosphonium bromide corresponding to (36).It seemed necessary to reduce the number of hydroxy-groups in the ring A component, and our attentions were therefore turned to the 2-dienediol (42). Access to compound (42) was first gained19 by a degradation of vitamin D2 (Scheme 12). Hydroxylation of the vitamin gives20 the trio1 (40); cleavage with 17 Vitamin OH (411ii, iii -OHHO” 0” HO’. HO’0&3 Reagents: i, KMnOl; ii, Pb(OAc),; iii, NaAlH,(OCH,CH,OMe), Scheme 12 J. V. Frosch, I. T. Harrison, B. Lythgoe, and A. K. Saksena,J. Chem. SOC.Perkin Trans. I, 1974, 2005. 8a Y. Wang, HA. Ting, J. J. Huang, Y.-C.Chow, and Y.-T.Huang, Actu Chim. Sin., 1958, 24,126.Synthetic Approaches to Vitamin D and its Relatives lead tetra-acetate and reduction of the products gave a mixture of the bicyclic alcohol (41) and the dienediol (42) from which the dienediol was separated by virtue of its water-solubility, and was obtained crystalline. From it we prepared a phosphonium bromide which appeared to have the desired structure (43). However, it did not take part in Wittig reactions with carbonyl compounds. As it was thought possible that the secondary hydroxy-group in (42) might be responsible for the difficulty, we prepared the simpler dienol(46) by a sequencelg in which the key reaction (Scheme 13) was the base-induced ring-opening of the ii,iii>% + COpH 0 i a OH (44) Reagents: i, LiNPri,, H30+; ii, CH,N,; iii, NaAlH,(OCH&H,OMe),; iv, Me$ N C1-;v, PPh,-NaI-Me,CO Scheme 13 &unsaturated &lactone (44)to give the 2-dienoic acid (45).From the 2-dienol (46) we prepared the phosphonium iodide (47). This compound, too, failed to participate in Wittig olefin syntheses. Although the cause of these failures is unclear, a way of circumventing them was found21 in turning from Wittig's method with phosphonium halides to the related Horner's method which uses phosphine oxides. In the two examples just discussed, the use of phosphine oxides seems essential, but we have also found that, in general,22 allylic diphenylphosphine oxides have marked advantages for the synthesis of conjugated dienes of defined geometry. The phosphine oxides are easily prepared from the relatively stable allylic chlorides, or the even more stable 2,6-dichlorobenzoates, by reaction with lithium diphenylphosphide, followed by oxidation with hydrogen peroxide.The high nucleophilicity of the diphenyl- a1 B. Lythgoe, T. A. Moran, M. E. N. Nambudiry, S. Ruston, J. Tideswell, and P. W. Wright,Tetrahedron Lett., 1975,3863. B. Lythgoe, T. A. Moran, M. E. N. Nambudiry, and S. Ruston, J. Chem. Sac.,Perkin Trans. I, 1976,2386. Lythgoe phosphide anion allows the reaction to proceed at low temperatures, minimizing the risk of loss of ally1 geometry. The phosphine oxides are easily purified. The elimination phase of the olefin synthesis takes place below 25"C, and the by- product, lithium diphenylphosphinate, is water-soluble and easily removed.Two examples of the use of this convenient method are outlined in Scheme 14. In the PPh20 i, ii __j Reagents: i, 2,ddichlorobenzoate treated with LiPPh,-THF; ii, H,O,-CHCI,; iii, BuLi ;ivycyclohexanone; v, compound (4) Scheme 14 Allylic phosphine oxides as precursors of conjugated dienes of defined geometry first, cis-crotyl alcohol (48) was converted into the phosphine oxide (49) and then into cis-crotylidenecyclohexane (50). In the second, the Z-dienol (46) was similarly converted into the phosphine oxide (51). The lithium derivative reacted without difficulty with Windaus and Grundmann's ketone (4) to give a single conjugated triene, which was shown to have the structure of 3-deoxyvitamin D2 (52).It was next necessary to obtain the Z-dienediol (42) by total synthesis.23 This presented two stereochemical problems. The first, that of securing the proper S-chirality at the secondary hydroxy-centre, was solved by using as the starting material the readily available S-cyclohex-4-ene-1 ,trans-2-dicarboxylic acid (53) (Scheme 15). It contains eight of the nine carbon atoms present in the