SUGAR EPOXIDES By F. H. NEWTH B.Sc. PH.D. (PARTINGTON RESEARCH LABORATORY PETROCHEMICALS LTD. URMSTON MANCHESTER) 1. Introduction AN ethylene oxide group is easily introduced into a sugar molecule by alkaline hydrolysis of an 0-arene(or a1kane)sulphonyl derivative or deoxyhalogeno-compound which has a vicinal hydroxyl group trans to the anionic group. This reaction occurs easily and with few exceptions and the epoxide is formed nearly quantitatively. The interest in sugar epoxides as distinct from other anhydro-derivatives lies in this reactive group. The oxide ring is opened by nucleophilic reagents to give usually two products with the trans-configuration and a substituent group which is derived from the reagent. This reaction has therefore provided a versatile method for the preparation of rare sugars from easily accessible ones and for the selective introduction of groups or atoms such as 0-alkyl amino and halogen into sugar molecules.The reaction is easily carried out but separation and characterisation of the products have often needed pro- longed study. It will be appreciated that much impetus has been given to the chemistry of sugar epoxides by the desire to synthesise naturally occurring sugars and two well-known examples are the synthesis of glucosamine (2-amino-2-deoxy-~-g~ucose) in 1939 and chondrosamine (2-amino-2-deoxy-~-galactose) in 1946. In the years following 1946 much attention was also given to sugar epoxides as intermediates in the prepara- tion of deoxy-sugars. The syntheses of the sugar components of the cardiac glycosides were successfully achieved but the chemistry was not so favour- able for 2-deoxyribose because of the direction of ring opening.Although the preparative aspect of the reactions of the sugar epoxides has been most closely studied there are important features in the field such as different rates of formation of the various oxides and the different proportions in which the products of ring fission are obtained. These have led to a consideration of the influence which other groups in the sugar molecule may have upon the formation and fission of the epoxide ring. In many cases it is possible to assign a particular conformation to the sugar ring with reasonable certainty and the chemistry of the sugar epoxides has now reached a stage where study of their reactions can reveal much about the intramolecular interactions which operate in the sugar molecule.2. Formation 2.1. Fischer’s Epig1ucosamine.-One of the earliest reactions in which a sugar epoxide must have been formed was an attempt to synthesise 30 NEWTH SUGAR EPOXDES 31 glucosamine from glucal.l3 4 6-Tri-O-acetyl-~-glucal 1 2-dibromide was converted into methyl 3 4 6-tri-O-acetyl-2-brorno-2-deoxy-~-~-glucoside (I). When this compound (or its chloro-analogue) was treated with ammonia the product was not 2-amino-2-deoxy-~-g~ucose or mannose but a substance which Fischer called methyl epiglucosaminide. This was later2 shown to be methyl 3-amino-3-deoxy-P-~-altroside (11). Fischer suspected this at the time and suggested that a 2 3-anhydro-ring might have been formed intermediately. CH,-OAc CH,-OH AcO 0”‘ - Epoxide -c HO QMe H2N (II) Br (1) It can of course be recognised now that the intermediate was methyl 2:3-anhydro-P-~-mannoside but it was not until 13 years later that the existence of 2 3-anhydrohexosides was established by concurrent work at Birmingham and St.Andrews. 2.2. 1 2-Anhydro-~-glucopyranose and 5 6-Anhydro-~-glucofuranose,- The first example of a sugar epoxide was given by BrigP in 1921. He found that when tetra-O-acetyl-/3-D-glucose was treated with phosphorus pentachloride 3 4 6-tri-O-acetyl-ZO-trichloroacetyl-~-~-glucosyl chlor- ide (111) was formed. Careful treatment of this compound with ammonia removed only the trichloroacetyl group forming 3 4 6-tri-O-acetyl-/3-~- glucosyl chloride (IV) and on further mild treatment with ammonia this was converted into 3 4 6-tri-O-acetyl-l 2-anhydro- a-D-glucose (V).This compound is a valuable synthetic intermediate. It reacts with methanol to give methyl 3 4 6-tri-O-acetyl-/3-~-glucoside and the 2-0-tosyl derivative (VI) was important in establishing the chemistry of the 2:3-epoxides. (VI) (VII) R=Br (I )o (VIII) R =OTs Ts = p€,H4Me-SO2 Fischer Bergmann and Schotte Ber. 1920,53 509. Haworth Lake and Peat J. 1939 271. Brigl 2. physiol. Chew. 1921 116 1 245. 