Syntheses of gibberellins A93and A94, naturalproducts detected in wheat grain

摘要

J. Chem. Soc. Perkin Trans. 1 1997 751 Syntheses of gibberellins A93 and A94 natural products detected in wheat grain Stuart Findlow,a Paul Gaskin,b Polly A. Harrison,a John R. Lenton,b Martin Penny a and Christine L. Willis *,a a School of Chemistry University of Bristol Cantockrsquo;s Close Bristol BS8 1TS UK b IACR-Long Ashton Department of Agricultural Sciences University of Bristol Long Ashton Bristol BS18 9AF UK Extracts of mature wheat grain have been analysed by GCndash;MS and found to contain two new metabolites. The syntheses of the two new gibberellins 1lsquor;-hydroxy-2lsquor;,3lsquor;-epoxyGA9 4 and 1lsquor;-hydroxy-2lsquor;,3lsquor;-epoxy- GA20 5 from the fungal gibberellins GA7 and GA3 are described. A range of protecting groups for the 7-carboxylic acid which can be removed under mild conditions are compared. The allyl ester proved most valuable in the manipulation of these multifunctional molecules and it is removed with tetrakis(triphenylphosphine)palladium(0) triphenylphosphine and potassium isobutyrate.The structures of the new natural products in wheat were confirmed to be 4 and 5 by comparison with the GCndash;MS data from the synthetic samples and are assigned the trivial names GA94 and GA93 respectively. We have described previously the synthesis of two gibberellins 1 and 2 containing a 1a-hydroxy-2b,3b-epoxide moiety in ring A.1 Only one of the known gibberellins GA6 3 has the 2b,3bepoxide function.2 Hydroxylation at C-1 occurs in both plant and fungal systems,3 however comparison of the GCndash;mass spectra of 1 and 2 with detected but uncharacterised gibberellins 4 revealed that neither compound is a natural product.On recent examination of extracts of mature wheat grain (Triticum aestivum cv Maris Huntsman) by GCndash;MS two new compounds were detected with similar but not identical mass spectral data to 1 and 2 (Table 1). A series of 1b-hydroxygibberellins have been identified previously by GCndash;MS in extracts of wheat and their structures confirmed by direct comparison with synthetic samples.5 Hence we proposed that the two new natural products were 4 and 5 with the 1b-hydroxy-2b,3b-epoxide functionality in ring A. We now describe the syntheses of 4 and 5 which confirm the structures of the natural products (now assigned the trivial names6 GA93 5 and GA94 4) and enable the metabolic grid to account for the presence of the 1b-hydroxy- GAs in wheat to be completed.5 Results and discussion Synthesis of 1lsquor;-hydroxy-2lsquor;,3lsquor;-epoxide 4 (Scheme 1) Gibberellin A7 7 was converted to the 1a-hydroxy-2b,3bepoxide 8 via a Payne-type rearrangement of the iododiol 6 as previously described.1 A method for inversion of the stereochemistry of the 1-hydroxy group was then required.The most obvious choice was an SN2 displacement of a suitable derivative of the 1a-alcohol. As with other sterically crowded systems 6 the direct inversion of a secondary alcohol under standard or modified Mitsunobu reaction conditions 7 has proven to be unsuccessful.8 Caesium acetate has been successfully used for the inversion of alcohols at C-2 and C-3 of gibberellins via the corresponding methanesulfonates.9 Thus 8 was converted to the methanesulfonate 9 under standard conditions. However treatment of 9 with caesium acetate in refluxing toluene simply returned starting material whereas in N,N-dimethylformamide (DMF) at 80 8C an intractable gum was obtained.The density of functionality surrounding C-1 (the lactone and epoxide) appeared to be interfering with selective attack of the nucleophile and so the use of a two-stage oxidation procedure for inversion of the alcohol was investigated. Several procedures for the oxidation of the 1a-alcohol 8 to the ketone 10 were compared. Although Fetizonrsquo;s reagent 10 and pyridinium dichromate (PDC)11 returned mainly starting material oxidation was successfully achieved under both Swern conditions 12 and with tetrapropylammonium perruthenate (TPAP) N-morpholine N-oxide (NMO) and molecular sieves 13 giving the required ketone 10 in 90 and 98 yield respectively.Ketone 10 was then reduced with sodium borohydride in methanol giving a 2 1 mixture (by 1H-NMR spectroscopy) of the 1a-alcohol 8 and the required 1b-alcohol 11. It has previously been shown that reduction of a ketone at C-3 of the gibberellins with aluminium isopropoxide in propan-2-ol gives predominantly the 3b-alcohol.14 However reduction of the 1-oxo-2b,3b-epoxide 10 under these conditions gave a mixture of 1a 1b-alcohols 8 and 11 in a disappointing 10 1 ratio. The mixture of alcohols proved to be inseparable by either flash chromatography or medium pressure chromatography. In an attempt to prepare derivatives of 8 and 11 which may be separated the mixture was treated with acetic anhydride in pyridine giving the acetates 12 and 13; these were also inseparable.Therefore it was apparent that a stereoselective method was required for ketone reduction in which hydride delivery was exclusively from the a-face at C-1. Bell and Turner 15 reported that reduction of 3-oxogibberellin 7-carboxylic acids with K-Selectride gives 95 yield of the 3balcohol (i.e. with delivery of hydride from the a-face at C-3). The stereochemical outcome has been rationalised in terms of steric and Coulombic inhibition of approach of the reagent to the b-face of the gibberellin by a borate complex with the 7-carboxylic acid. This hypothesis is supported by the fact that reduction of 3-oxogibberellin 7-methyl esters with K-Selectride 752 J. Chem. Soc. Perkin Trans. 1 1997 Scheme 1 Reagents and conditions i aq. KOH (0.8 mol dm23) THF 14 h room temp.then adjust to pH 9 I2 CH2Cl2 2 h; ii aq. KOH (0.1 mol dm23); iii CH2N2 MeOH; iv MsCl pyridine; v TPAP NMO mol. sieves; vi NaBH4 MeOH; vii Ac2O pyridine; viii Me2SO (COCl)2 Pri 2EtN CH2Cl2 Me2CO; ix K-Selectride KH2PO3 gives the 3a-alcohol in 93 yield.16 Thus the 1a-hydroxy-2b,3bepoxide 7-acid 1 was oxidised under Swern conditions to give the required keto acid 14 in 66 yield. Due to the increased polarity of the acid 14 compared with the methyl ester 10 it proved necessary to add 14 as a solution in acetone to the activated sulfonium species in dichloromethane. Finally reduction of keto acid 14 with K-Selectride proceeded with excellent stereofacial selectivity to give the required 1b-alcohol 4 as a crystalline solid in 24 overall yield from GA7. Synthesis of 1lsquor;,13-dihydroxy-2lsquor;,3lsquor;-epoxide 5 Following the successful synthesis of 4 it was proposed to use a similar approach for the preparation of the 13-hydroxylated compound 5.Gibberellin A3 was readily converted to the 1a-hydroxy-2b,3b-epoxide 2 as previously described (Scheme 2).1 However it was found that although oxidation of the 1-hydroxy 7-methyl ester 15 proceeded smoothly with either TPAP or under Swern conditions giving 16 (in 93 and 65 yield respectively) attempted oxidation of the corresponding 1-hydroxy 7-carboxylic acid 2 under similar conditions simply returned starting material. The problem was believed to be due to the polar nature of the dihydroxy acid compared with either the dihydroxy ester 15 or hydroxy acid 1. Further attempts to oxidise the dihydroxy acid 2 with PDC in DMF or using the Parikh modification of the Moffatt oxidation 17 gave an intractable gum.Reduction of the keto ester 16 with K-Selectride gave a 3 1 mixture of the 1b 1a-alcohol 17 and 15 (by 1H NMR spectroscopy) which proved to be inseparable by flash chromatography or MPLC. It was therefore apparent that the presence of the 7-carboxylic acid is essential to achieve good stereofacial selectivity in the reduction of the 1-ketone with K-Selectride. Since keto acid 18 could not be prepared via direct oxidation of the dihydroxy acid 2 methods for converting a 7-methyl ester to an acid in the presence of a 2b,3b-epoxide were investigated. Treatment of 8 with aqueous sodium hydroxide gave a complex mixture of products and reaction with sodium propanethiolate in hexamethylphosphoramide (HMPA)18 resulted in attack on the 2b,3b-epoxide as well as deprotection at C-7 giving 19 as the major product.Scheme 2 Reagents and conditions i aq. KOH (0.8 mol dm23) THF 14 h room temp. then adjust to pH 9 I2 CH2Cl2 2 h; ii aq. KOH (0.1 mol dm23); iii CH2N2 MeOH; iv TPAP NMO mol. sieves; v KSelectride KH2PO3 THF J. Chem. Soc. Perkin Trans. 