required dienediol; the ninth was introduced in a nitrile homologation step leading to the Q3 B. Lythgoe, R. Manwaring, J. R. Milner, T. A. Moran, M. E. N. Nambudiry, and J. Tideswell, J. Chem. SOC.,Perkin Trans. I, 1978, 387. Synthetic Approaches to Vitamin D and its Relative? i, ii, iii iv, v, vi R = CH2Ph (54) vii, viii.1 Reagents: i, CH,N, ; ii, LiAlH,; iii, Na-dioxan, PhCH,Br; iv, MeC,H,SO,Cl-pyridine; v, NaCN-Me,SO; vi, KOH-H,O-EtOH; vii, KI, on Na salt; viii, Ph,SnH; ix, H,-Pd; x, NaI-Me,CO on tosylate; xi, diazabicycloundecene; xii, LiNPr',, PhSSPh; xiii, PhCOC1-pyridine; xiv, NaIO,; xv, heat to 120 "C; KOH-HzO- MeOH Scheme 15 Ring A fragment for the vitamins D acid (54).When this acid was converted into the lactone (55) chirality was trans- ferred to provide that required at the secondary hydroxy-centre in the end- product (42). To solve the second problem, that of securing theZ-geometry of the trisubstituted double bond, we used a stereospecific sulphoxide thermolysis. Introd~ction~~of a phenylthio-group adjacent to the lactonic carbonyl group of (56) gave two epimers.The major epimer (57,which was that with the desired configuration, crystallized and was readily isolated; as it was also the more stable B. M. Trost and T. N. Salzmann, J. Am. Chem. SOC.,1973,95,6840. Lythgoe of the two, further amounts were obtained by base-catalysed equilibration of the residual mixture. It was then converted as shown into the Z-dienediol (42). The two hydroxy-groups in the dienediol(42) have widely different reactivities, which enable the preparation of the secondary monobenzoate (58). The synthe- sis25 of vitamin D2 from this compound is shown in Scheme 16. The mono- iii, iv OH PPh,O A& i, ii 4 Vitamin D, RO (58) (59) R = COPh (60) R = EtOCHMe Reagents: i, ClCH: NMe,+Cl--Me,NCHO; ii, LiPPh,-THF;H,O,; iii, compound (60) treated with BuLi, then with compound (4); iv, AcOH-H20 Scheme 16 Synthesis of vitamin D, benzoate was first converted into the crystalline phosphine oxide (59) and then into the acetal-protected analogue (60).The lithium derivative of the latter reacted with Windaus and Grundmann’s ketone (4) to give, after de-protection, crystalline vitamin D2 in ca. 60% yield. A similar reaction, employing the cl8 ketone (3,gave vitamin D3. In these syntheses, no significant amounts of vitamin D stereoisomers were produced. The S-configuration at C-3 and the 2-geometry of the 5,6-double bond, already present in the dienediol (42), were preserved in the vitamins. The 7,8-double bond, formed in the phosphine oxide olefin synthesis, had exclusively the natural E-geometry; the reasons for this have been discussed.12 Thus in its major aspects the synthetic route is highly stereoselective. 6 Synthesis of Des-AB-cholestane and Des-AB-ergostane Derivatives The partial syntheses, described in the preceding three Sections, of the 9,lO-seco- 25 B.Lythgoe, T. A. Moran, M. E. N. Nambudiry, J. Tideswell, and P. W. Wright, J. Chern. SOC.,Perkin Trans. I, 1978, 590. 463 Synthetic Approaches to Vitamin D and its Relatives sterols of the vitamin D3 series made use of the des-AB-cholestane derivatives (9,(22), and (30), whilst that of vitamin Dz made use of des-~~-ergost-22-en-8-one (4). In this Section synthetic routes to these compounds are described sothat the syntheses making use of them become total.All three of the des-AB-cholestane derivatives can be prepared14 by standard methods from simple compounds such as des-~~-cholest-8-ene (27), the 8/3,9a- diol (28), or, less easily, from the 8/3-01 (26). The 8/3-01 (26) had indeed already been obtained by a total synthesis in which Hagemann’s ester (61) (Scheme 17) was first elaborated to givez6 the trans-perhydroindan-l -one (62), the keto-group of which was then used as a means of adding, in a series of steps, the iso-octyl side-chain.7 However, the overall yield in the sequence of Scheme 17 was low (ca. 0.001 %). We therefore attempted to obtain the related 8/3,9a-diol(28) by a more convergent route. CO,Et(F P0 HO The