32 QUARTERLY REVIEWS Before considering this aspect however it is convenient to mention the conversion of 6-bromo-6-deoxy- 1 2-0-isopropylidene- a-~-glucofurano~e~ (VII) and I 2-0-isopropylidene-6-0-tosyl- a-D-glucofuranose6 (VIII) into 5 6-anhydro-1 2-0-isopropylidene- a-D-glucofuranose (IX) since this is a reaction not complicated by Walden inversion.2.3. 2 3- and 3 4Anhydrides.-In 1930 Helferich and Miillera isolated a crystalline methyl anhydro-#bhexoside after treating methyl 2 3 6- t~-~-acety~-4-0-tosy~-~-~-~ucoside (X) with sodium methoxide; this product was shown by Muller' to be methyl 3 4-anhydro-/3-~-galactoside (XI). In 1933 an anhydro-sugar was obtained from methyl 4:6-di-0- methyl-2 3-di-0-tosyl- a-D-glucoside (XII) and the alkaline conditions produced also some methyl 4 6-di-O-methyl- a-D-altroside* (XIII). The following year Haworth and his co-workersg described the formation of a methyl 2 3-anhydro-/3-~-hexoside from methyl 3 4 6-tri-0-acetyl-2-0- tosyl-fl-D-glucoside (VI) and they commented that the change of optical rotation to a high negative value on hydrolysis of the anhydride suggested the formation of an altrose derivative.bMc Q GMe TsO Me0 OMe Me0 OAc OH OTs HO (XI (Xi) (XI I) (XIII) Me PhCH- Me The manner in which these reactions were proceeding was now becom- ing obvious and Robertson and GriffithlO made an important contribution by showing that the products obtained from methyl 3-0-benzoyl-4 6-0- benzylidene-2-0-tosyl-u-~-glucoside (XIV) and methyl 2-0-benzoyl-4 6- O-benzylidene-3-O-tosyl-u-~-glucoside (XV) by treatment with sodium Freudenberg Toepfer and Anderson Ber. 1928 61 1751. Ohle and Vargha Ber. 1929 62 2435. 13 Helferich and Miiller Ber. 1930 63 2142. Miiller Ber. (a) 1934 67 421 ; (b) 1935 68 1094. * Mathers and Robertson J. 1933 1076. Haworth Hirst and Panizzon J. 1934 154. lo Robertson and Griffith J . 1935 1193. NEWTH SUGAR EPOXIDES 33 methoxide are methyl 2 3-anhydro-4 6-0-benzylidene-a-u-mannoside (XVI) and -a-D-alloside (XVII) respectively.In all these reactions the formation of an anhydro-ring has been ac- companied by inversion of configuration at that carbon atom to which the sulphonyloxy-group is attached. In the alkaline medium the anion of the vicinal trans-hydroxyl group can displace the toluene-p-sulphonyloxy- anion to form an epoxide and this can be represented as the intramolecular SN2 process (1) I I ( I I OTs If the vicinal group is cis as in methyl 2 3 6-tri-0-acetyl-4-0-methane- sulphonyl- p-D-galactose (XVIII) then only deacetylation occurs and the methanesulphonyloxy-group is not displaced.ll An interesting reaction examined by Peat and Wiggins12 was the treatment of methyl 3-0-tosyl- p-D-glucoside (XIX) with mild alkali.Methyl 2 3- (XX) (60 %) and methyl 3 4-anhydro-/%~-alloside (XXI) (25 %) were formed since the toluene-p- sulphonyloxy-group has two neighbouring hydroxyl groups which can cause its displacement. In addition to the two epoxides some methyl 3 6- anhydro-p-D-glucoside (XXII) [shown later13 to be the furanoside (XXIII)] was formed. Although at the time it was considered that this was one of the examples of hydrolysis of a toluene-p-sulphonyloxy-group with retention of configuration the 3 6-anhydride is a secondary product formed as Ohle and Wilke l4 pointed out by the anion from the 6-hydroxyl group attacking at position 3 in either (XX) or (XXI).15 ' OAc (XVIII) CH,*OH HO QMe OH O(IX) HO QMe @Me Ho@e H O z e OMe (XXI) (XXI I) (xx I I I) Ms=CH,-SO2 l1 Muller Moricz and Verner Ber.1939 72 745. l2 Peat and Wiggins J. 1938 1088. l3 Haworth Jackson and Smith J. 1940 620. l4 Ohle and Wilke Ber. 1938 71 2316. l5 Peat Adv. Carbohydrate Chern. 1946,2 37. 34 QUARTERLY REVIEWS 2.4. The Stereochemistry of Epoxide Formation.