ADVANCES XN CARBOHYDRATE CHEMISTRY VOLUME 10
This Page Intentionally Left Blank
t
Carbohydrate Chemistry Editor ME...
45 downloads
2669 Views
22MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
ADVANCES XN CARBOHYDRATE CHEMISTRY VOLUME 10
This Page Intentionally Left Blank
t
Carbohydrate Chemistry Editor MELVILLE L. WOLFROM Arsistant Editor R. STUART TIPSON Board of Advisors C. B. PURVES J. C. SOWDEN ROYL. WHISTLER
HERMANN 0. L. FISCHER
It. C. HOCKETT
W. W. PIGMAN
Board of Advisors for the British Isles J4. L. HIRST
STANLEY PEAT
MAURICESTACEY
Volume 10
1955
ACADEMIC PRESS INC., PUBLISHERS NEW YORK, N. Y.
Copyright, 1955, by ACADEMIC PRESS INC. 125 East 23rd Street New York 10, N. Y.
A12 Rights Reserved
No part of this book may be reproduced in any form by photostat, microfilm, or any other means without written permission from the publishers. Library of Congress Catalog Card Number: 45-11351
PRINTKD I N T H E UNITED STATES OF AMERICA
CONTRIBUTORS TO VOLUME 10 W. W. BINKLEY,Department of Chemistry, The Ohio State University, Columbus, Ohio*
G. P. ELLIS,Department of Chemistry, King's College, University of London, England A. B. FOSTER,Department of Chemistry, The University, Edgbaston, Birmingham, England
L. J. HAYNES,The University, Edinburgh, Scotland ,JOHN E. HODGE,Northern Utilization Research Branch, Agricultural Research Service, U . S . Department of Agriculture, Peoria, Illinois JOHN HONEYMAN, Department of Chemistry, King's College, University of London, England A. J. HUGGARD, Department Columbus, Ohio
0.f
Chemistry, The Ohio State University,
GEORGEG. MAHER,Research Laboratories, Clinton Foods Inc., Clinton, Iowa J. A. MILLS, Division of Biochemistry and General Nutrition, Commonwealth Scientific and Industrial Research Organization, Adelaide, Australia
F. H. NEWTH,The University, Cambridge, England W. J. POLGLASE, Department of Biochemistry, University of British Columbia, Vancouver, British Columbia, Canada 'Present address: The New York Sugar Trade Lahoratorg Inc., 113 Pearl Street, N e w York
V
4 , N . 1'.
This Page Intentionally Left Blank
PREFACE With this Volume, “Advances in Carbohydrate Chemistry” comes of age through the completion of a series of ten issues. Herein, modern conformational analysis is applied to the carbohydrate field by J. A. Mills (Adelaide), The ever-recurrent and always puzzling subject of nitrogen chemistry is elaborated in a discussion of the glycosylamines and their rearrangement products by G. P. Ellis with J. Honeyman (London) and by J. E. Hodge (Peoria). The preparation and reactivity of the useful glycosyl halides is presented by L. J. Haynes (Edinburgh) and F. H. Newth (Cambridge). W. W. Binkley (Columbus) summarizes the present status of column chromatographic technique as applied to the sugar group. Our series of chapters on the methyl ethers is augmented by G. G. Maher (Clinton, Iowa). Polysaccharide chemistry is represented by a chapter on the non-cellulosic components of wood from the pen of W. J. Polglase (Vancouver) and by one on the biochemically significant subject of heparin from A. B. Foster and A. J. Huggard (Birmingham and Columbus). These, together with an obituary of the late, esteemed Dr. E. G. V. Percival of Edinburgh, complete this Volume and are offered as a contribution to the summarizing of progress in the ever-growing subject of carbohydrate chemistry.
M. L. WOLFROM Columbus, Ohio
vii
This Page Intentionally Left Blank
CONTENTS CONTRIBUTORS TO VOLUME 1 0 . .. . . . . . . , , . . , . . . . . . . . , , . . . , . . . . . . . . . . . . . . . . . . . . .
v
PREFACE.. . . . . . . , . _ _, . . . . , , . . . . , , . . . . . . . . . . . , . . , , . . . . . , . . , , . . . . . . . . . . . . . . . . . vi EDMUNI) GEORGE VINCENT
!?ERrIVAL. . . . . . . . . .
. , . . , . . . . . . . . . . . . . . . . . . . . . . . . . . X ... lll
The Stereochemistry of Cyclic Derivatives of Carbohydrates BY J. A. MILLS,Department of Biochemistry and General Nutrition, Commonwealth Scientijic and Industrial Research Organization, Adelaide, Australia
I. Introduction . . . . . . . . . . . . . , . , . . , . _ _. ._ _ ,, . . . . . . . . . . . . . . . . . . . . . . _ _. . . . . . 2 11. Nomenclature and Methods of Illustration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 111. Steric Requirements of Reactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 IV. Fundamental Properties of Rings.. . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 V. Cyclic Acetals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 VI. Anhydro Compounds.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Column Chromatography of Sugars and Their Derivatives BY W. W. BINKLEY, Department of Chemistry, T h e Ohio Stale University, Columbus, Ohio
I. Introduction . . . . _ _, . . . . , . . . . . . . . . . . , . . . . . , . . , . . . . . . . . . . . . . . . . . . . . . . . . . 55 11. Adsorbents for the Column Chromatography of Sugars, Sugar Alcohols, 56 and Sugar Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Column Chromatography of Sugar Derivatives. . . . . . . . . . . . . . . . . , . . . . . . . 79 Glycosylamines BY G. P. ELLISAND JOHNHONEYMAN, Department of Chemisfry, King's College, University of London, England I. 11. 111. IV. V. VI. VII. VIII. IX.
Introduction ..... _ _. . . _ . . . . . . _ .. . . . . , , . . . . . . . . , . , . . . . . . . . . . . . . . . . . . . . Nomenclature . . . . . . . . . . . . . . . . . , . . . . . , . . . . _ .. _ _ _ . . . . _ .. _ . ._ _ _ . _ _. . .... Preparation . . . . , . . ,.. , . . . . . . . . . . . . , . . . . . . . . . . . . . . . , . . . , . . . . . . . . . . . . . . . . Physical Properties.. . . . . . . . . . , . . . . . . . . . , . , , . . . . . . , , , . . . . . . . . . . . . . . . . . . Structure. . . . _ .. , . . _ .. _ ., . . _ _ . _. .. . . . . . . . . . . . , . . . _ .. . . . . . _ . . . . . . . . _. . . Diglycosylamines . . , . . . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diitmide Derivatives of Aldoses.. . . , . . . . , , . . . . . . . . . . . . . . . . . . . . . . . . . . . . Uses of Glycosylamines.. . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tables of Properties of Glycosylamines.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,
95 96 97 101 102 120 121 124 125
The Amadori Rearrangement BY JOHN E. HODGE,Northern Utilization Research Branch, Agricultural Research Service, 7J. S . Department of Agriculture, Peoria, Illinois
I. Introduction... ............... . . ... ... ... . .......... . . . . . . . . . . . . . . . . . . . 169 ix
CONTENTS
X
I1 . Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Scope of the Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Experimental Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Mechanism of the Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Physical Properties of Amadori Rearrangement Products . . . . . . . . . . . . . . . VII . Chemical Properties of Amadori Rearrangement Products . . . . . . . . . . . . . . . VIII . Proof of Structure of Amadori Rearrangement Products . . . . . . . . . . . . . . . . IX . Retrospect and the Future . . . . . . . ..................... .. X . Tables of Compounds . . . . . . . . . . . . ............................
172 173 175 178 185 187 199 201 203
The Glycosyl Halides and Their Derivatives BY 1, . J . HAYNES, l’he University, Edinburgh, Scotland A N D F . H . NEWTH, The l J n i v w s i t y , Cambridge, England I . Introduction., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 . Preparation of Glycosyl Halide Derivatives . . . . . . . . . . . . . . . . . . . . . . . . 111. Structure of Glycosyl Halides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 IV . General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 V . Reaction Mechanism and Effect of Structure on Reaction Rates . . . . . . . . 234 V I . Reactions of the Poly-0-acylglycosyl Halides . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 VII . Tables of Properties of Some Glycosyl Halide Derivatives . . . . . . . . . . . . . . 246 The Methyl Ethers of the Aldopentoses and of Rhamnose and Fucose BY GEORGEG . MAHER,Research Laboratories, Clinton Foods I n c . , Clinton, Iowa Table I . ,. . . . . . . . . . . . . . . . . . . . . . . . . . ...................................... 257 Table I1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 Table I11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 Table IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 Table V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 Table V I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 Table V I I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268 Table VIII . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 The Methyl Ethers of D-Galactose BY GEORQEG . MAHER,Research Laboratories, Clinton Foods I n c . , Clinton, Iowa Table Table Table Table
I......................................................................