plan of the synthesis27 can be understood by reference to the dibasic ester (66) (Scheme 18).Cyclization was expected to afford a hydrindenone mixture which, after equilibration, would contain some 85 % afthe isomer (67) having the correct chirality at C-17. In the ester (66) ten of the nineteen skeletal carbon atoms (the north-east sector) have the same structural and stereochemical arrangement as those in R-dihydrocitronellic acid [cf. the orthoester (64)]. We proposed to use a derivative of that acid to provide that sector of the ester (66), so ensuring the correct chirality at C-20 in the ketone (67). An acceptable way of uniting the ten-carbon fragment with the future ring c was found in the orthoester variant28 of the Claisen rearrangement of allylic alcohols.Reaction of the orthoester (64) with the allylic alcohol29 (63) gave, after debenzoylation, a new allylic alcohol (65) which was then used in a second Claisen rearrangement to provide the dibasic ester (66). The rearrangements proceed stereospecifically with respect to the, cyclohexane ring, so that the m H. H. Inhoffen, S. Schutz, P. Rossberg, 0. Berges, K. H. Nordsiek, H. Plenio, and E. Horoldt, Chem. Ber., 1958, 91, 2626. I. J. Bolton, R. G. Harrison, and B. Lythgoe, J. Chem. SOC.(C), 1971, 2950. ** W. S. Johnson, L. Werthemann, W. R. Bartlett, T. J. Brocksom, T. Li, D. J. Faulkner, and M. R. Petersen, J. Am. Chem. Suc., 1970,92, 741. I. J. Bolton, R. G. Hamson, B. Lythgoe, and R. S. Manwaring, J. Chem.Soc. (C), 1971, 2944. Ly thgoe H I OH iii, iv, v1 vi, vii -viii-xii, iiI 17 Reagents: i, catalytic EtCOzH at 140 "C; ii, NaOH-H,O; iii, MeC(OMe),NMe,; iv, KOH- EtOH ; v, CH,N,; vi, NaH-Me,SO; vii, MeC,H,SO,H-H,O-AcOH; viii, ClC6H4C03H; ix, H2S04-H,0; x, Ac,O-pyridine; xi, HSCH,CH,SH-BF,.Et20; xii, Raney Ni; xiii, convert into compound (29) and treat with LiAiH, Scheme 18 Synthesis of des-Aa-cholestanes compound (63) is seen to contain in code the absolute configurations at C-13 and C-14 of the ketone (67), which was thus obtained relatively briefly from it in a yield of over 30 %. The application of standard methods then gave the 8/3,9a-dioI (28). This has been converted into the /3-epoxide (29) and the chloro-ketone (30) as shown in Scheme 9, and into the 8/3-01 (26) as shown in Scheme 18.Synthetic Approach to Vitamin D and its Relatives When a similar synthesis of the alcohol (41), corresponding to Windaus and Grundmann's ketone, was considered, it was apparent that the analogue of (67) would contain double bonds in both ring c and the side-chain, and that the selective functionalization of the first of them would present difficulties. The route was therefore modified so as to lead initially to the compound (70) in which the side-chain, though abbreviated, has facilities for later extension. This was effected30 as shown in Scheme 19. The allylic alcohol (63) was first + "qo EtO OEt I H ..~H,OH Scheme 19 allowed to react in a Claisen rearrangement with the cyclic orthoester (68).Such cyclic orthoesters are readily obtained by applying Meerwein's3I method to the appropriate 6-or, as in this case, y-lactone. The rearrangement product gave on debenzoylation the y-lactone (69) which was transformed, in a series of unexcep-tional steps, into the unsaturated primary alcohol (70). For identification, this compound was also prepared by a degradation of vitamin D2. The stereochemical results32 of the reactions of allylic alcohols with cyclic orthoesters, although occasionally surprising, can in general be interpreted in terms of the expected transition states. The initial reaction in Scheme 19 is stereospecific and, taking place by way of a boat-shaped transition state, results in the R-configuration shown at the new chiral centre adjacent to the lactonic carbonyl group of compound (69).This configuration was maintained during the 3o C. B. Chapleo, P. Hallett, B. Lythgoe, I. Waterhouse, and P. W. Wright, J. Chem. SOC., Perkin Trans. I, 1977, 1211. 