-The examples given so far illustrate the development of the subject and all appear straight- forward; in the light of the accepted views on inversion of configuration at a saturated carbon atom they are what would be expected. This part of carbohydrate chemistry is unfortunately completely lacking in quantita- tive data. In spite of this however inspection shows that although the experimental conditions chosen are often excessive some reactions are more difficult to carry out than others. The Haworth ring formulz admirably show configurational changes but do not help in finding the steric cause of different reactivities or the reason why a reaction follows a particular course.It is not until conformational drawings are made that intramolecular interactions can become apparent and a posteriori explana- tions can be given for such behaviour. In the base-catalysed formation of an epoxide from a trans-1 2-diol mono-0-toluene-p-sulphonate (reaction l) the intramolecular SN2 process in a six-membered ring requires the two groups to be in the diaxial position. The entering and the departing anion and the carbon atoms to which they are attached are then co-planar and this permits maximum participation. This condition is found in 1 6-anhydro-2-0-methane- sulphonyl-fl-D-galactose (XXIV) and 1 6-anhydro-4-0-tosyl-/3-~-mannose (XXV) and they are easily converted into the 2 3 3 (XXVI) and the 3 4- taZo-epoxide17 (XXVII) by mild alkali.The majority of toluene-p-sulphon- ales vicinal to a trans-hydroxyl group are however in the diequatorial 4 H O B HO 0 (XXIV) OMS (XXVI) mH - @ (x x v) (xxvr I) Ts 0 position as for example in the 2- (XIVa) and the 3-0-tosyl derivative and (XVa) which are smoothly converted into the manno- (XVI) and do-epoxide (XVII). The ease with which these compounds react suggests (X I Va) lR James Smith Stacey and Wiggins J . 1946 625. l7 Ham and Hudson J. Amer. Chem. SOC. 1942 64 925 2435. NEWTH SUGAR EPOXIDES 35 structural modification before epoxide formation and indeed in the analogous bimolecular ionic elimination from 1 2-di halides no reaction occurs when both trans-substituents are rigidly held in equatorial posi- tions.l* In a monocyclic system the diequatorial groups can pass into the diaxial position without much difficulty (conformation C1-+1C19) but in the bicyclic compounds (XIVa) and (XVa) the trans-fused ring containing the 4:6-O-benzylidene group confers rigidity on the chair form of the sugar ring and prevents this change.C(2) on the other hand can move downwards to give the boat form (XIVb) without disturbing the point of ring fusion and the two groups at C 0 and C(31 are now co-planar and in a position to react.20 n \ (OH) The hexose epoxides must be considered as analogous to 1 2-epoxycyclo- hexane which has been shown to have the half-chair conformation similar to that in cyclohexane.21 The manno-epoxide (XVI) can therefore be shown as (XVIa).22 It must then be recognised that the epoxide conformation is halfway in the chair-boat conversion and C(,) and C(,) in the boat form would have to move again in the reverse direction to‘become co-planar with C(l) and C(a).For this reason the true boat formmay never be reached. It is convenient however to formulate this extreme condition since in its attainment steric interactions are apparent which explain different activities and permit predictions. It is of interest to consider here the alkaline hydrolysis of methyl 3-O-tosyl-/3-~-glucoside~~ (XIX) to see whether conformational analysis can account for the predominance of the 2 3-ah-epoxide (XX) [allowing for the formation of the 3 6-anhydride] over the 3 4-ah-epoxide (XXI) in the reaction product.The form (XIXa) is the least hindered (all groups equatorial) and will be the “resting position” of the molecule. Form (XIXb) although it has the desired axial relation between the reacting groups must be ignored because of its state of extreme hindrance (all groups axial). Forms (XIXc d and e) represent three points in the cycle of positions possible in the flexible boat form23 in which the 2- 3- and 4-groups are axial (Reeves’s 2B B3 lS Barton and Rosenfelder J. 1951 1048. lS Reeves J. Amer. Chem. SOC. 1949 71 21 5. Newth J. 1956 441. a1 Ottar Acta Chem. Scand, 1947 1 283. Cookson Chem. and Ind. 1954,223 1512. 