I1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
273 274 277 279
Polysaccharides Associated with Wood Cellulose BY W . J . POLGLASE, Department of Biochemistry, University of British Columbia, Vancotcuer, British Columbia, Canada .................................... I . Introduction. . . . . . . . . . . . . . . . . I1 . Carbohydrate Constituents of Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I11 . Carbohydrates in Wood Cellulose Preparations . . . . . . . . . . . . . . . . . . . . . . . . . IV . Preparation and Composition of Wood Cellulose . . . . . . . . . . . . . . . . . . . . . . . . V . Fine Structure of Wood Cellulose and Associated Polysaccharides . . . . . . .
283 285 287 316 328
CONTENTS
xi
The Chemistry of Heparin BYA . B . FOSTER. Chemistry Department. T h e University. Birniingh.am. England A N D A . J . HUOGARD. Chemistry Department. T h e Ohio State University. Columbus. Ohio I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336 I1. The Discovery of Heparin., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336 111. The IV . The V . The VI . The
Isolation and Purification of Heparin . . . . . . . . . . . . . . . . . . . . . . . . . . . 337 . . . . . . . . 348 Structure of Heparin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anticoagulant Activity of Heparin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359 Biosynthesis of Mucopolysaccharides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
AUTHORINDEX .................................... SUBJECT
I N D E X .........................
CONTENTS O F
VOLUMES
1-9
........................................
. . . . . . . . 369
. . . . . . . . 391
........ 435
This Page Intentionally Left Blank
EDMUND GEORGEVINCENTPERCIVAL 1907-1 951 The death of E . G. V. Percival a t the early age of 43, when he was a t the height of his powers as a scientist, was a severe blow to carbohydrate chemistry. He was born on 10th November, 1907, in the Midlands of England a t Hinckley in the County of Leicester, being the younger son of Albert Henry and Elizabeth Percival. As a boy, he attended the King Edward VII Grammar School in the neighboring town of Coalville. His school record showed great promise, and in 1925 he entered the University of Birmingham where he studied for the honours degree in chemistry. His undergraduate career was one of unusual brilliance, and it was a foregone conclusion that he would obtain a First Class Honours in Chemistry a t the end of the course in 1928. A research career was clearly indicated for him, but, before choosing a field in which to specialize, Percival wisely decided to gain as wide an experience as possible. I n the fall of 1928, he joined William Wardlaw’s group of workers in the Birmingham University laboratories and took part in work on the chemistry of metallic co-ordination compounds. A year later, he gained a Research Fellowship awarded by the Canadian Pulp and Paper Association, tenable a t McGill University, Montreal, and there, in Harold Hibbert’s laboratory, he investigated various physicochemical problems concerned with addition compounds of cellulose. Before returning to England, he made an extensive tour of research centers in Canada and the United States, gaining much experience which was of value to him in his later work, and making, as was the case wherever he went, many lasting friendships. The experience in Montreal had given him a deep interest in carbohydrate chemistry, and this became his main field of work 011 his return to Birmingham University in October, 1930, as senior research assistant to W. N. Haworth (later Sir Norman Haworth). This period was one of rapid and successful progress, and a t this stage he decided to seek a post which would enable him to pursue a career in the academic world. The opportunity came three years later (1933), when he was appointed to a lectureship in organic chemistry in the University of Edinburgh. Here, in the Department presided over by James P. Kendall, F.R.S., Percival settled happily, and during the next 18 years he built up a research school in carbohydrate chemistry which had an international reputation. He found conditions so congenial in Edinburgh that he made no serious attempt to gain xiii
xiv
EDMUND GEORGE VINCENT PERCIVAL
promotion by moving elsewhere. In 1934, he married Ethel Elizabeth Kempson, herself a chemist, a graduate of the University of Birmingham where she had been a student in Sir Norman Haworth’s laboratories. After her arrival in Edinburgh, she took an active part in the development of Percival’s research activities, and assisted in the teaching of undergraduates. She and their two children survive him, and Mrs. Percival is COIItinuing, as a member of the University staff, some of the work initiated by her husband. Percival entered fully into the scientific and social life of both the University and the City of Edinburgh. He took a keen and sympathet,ic interest in student welfare and activities, and amongst his University duties was that of Director of Studies. He was elected t o membership in the Faculty of Science, and in 1948 was promoted to the status of Reader in the University. Shortly before his death he had been chosen to serve as a member of the Senatus Academicus. He gave his time and energy, with unselfish enthusiasm, to work for various scientific societies. For many years he had been interested in the work of the Pharmaceutical Societ,y and in British Chemical Abstracts. H e was concerned with work for the Institute of Brewing, he was one of the original pioneers in investigations on the chemistry of marine algal polysaccharides sponsored by the Scottish Seaweed Research Association, and he was an active member of this Association’s Chemical Advisory Committee. He served also on the local district committees of the Royal Institute of Chemistry and of the Society of Chemical Industry, and in addition acted for two years as Chairman of the Edinburgh and South East Scotland section of the Royal Institute. He was the local representative in Edinburgh of the Chemical Society (London), and at the time of his death was a member of the Society’s Council and of its Committee on Carbohydrate Nomenclature. His help was called for on all special occasions, and in 1948, when the Society of Chemical Industry held its annual general meeting in Edinburgh, and again in 1951, when he was Local Secretary for the Chemistry Section of the British Association for the Advancement of Science, his unselfish and unstinted efforts contributed in large part to the success of the meetings. It was characteristic of Percival that, to everything he undertook, he brought enthusiasm and intensity of purpose, whether it was in playing tennis or cricket, discussing painting (in which art he was deeply knowledgeable), walking on the Cheviot Hills, or in teaching and research. He had a brain which acted with lightning rapidity, and he had the power to carry through with accuracy (and with every attention to detail) a phenomenal amount of work in a very short time. He was a clear lecturer and expositor, and in consequence he was frequently invited to visit other universities and chemical societies. These visits, especially those which in-
EDMUND GEORGE VINCENT PERCIVAL
xv
volved traveling, he found highly congenial. He was a t his best as a teacher with advanced students of quick perception, who could follow without difficulty the logical but rapid development of his theme. Friendly, generous, sympathetic, and approachable, he was an inspiring leader for postgraduate research workers, who found him full of ideas and suggestions, and marveled a t his encyclopedic memory for details. All who had the privilege to work with him and enjoy his friendship will remember especially his kindliness, generosity, and unselfishness, his deep sense of loyalty, and his capacity to see and appreciate the humorous side of everyday happenings. Percival's outstanding ability was soon recognized, and awards came to him a t an early age-Fellowship of the Royal Institute of Chemistry in 1936, the D.Sc. degree of the University of Edinburgh in 1938, and, in 1941, election t o the Fellowship of the Royal Society of Edinburgh. Although Percival's main interest in chemistry lay in the carbohydrate field, his earliest research, carried out under William Wardlaw in Birmingham (1928-1929) was concerned with polynuclear co-ordination compounds of cobalt with amines.' During the following year, he held a Research Fellowship in Montreal, working with Hibbert on the constitution of sodacellulose, the hydrolysis of polysaccharides, and the absorption of aluminum ions on cellulose. It was demonstrated that a chemical compound is formed between the sodium hydroxide and the cellulose, containing about 15 % of alkali (or one molecule to each anhydro-D-glucose unit). Investigation of the rates of hydrolysis indicated that the lower rate for cellulose as compared wit,h that of starch is a reflection of the comparative rates of hydrolysis of cellobiose and maltose.2 I n the following three years (1930-33), working with Sir Norman Haworth, Percival began his studies in structural carbohydrate chemistry, and took part in the early work of the Birmingham School on the investigation of polysaccharides. Some of the most important results he obtained were concerned with the structures of starch and glycogen. Up to this time, strictly chemical evidence was lacking as to the presence of preformed maltose units in starch and glycogen, and concerning the nature of the ring present in the D-glucose residues. Percival developed a method of simultaneous deacetylation and methylation of the acetylated polysaccharides followed by acetolysis of the methyl derivatives. These yielded, amongst other products, a partially methylated maltose. The hexa-0methylbiose was oxidized to the bionic acid and subjected to further methylation. The resulting methyl octa-0-methylmaltobionate was then (1) E. G . V. Percival and W. Wardlaw, J . Chem. Soc., 1317, 1505, 2628 (1929). (2) A. C. Cuthbertson, H . Hibbert and E. G. V. Percival, J. Am. Chenz. SOC.,62, 3257, 3448 (1930); H. Hibbert and E. G . V. Percival, ibid., 62, 3995 (1930).