1 1997 753 In the light of these problems the approach to the synthesis of 5 had to be modified. A protecting group for the 7-acid was required which would effectively reduce the polarity of the dihydroxy acid 2 to enable oxidation of the 1-alcohol to a ketone and which then could be removed prior to reduction of the 1-ketone to the required 1b-alcohol. A range of protecting groups was examined (Scheme 3). Recently it has been reported that a cyanomethyl ester may be used to protect the 7-carboxylic acid of gibberellins.19 The acid 2 was heated to reflux with chloroacetonitrile to give the cyanomethyl ester 20 in quantitative yield.Oxidation of 20 under Swern conditions gave the keto epoxide 21 but attempts to remove the protecting group with either a mixture of potassium carbonatendash;potassium hydrogen carbonate in acetonendash; water or with potassium hydroxide gave an intractable mixture. Therefore although the protecting group was simple to put on and compatible with the oxidation conditions the route failed at the deprotection step the reaction conditions proving incompatible with the epoxy keto and lactone moieties in ring A. Another protecting group we considered was the 3- methylbut-2-enyl ester.20 The ester 22 was formed in 95 yield via a 1,3-dicyclohexylcarbodiimide (DCC)ndash;4-(N,N-dimethylamino) pyridine (DMAP) mediated coupling reaction.Oxidation of 22 with TPAP proceeded smoothly giving the keto ester 23 in 68 yield. Again however attempts to remove the ester to give the required 1-keto-2b,3b-epoxide 7-acid proved unsuccessful. Treatment of 23 with iodine in cyclohexane returned starting material when the reaction was conducted at room temperature and gave an intractable mixture at elevated temperatures. In this case it is possible that not only is the densely functionalised A ring sensitive to the reaction conditions but Scheme 3 also a Wagnerndash;Meerwein type rearrangement of the C/D rings may occur in the presence of iodine. Finally the allyl ester 24 was prepared from 2 using a DCCndash; DMAP mediated coupling with prop-2-en-1-ol (Scheme 4).Oxidation of alcohol 24 with TPAP successfully gave the ketone 25 (52 yield over the two steps). Removal of the protecting group was achieved with a mixture of potassium isobutyrate triphenylphosphine and tetrakis(triphenylphosphine)palladium( 0) 21 giving the required 1-keto-2b,3b-epoxide 18 in 35 yield. Reduction of 18 with K-Selectride gave the 1b-hydroxy- 2b,3b-epoxide 5 as a crystalline product in 5 overall yield from GA3. Comparison of 4 and 5 with the new metabolites in extracts of wheat grain and proposed biosynthesis. Mature grain of Triticum aestivum (cv Maris Huntsman) was extracted and the GAs were purified from the BuOH-soluble fraction as described in the Experimental section. The purified extract was methylated with diazomethane and derivatised with trimethylsilyl chloridendash;1,1,1,3,3,3-hexamethyldisilazanendash; pyridine prior to analysis by GCndash;MS.3 The extract contained two new metabolites which had not previously been identified.4 The synthetic samples of the 1b-hydroxy-2b,3b-epoxides 4 and 5 were separately derivatised as the trimethylsilyl ether methyl esters and the GCndash;MS data compared with those of the metabolites in wheat (Table 1).These data confirm the tentative assignments of the natural products and 4 is now assigned the trivial name GA94 and 5 is GA93. One of the most dramatic biological effects of gibberellins is their enhancement of stem elongation.3 The biological activities of GA93 and GA94 were assessed in a Tan-ginbozu dwarf rice immersion assay. It was found that in comparison with GA3 and GA7 neither GA93 nor GA94 is a potent stem elongation promoter (Table 2).Further bioassays are required to discover the significance of 1b-hydroxylation and the 2b,3b-epoxide in the role of GAs in plant growth and development in wheat. Experimental General experimental details have been described in a previous paper.22 For the numbering scheme used throughout the paper see structure 1. Unless otherwise stated all reactions were worked-up by the following standard procedure. The reaction mixture was poured into water and ethyl acetate. The pH was adjusted to 2 with 2 M hydrochloric acid and the products extracted with ethyl acetate. The combined organic extracts Scheme 4 Reagents and conditions i CH2 CHCH2OH DCC DMAP; ii TPAP NMO mol. sieves; ii Me2CHCO2K Ph3P Pd(PPh3)4 CH2Cl2 EtOAc; v K-Selectride KH2PO3 THF 754 J.Chem. Soc. Perkin Trans. 1 1997 Table 1 Comparison of the mass spectra of methylated trimethylsilylated derivatives of synthetic and wheat grain GA94 4 and GA93 5 and the 1ahydroxy- 2b,3b-epoxides 1 and 2 GA/source KRI Characteristic ions m/z ( base peak) 1a-Hydroxy-2b,3b-epoxy GA9 1 Synthetic 2606 432 (M+ 2) 417 (3) 385 (100) 288 (30) 201 (59) 145 (88) 73 (90) GA94 4 (1b-2b,3b-epoxy GA9) Synthetic 2505 432 (M+ 2) 414 (1) 403 (3) 400 (3) 382 (2) 370 (3) 356 (4) 313 (12) 310 (13) 301 (41) 300 (44) 283 (6) 268 (9) 255 (6) 241 (100) 240 (44) 221 (11) 211 (9) 132 (18) Wheat grain 2504 432 (M+ 2) 414 (1) 403 (3) 400 (3) 382 (2) 370 (2) 356 (4) 313 (12) 310 (12) 301 (39) 300 (41) 283 (6) 268 (10) 255 (6) 241 (100) 240 (41) 221 (11) 211 (9) 132 (17) 1a-Hydroxy-2b,3b-epoxy GA20 2 Synthetic 2646 520 (M+ 41) 473 (11) 376 (41) 303 (100) 235 (77) 207 (49) 73 (87) GA93 5 (1b-2b,3b-epoxy GA20) Synthetic 2663 520 (M+ 100) 505 (9) 491 (8) 461 (22) 447 (2) 401 (5) 389 (5) 376 (7) 347 (8) 329 (15) 305 (30) 303 (98) 279 (10) 235 (41) 208 (40) 207 (90) 194 (16) 193 (15) Wheat grain 2662 520 (M+ 77) 505 (9) 491 (8) 461 (22) 447 (2) 401 (4) 389 (6) 376 (7) 347 (9) 329 (15) 305 (31) 303 (100) 279 (9) 235 (40) 208 (37) 204 (89) 194 (16) 193 (24) were washed with water dried over anhydrous sodium sulfate and the solvent removed in vacuo.ent-2middot;,3middot;-Epoxy-10lsquor;-hydroxy-1lsquor;-methylsulfonyloxy-20-norgibberell- 16-ene-7,19-dioic acid 19,10-lactone 7-methyl ester 9 1a-Hydroxy-2b,3b-epoxide methyl ester 8 (76 mg 0.21 mmol) in pyridine (2.5 cm3) was stirred with methanesulfonyl chloride (26 ml) for 1 h at room temperature.The usual work-up was followed by purification by flash chromatography. Elution with 30 ethyl acetatendash;light petroleum gave methanesulfonate 9 as a foam (79 mg) (Found M+ 438.1349. C21H26O8S requires M 438.1348); dH 1.31 (s 18-H3) 2.61 (m 13-H) 2.72 (d J 11 6-H) 3.09 (d J 11 5-H) 3.16 (s OSO2Me) 3.25 and 3.29 (2 times; d each J 3.5 2- and 3-H) 3.72 (s CO2Me) 4.86 (br s 17-H) and 4.9 (br s 17-H and 1-H); m/z 438 (M+ 48) 406 (39) 402 (31) 378 (30) 360 (65) 254 (93) 221 (68) and 43 (100). Oxidation of hydroxy ester 8 Under Swern oxidation conditions. Oxalyl chloride (freshly distilled 0.28 mmol 0.25 ml) and dimethyl sulfoxide (7 mmol 0.51 ml) were stirred in dichloromethane (5 cm3) at 278 8C for 5 min. 1-Hydroxy-2,3-epoxide 8 (107 mg 0.30 mmol) was added dropwise in dichloromethane (5 cm3) and the reaction mixture stirred for 1.5 h at 278 8C.N,N-Diisopropylethylamine (5.6 mmol 1 ml) was added and the reaction allowed to warm to room temperature over 1.5 h. The usual work-up was followed by purification by flash chromatography. Elution with 40 ethyl acetatendash;light petroleum gave ent-2a,3a-epoxy-10b-hydroxy-1- oxo-20-norgibberell-16-ene-7,19-dioic acid 19,10-lactone 7- methyl ester 10 which was crystallised from ethyl acetatendash;light petroleum as needles (12 mg) mp 155ndash;157 8C (Found M+ 358.1412. C20H22O6 requires M 358.1416); dH 1.41 (s 18-H3) 2.80 (d J 11 6-H) 3.31 and 3.51 (2 times; d J 3.5 2- and 3-H) 3.39 (d J 11 5-H) 3.74 (s CO2Me) 4.99 and 5.26 (2 times; br s 17-H2); m/z 358 (M+ 48) 326 (40) 298 (13) 241 (22) 228 (100) 201 (19) and 91 (45).With tetrapropylammonium perruthenate (TPAP). Hydroxy ester 8 (104 mg 0.29 mmol) N-methylmorpholine N-oxide (51 mg 0.43 mmol) and molecular sieves (crushed and dried 