23 Reeves J. Amer. Chem. SOC. 1957 79 2261. 36 QUARTERLY REVIEWS and 1B conformation^^^). Form (XIXd) appears to be able to give equally the 2 3- and the 3 4-epoxide; form (XIXc) will lead to the 2 3-epoxide and (XIXe) to the 3:4-epoxide.A difference must be found therefore between the last two forms and probably lies in the 1:3-interactions OTs(,)/OMe(, and OTs(,)/CH,.OH( ;). In the alkaline reaction medium the "o- l&Me CHiOH (XIXa) OH HO OH (XIXb) TsO OMe CHiOH HO*H2C & HO.H,C 0 ) q f i HO OMe OH HO OH OH (x I xc) (XIXd) (X I Xc) anion of the 5-hydroxymethyl group will cause more hindrance to the departing toluene-p-sulphonyloxy-anion than the glycosidic methoxyl group and so the more favoured form would be (XIXc) leading to the 2:3-epoxide. It would be interesting now to know the composition of the products from the reaction of methyl 6-0-methyl-3-O-tosy1-~-~-glucoside and methyl 3- O-tosyl-fl-~-xyloside with alkali.There is a very striking difference in the reactivity of the U-tosyl deriva- tives of 1 6-anhydro-fi-~-altrose.~~ It was found that although 1 6- anhydro-3-O-tosyl-~-~-altrose (XXVIII) could be converted into an epoxide 1 6-anhydro-2-0-tosyl-/?-~-altrose (XXIX) and the 3 4-di-0- tosyl derivative (XXX) were quite resistant to alkaline hydrolysis. This behaviour can be explained when steric interactions are considered. In the (xxvr I I) (XXIX) otxx> y 2 - ? y2- ? HO mH OTS HOQ (XXV I I I a) OTS (XXVI I I b) boat form (XXVIIIb) which is the condition suitable for epoxide forma- tion there are interactions between OH( and Ocl) and between OH(, and C(s). On passing from the chair to the boat conformation there will be steric interaction between vicinal groups when one is axial and the other equatorial (cis) since they must move past each other.Thus in (XXVIIT 24 Idem ibid. 1950 72 1499. NEWTH SUGAR EPOXIDES 37 a+b) there will also be a OTS,,)/OH,~) passing interaction. Although the alkaline hydrolysis does not occur with great ease these combined steric factors are not sufficiently great to prevent reaction. In the ester (XXIX) on the other hand the interaction OH(,,/OH(4) will be less but when it is combined with the more severe interactions OTs( 2j0(1) and OTs(,)/C[g the total hindrance must be sufficiently great to prevent reaction. Similarly in the diester (XXX) there will be a very severe OTS(~)/OTS(~) passing inter- action and with the interactions OH(2)/0(1) and OH(2&,) there is enough hindrance again to prevent reaction or even attainment of the boat form.More weight is given to the concept of passing interaction when the alkaline hydrolysis of methyl 4 6-O-benzylidene-2-0-tosyl-a-~-gluco- sidelo (XXXI) and 1 5-anhydro-4 6-U-benzylidene-2-O-tosyl-~-glucit~l~~ (XXXII) is considered. The ester (XXXI) requires the temperature of boiling methanol and compound (XXXII) is hydrolysed easily at 0'. The only difference between the two compounds is the presence or absence of the 1-methoxyl group. If the first postulate that the diaxial condition must be attained by chair-boat transformation is correct the difference in reactivity must be due entirely to the 3fferent passing interactions OTs(,,/ This concept also provides an explanation for the difference in reactivity between methyl 4 6-0-benzylidene-2-0-tosyl-a- and -/hhgalactoside.20s26 OMe(1) and OTs(,/H(,).2.5. Epoxides from Di-Q-sulphonyl Compounds.-In addition to the reactions already described it is also possible to obtain an epoxide by alkaline hydrolysis of di-0-sulphonyl compounds. A well-known example of this is the formation of methyl 2 3-anhydro-4 6-0-benzylidene-a-~- alloside (XVII) from methyl 4 6-0-benzylidene-2 3-di-O-tosyl-a-~- glucoside (XXXIII). This reaction occurs easily with cold sodium methox- ide and the anhydro-glycoside (XVII) is formed quantitatively as the sole p r o d ~ c t . ~ ~ * ~ ~ The preparative value lies in the ease of formation of the di-0-tosyl derivative since only a protecting 0-benzylidene group need be introduced into the glucoside. The reaction is also of considerable theo- retical interest since one of the ester groups must undergo 0-S cleavage without difficulty in contrast to the well-known SN2 reaction (2) of alkyl sulphonates.This displacement at a sulphur atom which is very prevalent 8s Newth XVIth Int. Congr. Pure Appl. Chem. Paris 1957. 