Synthesis of cyclo-bis7alpha;, 12alpha;-diacetoxy-3beta;-dicyanomethyl-3alpha;-(4-methylenephenyl)cholanamide; a cholaphane with reduced flexibility and externally-directed functionality

摘要

919J. CHEM. soc. PERKIN TRANS. I 1993 Synthesis of Cy~lo-bis7a,l2a-diacetoxy-3~-dicyanomethyl-3a-(4-methylenephenyl)cholanamide; A Cholaphane with Reduced Flexibility and Externally-directed Functionality Anthony P. Davis" and Michael G. Orchard Department of Chemistry, Trinity CoJJege, Dublin 2, Ireland A second 'cholaphane' framework is manifested in the title compound 8, which has been prepared from methyl 3a.7a.12a-triacetoxycholanoate 9 in 23 overall yield. The synthesis involves the Knoevenagel condensation of ketone 11 with malononitrile to give dicyanomethylene derivative 12, equatorial-selective addition to the latter of an organocuprate derived from aryl bromide 15, conversion of the resulting adduct into amino acid 22, and cyclodimerisation. The framework of 8 has less conformational freedom and a better-defined cavity than the 'first-generation' cholaphanes 2, and also bears externally- directed functionality.During the past few years workers in this laboratory'-3 and elsewhere have been exploring the potential of cholic acid 1as a starting material in biomimetic/molecular recognition chemis- try. Among the factors which have stimulated this interest are (a) the low cost and ready availability of 1, (b) its rigid steroidal skeleton, (c) its nicely spaced array of differentiable functionality, (d) its chirality and (e) its curved profile Q(facilitating the design of frameworks with concave or toroidal surfaces). An early expression of the strategy was the design and (Me3SO2N/ synthesis of 'cholaphanes' 2,' the first macrocyclic host t ii .--I-/bsol; -8'-CF,CONH--/ CFSCONH0J Scheme 1 Reagents and conditions: i, CF,CO,H, (CF,CO),O; ii, H,, Pd-C substituent.There were two reasons why the latter might be desirable. Firstly, it could provide exo directed functionality which could be used to assist solubilisation (either in water, by using ionic groups, or in organic solvents by attaching flexible hydrocarbon chains) or to attach the macrocycle to some other unit (e.g. a polymer). Secondly, a relatively bulky group would constrain the aromatic ring to take up a conformation roughly parallel to the C2 axis of the macrocycle in contrast to the 'perpendicular' conformation favoured in cholaphanes 2 (vide 2 infra). This would provide a better-defined cavity with an increased 'hydrocarbon surface', particularly important for the molecules based on cholic acid. Subsequent work demonstrated proposed move to water-soluble cholaphanes capable of that cholaphanes 2 (R' = H, R2 = H or CH,Ph) are remark- hydrophobic binding.In addition, the loss of rotational freedom able in acting as receptors for carbohydrate nuclei in organic about the C(3)-aryl bond would somewhat reduce the overall solvents.2 The key step in the synthesis of 2 from 1 was the flexibility of the cholaphane framework. introduction of a p(aminomethy1)phenyl substituent in the 3a-Our task was thus to identify a structural fragment 4 which position of the steroid nucleus. As shown in Scheme 1, this was could be developed from the 3-OH of cholic acid with the same achieved by the chemoselective addition of organomanganese chemo- and stereo-selectivity as had been obtained previously reagent 3 to a 3-keto group followed by elimination and (Scheme 1).In particular, it would be convenient to protect the stereoselective hydrogenation from the (convex) p-face. C-24 carboxy group of 1 as a methyl ester, so that reagents Although successful, this sequence imposed a limitation on the which would react with this group were to be avoided if final structure in that it precluded the possibility of a 3p-possible. It seemed that an answer might fie in the sequence 920 J. CHEM. SOC. PERKIN TRANS. I 1993 X = (a) f