145 mg) were stirred at room temperature under nitrogen in dichloromethane (900 ml) and acetonitrile (100 ml). Tetrapropylammonium perruthenate (TPAP) (5 mg 0.015 mmol catalytic) was added and the reaction stirred for 2 h. To work-up the mixture it was passed through a short column of silica which was eluted with dichloromethane (100 cm3) and then ethyl acet- Table 2 Rice seedling bio-assay Second leaf Sheath/mm Total/mm Control GA3 GA7 GA93 GA94 15 40 50 16 21 23 55 64 26 33 ate (100 cm3). The ketone 10 was obtained in 100 yield the 1H NMR and mass spectral data being identical to those previously obtained.Reduction of keto ester 10 With sodium borohydride. Sodium borohydride (12.7 mg 0.34 mmol) in methanol (1 cm3) was added to a stirred solution of the epoxy ketone 10 (60 mg 0.17 mmol) in methanol (4 cm3). The mixture was stirred for 1 h at room temperature and then worked-up as usual. Purification by flash chromatography and elution with 25 ethyl acetate in light petroleum gave a 2 1 mixture of the 1a-hydroxy- and 1b-hydroxy-epoxides 8 and 11 (55 mg). The epimeric mixture showed as a single spot in three different solvent systems. dH(major 1a-alcohol 8; as previously described); dH(minor 1b-alcohol 11) 1.29 (s 18-H3) 2.61 (m 13-H) 2.69 (d J 10.5 6-H) 3.13 (d J 10.5 5-H) 3.41 (m 2-H and 3-H) 3.71 (s OCH3) 3.90 (d J 8 1-H) 4.86 and 4.98 ( 2 br s 17-H2).The trimethylsilyl derivative of 11 gave KRIdagger; 2505; m/z 432 (M+ 1) 400 (3) 370 (3) 356 (4) 324 (2) 300 (41) 283 (6) 268 (10) 241 (100) 240 (43) 221 (13) 211 (10) 195 (9) 132 (19) 119 (10) 93 (20) 73 (66) 55 (11) and 44 (30). Meerweinndash;Verleyndash;Ponndorf reduction. Aluminium foil (0.35 g 13.0 mmol) and mercuric chloride (3.5 mg 0.013 mmol) in propan-2-ol (15 cm3) were heated to reflux. Carbon tetrachloride (60 ml) was added and the mixture was heated at a gentle reflux for 3 h. Keto ester 10 (20 mg 5.59 times; 1022 mmol) in propan-2-ol (2 cm3) was added and the mixture was heated for a further 3 h with distillation of the acetone produced. The mixture was then cooled to room temperature and worked-up as usual. The crude product was purified by flash chromatography. Elution with 25 ethyl acetate in light petroleum gave the 1- alcohols 8 and 10 as a ~10 1 (a:b) epimeric mixture; spectroscopic data as previously described.ent-1middot;- and 1lsquor;-Acetoxy-2middot;,3middot;-epoxy-10lsquor;-hydroxy-20-norgibberell- 16-ene-7,19-dioic acid 19,10-lactones 7-methyl ester 12 and 13 The mixture of epimeric alcohols 8 and 10 (27 mg) in pyridine (750 ml) and acetic anhydride (700 ml) was stirred at room temperature for 2 h. The usual work-up gave a gum which was purified by flash chromatography. Elution with 10 ethyl acetate in light petroleum gave the epimeric acetates 12 and 13 in a ratio of 2 1 (a:b). The acetates also showed as one spot by TLC in three different solvent systems ethyl acetatendash;light petroleum (1 5) diethyl etherndash;hexane (3 10) and dichloromethanendash;methanol (9 1) (Found M+ 402.1671.C22H26O7 requires M 402.1679); dH(major 1a-acetate 13) 1.31 (s 18-H3) 2.15 (s OCOCH3) 2.62 (m 13-H) 2.72 (d J 11 6-H) 2.97 (d J 3.5 2- or 3-H) 3.08 (d J 11 5-H) 3.20 (d J 3.5 2- or 3-H) 3.72 (s OCH3) 4.86 and 4.97 (2 br s 17-H2) and 5.18 (d J 0.5 1-H); dH(minor 1bacetate 12) 1.32 (s 18-H3) 2.19 (s OCOCH3) 2.64 (m 13-H) 2.69 (d J 10.5 6-H) 3.28 (d J 10.5 5-H) 3.31 (d J 4 3-H) dagger; KRI (Kovats Retention Index) is described in detail in ref. 4. J. Chem. Soc. Perkin Trans. 1 1997 755 3.44 (t J 4 2-H) 3.72 (s OCH3) 4.86 and 4.97 (2 br s 17-H2) and 5.01 (d J 4 1-H); m/z 402 (M+ 13) 384 (5) 370 (74) 342 (13) 310 (55) 280 (37) 254 (100) 240 (55) 239 (78) 238 (80) 221 (51) 171 (35) 105 (38) and 91 (76). ent -2middot;,3middot;-Epoxy-10lsquor;-hydroxy-1-oxo-20-norgibberell-16-ene-7 19-dioic acid 19,10-lactone 14 Dichloromethane (5 cm3) and freshly distilled oxalyl chloride (0.24 cm3 2.9 mmol) were added to a 10 cm3 flame-dried roundbottomed flask under nitrogen at 278 8C .Dimethyl sulfoxide (0.49 cm3 7.22 mmol) was added dropwise to the solution and the reaction stirred for 5 min. 1a-Hydroxy 2b,3b-epoxide 1 (100 mg 0.29 mmol) was dissolved in a minimum of acetone and dichloromethane and added dropwise to the reaction mixture. After 1.5 h at 278 8C N,N-diisopropylethylamine (0.97 cm3 5.7 mmol) was added and the reaction mixture allowed to warm to room temperature for 1.5 h. The reaction was diluted with dichloromethane (2 cm3) and water (1 cm3) and then the usual work-up was carried out. Purification by column chromatography eluting with 40 ethyl acetatendash;light petroleum gave the epoxy ketone 14 as a white solid which was recrystallised from ethyl acetatendash;light petroleum as needles (66 mg) mp 155ndash; 156 8C (Found C 66.3; H 5.8.C19H20O6 requires C 66.28; H 5.81); dH 1.46 (s 18-H3) 2.82 (d J 11 6-H) 3.31 and 3.50 (2 times; d J 3.5 2- and 3-H) 3.35 (d J 11 5-H) and 4.94 (2 times; br s 17-H2); m/z 344 (M+ 38) 326 (30) 274 (32) 255 (32) 228 (56) 211 (28) and 91 (100). ent-2middot;,3middot;-Epoxy-1middot;,10lsquor;-dihydroxy-20-norgibberell-16-ene-7 19-dioic acid 19,10-lactone 4 A solution of the epoxy ketone 14 (100 mg 0.29 mmol) in tetrahydrofuran was added to dry powdered potassium dihydrogen orthophosphate (0.24 g 2.03 mmol) under nitrogen. The reaction was cooled to 278 8C and K-Selectride (potassium tri-sec-butylborohydride) (1.2 cm3 1 M solution in tetrahydrofuran) was added dropwise over 5 min stirring continuously.The solution was allowed to warm to room temperature over 2 h and then worked-up as usual. Purification by column chromatography eluting with 40 ethyl acetatendash;light petroleum gave GA94 4 which was crystallised from ethyl acetatendash; light petroleum as a white solid (55 mg) mp 142ndash;144 8C (Found C 65.5; H 6.2. C19H22O6 requires C 65.9; H 6.4); dH 1.36 (s 18-H3) 2.63 (m 13-H) 2.72 (d J 10.5 6-H) 3.14 (d J 10.5 5-H) 3.42 (2 times; s 2- and 3-H) 3.92 (d J 4 1-H) and 5.92 (2 times; br s 17-H2); dC 14.2 (C-18) 17.6 (C-11) 31.6 (C-12) 39.4 (C-13) 42.5 44.2 (C-14 and C-15) 47.2 and 47.6 (C-3 and C-9) 48.6 and 49.3 (C-4 and C-8) 51.4 51.9 (C-5 and C-6) 59.7 (C- 2) 69.2 (C-1) 92.4 (C-10) 107.7 (C-17) 156.2 (C-16) 172.2 and 176 (C-7 and C-19); m/z 346 (M+ 28) 328 (18) 310 (17) 287 (96) 240 (65) 195 (36) and 91 (100).Oxidation of 1middot;,13-dihydroxy-2lsquor;,3lsquor;-epoxide methyl ester 15 Under Swern oxidation conditions. Oxalyl chloride (freshly distilled 174 ml 1.99 mmol) and dimethyl sulfoxide (350 ml 4.99 mmol) were stirred in dichloromethane (5 cm3) at 278 8C for 5 min. Diol 151 (75 mg 0.20 mmol) in dichloromethane (5 cm3) was added dropwise and the reaction mixture stirred for 1.5 h at 278 8C. N,N-Diisopropylethylamine (690 ml 3.99 mmol) was added and the resulting yellow solution was allowed to warm to room temperature over 1.5 h. The usual work-up was followed by purification by flash chromatography. Elution with 70 ethyl acetatendash;light petroleum gave ent-2a,3a-epoxy-10b,- 13-dihydroxy-1-oxo-20-norgibberell-16-ene-7,19-dioic acid 19,- 10-lactone 7-methyl ester 16 as a gum (49 mg 65) (Found M+ 374.1360.C20H22O7 requires M 374.1366); dH 1.42 (s 18-H3) 2.80 (d J 10.5 6-H) 3.31 and 3.52 (2 times; d each J 3.5 2- and 3-H) 3.41 (d J 10.5 5-H) 3.75 (s CO2Me) 4.98 and 5.28 (2 times; br s 17-H2); m/z 374 (M+ 25) 346 (3) 342 (13) 315 (10) 304 (9) 275 (21) 231 (61) 135 (100) and 91 (97). With TPAP. To a stirred solution of diol 15 (50 mg 0.14 mmol) molecular sieves (4 Aring; vacuum oven dried 100 mg) and N-methylmorpholine N-oxide (24 mg 0.21 mmol) in dichloromethanendash;acetonitrile (2 cm3 200 ml) was added TPAP (2.4 mg 0.007 mmol). After 2 h the solvent was removed in vacuo and the resulting gum purified by flash chromatography. Elution with 70 ethyl acetatendash;light petroleum gave ketone 16 (48 mg) whose 1H NMR and mass spectral data were identical to those previously obtained.Reduction of keto ester 16 with K-Selectride Keto ester 16 (169 mg 0.45 mmol) in tetrahydrofuran (10 cm3) was added to dry powdered potassium hydrogen orthophosphate (374 mg 3.16 mmol). The mixture was cooled to 278 8C then K-Selectride (1 M