26 Wiggins J. 1944 522. 27 Richtmyer and Hudson J. Amer. Chem. SOC. 1941,63 1727. 2 38 QUARTERLY REVIEWS in carbohydrate chemistry has been called SN2S28 and must occur in those “isolated” secondary sulphonates which are hydrolysed with difficulty but with retention of config~ration.~~ Angyal and Gilham30 consider that ApSq0-R 4- R’O‘ - AreSOiO- 4- R’OR . . . . (2) the removal of the first sulphonyl group which will be the more accessible one will be facilitated by the inductive effect of the other sulphonyloxy- group (reaction 3).Ts? I Q- I 0 -5-9- t MeO“ - MaOTs t -7-v- - -<-‘$- + TsO- . . . . . .(3) OTs OTs If this is so the side reaction (4) should occur but the presence of the low- boiling dimethyl ether in the reaction product seems to have eluded investigators TsOMe f MeO’ -W Me20 t TSO- . - - - . (4) It is tempting to combine the reaction sequence (3) and postulate the concerted mechanism (9 but when the di-0-tosyl compound (XXXIIT) is treated with mild alkali for a short time the ah-epoxide (XVII) is formed together with some methyl 4 6-O-benzylidene-3-O-tosyl-a-~- glu~oside.~~ The formation of the mono-0-tosyl derivative supports Angyal and Gilham’s view and clearly hows the 2-sulphonyl group to be the more accessible. Before the discussion of epoxide formation from 2 3-di-0-sulphonyl compounds is continued the alkaline hydrolysis of 1 2-0-isopropylidene-5 6-di-0-tosyl- wD-glucofuranose (XXXIV) should be mentioned.Instead of 5 6-anhydro-1 2-0-isopropylidene-&-~- glucofuranose 3 6-anhydro- 1 2- 0-isopropylidene-5-0-tosyl- a-D-gluco- M a O t Gois sp7 ( 5 ) 0-CH CH,.OTs y 2 Ph CH-0 1 Q*Me TSO*$%.$? ”“*CQ? OTs 0-CMe -CMe (x x x I 1 I> (xxx I v) (XXXV) furanose (XXXV) is formed.32 It is not immediately obvious why the reaction should follow this course and not yield the 5 6-epoxide. 28 Cope and Shen ibid. 1956 78 5912. 2 9 Tipson Adv. Carbohydrate Chem. 1953 8 207. 30 Angyal and Gilham J. 1957 3691. 31 Honeyman and Morgan J. 1955 3660. 32 Ohle and Thiel Ber. 1933 66 525. NEWTH SUGAR EPOXIDES 39 There are several 2:3-di-O-tosyl derivatives and the problem is why should those of glucose and altrose give only one epoxide (allo- and manno-) whereas those of galactose give mixtures of the gulo- and the tab-epoxide.The factors which have been discussed in the preceding section must operate here and although the uncertainty about the exact mode of hydrolysis makes it difficult to be precise it is valuable to make a pre- liminary conformational appraisal. In the following discussion it is assumed that reaction (3) operates. 0-S fission may then occur in either diequatorial or diaxial systems but by the argument already developed the resulting anion must be in or approaching the diaxial position before epoxide formation. The same factors thenshouldaffect this reaction as are believed to influence the reaction of mono-0-toluene-p-sulphonates ; namely non-bonded interactions by axial substituents and passing interaction of two &-groups on change of conformation.It is however the first step in reaction (3) which determines the course of the reaction. It will of course be recognised that there may be an electronic influence from the acetal character of C(l) but in none of the reactions-epoxide formation or fission-has the Reviewer found any consistent indication that this is so. In methyl 4 6-0-benzylidene-2 3-di-O-tosyl-a-~-altroside (XXXVI) the reacting groups are diaxial and the manno-epoxide (XVI) is formed;33 it is unfortunate that the reaction conditions employed were too severe to give any indication of the ease of reaction. It can be inferred that the 2-toluene- O-CH2 P h T 0 / c H 2 Ph*CH-0 TsO QOM.O-OMe (XXXI t la) (xx XVI) Ph t-* Ph t - O P h T O l C H 2 oe" o& o& TsO TsO TsO OMe TsO (XXXVI I) Ts 0 Tso OMe p-sulphonyloxy-group is the more accessible since this must provide the anion to displace the 3-group. In compound (XXXIIIa) and 1 5-anhydro- 4 6-0-benzylidene-2 3-di-O-tosyl-o-glucitol (XXXVII) which has been shown to give also the a l l o - e p ~ x i d e ~ ~ ~ ~ ~ the primary attack must be also at OTs 21. The picture is complicated by the analogous galactose derivatives. (XXXVI I I) (XXXIX) 33 Robertson and Whitehead J. 1940 319. 