unctionalised (b) relative bulky Ar E 6 7 IE electron-withdrawing group M = metal Scheme 2 shown in Scheme 2. A Knoevenagel condensation5 could be used to generate a highly electron-deficient alkene 6,which should be subject to conjugate addition by an appropriate organometallic reagent (most probably an organocuprate) under conditions which would leave ester groups untouched.Precedent suggested that attack of the reagent should occur from the equatorial direction, giving the desired stereochemistry in adduct 7.6 As described el~ewhere,~ model studies on 4-tert-butylcyclo- hexanone confirmed that this sequence could be carried through with excellent yields and selectivities for a number of Knoevenagel reagents 5, using a higher-order cuprate derived from phenylmagnesium bromide and CuCN as the aryl donor.In this paper we give details of the application of one of these methods (E = CN) to the synthesis of a lsquo;second-generationrsquo; cholaphane 8 with externally-directed functionality and fewer conformational options than the earlier examples. a Results and Discussion The model studies referred to above left one major question unanswered. While the new methodology was effective for introducing an unsubstituted phenyl group, it was not clear that it could be extended to the introduction of thep-(aminomethyl)- phenyl required for the synthesis of 8. However, as the ketone 11 required as starting material for 8 was relatively accessible, it was decided to forgo further model investigations and work directly towards the macrocycle.As shown in Scheme 3, ketone 11 was prepared from methyl 3a,7a,12a-triacetoxycholanoate 9 rsquo;* oia selective acid-catalysed deacetylation of the equatorial 3a-OAc to give 10,rsquo; followed by oxidation with sodium periodate and catalytic quantities of ruthenium trichloride, after the method of Sharpless and co- workers.I2 The procedure was quite simple to carry out on a large scale, giving ketone 11 in an overall yield of 82. Condensation of 11 with malononitrile using the conditions developed by Mirek et (catalysis by acetic acid and ammonium acetate in refluxing benzene, with azeotropic removal of water) proceeded smoothly, giving the dicyanometh- ylene derivative 12 in 93yield after crystallisation.As shown in Scheme I, the sequence employed for the lsquo;first generationrsquo; cholaphanes 2 had relied on bis(trimethylsily1) protection in the p-(aminomethy1)phenyl organometallic. rsquo; It seemed reasonable to suppose that the same tactic might work in the present case, the only difference being that the reagent would need to be a higher-order cuprate rather than an organomanganese derivative. Thus, the aryl bromide 13 was prepared from p-bromobenzyl bromide and sodium hexameth- yldisilazide, and treated with magnesium in ether to give the corresponding Grignard reagent. In line with the model studies7 the latter was combined with cuprous cyanide in the ratio 2: 1, and an ethereal solution of dicyanomethylene derivative 12 was added to the mixture.Unfortunately, no addition product could be isolated. An analogous experiment employing the organolithium derived from 13, with dimeth- oxyethane as solvent, was similarly unsuccessful. It seemed quite likely that the problem was being caused by the lone pair on the nitrogen atom in these reagents which might be expected to have a high affinity for the copper and thus perturb the reactivity of the organometallic centre. Given the paucity of alternative N-protecting groups stable to strongly basic reagents,* and the presence of a free nitrogen lone pair in many of these alternatives, we decided to explore the lsquo;saferrsquo; option of an oxygen-containing reagent 14. Although perhaps long-winded, the substitution of the oxygen for a nitrogen at a later stage of the sequence seemed unlikely to cause serious prob- lems.