in tetrahydrofuran 2.26 cm3 2.25 mmol) was added dropwise over 5 min. The reaction was allowed to warm to 0 8C over 2 h with stirring. The usual workup was followed by purification by flash chromatography. Elution with 50 ethyl acetatendash;light petroleum with three drops acetic acid added gave a mixture of 1a-alcohol 15 and 1balcohol 17 (80 mg) in the ratio 1 3. Separation was attempted using medium pressure liquid chromatography eluting with 50 ethyl acetatendash;light petroleum (with three drops acetic acid added) which gave a 1 7 mixture (by 1H NMR) of the 1aalcohol 15 and the 1b-alcohol 17 as a white foam (13 mg) (Found M+ 376.1526.C20H24O7 requires M 376.1522); dH 1.31 (s 18-H3) 2.68 (d J 10 6-H) 3.18 (d J 10 5-H) 3.40 (m 2- and 3-H) 3.74 (s CO2Me) 3.91 (br s 1-H) 4.96 and 5.25 (2 times; br s 17-H2). GCndash;MS of the trimethylsilyl derivative of 17 gave KRI 2663; m/z 376 (M+ 38) 344 (78) 326 (37) 317 (34) 275 (32) 231 (100) 163 (67) and 135 (49). Further elution gave 1a-alcohol 15 (24 mg); 1H NMR and mass spectral data being identical to those previously assigned. The trimethylsilyl derivative of 15 gave KRI 2646. Treatment of hydroxy ester 8 with sodium propanethiolate Propanethiol (0.42 cm3) was added to sodium hydride (144 mg) (previously washed with light petroleum and dried under vacuum) in hexamethylphosphoramide (HMPA) (3 cm3) and the solution stirred for 2 h at room temperature under nitrogen.The complex formed was allowed to stand for 1 h before use. 1a-Hydroxy-2,3-epoxide methyl ester 8 (30 mg 0.08 mmol) was treated with the sodium propanethiolatendash;hexamethylphosphoramide complex (2 cm3) and the reaction stirred for 2 h at room temperature. The usual work-up was followed by puri- fication by column chromatography. Elution with 35 ethyl acetatendash;light petroleum gave a gummy mixture (116 mg) of which the major compound was hydroxy thioether 19; dH 0.90 (t J 7 SC2H4CH3) 1.12 (s 18-H3) 2.54 (apparent t J 7 2-H) 2.61 (d J 10 5-H) 3.23 (d J 10 6-H) 3.28 (d J 7 3-H) 3.98 (d J 7 1-H) 4.85 and 4.96 (2 times; br s 17-H2); m/z 580 (M+ 14) 478 (15) 390 (45) 375 (13) 291 (14) 241 (17) 217 (100) 191 (23) and 91 (9); GCndash;MS (trimethylsilyl derivative) m/z 580 (M+ 12) 565 (9) 549 (5) 478 (15) 460 (4) 390 (33) 217 (72) 191(18) and 73 (100).ent-2middot;,3middot;-Epoxy-1lsquor;,10lsquor;,13-trihydroxy-20-norgibberell-16-ene- 7,19-dioic acid 7-cyanomethyl ester 19,10-lactone 20 13-Hydroxyepoxy alcohol 2 (100 mg 0.28 mmol) in dichloromethane (3 cm3) and triethylamine (194 ml 1.12 mmol) were heated to reflux with chloroacetonitrile (21 ml 0.34 mmol) for 4 h. The reaction mixture was allowed to cool and diluted with dichloromethane (5 cm3). The usual work-up was followed by flash chromatography. Elution with 60 ethyl acetatendash;light petroleum gave hydroxy ester 20 as a yellow gum (116 mg); dH(CD3)2CO 1.24 (s 18-H3) 2.71 (d J 10 6-H) 2.97 (d J 10 5-H) 3.00 and 3.14 (2 times; d each J 3.5 2- and 3-H) 3.93 (br s 1-H) 4.72 (s CH2CN) 4.92 and 5.20 (2 times; br s 17-H2); m/z 401 (M+ 18) 344 (5) 326 (16) 231 (100) 163 (33) and 135 (17).756 J. Chem. Soc. Perkin Trans. 1 1997 ent-2middot;,3middot;-Epoxy-10lsquor;,13-dihydroxy-1-oxo-20-norgibberell-16- ene-7,19-dioic acid 7-cyanomethyl ester 19,10-lactone 21 Oxalyl chloride (freshly distilled 0.25 cm3 2.87 mmol) and dimethyl sulfoxide (0.51 cm3 7.19 mmol) were stirred in dichloromethane (5 cm3) at 278 8C for 5 min. Hydroxy ester 20 (100 mg 0.25 mmol) in dichloromethane (5 cm3) was added dropwise to the solution. After 1.5 h stirring N,N-diisopropylethylamine (1 cm3 5.78 mmol) was added and the solution became yellow. The reaction was allowed to warm to room temperature with stirring for 1.5 h. The usual work-up was carried out followed by purification by flash chromatography.Elution with 60 ethyl acetatendash;light petroleum (plus ten drops acetic acid) gave keto ester 21 as a gum (52 mg) (Found M+ 399.1302. C21H21NO7 requires M 399.1318); dH 1.43 (s 18-H3) 2.88 (d J 10.5 6-H) 3.53 and 3.54 (2 times; d each J 3.5 2- and 3-H) 3.42 (d J 10.5 5-H) 4.80 (s CH2CN) 5.01 and 5.31 (2 times; br s 17-H2); m/z 399 (M+ 12) 371 (2) 342 (3) 311 (9) 231 (69) 163 (38) and 135 (100). ent-2middot;,3middot;-Epoxy-1lsquor;,10lsquor;,13-trihydroxy-20-norgibberell-16-ene- 7,19-dioic acid 19,10-lactone 7-(3-methylbut-2-enyl) ester 22 To a solution of the 1a,13-dihydroxy-2,3-epoxide 2 (50 mg 0.14 mmol) in N,N-dimethylformamide (2 cm3) was added a catalytic amount of 4-(N,N-dimethylamino)pyridine (DMAP) (10 mg) and 3-methylbut-2-en-1-ol (30 ml 0.28 mmol). The reaction was cooled to 0 8C and 1,3-dicyclohexylcarbodiimide (DCC) (60 mg 0.28 mmol) was added.The reaction was allowed to warm to room temperature and stirred for 6 h. The reaction was filtered to remove any white precipitate which was washed with dichloromethane. The solvent was removed in vacuo and the crude product dissolved in dichloromethane and filtered again. The organic layer was washed sequentially with dilute hydrochloric acid and aqueous sodium hydrogen carbonate and then dried filtered and the solvents removed in vacuo. Purification by flash chromatography eluting with 40 ethyl acetatendash;light petroleum gave hydroxy ester 22 as a gum (57 mg) (Found M+ 430.2004. C24H30O7 requires M 430.1992); dH 1.30 (s 18-H3) 1.68 and 1.74 2 times; br s CO2CH2CHC(CH3)2 2.67 (d J 11 6-H) 3.04 and 3.18 (2 times; d each J 3.5 2- and 3-H) partially masking 3.03 (d J 11 5-H) 3.98 (br s 1-H) 4.61 (apparent t J 7 CO2CH2CHC(CH3)2 4.94 and 5.24 (2 times; br s 17-H2) and 5.41 (apparent t J 7 CO2CH2CHC(CH3)2; m/z 430 (M+ 10) 412 (5) 361 (15) 317 (8) 231 (10) 224 (24) 143 (21) and 69 (100).ent-2middot;,3middot;-Epoxy-10lsquor;,13-dihydroxy-1-oxo-20-norgibberell-16- ene-7,19-dioic acid 19,10-lactone 7-(3-methylbut-2-enyl) ester 23 Hydroxy ester 22 (67 mg 0.16 mmol) N-methylmorpholine Noxide (28 mg 0.24 mmol) and molecular sieves (crushed and dried 80 mg) were stirred in dichloromethane (2 cm3) and acetonitrile (0.5 cm3) at room temperature. TPAP (3 mg 0.008 mmol) was added and the reaction stirred for 2 h. Work-up was carried out by passing the reaction mixture through a short column of silica washing with dichloromethane and then ethyl acetate to give keto ester 23 as a gum (66 mg) (Found M+ 428.1841.C24H28O7 requires M 428.1835); dH 1.41 (s 18-H3) 1.72 and 1.76 2 times; br s CO2CH2CHC(CH3)2 2.77 (d J 10.5 6-H) 3.30 and 3.50 (2 times; d each J 3.5 2- and 3-H) 3.40 (d J 10.5 5-H) 4.11 (br s 1-H) 4.65 dd J 12 4 CO2CH2- CHC(CH3)2 4.96 and 5.27 (2 times; br s 17-H2) and 5.33 m CO2CH2CHC(CH3)2; m/z 428 (M+ 5) 410 (4) 359 (10) 224 (40) 143 (32) 99 (45) 69 (74) and 56 (100). ent-2middot;,3middot;-Epoxy-1lsquor;,10lsquor;,13-trihydroxy-20-norgibberell-16-ene- 7,19-dioic acid 19,10-lactone 7-(prop-2-enyl) ester 24 To a solution of the 1a,13-dihydroxy-2,3-epoxide 2 (100 mg 0.28 mmol) in N,N-dimethylformamide (4 cm3) was added a catalytic amount of DMAP (10 mg) and prop-2-en-1-ol (37 ml 0.28 mmol). The reaction was cooled to 0 8C and DCC (114 mg 0.55 mmol) was added.The reaction was allowed to warm to room temperature and stirred for 3 h. The reaction mixture was filtered to remove any white precipitate and the fitrate washed with dichloromethane. The solvent was removed in vacuo and the crude product dissolved in ethyl acetate and filtered again. The organic layer was washed sequentially with dilute hydrochloric acid and aqueous sodium hydrogen carbonate and then dried filtered and the solvents removed in vacuo. Purification by flash chromatography eluting with 50 ethyl acetatendash;light petroleum gave hydroxy ester 24 as a gum (83 mg) (Found M+ 402.1676. C22H26O7 requires M 402.1678); dH 1.31 (s 18-H3) 2.71 (d J 10.5 6-H) 3.06 and 3.19 (2 times; d each J 3.5 2- and 3- H) partially masking 3.04 (d J 10.5 5-H) 4.02 (br s 1-H) 4.65 (m CO2CH2CHCH2) 4.96 and 5.25 (2 times; br s 17-H2) 5.31 and 5.37 (2 times; m CO2CH2CHCH2) and 5.95 (m CO2CH2CHCH2); m/z 402 (M+ 23) 361 (22) 344 (36) 326 (100) 317 (38) 301 (46) 231 (94) and 163 (64).ent-2middot;,3middot;-Epoxy-10lsquor;,13-dihydroxy-1-oxo-20-norgibberell-16- ene-7,19-dioic acid 19,10-lactone 