34 Zissis and Richtmyer J. Amer Chem. Sac. 1955 77 5154, 40 QUARTERLY REVIEWS Methyl 4 6-0-benzylidene-2 3-di-O-tosyl-/3-~-galactoside (XXXVIII) gives only the talo-epo~ide~~ whereas the a-glycoside (XXXIX) gives a mixture of tab- and gulu-epo~ide.~~ In the former the primary attack must be at OTq3) and in the latter at both OTq,) and OTs,,).There does not appear at present to be any obvious reason for this behaviour. 2.6. Epoxides from Amino- and Nitrate Derivatives.-Deamination of aminodeoxy-compounds with nitrous acid occurs when there is a hydroxyl trans to the amino-group and an epoxide is formed. The reaction follows the course shown in (6). Methyl 2-amino-4 6-0-benzylidene-2-deoxy-a- D-altroside (XL) and methyl 3-amino-4 6-0-benzylidene-3-deoxy-a-~- altroside (XLI) are rapidly converted into the epoxides (XVII) and (XVI) by a solution of sodium nitrite in acetic acid.37 Sugar epoxides are equally easily formed in the same way from 4-amino-1 6-anhydro-4- deoxy-/h-mannose (XLII) and 6-amino-6-deoxy-l 2-0-isopropylidene- a-D-glucofuran ose3* (XLIII).During his studies on the sugar nitrates Honeyman has found that certain derivatives yield an epoxide on alkaline hydrolysis. The subject is complex and it can only be pointed out here that the derivatives of methyl 4 6-O-alkylidene-u-~-glucoside which yield the 2 3-allo-epoxide are the 2 3-dinitrate 3-nitrate 2-0-tosyl 3-nitrate and 3-0-tosyl 2-nitrate.39 S o r b and Reichsteh Helv. Chim. Acta 1945 28 1 662. Wiggins Nature 1946 157 300. 36 Gyr and Reichstein ibid. 1945 28 226. 38 Bashford and Wiggins ibid. 1950 165 566. 39Ansell and Honeyman J. 1952 2778; Honeyman and Morgan J. 1955 3660; Honeyman and Stening J. 1957,2278. "TH SUGAR EPOXIDES 41 3. Reactions Peat15 has very clearly described the stereochemistry of ring-opening by nucleophilic reagents (7).It is only necessary to add that with acidic reagents the same reaction occurs but is faster because protonation of the epoxide facilitates the movement of electrons (8). In the early work the reactions were nearly all with alkaline reagents and the products were .(7) Y I Y- *J 1 " -c-y- - ;c-c( ;c-c; 4 -c-c- I 0- 20' '03 6- * . * * ' [X may be HO- RO- NH,,RS- H- etc.; Y may be CI- Br- (RO),PO- etc.] sugar derivatives which could be easily characterised. The isolation of both isomers established the constitution of the epoxide although there appeared to be no reason for the predominance of one isomer over the other. In the last few years epoxide fission has been much discussed and the pattern of reaction is now clear.22~40~41 3.1.Ring-opening in Systems with a Rigid Conf~rmation.-Mills~~ first suggested the applicability to sugar epoxides of Furst and Plattner's rule that steroid epoxides break to give predominantly the axial isomer. The geometry of axial opening is shown at (a) (the small arrows indicate the direction of movement of the oxiran-carbon atoms) and it is obvious that there is a favourable co-planar transition state. Equatorial opening (b) is seen to be a very hindered process. When the conformation of a sugar epoxide is made rigid by a tram- fused 4:6-benzylidene group or a 1:6-anhydro-ring the product of ring scission contains almost exclusively the axial isomer. Thus with nucleo- philic reagents the epoxides shown in the annexed group of formulze (A) 40 Angyal Chem. a d Ind. 1954 1230. 41 Overend ibid.1955 995. 4a Mills cited by Newth and Homer J. 1953 989. 42 QUARTERLY REVIEWS give predominantly the products indicated. The very small amount of the other isomer which is usually formed shows that equatorial opening can occur to a limited extent. PhT0,5~2 ,o 0 ,o 0- OMe OMe“ CH 0 X = OMeI6 ( 4 0 . - N H l6 (Scheme A) x 3.2. Ring-opening in Systems with a Flexible Conformation.