*This issue is discussed in greater detail elsewhere, along with the development of a new silicon-based protecting group, the lsquo;benzostabasersquo; (BSB) group. In the present work a higher-order cuprate derived from BSB-protected p-bromoaniline was also tested with dicyanomethylene derivative 12, but was no more successful than the reagents derived from 13. OAC iii ___c 9 R=Ac 11 12 il, 10 R=H Scheme 3 Reugents and conditions: i, AcCl, CH,OH; ii, NaIO,, cat. RuCI,, CH,CN, CCl,, H,O; iii, CH,(CN),, NHiOAc-, AcOH, C,H, J. CHEM. soc. PERKIN TRANS. 1 1993 m*M)EuCN P = protectinggrolp M = Mg or Li 13 14 15 16 It appeared that a suitable O-protecting group P in reagent 14 would be tert-butyl, and we thus investigated the preparation of bromide 15 as starting material.The literature methods for this compound were not really satisfactory. The sulfuric acid- catalysed reaction of p-bromobenzyl alcohol and isobutene ' had been reported to give only 54 yield (and was in any case inconvenient experimentally), and the reaction of p-bromo- benzyl bromide with potassium tert-butoxide in DMF16 (for which no yield was quoted in the original reference) gave poor results in our hands. However, a modification of the latter was developed, in which the potassium tert-butoxide was prepared from potassium metal and an excess of dry tert-butanol under sonication, the p-bromobenzyl bromide added and the sonication continued for several hours.This resulted in a 92 yield of bromide 15, and gave material with a significantly higher melting point than previously recorded. Attempts to convert bromide 15 into the corresponding organolithium derivative with lithium metal were unsuccessful, possibly because of deprotonation of the benzylic CH, by the organolithium once it was formed. However the Grignard reagent 16 was produced smoothly from 15 and magnesium in THF under sonication (30-60 min).* We were pleased to find that, as shown in Scheme 4, the higher-order cuprate derived from 16 and cuprous cyanide (ratio 2 :1) reacted rapidly with 12 to give 17 in nearly quantitative yield. Only one stereoisomer could be detected, the structure being assigned by analogy with the model studies7 (and confirmed by the X-ray crystal structure of the final product).The tert-butyl protecting group was removed with trifluoroacetic acid (CH,Cl,, 50 "C, 5.5 h) giving trifluoroacetate 18, and the trifluoroacetate cleaved with aqueous ammonia (Et,O, 30 min) to give alcohol 19 in 90 overall yield from dinitrile 12. Conversion of the benzylic hydroxy to an amino group was achieved via azide 21. This could be formed directly from 19 using a Mitsunobu-type l7 procedure, following the method of Loibner and Zbiral Ph,P, diethyl azodicarboxylate (DEAD), HN,, benzene. However, the process was not consistently successful, and a more satisfactory method (as shown in Scheme 4) involved conversion of alcohol 19 into the corresponding mesylate 20 followed by treatment with tetramethylguanidinium azide in chloroform.The mesyla- tion step proved to be very sensitive to the exact procedure, the best results being obtained by addition of methanesulfonyl chloride to alcohol 19 in THF at 0 "C followed immediately by dropwise addition of di-isopropylethylamine. The overall conversion into azide 21 proceeded in a very satisfactory yield of * Reactions without sonication were much slower and gave complex product mixtures, as determined by aqueous quench followed by GC analysis. CAUTION: The interaction of azide ions and chlorinated solvents can lead to explosive side-products.20 As part ofa separate investigation we have found that, over a period of days at elevated temperatures, tetramethylguanidinium azide and chloroform can react to give significant quantities of a shock-sensitive, distillable liquid." 92 1 I, ll .