xvi
EDMUND GEORGE VINCENT PERCIVAL
hydrolyzed and the tetra-0-methyl-D-glucopyranose and tetra-0-methylD-glUCOniC 1,4-lactone were separated and identified, thus showing that the maltose structure must be present in starch and glycogen. Similar methods were applied to a di-0-methylxylan, in which the methoxyl groups were found to be attached to C2 and C3, showing that the linkage occurs a t C4 and that the D-xylose is present in the pyranose forrne3The methylation and hydrolysis technique was further applied to inulin, the methylated derivative of which gave rise to rather labile D-fructofuranose derivatives. In order to avoid unnecessary decomposition, the hydrolysis (after methylation) mas carried out using aqueous methanolic oxalic acid, and the yield of tetra-0-methyl-D-fructofuranose indicated the presence of one terminal nonreducing D-fructofuranose unit in 32 residues? One of Percival’s major contributions during his Birmingham period was the part he played, in collaboration with other members of the School, in determining the constitution of ascorbic acid (vitamin C). Oxidation studies were carried out which showed t,hat L-threonic acid is one of the products of oxidation, indicating ascorbic acid to be a derivative of L-gulose. The nature of the ring system was investigated by methylation studies, and vitamin C was proved to be the enolic form of 2-keto-~-gulonic1 , 4 - l a ~ t o n e . ~ Confirmation was obtained for this structure by the synthesis of vitamin C by the same group of workers. A t this point in his career, Percival moved to Edinburgh (1933), where he continued his work in the carbohydrate field. During the succeeding 18 years, his numerous publications covered a wide variety of subjects. From among the early work in Edinburgh may be ment,ioned the study of the compounds of alkali-metal hydroxides with sugars and polysaccharides, as a result of which he suggested that the alkali-suga,r complexes are definite compounds, and indicated how, by methylation, the points of attachment of the added hydroxides could be determined. Compounds of alkali-metal hydroxides with mono-, di-, and poly-saccharides were investigated.6 The methylation studies resulted in the production of partially methylated sugars whose identification required reference compounds, and Percival proceeded to study various mono-0-methylhexoses not previously charac(3) W. N. Haworth and E. G. V. Percival, J. Chem. SOC.,1342, 2850 (1931); 2277 (1932). (4) W. N. Hanorth, E. L. Hirst and E. G. V. Percival, J. Chew. SOC.,2384 (1932). ( 5 ) R. W. Herhert, E. L. Hirst, E. G. V . Percival, R. J. W. Reynolds and F. Smith, J. Chem. SOC.,1270 (1933) ; R . G. Ault,, D. I<. Baird, H . C. Carrington, W. N . Haworth, R. W . Herhert, E. L. Hirst, E. G. V. Percival, F. Smith and M. St,acey, ibid., 1419 (1933). 1160 (1934);648 (1935); E. G. V. Percival and (6) E. G. V. Percival, .J. Chem. SOC., G. G. Ritchie, i b i d . , 1765 (1936); W. J . Heddle and E. G. V. Percival, i b i d . , 1690 (1938).
EDMUND GEORGE VINCENT PERCIVAL
xvii
t,erized; for example, 4-O-methyl-~-glucoseand 2- and 6-O-methyl-~-galactose. He also devoted much attention to the structure of sugar osazones. By the methylation method, he decided that the previous assumption that osazones possess a straight-chain structure was erroneous, and that sugar osazones could exist in a ring form. I n the case of D-arabinohexose phenylosazone (“glucosazone”), the ring was shown to be of the 2 , 6 type. I n the course of these studies, the hitherto-unknown 5-0-methylD-arabino-hexose phenylosazone was characterized for the first time by a single methylation of D-arabino-hexose phenylosazone, and the work was also extended t o include the structures of several new anhydrohexose osazones of the Diels type.’ An entirely different approach to the structure of methylated and free sugars was made by the application of Boeseken’s method, involving measurements of the conductivity of sugars in boric acid solution. I n this way, the respect,ive configurations assigned to the hydroxyl groups on C l and C2 in a- and P-D-glucose were confirmed, the conductivity being considerably influenced by the cis or trans arrangement of the hydroxyl groups. The method was extended and applied to such methylated sugars as 3,4,6tri-0-methyl-a-D-mannose, where the hydroxyl groups on C1 and C2 were shown to have the trans arrangement.8 The main portion of Percival’s work in the years following 1936 was concerned with the structure of polysaccharides; in particular, of plant mucilages and seaweed polysaccharides. At this time, the polysaccharides of marine algae had been but little investigated; Percival was among the first in this field and became a recognized authority on the subject. One of the earliest studies he embarked on was that of the structure of agar, which presented a.n extremely difficult and complex problem. The results of the preliminary work on acetylation and methylation indicated that the basic structure consists of o-galactose units linked directly or indirectly through C l and C3,as shown by the isolation of the hitherto-unknown 2,4,6-tri0-methyl-a-D-galactose and of an unknown portion giving strong ketose reactions. Later work proved the presence also of 3,6-anhydro-L-galactose, but the complexity of the molecule precluded t,he solution of the detailed structure atj that t,ime. This study also entailed the synthesis of several anhydrogalactose derivatives used in the invest,igat>ionas reference compounds.$ At about the same time, work was commenced on other algal (7) Elizabeth E. Pereival and E. G. V. Percival, J . Chem. SOC.,1398 (1935); 1320 (1937); 750 (1941); F:. G. V. Percival, ibid., 1770 (1936); 1384 (1938); 783 (1945); J. R. Muir and E. ‘2. V. Percival, ibitl., 1479 (1940); W. J. Heddle arid E. G. V. Percival, ibid., 1511 (1940). (8) H. T . Macpherson and E. G . V. Percival, J . Chem. SOC.,1920 (1937). (9) €3. G. V. Percival and J. C. Somerville, J. Che,m. SOC., 1615 (1937); I. A. Forbes and E. G. V. Percival, ibid., 1844 (1939); T. L. Cottrell and E. G. V. Percival, ibid., 749 (1942).
xviii
EDMUND GEORGE VINCENT PERCIVAL
polysaccharides, notably those forming the mucilaginous extracts of Chondrus crispus and Gigartina slellata, collectively known as Irish moss. The polysaccharides from the two algae are essentially similar, in that both contain a considerable number of ester sulfate groups and consist of galactopyranose residues linked through C1 and C3, a portion of the galactose being in the L-form in the Chondrus polysaccharide. The complete elucidation of the structure proved difficult owing to the presence of the sulfate groups which interfere with methylation.1° In view of these difficulties and of the need for more knowledge of carbohydrate sulfuric esters, Percival then investigated the sulfates of D-glucose and D-galactose, their rates of hydrolysis with reference to the position of the sulfate group in the molecule, and the formation of 3,6- and 5,6-anhydro sugars on alkaline hydrolysis.]' With the collaboration of the newly-formed Scottish Seaweed Research Association, Percival commenced a further series of investigations in the marine algal field,lla the earlier work being concerned with the development of specific methods of analysis for the various constituents present in seaveed. Procedures were worked out for the estimation of D-mannitol, alginic acid, laminarin, and fucoidin.12 At the same time, structural studies were instituted on the last three polysaccharides.*3 Laminarinlla was shown to consist essent>iallyof (3 1) linked P-D-glucose units, and to occur in two forms, a soluble and an insoluble, depending on the species of seaweed used for extraction. I n fucoidin,lla the principal sugar is L-fucose, along with sulfate groups, and a branched structure was proposed in which 2 + 1a-L links predominate. Among other seaweed polysaccharides, the xylan from Rhodymenia palmata was found to be of unusual interest in that it was shown to contain both (3 --f 1) and (4--+ 1) linked D-XylOSe residues as part of the main structure of the polysaccharide. Percival had had previous experience with xylans while investigating the structures of the mucilages
-
(10) J. Buchanan, Elizabeth E. Percival and E. G. V. Percival, J . Chem. SOC., 51 (1943); E. T. Dewar and E. G. V. Percival, ibid., 1622 (1947); R . Johnston and E. G. V. Percival, ibid., 1994 (1950). (11) Elizabeth E. Percival and E. G. V . Percival, J . Chem. Soc., 1585 (1938); 874 (1945); E. G. V. Percival and T. H. Soutar, ibid., 1475 (1940); R. B. Duff and E. G. V. Percival, ibid., 830 (1941); 1675 (1947); E. G. V. Percival, ibid., 119 (1945); R. B. Duff, J . Chem. SOC.,1597 (1949). (Ila) See also, T. Mori, Advances in Carbohydrate Chem., 8, 315 (1953). (12) M. Christine Cameron, E. G. V. Percival and A. G. Ross, J . SOC.Chem. Znd. (London), 67, 161 (1948); E. G. V. Percival and A . G. Ross, ibid., 67, 420 (1948); W. A. P. Black, W. J . Cornhill, E. T. Dewar, E G. V. Percival and A. G . Ross, ibid., 69, 317 (1950). (13) E. G. V. Pereivsl a n d A. G . Ross, .I. Cl,eni. Sot-., 717 (1950); 720 (1951); J. Conchie and €2. G. V. Percival, itlid., 827 (1950) ; J . J. Connell, E. L. Hirst and E. G. V . Percival, ibid., 3494 (1950).