-The monocyclic sugar epoxides have a flexible conformation and can exist in two forms (9).22 If it is accepted that axial attack is the rule either *,=* . . . . . . . . (9) form may react and the products can then change into their most stable conformation^.^^ From this point of view it is unnecessary to postulate “exceptions” to Furst and Plattner’s rule. The predominant isomer will have its origin in the more stable form of the epoxide; the proportion of each isomer will reflect the energy difference between the two conforma- tions of the epoxide.It is not possible to predict at present which conforma- tion will be the more stable. Charalambous and P e r ~ i v a l ~ ~ examined the fission of methyl 2 3- and 43 Peat and Wiggins J. 1938 1810. 44 Grob and Prins Helv. Chim. Acta 1945 28 840; Jeanloz Prins and Reichstein 46 Harvey Michalski and Todd f. 1951 2271. 46 Prins J. Amer. Chem. Sac. 1948 70 3955. 47 Myers and Robertson ibid. 1943 65 8; Wiggins f. 1947 18. 48 Charalambous and Percival J. 1954 2443. ibid. 1946 29 371. NEWTH SUGAR EPOXIDES 43 3 4-anhydro-6-deoxy-a-~-taloside and their 2- and 4-0-methyl derivatives by sodium methoxide. The course of the reactions of the two 0-methyl derivatives is shown in the following formula= (B) and the axial opening of the stable conformations of the epoxides (eq’ eq’ ax‘) accounts for the products which were isolated.The unrnethylated OMe ( 7 d O M e O 4 - O M e omMe Me OMe (Scheme B) anhydrides in contrast --c MeoMe Hb bMe gave predominantly the alternative isomers and for comparison the “un- stable” (ax‘ ax‘ eq’) conformations are shown in (C). It is however from these two conformations that the products must originate. The only reasonable explanation must involve the free hydroxyl group and it is suggested that hydrogen-bonding between this and the lactol-oxygen atom makes the conformations in (C) the more stable (cf. 1 6-anhydrides for similar atomic distances). M e ~ T e JMe H o o M e -c H o e e OMc 0 OH H O ‘ 0 H 2 pts M e a H 0’ ocyMe OH Ipt.(Scheme C) c This suggested role of the hydroxyl group adequately explains the persistent formation of xylose derivatives (XLV) from methyl 2 3-anhydro- P-D(and L)-ribopyranoside (XLIV). 3.3. Epoxide Migration.-It has been seen that trans-opening of an epoxide ring occurs by attack of a nucleophilic reagent on one of the oxiran-carbon atoms from the side opposite to the oxygen atom. This reagent may be within the molecule itself as in the formation of methyl 44 QUARTERLY REVIEWS 3 6-anhydro-fl-~-glucoside (XXII) from (XX) or (XX1).12 If there is a hydroxyl group adjacent to the epoxide but trans to the ring an intra- molecular displacement by the hydroxyl anion can also occur with the formation of a second epoxide (Scheme 10).This migration was first a G-> -0- ?J postulated by Lake and Peat5* to explain the formation of methyl 3:4- anhydro-/3-D-altroside (XLVIII) as well as methyl 2 3-anhydro-P-~- mannoside (XLVII) from methyl 2-0-tosyl-P-~-glucoside (XLVI). The reaction was also assumed to occur when 1 6-anhydro-3-O-tosyl-#%~- altrose (XXVIII) was found to give not 1 6-2 3-dianhydro-P-~-mannose but 1 6-2 3-dianhydro-/3-~-altrose~~ (XLIX). CHiOH CH2-OH CH2-OH CH,-0 HO Icsp” - “o&e - o e e o&$ OTs (x LV I) (XLVI I) (XLVI I I) (XLIX) Epoxide migration in the inositol series has recently been demonstrated3* in the conversion of (lS)-l 2-anhydroalloinositol (L) into ( 1 9 - 1 2- anhydroneoinositol (LI) by very mild alkali. At equilibrium there is present 10% of the compound (L) and 90% of its isomer (LI).The latter is more stable by about 1-3 kcal./mole and in its preferred conformation has only one axial hydroxyl group. In Scheme (D) it can be seen that it is an obvious requirement that the attacking hydroxyl anion shall be in an axial position; the displaced anion will then be equatorial. For the reverse reaction there must be a conforma- tional shift and axial attack can again occur. It is necessary to consider the non-bonded interactions in all four forms when attempting to predict the 4B Honeyman J. 1946 990. Mukherjee and Todd J. 1947 969. 61 Baker and Schaub J. Org. Chem. 1954,19 646. 