-.12 ~111 17 R = BU' 18 R=CF3CO "619 R=H E20 R = CH3SO2 21 vii.vill Ix1 22 Scheme 4 Reagents and conditions: i, 16 (2 equiv.), CuCN (1 equiv.), THF; ii, NH4Cl aq.; iii, CF3C02H, CH,CI,, 50 "C, h; iv, NH, aq., Et,O, 30 min; v, CH,SO,Cl, PriNEt, THF, 0 "C; vi, (Me2N),CNHl- Ni, CHCI,, 30 min; vii, Ph,P, THF, CH30H, H20; viii, LiOH; ix, HCl aq. 93. The azide could be reduced to the corresponding amine using catalytic hydrogenation or triphenylphosphine in wet THF." The latter proved more successful, giving the amine in SS-SS yields.However, it was found to be preferable to perform the reduction and hydrolysis of the C-24 methyl ester concurrently in a one-pot procedure, giving amino acid 22 in 87 yield. The remaining step in the synthesis of 8 was, of course, the cyclodimerisation of 22 via formation of two amide bonds. In the synthesis of 2 the analogous transformation had been accomplished using two methods. ' Firstly, direct treatment of the amino acid with the condensing agent diethyl phosphoro- cyanidate (EtO),P(O)CN, DEPC 23 in CHC1,-DMF in the presence of K,HPO, had given yields of (crystalline) macrocycle of up to 32, and secondly a more lengthy (but more efficient) approach via an N-tert-butoxycarbonyl-pro-tected pentafluorophenyl ester had given up to 55 yield.In the present work only the former was tried, a minor modification being the use of THF as solvent. With an 8.6 mmol dmP3 concentration of amino acid 23,the method resulted in a 42 crystalline yield of macrocycle 8. On the basis of our earlier experience, it is quite likely that this yield could be improved by further investigations. Even so, the overall yield of 8from ketone 11was 28. Not unexpectedly, for a rather rigid molecule of this size, the macrocycle was relatively insoluble in chloroform, dissolving to the extent of ca. 1 mg ~m-~.The 'H NMR spectrum in CDCI, showed a feature which was also characteristic of the earlier series of cholaphanes 2, namely the splitting of the benzylic protons into an AB quartet (further coupled to NH); the corresponding protons appeared as a singlet in the spectra of the acyclic precursors. Final characterisation of 8 was provided by X-ray crystallography, as reported in our preliminary communica- ti~n.~As expected, the structure showed the aryl spacer groups in an 'inward-facing' conformation (as illustrated in the formula), supporting our assumption that the exo-directed CH(CN)2 groups would be bulky enough to favour such an orientation.However, in order to establish that the new linkage would have less freedom of rotation than that in 2, it was necessary to resort to computational methodology. The MM2 molecular mechanics method, as implemented in MacroModel V3.1X, was applied to the model systems 23 and 24.In the case FN H YNI 23 24 of 23, the favoured conformation was as shown, with the aromatic ring eclipsing the adjacent C-H. Rotation of the C-Ar bond through 60" in either direction exacted an energetic penalty of ca. 5.9 kJ molt'. A similar distortion of 24 from its preferred conformation (shown) was significantly more difficult, requiring ca. 13.4 kJ mol-'.* In conclusion, we have developed a practical and high- yielding synthesis of a second cholaphane framework with lower conformational freedom than the earlier system, a better-defined cavity and externally-directed functionality. As discussed el~ewhere,~ work on a model system has suggested that further modification of the dicyanomethyl group may not be practicable, so that the last of these 'advantages' may prove to be of limited worth.However, the other features are likely to be valuable in the development of, for example, new host molecules with more predictable complexation behaviour. Experimental 'H NMR spectra were recorded on a Bruker WP 80 spectrometer at 80 MHz or, where indicated, on a Bruker MSL 300 instrument at 300 MHz. (CH,),Si was used as the internal standard. J-Values are given in Hz. IR spectra were recorded on a Perkin-Elmer 883 spectrophotometer. Solvents were purified by standard procedure^.