7-(prop-2-enyl) ester 25 Dihydroxy ester 24 (82 mg 0.20 mmol) N-methylmorpholine N-oxide (36 mg 0.31 mmol) and molecular sieves (crushed and dried 100 mg) were stirred in dichloromethane (2 cm3) and acetonitrile (0.2 cm3) at room temperature. TPAP (3.5 mg 0.01 mmol) was added and the reaction stirred for 3 h. Work-up was carried out by passing the reaction mixture through a short column of silica washing with dichloromethane and ethyl acetate to give keto ester 25 as a gum (56 mg) (Found M+ 400.1512.C22H24O7 requires M 400.1522); dH 1.42 (s 18-H3) 2.81 (d J 10.5 6-H) 3.31 and 3.51 (2 times; d each J 3.5 2- and 3-H) 3.41 (d J 10.5 5-H) 4.64 (m CO2CH2CHCH2) 4.98 and 5.28 (2 times; br s 17-H2) 5.32 and 5.39 (2 times; m CO2CH2CHCH2) and 5.91 (m CO2CH2CHCH2); m/z 400 (M+ 28) 359 (33) 342 (21) 315 (10) 301 (13) 224 (59) 143 (40) 99 (48) and 56 (100). ent-2middot;,3middot;-Epoxy-10lsquor;,13-dihydroxy-1-oxo-20-norgibberell-16- ene-7,19-dioic acid 19,10-lactone 18 Isobutyric acid (125 ml 0.79 mmol) was stirred in aqueous potassium hydroxide (1.5 M 2 cm3) for 0.5 h. The solvent was removed in vacuo using toluene to form an azeotrope. The resulting white powder was taken up in ethyl acetate (2 cm3) and stirred at room temperature with tetrakis(triphenylphosphine)- palladium(0) (18 mg 0.02 mmol) and triphenylphosphine (7 mg 0.03 mmol).Keto ester 25 (210 mg 0.53 mmol) was added to the reaction mixture in dichloromethane (2 cm3) and ethyl acetate (1 cm3) and stirring continued for 4 h. The usual workup was carried out then the combined organic phase was extracted with aqueous sodium hydrogen carbonate (2 times; 75 cm3). The aqueous extract was then acidified to pH 2 and extracted with ethyl acetate. The combined organic layers were dried filtered and the solvent removed in vacuo. Elution of a flash column with 70 ethyl acetatendash;light petroleum gave keto acid 18 as a cream foam (67 mg) (Found M+ 360.1205. C19H20O7 requires M 360.1209); dH(CD3)2CO 1.41 (s 18-H3) 2.82 (d J 10.5 6-H) 3.44 and 3.74 (2 times; d each J 3.5 2- and 3-H) partially masking 3.41 (d J 10.5 5-H) 4.90 and 5.23 (2 times; br s 17-H2); m/z 360 (M+ 39) 342 (18) 318 (8) 290 (8) 272 (12) 231 (61) 163 (48) and 135 (100).ent-2middot;,3middot;-Epoxy-1middot;,10lsquor;,13-trihydroxy-20-norgibberell-16-ene- 7,19-dioic acid 19,10-lactone 5 1-Keto-13-hydroxy-2,3-epoxide 18 (67 mg 0.19 mmol) in tetrahydrofuran (10 cm3) was added to dry powdered potassium hydrogen orthophosphate (154 mg 1.3 mmol). The mixture was cooled to 278 8C then K-Selectride (1 M in tetrahydrofuran 2.26 cm3 2.25 mmol) was added dropwise over 5 min. The reaction was allowed to warm to 0 8C over 2 h with stirring. The usual work-up was carried out. The organic phase was extracted with aqueous sodium hydrogen carbonate (3 times; 20 cm3) and then the aqueous extract was acidified to pH 3 and J. Chem. Soc. Perkin Trans. 1 1997 757 extracted with ethyl acetate (3 times; 30 cm3). Purification by flash chromatography eluting with 70 ethyl acetatendash;light petroleum (plus three drops acetic acid) gave the 1b-hydroxy-2b,3bepoxide GA93 5 which was crystallised from ethyl acetatendash;light petroleum as a white powder (26 mg) mp 145ndash;146 8C (Found M+ 362.1361.C19H22O7 requires M 362.1366); dH(CD3)2CO 1.26 (s 18-H3) 2.61 (d J 10 6-H) 3.21 (d J 10 5-H) 3.36 (m 2- and 3-H) 3.91 (br s 1-H) 4.88 and 5.19 (2 times; br s 17-H2); dC(CD3)2CO 14.8 (C-18) 17.8 (C-11) 39.8 (C-12) 43.5 46.2 (C-14 and C-15) 47.2 and 47.6 (C-3 and C-9) 49.1 and 49.6 (C-4 and C-8) 51.4 51.6 (C-5 and C-6) 59.7 (C-2) 65.2 (C-1) 78.1 (C-13) 94.2 (C-10) 106 (C-17) 159 (C-16) 174 and 177 (C-7 and C-19); m/z 362 (M+ 45) 344 (100) 3 (15) 231 (79) 163 (45) 135 (63) and 91 (48). Tan-ginbozu dwarf rice immersion assay Tan-ginbozu dwarf rice seeds were soaked in water for 2 days at 28 8C under constant light (the water was changed at 12 h intervals) by which time the coleoptile had emerged.The germinated seeds were selected for uniformity and placed in groups of six in cylindrical vials (18 mm diameter and 50 mm depth) which contained sterile water (1 cm3) and a methanolic solution of the substrate (10 ml). Gibberellins A7 A93 and A94 at a concentration of 10 mg per 10 ml of methanol were tested in duplicate. Two vials containing sterile water (1 cm3) and methanol (10 ml) were used as control standards. The seeds were allowed to grow at 28 8C under constant lighting. After 5 days the length of the second leaf-sheath and the total lengthmdash;from seed to longest leaf-tipmdash;was recorded. Acknowledgements We are grateful to the EPSRC for financial support to P.A. H. and M. P. IACR receives grant-aided support from the BBSRC. References 1 M. Penny C. L. Willis A. S. Batsanov and J. A. K. Howard J. Chem. Soc. Perkin Trans 1 1993 541. 2 J. MacMillan J. C. Seaton and P. J. Suter Tetrahedron 1962 18 349. 3 L. N. Mander Chem. Rev. 1992 92 573. 4 P. Gaskin and J. MacMillan GCndash;MS of Gibberellins and Related Compounds Methodology and a Library of Reference Spectra Cantockrsquo;s Enterprises Bristol 1991. 5 P. Gaskin P. S. Kirkwood J. R. Lenton J. MacMillan and M. E. Radley Agric. Biol. Chem. 1980 44 1589; N. Murofushi M. Sugimoto K. Itoh and N. Takahashi Agric. Biol. Chem. 1979 43 2179; P. S. Kirkwood and J. MacMillan J. Chem. Soc. Perkin Trans. 1 1982 689; C. L. Willis Tetrahedron Lett. 1990 31 6437; M.Penny C. L. Willis P. Gaskin and J. R. Lenton Phytochemistry 1993 33 951; M. Penny C. L. Willis P. Gaskin and J. R. Lenton Phytochemistry 1994 37 1063. 6 J. MacMillan and N. Takahashi Nature (London) 1968 217 170. 7 For example J. D. Winkler and P. M. Herohberger J. Am. Chem. Soc. 1989 11 4852; O. Mitsunobu M. Ebina and T. Orgihara Chem. Lett. 1982 373. 8 O. Mitsunobu Synthesis 1981 1; M. Saiah M. Bessodes and K. Antonakis Tetrahedron Lett. 1992 33 4317; S. Martin and J. A. Dodge Tetrahedron Lett 1991 32 3017. 9 A. Chu and L. N. Mander Tetrahedron Lett. 1988 29 2727; C. L. Willis Tetrahedron Lett. 1987 28 6705. 10 F. J. Kakis M. Fetizon N. Douchkine M. Golfier P. Mourgues and T. Prange J. Org. Chem. 1974 39 523. 11 E. J. Corey and G. Schmidt Tetrahedron Lett. 1979 399; J. Herscovici and K.Antonakis J. Chem. Soc. Chem. Commun. 1980 561. 12 A. J. Mancusso and D. Swern Synthesis 1981 165. 13 W. P. Griffith S. V. Ley G. P. Whitcombe and A. D. White J. Chem. Soc. Chem. Commun. 1987 1625. 14 D. H. Bowen C. Cloke D. M. Harrison and J. MacMillan J. Chem. Soc. Perkin Trans. 1 1975 83. 15 R. A. Bell and J. V. Turner Tetrahedron Lett. 1981 22 4871. 16 L. Lombardo L. N. Mander and J. V. Turner Aust. J. Chem. 1981 34 745. 17 J. R. Parikh and W. von Doering J. Am. Chem. Soc. 1967 89 5505. 18 P. A. Bartlett and W. S. Johnson Tetrahedron Lett. 1970 4459. 19 H. M. Hugel K. V. Bhaskar and R. W. Logmore Synth. Commun. 1992 22 693. 20 J. Cossy A. Albouy M. Scheloske and D. G. Pardo Tetrahedron Lett. 1994 35 1539. 21 P. D. Jeffery and S. W. McCombie J. Org. Chem. 1982 47 587.22 K. Hanson J. A. Orsquo;Neill T. J. Simpson and C. L. Willis J. Chem. Soc. Perkin Trans. 1 1994 2493. Paper 6/04983D Received 16th July 1996 Accepted 24th October 1996 J. Chem. Soc. Perkin Trans. 