31 H. Meerwein, P. Borner, 0.Fuchs, H. J. Sasse, H. Schrodt, and J. Spille, Chem. Ber., 1956, 89, 2060. m R. J. Cave, B. Lythgoe, D. A. Metcalfe, and I. Waterhouse, J. Chem. SOC.,Perkin Trans. 1, 1977, 1218. Lythgoe rest of the synthesis, so that the 17p-configuration in compound (70) is a result of asymmetric induction rather than, as in compound (28) of Scheme 18, a result of equilibration. The primary alcohol (70) was converted33 by standard methods into the benzoyloxyaldehyde (77).The proper way of completing the synthesis then appeared to lie in a use of Schlo~ser’s3~modified Wittig reaction, in order to obtain the new double bond in the trans-configuration. However, although the required optically active halide (75) proved readily available, yields of the corresponding phosphonium halide were disappointing. By contrast, reaction with sodium thiophenoxide gave, in excellent yield, a thio-ether which was oxidized quantitatively to give the sulphone (76). Metallated alkyl aryl sulphones react with carbonyl compounds to give, after acetylation or benzoylation, diastereoisomeric p-acyloxy-sulphones. On treat- ment with sodium amalgam in methanol these undergo loss of both functional groups, forming 0lefins.3~ In a study36 of the stereochemistry of the reaction, when used for the formation of acyclic internal disubstituted olefins, we observed (Scheme 20) first that for unbranched olefins, e.g.(73, the trans: cis ratio of the OCOPh RCHzCHLi IS02Ph -t OHCCH,R’ I +RCH,CH -CHCH,R‘ I S0,Ph + RCH,CH=CHCH,R’ trans : cis = 80:20 (71) (72) S0,Ph +Li + (74) Scheme 20 Stereochemistry of the p-benzoyloxysulphone elimination product is ca. 80: 20 for each of the intermediate acyloxysulphone diastereo- isomers (71). Secondly, when substituents are introduced into the starting materials at positions adjacent either to the aldehyde group, or to the methylene 99 B. Lythgoe, D. A. Roberts, and I. Waterhouse, J. Chem. SOC.,Perkin Trans.1, 1977, 2608. 34 M. Schlosser and K. F. Christmann, Liebigs Ann. Chem., 1967, 708, 1. ss M. Julia and J.-M. Paris, Tetrahedron Lett., 1973, 4833. so P. J. Kocienski, B. Lythgoe, and S. Ruston, J. Chem. SOC.,Perkin Trans. I, 1978, 829; see also ref. 12. 467 Synthetic Approach to Vitamin D and its Relatives group bearing the sulphonyl function, the trans-selectivity of the reaction is markedly increased. Two examples will illustrate this effect. The reaction partners isobutyl phenyl sulphone and 2-ethylbutyraldehyde, both of which contain substituents of the type in question, gave rise to an olefin consisting almost exclusively of the trans-isomer (73). Similarly, 2-2-methylbut-2-enyl phenyl sulphone and E-Zmethylbut-2-enal gave rise almost exclusively to (ZE,4E,62)-3,6-dimethylocta-2,4,6-triene(74),in which the original geometries of the starting materials were preserved, and the new, central double bond had the trans- configuration.The reasons for this trans-selectivity have been discussed.12 These observations formed the basis for the stereoselective construction3’ (Scheme 21) of the trans-double bond in the alcohol (41)which corresponds to OAc CHO i, ii (75) x = I (76) X = SOzPh OCOPh 6COPh (77) iii, iv* I I OH Reagents: i, Li derivative of (76); ii, Ac20;iii, Na/Hg-MeOH-EtOAc; iv, KOH-EtOH-H,O; v, Cr0,-pyridine Scheme 21 Synthesis of Wiiidaus and Grundmann’s ketone Windaus and Grundmann’s ketone (4),starting from the benzoyloxyaldehyde (77)and the optically active sulphone (76).By virtue of this work the synthesis of vitamin Dz described in the preceding section became formally total.7 la-Hydroxyvitamin D3 During the past decade important advances have been made in understanding P. J. Kocienski, B. Lythgoe, and D. A. Roberts, J. Chem. Soc., Perkin Trans. I, 1978, 834. Lythgoe the mode of action of vitamins D.38The full physiological activity of vitamin D3 is developed only after its metabolic hydroxylation, at first in the liver to give 25-hydroxyvitamin D3, and then in the kidneys to give la,25-dihydroxyvitamin D3 (79) (Scheme 23, the hormonal form, which is very potent