^^ Analytical thin-layer chromatogra- phy (TLC) was performed on aluminium sheets coated with silica gel 60(0.2 mm layer thickness). Steroidal compounds were visualised by charring over a flame.Gas chromatography (GC) was performed using a Varian 3300 instrument fitted with a 25 m OV1 capillary column. Silica gel 60, 400-230 mesh (Merck) *Global minima were found using the Polak-Ribiere Conjugate Gradient (PRCG) algorithm. The energies of the other conformations were estimated after minimisation by Steepest Descents (SD), as the PRCG method merely relocated the global minima. J. CHEM. SOC. PERKIN TRANS. 1 1993 was used for flash chromatography and, where indicated, filtration of reaction products. Reactions involving sonication were performed in a B amp;T laboratory ultrasonic cleaning bath. Methyl 7a, 12a-Diacetoxy-3-oxocholan-24-oate11.-A soh-tion of the triacetate 9 ''(1 5.49 g, 28.2 mmol) in methanol (1 50 cm3) was treated with acetyl chloride (10 cm3)." After 15 min the mixture was evaporated under reduced pressure to give a white foam of the alcohol 10 (13.80 g) (pure by TLC).This material was then dissolved in acetonitrile-carbon tetrachloride (1 :I ;100 cm3) and stirred vigorously with sodium periodate (24.0 g, 112 mmol) in water containing ruthenium trichloride hydrate (150 mg, 0.7 mmol).12 After 75 min the mixture was extracted with dichloromethane (250 cm3) which was then evaporated under reduced pressure to give a grey powder (14.05 g). This was dissolved in and eluted with benzene-ethyl acetate (6:4), through a silica gel plug to give a white solid (12.99 g), which was crystallised from benzene to give ketone 11(1 1.74 g, 82), m.p. 189-190deg;C (lit.,' 190-191 "C); TLC R, 0.36 in hexane-ethyl acetate (1 :1). Methyl 7a, 1 2a-Diacetoxy-3-dicyanomethylenecholan-24-oate 12.-The ketone 11 (10.65 g, 21.1 mmol) and malononitrile (1.452 g, 21.97 mmol) were dissolved in dry benzene (23 cm3) inside a 50 cm3 round bottom flask fitted with a Dean and Stark apparatus.To this was added ammonium acetate (0.35 g) and glacial acetic acid (1 an3). The reaction mixture was placed under an atmosphere of argon before being lowered into an oil- bath held at 100 "C. After 15 min, analysis by TLC implied the reaction was complete, so the mixture was allowed to cool to room temperature and then partitioned between chloroform- ether (1 :1;20 cm3) and water (100 an3).After the aqueous layer had been removed some product crystallised out of the organic layer, so more chloroform (10 an3)was added to redissolve it.The organic layer was washed successively with water (2 x 50 cm3) and brine (20 cm'), and the solvent removed under reduced pressure. The crude product was crystallised from ethyl acetate to give white needles of the unsaturateddinitrile 12 (10.86 g, 93), m.p. 217-218 "C (Found: C, 69.6; H, 7.7; N, 4.95. C32H44N20, requrires C, 69.54; H, 8.02; N, 5.07); v,,,(film from THF)/cm-' 2233 (conj CN), 1739 (CO) and 1597 (conj. alkene); G,(CDCl,) 0.76 (3 H, s, 18-H), 0.81 (3 H, d, J 6,21-H), 0.99 (3 H, s, 19-H), 2.09 (3 H, s. OCOMe), 2.10(3 H, s, OCOMe) 3.66 (3 H, s, C02Me), 5.02 (1 H, m, 7P-H) and 5.12 (1 H, m, 12P- H); TLC R, 0.52 in hexane-ethyl acetate (1 :1).tert-Butyl p-Bromobenzyl Ether 15-Fresh potassium wire (2.49 g, 89.2 mmol) was introduced into a flask containing dry tert-butanol (80 cm3) under an atmosphere of argon. As the reaction proceeded the potassium reacted more slowly with the butanol, so after 45 min the flask was lowered into an ultrasonic bath and sonicated until the potassium had dissolved (1 5 rnin). Solid p-bromobenzyl bromide (1 8.60 g, 74.4 mmol) was added and the reaction mixture was sonicated for 2 h giving a milky white suspension. The flask was allowed to stand overnight, then sonicated for a further 6 h after which analysis by TLC implied that nop-bromobenzyl bromide remained. The reaction mixture was left standing for a further 24 h before being