EDMUND GEORGE VINCENT PERCIVAL
xix
from t,he seeds of various meinhers of t'he plantain family (Plantago lanccolata, Plantago arenaria, and Plantago ov,uta). All t'hese sylans are complex in st,ructure and contain, in addit>ion t o D-xylose, ot,her pentoses, hexoses, and uronic acid residues; for example, the mucilage from Plantago arenaria contains D-xylose, L-arabinose, u-galactose, and wgalacturonic acid. The mode of' attachment of the uronic acid is not clear, and the struct,ure of the main D-xylopyranose chain includes various types of linkage and is highly branched. I n the mucilage from Plantago lanceolnta, for example, the following types of D-xylose residue were foundL4to occur as building unit,s: D-xylp 1, 3 D-xylp 1, 4 D-xylp 1, :D-xylp 1, iD-xylp 1, and :D-xylp ;. Further studies were carried out on the xylans from esparto and from pear cell-wall. I n the former, the xylan was shown to be free from L-arabinose and t o consist of a singly-branched molecule, the main chain containing about 75 D-xylopyranose units linked through the 4 position, the branching point occurring on C3. For the pear cell-wall polysaccharide, the general structure was found similar, but with the modification that t3he terminal residue consists of a D-glucuronic, acid unit and the main chain is rather longer (ca. 115 D-xylopyranose Percival also applied himself to the study of t.he polysaccharides from lichens, as, for example, tjhe complex product of alkaline extraction of Iceland moss (Cetraria islandica). As a result of this work, he concluded tJhat t,he polysaccharide consists of p-D-glucose residues united by various linkages: 1,2, 1,3, 1,4, and 1 , G , and includes also terminal D-galacto- and D-gluco-pyranose end groups. It was not possible to decide whether these linkages all occur in one polysaccharide.16 I n the later stages of his career, Percival published papers dealing with barley starch1? and wood starches,L8the former being found to be, in the main respects, similar to other cereal starches. Fructans also attracted his attention, and those from couch grass and perennial rye grass were particularly investigated; it was shown that the fructan from couch grass has both 2 , l and 2,G linkages in the molecule, whereas that from rye grass is essentJially a 2,G-linked, straight-chain p o l y s a ~ c h a r i d eFinally, .~~ mention may be made of the inulin from dahlia tubers, in which the presence of (14) J. Mullan and E. G. V . Percival, J . Chem. Soe., 1501 (1940); W . A. G. Nelson and E. G. V. Percival, i b i d . , 58 (1942); R. A. Laidlaw and E. G. V. Percival, ibid., 1600 (1949); 528 (1950); E. G. V. Percival and I. C. Willox, i b i J . , 1608 (1949). (15) S. K. Chanda, E. L. Hirst and E. G. V . Percival, J . Chem. SOC.,1240 (1951). (16) H. Granichstadten and E. G. V . Percival, J . Chem. SOC.,54 (1943). (17) I. C. MacWilliam and E. G. V . Percival, J . Chem. SOC.,2259 (1951). (18) W. G. Campbell, J. L. Frahn, E. L. Hirst, D. F . Pnckman and E. G . V. Perc i v d , J . Chem. SOC.,3489 (1951). (19) P. C . Arni and E. G. V . Percival, J . Chem. SOC.,1822 (1951).
xx
EDMUND GEORGE VINCENT PERCIVAL
u-glucose as an integral part of the molecule was demonstrated and possible structures were suggestedz0in which one n-glucose residue occurred as aii end group, linked to D-fructose by a sucrose type of link, whilst the second u-glucose residue was in the middle of the chain, linked through C1 and C3. Oiily a few days before his death, Percival took a prorninerit part in a conference 011 grass, sponsored by the Nutrition Society, a t which he read a p a p e P 011 the “Carbohydrate Coiistituerits of Herbage.” In this paper, he emphasized the importance of the fructaris as reserve carbohydrates in the metabolism of the plant, arid expressed the opinion that they would be found to play a considerable part, in the chemistry of the preservation of grass in the form of silage. E. L. HIRST A. G. Ross (20) E. I,. Hirst, D I. McGilvrag and E. G. V. Percival, J . Cheni. Soc., 1297 (1950). (21) E. G. V. Percivnl, Brit. J . Nutrztion, 6, 104 (1952).
THE STEREOCHEMISTRY OF CYCLIC DERIVATIVES OF CARBOHYDRATES* BY J . A . MILLS Division of Biochemistry and General Nutrition. Commonwealth Scientific and Industrial Research Organization. Adelaide. Australia
CONTENTS 1. 1ntroduct)ion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Nomenclature and Methods of Illustration . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . Structural Formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 . Steric Requirements of Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Reversible and Irreversible Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . Reactions with Specific Steric Requirements . . . . . . . . . . . . . . . . . . . . . I V . Fundamental Properties of Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 . Conformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . Monocyclic Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a . Five-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b . Sis-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c . Other Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3. Structures Containing Fused Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a . Two Five-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b . Two Sis-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c . One Five-membered and One Six-membered Ring . . . . . . . . . . . . . . . . . . . . d . Other Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 . Bridged Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 . The Course of Irreversible Ring Closures . . . . . . . . . . . . . . . . . . . . . . . . . . V . Cyclic Acetals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. General Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . Monocyclic Acetals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a . Five-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b . Six-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 . Acetals Containing Fused Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a . Fused Five-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b . Fused Six-membered Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V I . Anhydro Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. The 1,4:3,6-Dianhydrohesitols. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 . 1 , 6-Anhydro Derivatives of Pyranose Sugars . . . . . . . . . . . . . . . . . . . . . 3 . The Scission of Epoxide Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~~~~~
~
~
~
*This review is based on literature available t o the author up t o July. 1954.
2 3 3 4 6 6 8 10 . 10 12 12 14 17 17 17 18 20 20 21 23 25 25 26 26 28 31 31
36 46 46 40 51
2
J. A . MILLS
I . INTRODUCTION The chief aim of this article is to make a comparison between carbohydrates and analogous alicyclic compounds, with emphasis on the stereochemistry of the ring systems involved. In alicyclic chemistry, notable additions have recently been made to our knowledge of the stereochemistry This new information of saturated rings, particularly six-membered already has had a considerable impact on carbohydrate chemistry, but has 6, The wider application attempted been restricted t o a few here will, it is hoped, assist in the coordination of considerable parts of the vohminous and diverse information available about carbohydrates, and may possibly lead to further development of stereochemical theory. I t is of interest to recall the development of the chemistry of terpenoids and steroids. Initially, these compounds were studied as problems in structural chemistry, but it was later realized14 that terpenoids of established structure provided first-class material for the study of the mechanisms of reactions, their availability in optically active forms being a notable advantage. Stereochemical theory has also greatly benefited in recent years from studies on steroids, largely because of the wider range of structural types available through the complexity of steroid molecules. Carbohydrates also possess the desirable features of optical activity, crystallinity, and availability, and display a diversity in structural types and reactions that cannot be matched in other fields. The writer, whose interests have lain mainly in the alicyclic field, has been greatly impressed by the scope that s t
(I) C. W. Beckett, K. S. Pitzer and R . Spitser, J . Am. Chem. Soc., 69,2488 (1947). (2) 0. Hassel and B. Ottar, dcta Chenr. Scand., 1, 929 (1947). (3) 0. Hassel, Research (London), 3, 504 (1950). (4) D. H. R. Barton, Experientia, 6, 316 (1950). (5) S. J. Angyal and J. A. Mills, Revs. Pure and A p p l . Chem. (Australia), 2, 185 (1952). (6) D. H. R. Barton, J. Cheni. Soc., 1027 (1953). (7) 0. Hassel, Quart. Revs. (London), 7, 221 (1953). (8) H. C. Brown, J. H. Brewster and H. Shechter, J . A m . Chern. Soc., 76, 467 (1954). (9) Further discussions are given by (a) R. B. Turner, in “Natural Products Related to Phenanthrene,” L. F. Fieser and Mary Fieser, eds., Reinhold Publishing Corp., New York, N. Y., 3rd Edition, 1949, p. 620; (b) V. Prelog, J . Chem. Soc., 420 (1950); (c) W. Klyne, in “Progress in Stereochemistry,” W. Klyne, ed., Academic Press Inc., New York, N. Y., Butterworths Publications Ltd., London, 1954, p. 36. (10) R . E. Reeves, J . A m . Chem. SOC.,72, 1499 (1950); and preceding papers in the series. (11) R. E. Reeves, Advances in Carbohydrate Chem., 6, 107 (1951). (12) R. J. Dimler, Advances in Carbohydrate Chem., 7, 37 (1952). (13) R. C. Cookson, Chemistry & Industry, 223 (1954). (14) See, for example, W. Huckel, Die Chemie, 66, 227 (1942).