and rapidly acting. (78) X = OH, Y = H (80)25-Hydroxyvitamin D, 1 a-Hydroxyvitamin D, (79) x = Y = OH 1 a,25-Dihydroxyvitamin D, scbeme 22 The dihydroxyvitamin (79) is clinically effective in those cases of osteodystrophy where, because the kidneys fail to effect the la-hydroxylation step, the vitamin itself is ineffective.la-Hydroxyvitamin D3 (80) is similarly effective, since it can be converted in vivo into the hormone (79). These and related observations have greatly stimulated experiments on the chemical preparation of hydroxylated vitamin D derivatives. The hormone (79) was first obtained in well characterized and crystalline form by the routea9 outlined in Scheme 23. Its central feature was the novel and elegant sequence which was used to effect the la-hydroxylation step (ii); step (i) derives from earlier work, and steps (iii) and (iv) use methods parallel to those normally used for the preparation of vitamin D3 from cholesterol.Most other routes to the hormone (79) follow a similar synthetic strategy, although starting materials, methods, and sequence may be different. It was of interest to consider whether the approaches described in earlier Sections of this lecture have relevance to the preparation of the hormone (79). For vitamin D3 itself, total synthesis is so much longer than partial synthesis from cholesterol that it is ineffective as a preparative method. When, however, s8 For reviews, see inter al. H. F. DeLuca and H. K. Schnoes, Ann. Rev. Biochem ,1976,45, 631; P. E. Georghiou, Chem. SOC.Rev., 1977, 6, 83. s9 D. H. R. Barton, R. H. Hesse, M.M. Pechet, and E. Rizzardo, J. Chem. SOC.,Chem. Commun., 1974,203. Synthetic Appraach to Vitamin D and its Relatives Cholesterol 25-H ydroxycholesteroI iii 1 a,25-Dihydroxycholesterol1iii I a,25-Dihydroxy-7-dehydrocholesterol(as triacetate) 1iv 1 a,25-Dihydroxyvitamin D, Scheme 23 as in Scheme 23 and its relatives, a starting material must be modified at two separate sites in order to prepare it for the subsequent transformations (iii) and (iv), an element of linearity is introduced which can diminish, although it may not completely abolish, the advantage of partial synthesis. Moreover, total synthesis is often advantageous for the preparation of labelled compounds, use- ful in biological work. We therefore explored synthetic routes to la-hydroxyvitamin D3 (80).The first, an extension of the work of Scheme 10, proceeded by way of la-hydroxy-precalciferols, and required as the ring A component the enyne (83). This was prepared from the lactone (14) as shown in Scheme 24. The lactone bridge directs the attack of m-chloroperbenzoic acid to the opposite face of the molecule, so that the epoxide (81) is the major product, thus providing the correct con- figuration at the second hydroxy-centre in the dihydroxy-ester (82), and in the enyne (83), which were then obtained by standard methods. The lithium derivative of the dnyne (83) and the chloro-ketone (30) were used40 (Scheme 25) to obtain, in analogy with earlier work, the diacetoxyenynene (84), which was converted successively into lo~-hydroxyprecalciferol3(85) and la-hydroxyvitamin D3 (80).In a second shown in Scheme 26, a la-hydroxytachystero13 derivative (88) was first prepared by a Julia olefin synthesis using the components (86) and (87). The p-tolyl sulphone, from which the lithium derivative (86) is derived, was obtained from the allylic alcohol (19); the protected aldehyde (87) was obtained by standard methods from the dihydroxy-ester (82). Tachysterol derivatives such 40 R. G. Harrison, B. Lythgoe, and P. W. Wright, J. Chem. Sac.,Perkin Trans. I, 1974, 2654. P. 3. Kocienski and B. Lythgoe, J. Chem. SOC.,Perkin Trans. I, 1980, 1400. 470 Lythgoe OH I I HO C0,Me ?Ac YSiMe:, I I CHO C:CHMe,SiOfiAcO a-(83) Scheme 24 Li I CRH,, + Me,SiO'&OSiMe,-+ @ 111 17 Scheme 25 1 a-Hydroxyvitamin D3 by way of 1a-hydroxyprecalciferal, as (88) can be converted very efficiently by fluorenone-sensitized irradiation42 into the corresponding precalciferol derivatives.Following this conversion, the 4p A. E. C. Snoeren, M. R. Daha, J. Lugtenburg, and E. Havinga, Red. Trav. Chim.Pays-Bas, 1970, 89, 261. 