partitioned between ether (300 cm3) and water (1.2 dm3).The organic layer was washed with water (1 dm3 then 2 x 500 cm3) and then with brine (2 x 25 cm3). The solution was filtered through sand and evaporated under reduce pressure to give 18.18g of off-white crystals. This was distilled under vacuum to give a viscous liquid which on cooling gave tert-butyl ether 15as a colourless crystalline solid (16.71 g, 92), b.p. 89-91 "C at 0.2 mmHg; m.p. 58-59 "C (lit.,15 48-50 "C);S,(CDCI,) 1.28 (9 H, s, tert-butyl), 4.38 (2 H, s, CH,) and 7.20,7.44 (4 H, ABq, JAB8.3, aromatic); TLC R, 0.08 in hexane. J. CHEM. SOC. PERKIN TRANS. I 1993 Methyl 7a, 12a-Diacetoxy-3P-dicyanomethyl-3a-(p-hydroxy-methylphenyl)-cholan-24-oate19.-Magnesium turnings (1.34 g, 55.1 mmol) were flame dried under vacuum (0.1 mmHg) and cooled under an atmosphere of argon.To this was added a solution of tert-butyl p-bromobenzyl ether 15 (9.73 g, 40.02 mmol) in dry THF (80 cm3). The flask was lowered into an ultrasonic bath, the reaction was initiated with 1,2-dibro-moethane (0.2 cm3, 2.3 mmol) and the formation of the Grignard reagent was followed by GC,which implied that the reaction was ca. 98 complete after 30 min. The resultant solution was stirred in an ice bath, solid copper(1) cyanide (1.73 g, 19.3 mmol) was added under a flow of argon, and then the mixture was warmed to 10 "C. After waiting 15 rnin for all of the copper cyanide to dissolve, a solution of the unsaturated dinitrile 12 (9.84 g, 17.8 mmol) in dry THF (70 cm3) was added.After approximately 5 min the reaction was quenched with a saturated solution of ammonium chloride (2 cm3) and diluted with ether (50 cm3). The slurry was filtered and the filtrate was evaporated under reduced pressure to give inorganic salts and an oil. This was partitioned between water and ether and the organic layer re-filtered and evaporated to give a yellow foam (15.5 g). The foam was triturated in hexane (30 cm3), which extracted most of the tert-butyl benzyl ether from the insoluble product, leaving crude tert-butyl ether 17 (12.92 g, ca. 17.8 mmol). The above material was dissolved in dichloromethane (1 8 cm') containing trifluoroacetic acid (10 cm3, 70 mmol) and the solution was heated to 50deg;C.After 5f h, analysis by TLC* implied that no tert-butyl ether 17 remained, a small amount of the desired alcohol 19 was present, and the majority of the material was the trifluoroacetate 18. The reaction mixture was then partitioned between ether (60 m3)and aqueous ammonia (5 mol dm-3, 200 an3)in a separating funnel. After 30 min analysis by TLC showed only the presence of the benzyl alcohol 19. The aqueous layer was removed and the organic layer washed with water (2 x 100 cm3) and brine (2 x 50 cm3), and then passed through a silica gel plug (3 cm diameter x 4 cm length) and eluted with ethyl acetate (50 cm'). This gave, on evaporation under reduced pressure, an off-white solid (1 1.24 g) which was dissolved in l,l, 1-trichloroethane and caused to precipitate by adding hexane+ther (1 : 1; 20 cm').The precipitate was washed with ether to give, after evacuation (0.1 mmHg; 3 h), the benzyl alcohol 19 (10.16 g), m.p. 109-1 11 "C (Found: C, 61.55; H, 7.0; N, 3.35. C39H,2N20, with 1 molecule of CH3CC13 requires C, 62.00; H, 6.98; N, 3.53);1 v,,,(film from THF)/cm-' 3439 (OH), 2254w (CN) and 1740 (CO); SH(CDCI3)0.73 (3 H, S, 18-H), 0.77(3 H,d,J6.5,21-H), 1.04(3 H, s, 19-H), 2.00(3 H,s,OCOMe),2.01(3 H, s,OCOMe), 3.64(3 H, s, CO,Me), 4.37 1 H, s, CH(CN),, 4.75 (2 H, d, J4.6, with D20 shake goes to 2 H, s, benzyl), 4.96 (1 H, m, 7P-H), 5.05 (1 H, m, 12P-H) and 7.44,7.57 (4 H, ABq, JAB 11, aromatic). Evaporation and flash chromatography hexane-ethyl acetate (1 :l) of the mother liquor from the crystallisation gave more of the benzyl alcohol 19 (0.42 g), so that the total overall yield from unsaturated dinitrile 12 was 10.58 g (90).Methyl 7a,12a-Diacetoxy-3a-(p-azidomethylphenyl)-3~-dicyanomethyl-chofan-24-oate21 .