CYCLIC DERIVATIVES OF CARBOHYDRATES
3
carbohydrate chemistry offers for elaborating and testing stereochemical hypotheses. Important contributions have, of cvursc, already come from this source, but usually as by-products of the development of carbohydrate chemistry, rather than through the deliberate choice of carbohydrate compounds as models for experimental study. To permit full advantage to be taken of the opportunities offered, a more detailed knowledge of the stereochemistry of cyclic carbohydrates is needed. The simple pyrariose and furanose sugars pose formidable problems of interactions between neighboring groups, but this difficulty largely disappears in their bicyclic and tricyclic derivatives. Accordingly, attentjion is confined mainly to compounds containing several rings, about which much information has been collected in recent reviews on cyclic acetalsI6 and anhydro I7 Research on these topics may seem to have lost some of its compounds,16~ initial impetus, but it will be seen that they still offer scope for further work of fundamental importance. Some of the conformational assignments are necessarily tentative a t present, and the opportunity is taken to point out directions in which more decisive evidence might be found. 11. NOMENCLATURE AND METHODS OF ILLUSTRATION 1 . Definitions
The convenient system of nomenclature for cyclic acetals introduced by Barker and Bourne15 will be used here. Rings formed by the engagement of hydroxyl groups attached to carbon atoms that are adjacent, separated by one atom, or separated by two atoms, are called a-,p-, or y-rings, respectively, and, if both hydroxyl groups are secondary, the designation C for two hydroxyl groups on t\he same side of the Fischer projection of the acyclic polyalcohol, or T for two groups on opposite sides, is added. Those parts of a polyol carbon chain not incorporated into an acetal ring, which appear as substituents in the ring, will be called residues. 2,4-O-Methylene-~-glucitolcontains a pC-ring, with the residues -CH20H and -CHOH-CH20H as substituents. The term fused rings wiH-bd used to denote the type of union found in decalin, where two rings have two atoms in common; the two atoms comprise a ring juncticn and are situated a t angular positions. Bridged rings are rings in which non-adjacent ring-atoms are joined by a bridge containing contains two one or more atoms. Methyl 3,6-anhydro-a-~-glucofuranoside (15) S.A. Barker and E. J. Bourne, Advances in Carbohydrate Chem., 7,137 (1952). (16) S.Peat, Advances in Carbohydrate Chem., 2, 37 (1946). (17) L.F.Wiggins, Advances in Carbohydrate Chem., 6, 191 (1950). (18) S. A. Barker and E. J. Bourne, J . Chem. SOC.,905 (1952).
4
J. A. MILLS
fused five-membered rings, whereas methyl 3 ,6-anhydro-a-n-glucopyranoside contains a bridged riug, and may be regarded as a six-membered ring bridged by two atoms or as a seven-membered ring bridged by one atom. 2. Structural Formulas
There is no settled convention for the representation of polyacetals and other complex cyclic derivatives of carbohydrate^.^^ To facilitate the comparison of polycyclic derivatives of carbohydrates with alicyclic compounds, the carbohydrate derivatives will be drawn in perspective, with single rings and fused rings placed in the plane of the paper, and with darkened or broken lines used to show the orientation of substituents and as is cusbridges above or below the plane of the rings, tomary for terpenes arid steroids. The convenience of the proposed usage is seen on comparing the Fischer projection (I) and the new representation (11) of 1,2:3,5-di-O-isopropylidene-a-D-glucofuranose; I1 shows clearly the nature of the ring-junctions and the relative positions of substituents, which are the factors controHing the stability and reactivity of the compound. The steroid type of formula is suitable for showing the configurations at new asymmetric centers in cyclic acetals, if these are known. Hydrogen atoms in the angular positions
I
CH,OH I
I1
111
a t ring-junctions must always be shown, but those a t points of substitution (as a t C5 in 11) may be omitted. The diagrams are more suitable for setting in type if drawn with bonds at angles of 45” and 90” only, but the true shape of the molecules is approached more closely by the representation 111. Fischer projections are still essential for showing genetic relations between carbohydrates, but are definitely unsuitable for precise stereochemical deductions; and reliance on them may have been responsible for some (19) W. W. Pigman and R. M. Goepp, J r . , “Chemistry of the Carbohydrates,” Academic Press Inc., New York, N. Y., 1948, p. 53. (20) J. A. Mills, Ph.D. Thesis, Cambridge, Engl., 1953. (21) J. A. Mills, Chemistry & Industry, 633 (1954).
CYCLIC DERIVATIVES OF CARBOHYDRATES
5
errors found in the literature. Formulas similar to I1 have already been used and for derivatives of D-glucofuranuronofor 1 4: 3 6-dianhydrohe~itols'~ r-lactone,22and the writer believes that they should be regularly used for depicting complex carbohydrate derivatives containing fused or bridged rings. Recourse to accurate models is necessary in studying intricate stereochemical problems, but with practice much information may be read from properly drawn diagrams of the type of 11. Because the plane of the paper is the plane of reference, the new formulas may be rotated in the plane of the paper without alteration of their significance, although this may not be true of all perspective formulas.23 Haworth perspective formulas24 are suitable for single pyranose and furanose rings and for some simple bridged structures, but when it is necessary to show conformations of six-membered rings, the non-planar "chair)' and "boat" forms will be drawn in perspective. A method has been devisedz5 for drawing chair rings with lines at angles of 45" and 90" only, suitable for setting in type; it is, however, not readily applicable to some bicyclic structures with cis ring-junctions. Two veiy useful rules for comparing Fischer projections with the new formulas may readily be deduced. (1) In a single ring, of any kind or size, resulting from closure a t two secondary hydroxyl groups on the same side. of a Fischer projection, the two residues will be in cis arrangement, whereas two hydroxyl groups on opposite sides of a Fischer projection will lead to a trans arrangement of residues. ( 2 ) When two adjacent carbon atoms of a Fischer projection become the ring-junction of a pair of fused rings, the ring-junction will be trans if the hydroxyl groups attached to the two carbon atoms are on the same side of the Fischer projection, but cis if the hydroxyl groups are on opposite sides. )
)
As an example, formula V will be transformed to a Fischer projection. Inspection
CH,OH
IV (22) (23) (24) (25)
1
V
K.-C. Tsou and A. M. Seligman, J. A m . Chem. Soc., 74, 5605 (1952). C . S . Hudson, Advances i n Carbohydrate Chent., 3, 1 (1948). H. D. K. Drew and W. N. Hnworth, .I. Chem. SOC.,2303 (1926). R . S. Cahn, J . Chem. Soc., 695, footnote (1952); 3520 (1951).
6
J. A . MILLS
of the steric arrangement a t C2 in V shows that i t is derived from a hydroxyl group on the left of a Fischer projection. Considering ring B only, it is seen t h a t the substituents C5 (in ring A) and C1 are trans; therefore the 2,4-ring is a OT-ring, and the hydroxyl group a t C4 is on the right of the Fischer projection. The ring-junction is cis, so the hydroxyl groups a t C3 and C4 are on opposite sides of the projection. By inspection, or by considering the substituents in ring A , the configuration a t C5 is (IV). Similar determined. Formula V represents 2,4:3,5-di-O-methylene-~-mannitol reasoning may be used t o derive steroid-type formulas from Fischer projections.
To assist comparisons between diagrams drawn according to different conventions, the ring-oxygen atom will, as far as is practicable, be placed at the top in furanose rings, and at the top right-hand corner in pyranose rings. The important class of bicyclic acetals represented by V will always be depicted with the orientation of oxygen atoms shown in V.