1 1997 751 Syntheses of gibberellins A93 and A94 natural products detected in wheat grain Stuart Findlow,a Paul Gaskin,b Polly A. Harrison,a John R. Lenton,b Martin Penny a and Christine L. Willis *,a a School of Chemistry University of Bristol Cantockrsquo;s Close Bristol BS8 1TS UK b IACR-Long Ashton Department of Agricultural Sciences University of Bristol Long Ashton Bristol BS18 9AF UK Extracts of mature wheat grain have been analysed by GCndash;MS and found to contain two new metabolites. The syntheses of the two new gibberellins 1lsquor;-hydroxy-2lsquor;,3lsquor;-epoxyGA9 4 and 1lsquor;-hydroxy-2lsquor;,3lsquor;-epoxy- GA20 5 from the fungal gibberellins GA7 and GA3 are described.A range of protecting groups for the 7-carboxylic acid which can be removed under mild conditions are compared. The allyl ester proved most valuable in the manipulation of these multifunctional molecules and it is removed with tetrakis(triphenylphosphine)palladium(0) triphenylphosphine and potassium isobutyrate. The structures of the new natural products in wheat were confirmed to be 4 and 5 by comparison with the GCndash;MS data from the synthetic samples and are assigned the trivial names GA94 and GA93 respectively. We have described previously the synthesis of two gibberellins 1 and 2 containing a 1a-hydroxy-2b,3b-epoxide moiety in ring A.1 Only one of the known gibberellins GA6 3 has the 2b,3bepoxide function.2 Hydroxylation at C-1 occurs in both plant and fungal systems,3 however comparison of the GCndash;mass spectra of 1 and 2 with detected but uncharacterised gibberellins 4 revealed that neither compound is a natural product.On recent examination of extracts of mature wheat grain (Triticum aestivum cv Maris Huntsman) by GCndash;MS two new compounds were detected with similar but not identical mass spectral data to 1 and 2 (Table 1). A series of 1b-hydroxygibberellins have been identified previously by GCndash;MS in extracts of wheat and their structures confirmed by direct comparison with synthetic samples.5 Hence we proposed that the two new natural products were 4 and 5 with the 1b-hydroxy-2b,3b-epoxide functionality in ring A. We now describe the syntheses of 4 and 5 which confirm the structures of the natural products (now assigned the trivial names6 GA93 5 and GA94 4) and enable the metabolic grid to account for the presence of the 1b-hydroxy- GAs in wheat to be completed.5 Results and discussion Synthesis of 1lsquor;-hydroxy-2lsquor;,3lsquor;-epoxide 4 (Scheme 1) Gibberellin A7 7 was converted to the 1a-hydroxy-2b,3bepoxide 8 via a Payne-type rearrangement of the iododiol 6 as previously described.1 A method for inversion of the stereochemistry of the 1-hydroxy group was then required.The most obvious choice was an SN2 displacement of a suitable derivative of the 1a-alcohol. As with other sterically crowded systems 6 the direct inversion of a secondary alcohol under standard or modified Mitsunobu reaction conditions 7 has proven to be unsuccessful.8 Caesium acetate has been successfully used for the inversion of alcohols at C-2 and C-3 of gibberellins via the corresponding methanesulfonates.9 Thus 8 was converted to the methanesulfonate 9 under standard conditions.However treatment of 9 with caesium acetate in refluxing toluene simply returned starting material whereas in N,N-dimethylformamide (DMF) at 80 8C an intractable gum was obtained. The density of functionality surrounding C-1 (the lactone and epoxide) appeared to be interfering with selective attack of the nucleophile and so the use of a two-stage oxidation procedure for inversion of the alcohol was investigated. Several procedures for the oxidation of the 1a-alcohol 8 to the ketone 10 were compared. Although Fetizonrsquo;s reagent 10 and pyridinium dichromate (PDC)11 returned mainly starting material oxidation was successfully achieved under both Swern conditions 12 and with tetrapropylammonium perruthenate (TPAP) N-morpholine N-oxide (NMO) and molecular sieves 13 giving the required ketone 10 in 90 and 98 yield respectively.Ketone 10 was then reduced with sodium borohydride in methanol giving a 2 1 mixture (by 1H-NMR spectroscopy) of the 1a-alcohol 8 and the required 1b-alcohol 11. It has previously been shown that reduction of a ketone at C-3 of the gibberellins with aluminium isopropoxide in propan-2-ol gives predominantly the 3b-alcohol.14 However reduction of the 1-oxo-2b,3b-epoxide 10 under these conditions gave a mixture of 1a 1b-alcohols 8 and 11 in a disappointing 10 1 ratio. The mixture of alcohols proved to be inseparable by either flash chromatography or medium pressure chromatography.In an attempt to prepare derivatives of 8 and 11 which may be separated the mixture was treated with acetic anhydride in pyridine giving the acetates 12 and 13; these were also inseparable. Therefore it was apparent that a stereoselective method was required for ketone reduction in which hydride delivery was exclusively from the a-face at C-1. Bell and Turner 15 reported that reduction of 3-oxogibberellin 7-carboxylic acids with K-Selectride gives 95 yield of the 3balcohol (i.e. with delivery of hydride from the a-face at C-3). The stereochemical outcome has been rationalised in terms of steric and Coulombic inhibition of approach of the reagent to the b-face of the gibberellin by a borate complex with the 7-carboxylic acid. This hypothesis is supported by the fact that reduction of 3-oxogibberellin 7-methyl esters with K-Selectride 752 J.Chem. Soc. Perkin Trans. 1 1997 Scheme 1 Reagents and conditions i aq. KOH (0.8 mol dm23) THF 14 h room temp. then adjust to pH 9 I2 CH2Cl2 2 h; ii aq. KOH (0.1 mol dm23); iii CH2N2 MeOH; iv MsCl pyridine; v TPAP NMO mol. sieves; vi NaBH4 MeOH; vii Ac2O pyridine; viii Me2SO (COCl)2 Pri 2EtN CH2Cl2 Me2CO; ix K-Selectride KH2PO3 gives the 3a-alcohol in 93 yield.16 Thus the 1a-hydroxy-2b,3bepoxide 7-acid 1 was oxidised under Swern conditions to give the required keto acid 14 in 66 yield. Due to the increased polarity of the acid 14 compared with the methyl ester 10 it proved necessary to add 14 as a solution in acetone to the activated sulfonium species in dichloromethane. Finally reduction of keto acid 14 with K-Selectride proceeded with excellent stereofacial selectivity to give the required 1b-alcohol 4 as a crystalline solid in 24 overall yield from GA7.Synthesis of 1lsquor;,13-dihydroxy-2lsquor;,3lsquor;-epoxide 5 Following the successful synthesis of 4 it was proposed to use a similar approach for the preparation of the 13-hydroxylated compound 5. Gibberellin A3 was readily converted to the 1a-hydroxy-2b,3b-epoxide 2 as previously described (Scheme 2).1 However it was found that although oxidation of the 1-hydroxy 7-methyl ester 15 proceeded smoothly with either TPAP or under Swern conditions giving 16 (in 93 and 65 yield respectively) attempted oxidation of the corresponding 1-hydroxy 7-carboxylic acid 2 under similar conditions simply returned starting material. The problem was believed to be due to the polar nature of the dihydroxy acid compared with either the dihydroxy ester 15 or hydroxy acid 1.Further attempts to oxidise the dihydroxy acid 2 with PDC in DMF or using the Parikh modification of the Moffatt oxidation 17 gave an intractable gum. Reduction of the keto ester 16 with K-Selectride gave a 3 1 mixture of the 1b 1a-alcohol 17 and 15 (by 1H NMR spectroscopy) which proved to be inseparable by flash chromatography or MPLC. It was therefore apparent that the presence of the 7-carboxylic acid is essential to achieve good stereofacial selectivity in the reduction of the 1-ketone with K-Selectride. Since keto acid 18 could not be prepared via direct oxidation of the dihydroxy acid 2 methods for converting a 7-methyl ester to an acid in the presence of a 2b,3b-epoxide were investigated.Treatment of 8 with aqueous sodium hydroxide gave a complex mixture of products and reaction with sodium propanethiolate in hexamethylphosphoramide (HMPA)18 resulted in attack on the 2b,3b-epoxide as well as deprotection at C-7 giving 19 as the major product. Scheme 2 Reagents and conditions i aq. KOH (0.8 mol dm23) THF 14 h room temp. then adjust to pH 9 I2 CH2Cl2 2 h; ii aq. KOH (0.1 mol dm23); iii CH2N2 MeOH; iv TPAP NMO mol. sieves; v KSelectride KH2PO3 THF J. Chem. Soc. Perkin Trans. 