-A solution of the alcohol 19 * Both rert-butyl ether 17 and trifluoroacetate 18 had R,0.55in hexane- ethyl acetate (I : I). However, TLC analysis was possible after treatment of samples with aqueous ammonia (as in the main procedure), converting 18 into alcohol 19 (R,0.30).t Crystals of compounds with 3a-aryl-3P-dicyanomethyl substitution tended to be quite extensively solvated, as is often the case for rigid molecules with irregular structures. It was not generally possible to remove all the solvent before microanalysis. 923 (5.08 g, 7.69 mmol) in dry THF (50 cm3) was stirred vigorously at 0deg;C under an atmosphere of argon.To this was added methanesulfonyl chloride (2.4 cm3, 3 1 mmol) immediately followed by diisopropylethylamine (2.7 cm3, 15.5 mmol), dropwise over 3 min. The light yellow mixture was diluted with ether (50cm3) then washed with water (3 x 100cm3) and brine. Evaporation of the solvent under reduced pressure gave the methanesulfonate 20 (5.67 g) as a white foam. To this was added tetramethylguanidinium azide (2.93 g, 15.4 mmol), l9 and the resulting mixture was dissolved in chloroform (1 5 cm3).$ After 30 rnin the solution was diluted with ether (50 cm3), washed with water (3 x 50 cm3) then brine, and passed through a silica plug which was then eluted with ether-chloroform (1 :1).Evaporation of the solvent gave the azide 21 (4.91 g, 93). Crystallisation from ether afforded an analytical sample, m.p. 105-1 10 "C(Found: C, 67.75; H, 7.8; N, 9.75. C3,HslNs0, with 0.5 mol of H,O requires C, 67.41; H, 7.54; N, 10.08);t v,,,(film from THF)/cm-' 2252w (CN), 2101s (N3), 1730br (CO) and 1600 (aromatic); S,(CDCl,) 0.73 (3 H, s, 18-H), 0.77 (3 H, d, J 6.5,21-H),1.04(3H,s,19-H),1.98(3H,s,OCOMe),2.00(3H,s, OCOMe),3.64(3 H,s,C02Me),4.401 H,s,CH(CN),,4.42(2 H, s, benzyl), 4.96 (1 H, m, 7P-H), 5.05 (1 H, m, 12P-H) and 7.40, 7.58 (4 H, ABq, JAB 8.5, aromatic); TLC R, 0.62 in hexane+thyl acetate (1 :1). Cyclo-bis7a, 1 2a,-diacetoxy-3 P-dicyanomethyl-3a-(4-methyl-enepheny1)cholanamidel 8.-The azido ester 21 (1.868 g, 2.72 mmol) and triphenylphosphine (1.42 g, 5.44 mmol) were stirred in THF-methanol-water (2: 2: 1; 15 cm') for 1 h at room temperature.Lithium hydroxide monohydrate (0.457 g, 10.9 mmol) and THF-methanol-water (2 :2 :3; 7 cm3) were added. After stirring for 1.5 h the mixture was neutralised with aqueous hydrochloric acid (2.18 cm3; 5 mol dm-3). Evaporation of the solvent under reduced pressure gave an off-white gum which was dissolved in chloroform-methanol (1 :1) and put onto a silica gel column. Elution with chloroform washed out the triphenylphosphine oxide, and then elution with chloroform- methanol ( 10:1) grading to (2 :l) gave amino acid 22 (1.264 g), followed by a mixture of 22 and lithium chloride.The lithium chloride was removed from the latter by washing a THF- chloroform solution of the mixture with water, after which evaporation and flash chromatography gave a further sample of amino acid 22 (0.280 g, 87 in total); v,,,(film from THF.)/m-' 3440-2650br (NH), 2254w (CN) and 1735 (CO); TLC R, 0.14 in chloroform-methanol (5 :1). To a solution of the above material (0.196 g, 0.30 mmol) in dry THF (35 cm3), stirred under an atmosphere of argon, was added diethyl phosphorocyanidate (DEPC) 23 (1 15 mm3, 0.759 mmol). After 10 min, dipotassium hydrogen phosphate C0.36 g, 2.0 mmol; previously flame dried under vacuum (0.1 mmHg), and cooled under argon was added as a slurry in dry THF (5 cm3) to the reaction mixture.After stirring for 12 days the solvent was removed under reduced pressure and the residue partitioned between water and chloroform. The solvent was evaporated from the organic layer under reduced pressure to give a viscous oil which was dissolved in a minimum volume of ethyl acetate-methanol and subjected to flash chromatography using hexane-ethyl acetate (1 :5) as