111. STERIC REQUIREMENTS OF REACTIONS I . Reversible and Irreversible Rbactions Carbohydrate chemistry provides many examples of reactions that could proceed by alternative pathways, to afford isomeric products that differ in configuration or structure. The effect of the configuration of the reactants on the course of such reactions may depend very largely on whether the reactions are reversible or irreversible. If all pathways are irreversible, the composition of the products depends on the relative speed along the pathways; it is kinetically controlled. These speeds are determined by entropy changes and by the stereochemistry of the traiisition states26concerned; the latter is often related to the stereochemistry of the reactants in a complex way, and has still to be elucidated for many reactions. On the other hand, if each pathway of reaction is reversible, and i f a true equilibrium is established, the composition of the products is quite independent of the mechanism of reaction, being determined solely by the relative thermodynamic stabilities of the constituents, even though the speed of the individual reartions may be very different. A study of such reversible reactions is therefore suitable for assessing the relative stabilities of isomeric compounds. Carbohydrate chemistry is especially rich in examples of ring structures formed by reversible reactions, such as g l y c ~ s i d e sand ~ ~ cyclic acetals,28 and provides a more convenient field for examining the factors governing the stability of rings than does alicyclic chemistry, in which ring closures are usually irreversible. Furthermore, the prominent reversible reactions (26)For reviews and references, see (a) E. D. Hughes, Quart. Revs. (London), 2, 107 (1948); (b) C. R. Ingold, “Structure and hfechanism in Organic Chemistry,” Cornell University Press, Ithsca, N . Y , 1953,pp. 40-49. (27)1’. A. Levene, A. I,. R:LJmond and It. T. Dillon, J . B i d . (‘hem ., 96, 699 (1932). (28) R. M. Hann and C.S. Hudson, J. Ant. C’hem. Soc., 66, 1909 (1944).
CYCLIC DERIVATIVES OF CARBOHYDRATES
7
of carbohydrates are nearly all catalyzed by acids, whereas the equilibria most studied in alicyclic chemistry, such as epimerizatioii of secondary hydroxyl groups4~ 2 9 , ao and of ring 29 are base-catalyzed. Epimerizations a t a positions in aldonic acids3' and 2 , 4 :3 ,&di-O-methylenehexaric acids5" are well established, but base-catalyzed epimerizatioii of secondary hydroxyl groups in the rings of g!ycosides and anhydro derivatives does not seem to have been investigated. Irreversible reactions of carbohydrates include the formation of glycosides and 1-deoxy-1-thioglycosidesfrom acylglycosyl halides33,34 and thioa c e t a l ~ , respectively, ~~-~~ esterifications by acyl and sulfonyl halides or anhydrides in ~ y r i d i n e ,40~displacements ~ of sulforiyloxy groups by various reagents,3Yand most examples of formation and scission of anhydro rings.l6V 17939 It is more difficult to define the influence of stereo effects on the course of these reactions; and some of the results obtained await explanation. I n using the results of reversible reactions in stereochemical analysis, a decision is often difficult as to whether a true equilibrium had in fact been established in experiments not specifically designed for quantitative studies. Recent syntheses of cyclic acetals are examples. When reducing sugars are condensed with acetone or acetaldehyde in the preseiice of strong acids, it is possible that decompositions caused by the catalyst will proceed more rapidly than the establishment of an equilibrium. This risk is less in reactions of alditols with benzaldehyde or formaldehyde, but again a true equilibrium may not result, because of the crystallization of insoluble acetals, or through interruption of the experiment before the rather difficultly hydrolyzable methylene acetals have come to equilibrium. The processing of reaction mixtures is often very incomplete. These uncertainties have been 1
(29) W. Huckel, Ahn., 633, 1 (1938). (30) W. von E. Doering and T. C. Aschner, J. Am. Chem. SOC.,71, 838 (1949), and references cited there. (31) E. Fischer, Ber., 23,799 (1890); W. N . Hnworth and C. W. Long, J . Chem. SOC., 345 (1929). (32) W. N. Haworth, W. G. M. Jones, M. Stncey and L. F. Wiggins, J . Chem. SOC., 61 (1944). (33) E. Pacsu, Advances in Carbohydrate Chem., 1, 77 (1945). (34) W. L. Evans, D. D. Reynolds and E. A. Talley, Advances in Carbohydrate Chem., 6, 27 (1951). (35) J. W. Green and E. Pacsu, J . Am. Cheni. SOC.,69, 1205, 2569 (1937). (36) E. Pacsu, J . Am. Chem. SOC.,61, 1671 (1939). (37) M. L. Wolfrom, D. I. Weisblat and A. R. Hanze, J . Am. Chem. Soc., 66, 2065 (1944). (38) A. L. Raymond, Advances in Carbohydrate Chem., 1, 129 (1945). (39) R. S. Tipson, Advances i n Carbohydrate Chem., 8, 107 (1953). (40) J. M. Sugihara, Advances in Carbohydrate Cheni.,8, 1 (1953).
8
J. A . MILLS
kept in mind when using data on cyclic acetals for the generalizations of Section V, and most weight has been assigned to the results of the experiments that seem to have more nearly approached the ideals of a true equilibrium and complete subsequent processing. 2. Reactions with Specific Steric Requirements
That the bimolecular displacement reaction (SN2) requires a collinear arrangement of three atomic centers is well known. Although few kinetic studies have been made, SN2reactions are probably fairly common in carbohydrate chemistry, especially intramolecular reactions leading to the establishment of anhydro rings. If the required steric arrangement is not attainable, the reaction may be entirely prohibited.16 The closure and scission of ethylene oxide rings is a special case of the intramolecular SN2reaction,16 and would seem to call for a tranx =-Cc,.- sition state with a specific arrangement (A) of four atomic ,. \o,.,, centers.l3. l 6 It is true that a trans arrangement of groups16 is necessary for the reaction, but this will not be sufficient A for ensuring reaction if molecular rigidity prevents the attainment of the transition state, as is shown by the following example. The reaction of 5,6-anhydro-l, 2-O-isopropylidene-cr-~-glucofuranose (VI) with bases is a fruitful source16r41 of D-glucose derivatives having a substituent a t C6. However, even when the opening of the anhydro ring of VI, by external attack of an anion, proceeds with difficulty, there is apparently little tendency for an intramolecular displacement by the free hydroxyl group a t C3; there is no mention4I of the formation of 3,6-a11hydro-1 ,2-O-isopropy~idene-cu-~-g~ucofuranose (VII), although this42 is a well-defined solid, and would have been readily recognized. On the other (VIII) reacts hand, 1,2-O-isopropylidene-5,6-di-O-tosyl-cr-D-glucofuranose
'
VI
VII
VIII
very readily with base, giving42the 5-0-tosyl derivative of VII. Scission of ethylene oxide rings through intramolecular attack by adjacent hydroxyl groups has often been observed,I6 and the failure of such a n attack to occur with VI must be ascribed to steric factors. Approach of C6 to the hydroxyl group a t C3 may occur with about equal ease for VI and VIII, but whereas (41) H. Ohle and K. Tessmar, Be?.., 71, 1843 (1938). (42) H. Ohle, L. von Vargha and H. Erlbach, Ber., 61, 1211 (1928).