47 1 Synthetic Approach to Vitamin D and its Relatives CHO 1a-Hydroxyprecalciferol bis-( R)ethe r t-(80) (88) R = SiMe,But; Ar = p-MeC6H4 Scheme 26 1a-Hydroxyvitumin Dsby way of u 1a-hydroxy?achysleruZ9derivative la-hydroxyvitamin (80) was obtained in a yield of 57% from the allylic alcohol (19) or 12.8 % overall from cholesterol. This compares favourably with many of the yields reported in the current literature; admittedly, however, more steps were needed in our method.As in the synthesis of tachysterols (Scheme 7), and for similar reasons, the 6-trans-compound (88) was formed essentially stereospecifically. It would clearly be of great value if a similar addition4imination sequence were available for the stereoselective construction of an internal disubstituted double bond with cis-geometry. This result can of course be achieved by use of the ‘salt-free’ Wittig method, but only in those cases where the phosphorane used is not allylic; it can not be used to construct the central double bond in such conjugated trienes as natural 152-phytoene (93) or precalciferol~. We were not able to devise a direct Lythgoe method for the purpose in question, but a method was found43 for constructing an internal acetylenic link which, by adding a semihydrogenation step, permitted the desired result to be achieved indirectly. The method is illustrated by the synthesis44 of the conjugated triene (92) (Scheme 27).0 c (93) R = E-geranyl Scheme 27 Synthesis of conjugated enynenes 43 B. Lythgoe and 1. Waterhouse, Tetrahedron Lett., 1978, 2625. For a related synthesis, see P. A. Bartlett, F. R. Green 111, and E. H. Rose,J. Am. Chem. SOC.,1978, 100, 4852. 44 B. Lythgoe and I. Waterhouse, J. Chem. SOC.,Perkin Trans. 1, 1979, 2429. Synthetic Approach to Vitamin D and its Relatives Reaction of the magnesium bromide derivative of E-geranyl phenyl sulphone with methyl E-geranate gave the /%oxo-sulphone (89) ; its lithium enolate reacted with diethyl phosphorochloridate to give the enol phosphate (90).When this was treated with sodium amalgam in tetrahydrofuran-dimethyl sulphoxide, 1,Zelimination of both functional groups took place giving the enynene (91), in which the E-geometry of both starting materials was preserved. Semihydrogena- tion over Lindlar’s catalyst then gave the conjugated E,Z,E-triene (92), which is an analogue of 152-phytoene (93). In an exactly similar manner, the bis-t-butyldimethylsilyl ether of the ester (82) and the p-tolyl sulphone corresponding to (86) were used to ~repare4~ the conjugated enynene (84) from which, as previously described, 1 a-hydroxy- precalciferol3 and 1 a-hydroxyvitamin D3 were obtained in turn.Although these experiments have not so far been extended to yield la,25- dihydroxyvitamin D3, they are clearly serviceable for that purpose, and it may be useful to point to two routes as the most promising. The first, analogous to that of Scheme 16, constructs the 7,8-double bond of the metabolite by interaction of the diphenylphosphine oxide corresponding to an allylic alcohol (96) with a suitably protected ketone corresponding to the alcohol (94). This latter alcohol has been obtained by total synthesis33 (although probably not by the most efficient available route) as well as from vitamin D3.46 A second route to the metabolite (79) requires the preparation of the allylic alcohol (95) and the use of a suitable phosphorus or sulphur derivative of it in an olefin synthesis with the aldehyde (87).As in Scheme 26, the resulting tachysterol derivative would then be converted into the vitamin D analogue. QH (96) Scheme 28 B. Lythgoe and I. Waterhouse, J. Chem. SOC.,Perkin Trans. I, 1980, 1405. 46 Z. Cohen, E. Berman, and Y. Mazur, J. Org. Chem., 1979 44, 3077. 474 I wish to thank sincerely the numerous and skilful collaborators, named in the references, whose experiments form the basis for this lecture.