1 1997 753 In the light of these problems the approach to the synthesis of 5 had to be modified. A protecting group for the 7-acid was required which would effectively reduce the polarity of the dihydroxy acid 2 to enable oxidation of the 1-alcohol to a ketone and which then could be removed prior to reduction of the 1-ketone to the required 1b-alcohol.A range of protecting groups was examined (Scheme 3). Recently it has been reported that a cyanomethyl ester may be used to protect the 7-carboxylic acid of gibberellins.19 The acid 2 was heated to reflux with chloroacetonitrile to give the cyanomethyl ester 20 in quantitative yield. Oxidation of 20 under Swern conditions gave the keto epoxide 21 but attempts to remove the protecting group with either a mixture of potassium carbonatendash;potassium hydrogen carbonate in acetonendash; water or with potassium hydroxide gave an intractable mixture. Therefore although the protecting group was simple to put on and compatible with the oxidation conditions the route failed at the deprotection step the reaction conditions proving incompatible with the epoxy keto and lactone moieties in ring A.Another protecting group we considered was the 3- methylbut-2-enyl ester.20 The ester 22 was formed in 95 yield via a 1,3-dicyclohexylcarbodiimide (DCC)ndash;4-(N,N-dimethylamino) pyridine (DMAP) mediated coupling reaction. Oxidation of 22 with TPAP proceeded smoothly giving the keto ester 23 in 68 yield. Again however attempts to remove the ester to give the required 1-keto-2b,3b-epoxide 7-acid proved unsuccessful. Treatment of 23 with iodine in cyclohexane returned starting material when the reaction was conducted at room temperature and gave an intractable mixture at elevated temperatures. In this case it is possible that not only is the densely functionalised A ring sensitive to the reaction conditions but Scheme 3 also a Wagnerndash;Meerwein type rearrangement of the C/D rings may occur in the presence of iodine.Finally the allyl ester 24 was prepared from 2 using a DCCndash; DMAP mediated coupling with prop-2-en-1-ol (Scheme 4). Oxidation of alcohol 24 with TPAP successfully gave the ketone 25 (52 yield over the two steps). Removal of the protecting group was achieved with a mixture of potassium isobutyrate triphenylphosphine and tetrakis(triphenylphosphine)palladium( 0) 21 giving the required 1-keto-2b,3b-epoxide 18 in 35 yield. Reduction of 18 with K-Selectride gave the 1b-hydroxy- 2b,3b-epoxide 5 as a crystalline product in 5 overall yield from GA3. Comparison of 4 and 5 with the new metabolites in extracts of wheat grain and proposed biosynthesis. Mature grain of Triticum aestivum (cv Maris Huntsman) was extracted and the GAs were purified from the BuOH-soluble fraction as described in the Experimental section.The purified extract was methylated with diazomethane and derivatised with trimethylsilyl chloridendash;1,1,1,3,3,3-hexamethyldisilazanendash; pyridine prior to analysis by GCndash;MS.3 The extract contained two new metabolites which had not previously been identified.4 The synthetic samples of the 1b-hydroxy-2b,3b-epoxides 4 and 5 were separately derivatised as the trimethylsilyl ether methyl esters and the GCndash;MS data compared with those of the metabolites in wheat (Table 1). These data confirm the tentative assignments of the natural products and 4 is now assigned the trivial name GA94 and 5 is GA93. One of the most dramatic biological effects of gibberellins is their enhancement of stem elongation.3 The biological activities of GA93 and GA94 were assessed in a Tan-ginbozu dwarf rice immersion assay.It was found that in comparison with GA3 and GA7 neither GA93 nor GA94 is a potent stem elongation promoter (Table 2). Further bioassays are required to discover the significance of 1b-hydroxylation and the 2b,3b-epoxide in the role of GAs in plant growth and development in wheat. Experimental General experimental details have been described in a previous paper.22 For the numbering scheme used throughout the paper see structure 1. Unless otherwise stated all reactions were worked-up by the following standard procedure. The reaction mixture was poured into water and ethyl acetate. The pH was adjusted to 2 with 2 M hydrochloric acid and the products extracted with ethyl acetate.The combined organic extracts Scheme 4 Reagents and conditions i CH2 CHCH2OH DCC DMAP; ii TPAP NMO mol. sieves; ii Me2CHCO2K Ph3P Pd(PPh3)4 CH2Cl2 EtOAc; v K-Selectride KH2PO3 THF 754 J. Chem. Soc. Perkin Trans. 1 1997 Table 1 Comparison of the mass spectra of methylated trimethylsilylated derivatives of synthetic and wheat grain GA94 4 and GA93 5 and the 1ahydroxy- 2b,3b-epoxides 1 and 2 GA/source KRI Characteristic ions m/z ( base peak) 1a-Hydroxy-2b,3b-epoxy GA9 1 Synthetic 2606 432 (M+ 2) 417 (3) 385 (100) 288 (30) 201 (59) 145 (88) 73 (90) GA94 4 (1b-2b,3b-epoxy GA9) Synthetic 2505 432 (M+ 2) 414 (1) 403 (3) 400 (3) 382 (2) 370 (3) 356 (4) 313 (12) 310 (13) 301 (41) 300 (44) 283 (6) 268 (9) 255 (6) 241 (100) 240 (44) 221 (11) 211 (9) 132 (18) Wheat grain 2504 432 (M+ 2) 414 (1) 403 (3) 400 (3) 382 (2) 370 (2) 356 (4) 313 (12) 310 (12) 301 (39) 300 (41) 283 (6) 268 (10) 255 (6) 241 (100) 240 (41) 221 (11) 211 (9) 132 (17) 1a-Hydroxy-2b,3b-epoxy GA20 2 Synthetic 2646 520 (M+ 41) 473 (11) 376 (41) 303 (100) 235 (77) 207 (49) 73 (87) GA93 5 (1b-2b,3b-epoxy GA20) Synthetic 2663 520 (M+ 100) 505 (9) 491 (8) 461 (22) 447 (2) 401 (5) 389 (5) 376 (7) 347 (8) 329 (15) 305 (30) 303 (98) 279 (10) 235 (41) 208 (40) 207 (90) 194 (16) 193 (15) Wheat grain 2662 520 (M+ 77) 505 (9) 491 (8) 461 (22) 447 (2) 401 (4) 389 (6) 376 (7) 347 (9) 329 (15) 305 (31) 303 (100) 279 (9) 235 (40) 208 (37) 204 (89) 194 (16) 193 (24) were washed with water dried over anhydrous sodium sulfate and the solvent removed in vacuo.ent-2middot;,3middot;-Epoxy-10lsquor;-hydroxy-1lsquor;-methylsulfonyloxy-20-norgibberell- 16-ene-7,19-dioic acid 19,10-lactone 7-methyl ester 9 1a-Hydroxy-2b,3b-epoxide methyl ester 8 (76 mg 0.21 mmol) in pyridine (2.5 cm3) was stirred with methanesulfonyl chloride (26 ml) for 1 h at room temperature.The usual work-up was followed by purification by flash chromatography. Elution with 30 ethyl acetatendash;light petroleum gave methanesulfonate 9 as a foam (79 mg) (Found M+ 438.1349. C21H26O8S requires M 438.1348); dH 1.31 (s 18-H3) 2.61 (m 13-H) 2.72 (d J 11 6-H) 3.09 (d J 11 5-H) 3.16 (s OSO2Me) 3.25 and 3.29 (2 times; d each J 3.5 2- and 3-H) 3.72 (s CO2Me) 4.86 (br s 17-H) and 4.9 (br s 17-H and 1-H); m/z 438 (M+ 48) 406 (39) 402 (31) 378 (30) 360 (65) 254 (93) 221 (68) and 43 (100). Oxidation of hydroxy ester 8 Under Swern oxidation conditions.Oxalyl chloride (freshly distilled 0.28 mmol 0.25 ml) and dimethyl sulfoxide (7 mmol 0.51 ml) were stirred in dichloromethane (5 cm3) at 278 8C for 5 min. 1-Hydroxy-2,3-epoxide 8 (107 mg 0.30 mmol) was added dropwise in dichloromethane (5 cm3) and the reaction mixture stirred for 1.5 h at 278 8C. N,N-Diisopropylethylamine (5.6 mmol 1 ml) was added and the reaction allowed to warm to room temperature over 1.5 h. The usual work-up was followed by purification by flash chromatography. Elution with 40 ethyl acetatendash;light petroleum gave ent-2a,3a-epoxy-10b-hydroxy-1- oxo-20-norgibberell-16-ene-7,19-dioic acid 19,10-lactone 7- methyl ester 10 which was crystallised from ethyl acetatendash;light petroleum as needles (12 mg) mp 155ndash;157 8C (Found M+ 358.1412. C20H22O6 requires M 358.1416); dH 1.41 (s 18-H3) 2.80 (d J 11 6-H) 3.31 and 3.51 (2 times; d J 3.5 2- and 3-H) 3.39 (d J 11 5-H) 3.74 (s CO2Me) 4.99 and 5.26 (2 times; br s 17-H2); m/z 358 (M+ 48) 326 (40) 298 (13) 241 (22) 228 (100) 201 (19) and 91 (45).With tetrapropylammonium perruthenate (TPAP). Hydroxy ester 8 (104 mg 0.29 mmol) N-methylmorpholine N-oxide (51 mg 0.43 mmol) and molecular sieves (crushed and dried 145 mg) were stirred at room temperature under nitrogen in dichloromethane (900 ml) and acetonitrile (100 ml). Tetrapropylammonium perruthenate (TPAP) (5 mg 0.015 mmol catalytic) was added and the reaction stirred for 2 h. To work-up the mixture it was passed through a short column of silica which was eluted with dichloromethane (100 cm3) and then ethyl acet- Table 2 Rice seedling bio-assay Second leaf Sheath/mm Total/mm Control GA3 GA7 GA93 GA94 15 40 50 16 21 23 55 64 26 33 ate (100 cm3).The ketone 10 was obtained in 100 yield the 1H NMR and mass spectral data being identical to those previously obtained. Reduction of keto ester 10 With sodium borohydride. Sodium borohydride (12.7 mg 0.34 mmol) in methanol (1 cm3) was added to a stirred solution of the epoxy ketone 10 (60 mg 0.17 mmol) in methanol (4 cm3). The mixture was stirred for 1 h at room temperature and then worked-up as usual. Purification by flash chromatography and elution with 25 ethyl acetate in light petroleum gave a 2 1 mixture of the 1a-hydroxy- and 1b-hydroxy-epoxides 8 and 11 (55 mg). The epimeric mixture showed as a single