CYCLIC DERIVATIVES O F CARBOHYDRATES
9
reaction of VIII involves only a simple displacement a t C6, reaction of VI will require a specific arrangement of C6, C5, and the attacking and displaced oxygen atoms, which apparently is unattainable. It is interesting to note that 1,4-anhydro-6-0-tosylsorbitol affords43I ,4:3,6-dianhydrosorbito1 on treatment with base; formation of the 5 ,6-anhydro compound may reasonably be postulated as an intermediate stage, and it seems that the greater flexibility of t,he single five-membered ring, compared with that of the bicyclic system of VI, permits the establishment of the transition state required for conversion of the 5,6-anhydro to the 3,6-anhydro compound. The bimolecular elimination reaction (E2) also requires a specific ar~ B, the groups X and Y are rangement (B) of four atomic ~ en ters.4In coplanar with the two carbon atoms and antiparallel (anti-trans, or “true
trans” relation). If the configuration of the reactant prevents the establishment of the transition state B, eliminat,ion by the E2 mechanism will not occur. It is doubtful whether this mode of elimination has been kinetically established for any carbohydrates, but for this, as for other stereospecific reactions, they provide a potentially useful testing ground. Some interesting results have recently been obtained45in studies on elimination reactions of esters of hydroxy acids. The mechanism is not fully established, but probably is of the bimolecular type. An especially interesting observation is that sodium iodide promotes the removal of two vicinal sulfonyloxy groups by a process of cis-elimination; a series of elimination reactions of this type is known in carbohydrate chemistry,39but apparently does not yet include an example from which the stereochemistry of the reaction could be deduced. The spontaneous decomposition of the diazonium ions obtained on treating non-aromatic primary amines wit,h nitrous acid in weakly acidic solutions (deamination reaction) is an irreversible reaction possessing great driving force under mild conditions. The steric consequences of the reaction in the aliphatic and alicyclic fields have proved to be quite diverse,4°-6n (43) (44) (45) (1953). (46)
V. G. Bashford and I,. F. Wiggins, J . Chem. Soe., 299 (1948). See ref. 26(b), p. 465. R. P. Linstead, L. N. Owen and R. F. Webb, J . Chem. SOC.,1211, 1218, 1225 Phyllis Brewster, F. Hiron, E. D. Hughes, C. K. Ingold and P. A. D. S. Rao,
10
J. A . MILLS
but amenable to classification. Its application to carbohydratesl6<4 3 , 62 has already afforded interesting results, including cyclizations and changes in the size of rings. Most of the results obtained so far can be interpreted with the aid of information already gained in other fields, but further use of this reaction in t,he carbohydrate field may well be interesting and profitable. 51e
IV. FUNDAMENTAL PROPERTIES OF RINGS The more important stereochemical features of small alicyclic rings will be mentioned in this Section, with a comparative summary of the properties of the corresponding het,erocyclic rings found in carbohydrates. The stereochemistry of selected carbohydrate derivatives will be discussed in detail in the following Sections, with some amplification of the simple theory outlined here. 1. Conformations
The full importance of the precise shape of molecules in determining stability and chemical reactivity has only recently been recognized. Calculations of the shape of various rings had previously been made, and pyranose and furanose rings had been examined t h e ~ r e t i c a l l yand ~ ~ by x-ray analys ~ s ,55~ but ~ , the significance of the calculated shapes, and methods for selecting the more probable shapes when alternatives were possible, could not be properly assessed until the importance of interactions between nonbonded atoms was realized. The classical theories of steric hindrance related largely to restrictions on rotation in rigid molecules containing double bonds or aromatic rings, and restrictions on rotation above single bonds caused by the collision of bulky groups, as in optically active biphenyls. It is now knownz6,6 6 that steric effects are equally important in cases where rotation about single bonds had previously been considered to be unrestricted. This was first proved Nature, 166, 179 (1950); but see also M. B. Watson and G. W. Youngson, J . Chem. Soc., 2145 (1954). (47) G. E. McCasland, J . Am. Chem. SOC.,73, 2293 (1951). (48) J. A. Mills, J . Chem. Soc., 260 (1953). (49) J. D. Roberts and R. H. Maeur, J . Am. Chem. Soc., 73, 2509 (1951). (50) R. J . W. Cremlyn, D. L. Garmaise and C. W. Shoppee, J . Chem. Soc., 1847 (1953). (51) V . G . Bashford and L. F. Wiggins, Xature, 166, 566 (1950). (52) A . B. Foster, E. F. Martlew and M. Stacey, Chemistry & Industry, 825 (1953). (53) H. S . Isbell, J . Research Natl. BUT.Standards, 18, 505 (1937). (54) C. A. Beevers and W. Cochran, Nature, 167, 872 (1946). (55) 0. L. Sponsler and W. H. Dore, Ann. Rev. Biochem., 6, 63 (1936). (56) J. C. McCoubrey and A. R. Ubbelohde, Quart. Revs. (London), 6,364 (1951).
CYCLIC DERIVATIVES OF CARBOHYDRATES
11
for simple aliphatic molecules; it has been demonstrated spec%roscopically, by electron diffraction measurements, and in other ways.66 For ethane, the most stable state is the staggered state, in which each C-H bond of one methyl group bisects the angle between two C-H bonds of the other group, when the molecule is viewed along the C-C axis; the least stable is the eclipsed state, in which C-H bonds of the two groups -’ -1e in pairs when viewed in this way. Rotation about the C-C bond in etriane requires a passage through successive eclipsed states, and there is a potential barrier opposing this rotation, amounting to about 3 kral./ mole.57The nature of the potential barrier is not fully understood, but seems56t o be due mainly to repulsions between the hydrogen atoms of the two methyl groups. Repulsion will be greatest when the molecule is in the eclipsed state, as this causes the closest approach of hydrogen atoms. When larger atoms or groups are substituted for hydrogen atoms in ethane, there may be several possible staggered states. I t is probable56 that the larger groups tend to take up posiLions as far apart as possible. For long polymethylcne chains a planar, zig-zag arrangement of the carbon “backbone” is particularly favorable. The term conformations is usually applied to those possible configurations of molecules that represent different stages in rotation about single bonds but do not involve significant departures from normal interatomic distances and valence angles. Individual conformations cannot be isolated and studied as pure species, except in the crystalline state. Kevertheless, it is unlikely that an assembly of molecules in the unexcited state will a t any instant contain more than a very small fraction of its membership in highly unfavorable confcpations, and the physical and chemical properties of the assembly will correspond largely to those to be expected of the favorable conformations. The aims of conformational analysis are to devise a set of reliable rules for assessing the relative stabilities of different conformations of molecules in which rotation or oscillation about single bonds is possible and to correlate conformations with chemical and physical properties. It must be emphasized that a molecule need not react in a particular conformation, even though this conformation is far more stabIe than others. If the steric requirements of a reaction demand an unfavorable conformation of the reactant, the reaction will usually still occur (perhaps very sluggishly), especially if the driving force of the reaction is great. For cyclopentane and larger saturated alicyclic rings, which do not require distortion of the carbon valence angles, several conformations are possible, but far fewer than for the acyclic analogs. The potential barriers between stable conformations in cyclohexane and the slightly larger rings (57) J. G. Aston, S. Isserow, G. J. Sanse and R. M. Kennedy, J. Chem. Phys., 12, 336 (1944).
12
J . A . MILLS
will be greater than in acyclic compounds, because change from one conformation to another requires distortion of valence angles as ~ las lpassage through eclipsed states. The factors to be considered'-s in assessing conformational stability in rings are the same as for acyclic compounds: the number of pairs of eclipsed bonds should be as small as possible, and the spatial separation of groups, especially bulky groups, should be as great as possible. Barker, Bourne and Whiffen58have shown that, if the preferred conformation of glycitols is that with the planar, zig-zag arrangement of carbon atoms characteristic of long-chain polymethyleiie compounds, it is 18 of cyclicpossible to explain the main features of the observed patternxb# acetal formation. Brown has drawn attentions to the existence of a potential barrier to the rotation of the methyl groups in dimethyl ether which is comparable in magnitude to the barrier observed for propane.b7He has suggested that this may be attributed to provision by the unshared pairs of electrons, on the oxygen atom, of a source of repulsive force similar to that of a C-H bond, and may have an important bearing on the conformations of heterocyclic rings containing oxygen atoms.8 Further examples of this effect are considered subsequently; it seems to be real, but less in magnitude than the repulsions due t o C-H bonds. 2. Monocyclic Compounds a. Five-membered Rings.-If cyclopentane were perfectly planar, the carbon valence angles would be distorted from 109.46' to log", and all C-H bonds would be in eclipsed positions. The ring is known to be puckered,69 to an extent that seems to diminish non-bonded interactions a t the expense of introducing a slight, valence-angle strain. It has been found that trans1,2-dimethylcyclopentane is rather more stable than the cis isomer, and cis-1 ,3-dimethylcyclopentane slightly more stable than the trans isomer; HaresnapeGohas shown that these orders of stability are to be expected from lion-bonded interactions and the nature of the puckering in the ring. The same order for the 1 ,Zdimethyl derivatives, but not necessarily for the 1,3-, would be predicted for planar cyclopentane rings. For many purposes, such as formation of cyclic derivatives of glycols,61-62 viciiial cis groups attached to the cyclopeiitane ring may be considered to (58) S. A. Barker, E. J. Bourne and D. H. WhifTen, J . Chem. SOC.,3865 (1952). (59) J. E. Kilpatrick, K. S. Pitzer and R. Spitzer, J . Am. Chem. SOC.,69, 2483 (1947); F. A. Miller and R. C. Inskeep, J . Chem. Phys., 18, 1519 (1950); seealso Ref. 26(b), pp. 52-56. (60) J . N . Haresnape, Chemistry & Industry, 1091 (1953). (61) V. Prelog, K. Schenker and H. H. Ciinthard, Helv. Chim. Acta, 36, 1598 (1952). (62) B. Englund, J . prakt. Chem., 122, 121 (1929).