63 Kent Stacey and Wiggins J. 1949 1232. 63 Allerton and Overend J. 1951 1480. 64 Lake and Peat J. 1939 1069. NEWTH SUGAR EPOXIDES 45 direction of equilibrium.It is interesting that the conformational shift cannot occur in 1 6-3 4-dianhydro-fl-~-altrose (XLIXa) and it is very doubtful whether its reconversion into 1 6-2 3-dianhydro-j3-~-mannose is possible. 0-J ‘ 0 JF Jf (X LI Xa) (Scheme D) 3.4. 3:4-AnhydrogaIactose.-In 1935 Oldham and Robertsons5 isolated mono-0-isopropylidene derivatives of galactose and gulose from the reaction of the 3 4-anhydro-derivative of methyl a-D-galactoside (LII). This apparently anomalous cis- and trans-opening of the oxide ring was re-investigated by Labaton and Newths6 who confirmed the earlier work and examined the action of hydrochloric acid on the anhydro-sugar. It was however their assignment of a 3 6-structure to the benzylidene derivative of one of the chlorohydrins believed to be methyl 4-chloro-4-deoxy-a-~- glucoside that stimulated further investigation.Buchanan5’ saw that the syrupy anhydride (LII) could be a mixture of the guto- and galacto- epoxides owing to epoxide migration and that some of the earlier products could be derived from the gulo-epoxide. By treating methyl 2 3-anhydro- 4 6-0-benzyl.idene-a-~-gu~oside~~ with hydrochloric acid he showed that “methyl 3 6-0-benzylidene-4-chloro-4-deoxy-a-~-glucoside”~~ was methyl 4 6-0-benzylidene-2-chloro-2-deoxy-a-~-idoside (LIII) and this con- firmed the presence of the gulo-epoxide in the mixture designated 0.11). There was now an obvious uncertainty about the identity of “methyl 3-chloro-3-deoxy-a-~-guloside”. Buchanan re-examined this compound 66 Oldham and Robertson J. 1935 685. 66 Labaton and Newth J. 1953 992.67 Buchanan J. 1958,995; Chem. and Ind. 1954 1484. 46 QUARTERLY REVIEWS and observed a very slow consumption of 1 mol. of periodate. It was therefore methyl 4-chloro-4-deoxy-a-~-glucoside (LIV). This was con- firmed when authentic methyl 3 4-anhydro-a-~-galactoside was treated with hydrochloric and one of the chlorohydrins was identical with Labaton and Newth's compound; the other by virtue of its stability to periodate was methyl 3-chloro-3-deoxy-~-~-guloside (LV). To explain the formation of 0-acetyl-3 4-isopropylidene-a-~-galacto- side in the Oldham and Robertson reaction acid-catalysed cis-opening of the epoxide by acetone was suggested.56 Buchanan5' showed however that this was not the 2-0-acetate but the 6-0-acetate (LVI) and its forma- tion was explained when he treated pure and authentic methyl 2-0-acetyl- 3 4-anhydro-6-O-trityl-a-~-galactoside (LII) and methyl 4-0-acetyl-2 3- anhydro-6-O-trityl-a-~-guloside (LVII) with anhydrous hydrogen chloride in acetone.58 The former was converted into methyl 0-acetyl-4 6-0- isopropylidene-a-D-guloside (LVIII) and the latter into 6-0-acetyl-3 4- 0-isopropylidene-a-D-galactoside (LVI).This was explained by the directive influence of the neighbouring trans-0-acetyl group on the epoxide fission through the carbonium-type intermediates shown in (LII)+(LVIII) and (LVII)+( LVI) . HO OH (Llll) (LI v) (LV) Me H Ac CMe t &Vlll) Me H 3.5. Reaction with Grignard Reagents.-The epoxides (XVI) and (XVII) have been very completely examined. With alkyl- and phenyl-magnesium halides the products are the halogenohydrins and these are also formed with magnesium halides.Diethylmagnesium and diphenylmagnesium give 68 Buchanan J, 1958 2511. NEWTH SUGAR EPOXIDES 44 C-ethyl and C-phenyl derivatives. The reactions which are shown in the batch of formulze below do not show a consistent pattern and more varied examples are required before the mechanism of reaction can be fully under stood. (X= I Br ,or Cl) 0-CH 0-CH 0-CH Ph *C I H-0 o O M e A X Ph-iH-O&Mrz z P h - l H - O o O M e Et (X=Bror I) (XVO Reagents 1 (a) MgX262 (ii) EtMgBrso; EtMgl PhMgBr. 2 MgMelSS. 3 (a) MeMgP 6s Newth Richards and Wiggins J. 1950 2356. 6o Richards and Wiggins J. 1953 2442. 61 Richards J. 1954 4511. 6a Richards Wiggins and Wise J. 1956 496. 63 Foster Overend Stacey and Vaughan J. 1953 3308. 64 Richards J. 1955 2013. EtMgl PhMgl; (b) MgBP2 Mgl (not MgCI,). 4 MgPh,. 5 MgEt,63.