eluent. Evaporation of the solvent and recrystallisation of the product from chloroform- methanol-THF gave the cholaphane 8 (80 mg, 42) as fine white needles, which became a fine white powder after evacuation (0.1 mmHg; 50 "C) for 6 h, m.p. 253-255 "C (Found: CAUTION The interaction of azide ions and chlorinated solvents can lead to explosive side-products." As part of a separate investigation we have found that, over a period of days at elevated temperatures, tetramethylguanidinium aide and chloroform can react to give significant quantities of a shock-sensitive, distillable liquid.21 C, 70.3; H, 7.85; N, 6.3.C76H,,N6010 with 2 mol Of H20 requires C, 70.67; H, 7.96; N, 6.5 I);* v,,,(film from THF)/cm-' 3329 (NH), 2254w (CN), 1741 and 1663 (CO); 6,(300 MHz; CDC1,)0.744(6H,s, 18-H),0.778(6H,d, J4.8,21-H), 1.042(6 H, s, 19-H), 1.986 (6 H, s, OCOMe), 2.000 (6 H, s, OCOMe), 2.344(2H,br,d,J14,4P-H),2.530(2H,t,J14,4a-H),4.3662 H, s, CH(CN),, 4.342 (2 H, dd, J 14,6, CHN), 4.5 15 (2 H, dd, J 14,6, CHN), 4.959 (2 H, m, 7P-H), 5.759 (2 H, t, J6, NH) and 7.455,7.536 (8 H, ABq, JAB8.4, aromatic); TLC R, 0.45 in ethyl acetate.Acknowledgements Financial support was provided by EOLAS, the Irish Science and Technology agency. We are grateful to Diamalt GmbH for gifts of cholic acid and methyl cholate. * Crystals of compounds with 3a-aryl-3P-dicyanomethyl substitution tended to be quite extensively solvated, as is often the case for rigid molecules with irregular structures. It was not generally possible to remove all the solvent before microanalysis. References 1 R. P. Bonar-Law and A. P. Davis, J. Chem. Soc., Chem. Commun., 1989, 1050. 2 R. P. Bonar-Law, A. P. Davis and B. A. Murray, Angew. Chem., Int. Ed. Engl., 1990, 29, 1407; K. M. Bhattarai, R. P. Bonar-Law, A. P. Davis and B. A. Murray, J.Chem. Soc., Chem. Commun., 1992,752. 3 A. P. Davis and M. G. Orchard, J. Chem. Soc., Chem. Commun., 1991,612. 4 See e.g. C. J. Burrows and R. A. Sauter,J. Inclusion Phenomena, 1987, 5,117; R. P. Bonar-Law and J. K. M. Sanders,J. Chem. Soc., Chem. J. CHEM. SOC. PERKIN TRANS. I 1993 Commun., 1991, 574; J.4. Kikuchi, C. Matsushima, K. Suehiro, R. Oda and Y. Murakami, Chem. Lett., 1991, 1807. 5 G. Jones, Org. React., 1967, 15, 204. 6 H. 0.House, W. L.. Respass and G. M. Whitesides, J. Org. Chem., 1966, 31, 3128; D. Nasipuri, A. Sarker and S. K. Konar, J. Org. Chem., 1982,47,2840. 7 A. P. Davis, T. J. Egan, M. G. Orchard, D. Cunningham and P. McArdle, Tetrahedron, 1992,48, 8725. 8 B. H. Lipshutz, R. S. Wilhelm and J. A. Kozlowski, Tetrahedron, 1984,40,5005.9 A. S. Jones, M. Webb and F. Smith, J. Chem. Soc., 1949,2164. 10 J. F. Baker and R. T. Blickenstaff, J. Org. Chem., 1975,40,1579; G. Hofle, W. Steglich and H. Vorbruggen, Angew. Chem. Int. Ed. Engl., 1978, 17, 569. 11 J. R. Dias and R. Ramachandra, Synth. Commun., 1977,7,293. 12 P. H. J. Carlsen, T. Katsuki, V. S. Martin and K. B. Sharpless, J. Org. Chem., 1981,46,3936. 13 J. Mirek, M. Adamczyk and M. Mokrosz, Synthesis, 1980,296. 14 R. P. Bonar-Law, A. P. Davis and B. J. Dorgan, Tetrahedron Lett., 1990,31, 6721; R. P. Bonar-Law, A. P. Davis, B. J. Dorgan, M. T. Reetz and A. Wehrsig, Tetrahedron Lett., 1990,31,6725. 15 M. Nassal, Liebigs Ann, Chem., 1983, 1510. 16 P. T. Lansbury and V. A. Pattison, J. Org. Chem., 1962,27, 1933. 17 0.Mitsunobu, Synthesis, 1981, 1. 18 H. Loigner and E. Zbiral, Helv. Chim. Acla, 1976,59,2100. 19 A. J. Papa, J. Org. Chem., 1966,31, 1426. 20 L. Bretherick, Chem. and Eng. News, 1986,64 (51), 2. 21 A. P. Davis and M. G. Orchard, unpublished work. 22 M. Vaultier, N. Knouzi and R. Carrie, Tetrahedron Lett., 1983, 24, 763. 23 T. Shioiri, Y. Yokayama, Y. Kasai and S. Yamada, Tetrahedron, 1976,32,2211. 24 D. D. Perrin, W. L. F. Armarego and D. R. Perrin, PuriJicationof Laboratory Chemicals, 2nd edn., Pergamon Press, Oxford, 1980. Paper 2/06854K Received 24th December 1992 Accepted 14th Junuary 1993