spot in three different solvent systems. dH(major 1a-alcohol 8; as previously described); dH(minor 1b-alcohol 11) 1.29 (s 18-H3) 2.61 (m 13-H) 2.69 (d J 10.5 6-H) 3.13 (d J 10.5 5-H) 3.41 (m 2-H and 3-H) 3.71 (s OCH3) 3.90 (d J 8 1-H) 4.86 and 4.98 ( 2 br s 17-H2).The trimethylsilyl derivative of 11 gave KRIdagger; 2505; m/z 432 (M+ 1) 400 (3) 370 (3) 356 (4) 324 (2) 300 (41) 283 (6) 268 (10) 241 (100) 240 (43) 221 (13) 211 (10) 195 (9) 132 (19) 119 (10) 93 (20) 73 (66) 55 (11) and 44 (30). Meerweinndash;Verleyndash;Ponndorf reduction. Aluminium foil (0.35 g 13.0 mmol) and mercuric chloride (3.5 mg 0.013 mmol) in propan-2-ol (15 cm3) were heated to reflux. Carbon tetrachloride (60 ml) was added and the mixture was heated at a gentle reflux for 3 h. Keto ester 10 (20 mg 5.59 times; 1022 mmol) in propan-2-ol (2 cm3) was added and the mixture was heated for a further 3 h with distillation of the acetone produced.The mixture was then cooled to room temperature and worked-up as usual. The crude product was purified by flash chromatography. Elution with 25 ethyl acetate in light petroleum gave the 1- alcohols 8 and 10 as a ~10 1 (a:b) epimeric mixture; spectroscopic data as previously described. ent-1middot;- and 1lsquor;-Acetoxy-2middot;,3middot;-epoxy-10lsquor;-hydroxy-20-norgibberell- 16-ene-7,19-dioic acid 19,10-lactones 7-methyl ester 12 and 13 The mixture of epimeric alcohols 8 and 10 (27 mg) in pyridine (750 ml) and acetic anhydride (700 ml) was stirred at room temperature for 2 h. The usual work-up gave a gum which was purified by flash chromatography. Elution with 10 ethyl acetate in light petroleum gave the epimeric acetates 12 and 13 in a ratio of 2 1 (a:b). The acetates also showed as one spot by TLC in three different solvent systems ethyl acetatendash;light petroleum (1 5) diethyl etherndash;hexane (3 10) and dichloromethanendash;methanol (9 1) (Found M+ 402.1671.C22H26O7 requires M 402.1679); dH(major 1a-acetate 13) 1.31 (s 18-H3) 2.15 (s OCOCH3) 2.62 (m 13-H) 2.72 (d J 11 6-H) 2.97 (d J 3.5 2- or 3-H) 3.08 (d J 11 5-H) 3.20 (d J 3.5 2- or 3-H) 3.72 (s OCH3) 4.86 and 4.97 (2 br s 17-H2) and 5.18 (d J 0.5 1-H); dH(minor 1bacetate 12) 1.32 (s 18-H3) 2.19 (s OCOCH3) 2.64 (m 13-H) 2.69 (d J 10.5 6-H) 3.28 (d J 10.5 5-H) 3.31 (d J 4 3-H) dagger; KRI (Kovats Retention Index) is described in detail in ref. 4. J. Chem. Soc. Perkin Trans. 1 1997 755 3.44 (t J 4 2-H) 3.72 (s OCH3) 4.86 and 4.97 (2 br s 17-H2) and 5.01 (d J 4 1-H); m/z 402 (M+ 13) 384 (5) 370 (74) 342 (13) 310 (55) 280 (37) 254 (100) 240 (55) 239 (78) 238 (80) 221 (51) 171 (35) 105 (38) and 91 (76).ent -2middot;,3middot;-Epoxy-10lsquor;-hydroxy-1-oxo-20-norgibberell-16-ene-7 19-dioic acid 19,10-lactone 14 Dichloromethane (5 cm3) and freshly distilled oxalyl chloride (0.24 cm3 2.9 mmol) were added to a 10 cm3 flame-dried roundbottomed flask under nitrogen at 278 8C . Dimethyl sulfoxide (0.49 cm3 7.22 mmol) was added dropwise to the solution and the reaction stirred for 5 min. 1a-Hydroxy 2b,3b-epoxide 1 (100 mg 0.29 mmol) was dissolved in a minimum of acetone and dichloromethane and added dropwise to the reaction mixture. After 1.5 h at 278 8C N,N-diisopropylethylamine (0.97 cm3 5.7 mmol) was added and the reaction mixture allowed to warm to room temperature for 1.5 h. The reaction was diluted with dichloromethane (2 cm3) and water (1 cm3) and then the usual work-up was carried out.Purification by column chromatography eluting with 40 ethyl acetatendash;light petroleum gave the epoxy ketone 14 as a white solid which was recrystallised from ethyl acetatendash;light petroleum as needles (66 mg) mp 155ndash; 156 8C (Found C 66.3; H 5.8. C19H20O6 requires C 66.28; H 5.81); dH 1.46 (s 18-H3) 2.82 (d J 11 6-H) 3.31 and 3.50 (2 times; d J 3.5 2- and 3-H) 3.35 (d J 11 5-H) and 4.94 (2 times; br s 17-H2); m/z 344 (M+ 38) 326 (30) 274 (32) 255 (32) 228 (56) 211 (28) and 91 (100). ent-2middot;,3middot;-Epoxy-1middot;,10lsquor;-dihydroxy-20-norgibberell-16-ene-7 19-dioic acid 19,10-lactone 4 A solution of the epoxy ketone 14 (100 mg 0.29 mmol) in tetrahydrofuran was added to dry powdered potassium dihydrogen orthophosphate (0.24 g 2.03 mmol) under nitrogen.The reaction was cooled to 278 8C and K-Selectride (potassium tri-sec-butylborohydride) (1.2 cm3 1 M solution in tetrahydrofuran) was added dropwise over 5 min stirring continuously. The solution was allowed to warm to room temperature over 2 h and then worked-up as usual. Purification by column chromatography eluting with 40 ethyl acetatendash;light petroleum gave GA94 4 which was crystallised from ethyl acetatendash; light petroleum as a white solid (55 mg) mp 142ndash;144 8C (Found C 65.5; H 6.2. C19H22O6 requires C 65.9; H 6.4); dH 1.36 (s 18-H3) 2.63 (m 13-H) 2.72 (d J 10.5 6-H) 3.14 (d J 10.5 5-H) 3.42 (2 times; s 2- and 3-H) 3.92 (d J 4 1-H) and 5.92 (2 times; br s 17-H2); dC 14.2 (C-18) 17.6 (C-11) 31.6 (C-12) 39.4 (C-13) 42.5 44.2 (C-14 and C-15) 47.2 and 47.6 (C-3 and C-9) 48.6 and 49.3 (C-4 and C-8) 51.4 51.9 (C-5 and C-6) 59.7 (C- 2) 69.2 (C-1) 92.4 (C-10) 107.7 (C-17) 156.2 (C-16) 172.2 and 176 (C-7 and C-19); m/z 346 (M+ 28) 328 (18) 310 (17) 287 (96) 240 (65) 195 (36) and 91 (100).Oxidation of 1middot;,13-dihydroxy-2lsquor;,3lsquor;-epoxide methyl ester 15 Under Swern oxidation conditions. Oxalyl chloride (freshly distilled 174 ml 1.99 mmol) and dimethyl sulfoxide (350 ml 4.99 mmol) were stirred in dichloromethane (5 cm3) at 278 8C for 5 min. Diol 151 (75 mg 0.20 mmol) in dichloromethane (5 cm3) was added dropwise and the reaction mixture stirred for 1.5 h at 278 8C. N,N-Diisopropylethylamine (690 ml 3.99 mmol) was added and the resulting yellow solution was allowed to warm to room temperature over 1.5 h. The usual work-up was followed by purification by flash chromatography.Elution with 70 ethyl acetatendash;light petroleum gave ent-2a,3a-epoxy-10b,- 13-dihydroxy-1-oxo-20-norgibberell-16-ene-7,19-dioic acid 19,- 10-lactone 7-methyl ester 16 as a gum (49 mg 65) (Found M+ 374.1360. C20H22O7 requires M 374.1366); dH 1.42 (s 18-H3) 2.80 (d J 10.5 6-H) 3.31 and 3.52 (2 times; d each J 3.5 2- and 3-H) 3.41 (d J 10.5 5-H) 3.75 (s CO2Me) 4.98 and 5.28 (2 times; br s 17-H2); m/z 374 (M+ 25) 346 (3) 342 (13) 315 (10) 304 (9) 275 (21) 231 (61) 135 (100) and 91 (97). With TPAP. To a stirred solution of diol 15 (50 mg 0.14 mmol) molecular sieves (4 Aring; vacuum oven dried 100 mg) and N-methylmorpholine N-oxide (24 mg 0.21 mmol) in dichloromethanendash;acetonitrile (2 cm3 200 ml) was added TPAP (2.4 mg 0.007 mmol). After 2 h the solvent was removed in vacuo and the resulting gum purified by flash chromatography.Elution with 70 ethyl acetatendash;light petroleum gave ketone 16 (48 mg) whose 1H NMR and mass spectral data were identical to those previously obtained. Reduction of keto ester 16 with K-Selectride Keto ester 16 (169 mg 0.45 mmol) in tetrahydrofuran (10 cm3) was added to dry powdered potassium hydrogen orthophosphate (374 mg 3.16 mmol). The mixture was cooled to 278 8C then K-Selectride (1 M in tetrahydrofuran 2.26 cm3 2.25 mmol) was added dropwise over 5 min. The reaction was allowed to warm to 0 8C over 2 h with stirring. The usual workup was followed by purification by flash chromatography. Elution with 50 ethyl acetatendash;light petroleum with three drops acetic acid added gave a mixture of 1a-alcohol 15 and 1balcohol 17 (80 mg) in the ratio 1 3.Separation was attempted using medium pressure liquid chromatography eluting with 50 ethyl acetatendash;light petroleum (with three drops acetic acid added) which gave a 1 7 mixture (by 1H NMR) of the 1aalcohol 15 and the 1b-alcohol 17 as a white foam (13 mg) (Found M+ 376.1526. C20H24O7 requires M 376.1522); dH 1.31 (s 18-H3) 2.68 (d J 10 6-H) 3.18 (d J 10 5-H) 3.40 (m 2- and 3-H) 3.74 (s CO2Me) 3.91 (br s 1-H) 4.96 and 5.25 (2 times; br s 17-H2). GCndash;MS of the trimethylsilyl derivative of 17 gave KRI 2663; m/z 376 (M+ 38) 344 (78) 326 (37) 317 (34) 275 (32) 231 (100) 163 (67) and 135 (49). Further elution gave 1a-alcohol 15 (24 mg); 1H NMR and mass spectral data being identical to those previously assigned. The trimethylsilyl deriva