CYCLIC DERIVATIVES OF CARBOHYDRATES
13
be eclipsed, and vicinal trans groups to subtend a projected angle63of 120". An angle of 180" seems to be readily attained with vicinal trans groups, as elimination reactions and epoxide formation proceed normally; only a slight distortion beyond that of the normal puckering is required. The precise shape of tetrahydrofuran is less well known, but the ring is probably puckered, although to a lesser degree than that of cyclopentane. It has fewer eclipsed C-H bonds than has cyclopentane, arid the magnitude of the repulsive forces tending to distort the ring from a plane will depend on the unknown value of the effects contributed by the unshared electrons on the oxygen atom. An x-ray investigation of sucrose54showed a furanose ring markedly distorted from a plane, but part of this distortion may be due t o forces in the crystal lattice and part to the bulky nature of the substituents. For most purposes, the tetrahydrofumn ring may be regarded a8 a plane, fairly readily distortable to meet, the needs of reactiorp: A trans arrangement of vicinal substituents should be more stable than a cis arrangement, but for non-vicinal substituents the order of stability is not readily predictable. The 1 ,S-dioxolane ring (IX), important in cyclic acetals, is probably nearly planar; some small changes in valence angles or interatomic distances are needed on closing the ring.58There are only two pairs of eclipsed C-H bonds, but here, also, the contributions from the unshared electrons on oxygen to the repulsions tending to force the ring out of a plane cannot be accurately assessed. The occurrence of cis isomers, only, in fusions of this ring to five- or six-membered rings is evidence for a preferred planar conformation of the I ,3- IX dioxolane ring. I n cyclic acetals, IX may be substituted at C2, C4, and C5. A trans arrangement of substituents a t C4 and C5 will be more stable than a cis arrangement, as is clearly shown by the quantitative m e a s ~ r e r n e n t son ~~ the cyclic 0-isopropylidene derivatives of erythro- and threo-1 ,2-diphenyl1,2-ethaaediol, and the qualitative evidence given in Section V. The order of stability for isomeric derivatives with substituents at C2 and C4 (or C5) cannot be predicted by the method used60 for the 1,3-dimethylcyclopentanes, since that takes into account the puckering of the cyclopentane ring. Unsaturation in a five-membered ring alters the position slightly. A double bond exocyclic to a cyclopentane ring, as in cyclopentanone, represents a highly stable arrangement, whereas a similar arrangement in a cyclohexane ring introduces an element of instability.s If this is also true of heterocyclic rings containing oxygen atoms, it explainsS why aldonic acids
c-
(63) Defined in ref. 11. (64) P. H. Hermans, Z . p h y s i k . Chem., 113, 337 (1924).
14
J. A . MILLS
tend to form 1,4- rather than l15-lactones. A carbon atom of a cyclopentane ring passes more readily t80the trigoiial state (formation of a carboriium ion) than an atom in a cyclohexane ring, and if this also holds for t)he heterocyclic rings of sugars, it may8 explain why furanosides are formed arid hydrolyzed more rapidly than pyranosides. b. Six-membered Rings.-The stereochemistry of the cyclohexane ring is known in great detail.‘-8*65 Of the possible conformations, the chair form (X)66 is invariably encountered as the predominant species, unless the ring
X
XI
is constrained, by bridging, to assume the boat form (XI).66 Valence-angle strain is absent in both forms, but, whereas in the chair the adjacent C-H bonds are all in fully staggered arrangement, the boat has four sets of eclipsed bonds, as well as a very close approach of two atoms or groups (A and Q1 in XI) a t the ends of the boat. Bonds to substituents in the chair ring are geometrically of two sorts. Six bonds, three on each side, are parallel and normal to the mean plane of the ring, and are now called6’ “axial” ( a ) bonds (formerly termed E or “polar”). The other six, “equatorial” (e) bonds (formerly designated K ) , are directed outward, making angles of 60” and 120” alternately with the vertical, axial bonds. Each equatorial bond is parallel to two sides of the ring. For the majority of substituted cyclohexanes, two alternative chair conformations are possible: all equatorial bonds in the first become axial bonds in the second, and vice versa. It has beeii found that substituents tend to occupy equatorial rather than axial positions, and that, if there are several substituents, the greater number of substituents, or the larger substituents, will be equatorial. The chief reason for the greater stability of conformations with equatorial substituents is the repulsion between an axial substituent and the axial hydrogen atoms or other groups on the same side of the ring. A conformation with two axial groups larger than hydrogen atoms on the (65) D. H. R. Barton, Chemistry & Industry, 664 (1953). (66) Hydrogen atoms are omitted. The diagrams are true for parallelism of bonds, but not necessarily for interatomic distances. The positions for substituent groups were chosen purely for convenience of exposition. (67) D. H. R. Barton, 0. Hassel, K. S. Pitzer and V. Prelog, Nature, 172, 1096 (1953).
CYCLIC DERIVATIVES O F CARBOHYDRATES
15
same side of the ring (such as B and Qz in X) is very unfavorable, as the van der Waals barriers of the groups will overlap. Substituents in the cis-l,3 relation will therefore be equatorial if possible. Two groups in le, 2a relation (cis) are exactly the same distance apart in the chair ring as the same two groups in le, 2e relation (trans): in X, group P is equidistant from the two groups Q', Q2. The IP, 212 arrangement is, however, the more stable. If the requirements of a reaction force two adjacent bonds t o coplanarity, it is easier for this to occur with le, 2a bonds (cis) than with l e , 2e bonds (trans), because the latter deformation tends to crowd together the axial hydrogen atoms.2*68 Vicinal, cis-glycol groups in six-membered rings afford cyclic acetals, whereas the trans isomers normally do not. An alternative conformation for vicinal trans groups is la, 2a (groups A and B in X), which is unstable relative to the le, 2e arrangement, but has the groups antiparallel, and is therefore a favorable conformatioil for reactions requiring this specific geometry. It may be noted that a complete conversion of the ring is not needed in order to change conformation le, 2e to l a , 2a: conversion to the appropriate boat form is sufficient (compare X and X I for groups P, Ql), and this may happen in relatively rigid structures for which complete conversion of the ring is impossible. Even when cpmplete interconversion of chair conformations occurs, a boat form is probably5 an intermediate stage. Because of the differing degrees of hindrance by neighboring atoms or groups, groups in equatorial positions are more accessible than those in axial positions, and this accounts for the observed order of reactivity of axial and equatorial substituents in such reactions as esterification, hydrolysis, and oxidation of alcohols to ketone^.^^ Barton has shown6 66 that an approximate calculation of the relative accessibility at various positions in the steroid nucleus is possible, and the calculations agree with experimental values. Saturated heterocyclic rings containing oxygen, nitrogen, or sulfur atoms also exist preferentially in stable, chair conformations resembling that of cyclohexane. Physical measurements, such as those of electron diffraction and dipole moments, have shown this conformation for p - d i ~ x a n e , ~ ~71. s - t r i ~ x a n e ,71~ ~~*a r a l d e h y d e ,73~ ~~ - t r i t h i a n e ,and ~ ~ 2,4,6-trimethyl-s-tri9
(68) S. J. Angyal and C. G. Macdonald, J . Chem. SOC.,686 (1952). (69) 0. Hassel and H. Viervoll, Acta Chem. Scand., 1, 149 (1947). (70) K.E. Calderbank and R . J. W. Le FBvre, J . Chem. SOC.,199 (1949). (71) M.Kimura and K. Aoki, J . Chem. SOC.Japan, Pure Chem. Sect., 72, 169 (1951);Chem. Abstracts, 46,3341 (1952). (72) R. J. W. Le FBvre, Joan W. Mulley and B. M. Smythe, J . Chem. Soc., 290
(1950). (73) E. Lippert and R. Mecke, Z , Elektrochem., 66, 366 (1951).
16
J. A . MILLS
thiane.69A chair ring is consistent with the chemical properties of 2,6disubstituted p - d i o x a n e ~ The . ~ ~ stereochemistry of alkaloids and related compounds has been extensively d i s c u ~ s e d , ~and ~ - ~the ~ experimental evidence is indicative of a stable chair form for the piperidine ring. In view of the evidence adduced for these diverse ring systems, it is reasonable to assume that a chair form is preferred for the m-dioxane ring (XII) of cyclic acetals, and this agrees with a calculations0 from infrared absorption frequencies. The pyranose ring is always encountered in a chair form unless constrained to a boat form by bridges.2*6 * l1 This should hold for tetrahydropyran rings in general. X-ray crysXII tallographic data54s 5 5 on the pyranose ring have been confirmed by the evidence from formation of complexes with cuprammonium ions.l0.l1 Hassel and Ottar2 had recognized the probability that the pyranose ring would be conformationally analogous to cyclohexane, and they showed that this assumption explained several features of carbohydrate chemistry. Reeves extended the theoretical interpretation of pyranose ring conformations, and has listed101 11 “instability factors” to be used in assessing the relative stability of alternative chair conformations for pyranose rings. His treatment of the subject differs somewhat from that given to derivatives of cyclohexane by other workers. Following Hassel and Ottar? he regarded as unfavorable the simultaneous occurrence of the terminal primary alcoholic group and a hydroxyl group (at C1 or C3) in axial positions on the same side of the ring, yet he did not assign an unfavorable weighting to the presence of two axial hydroxyl groups (at C2 and C4, or Cl and C3) on the same side of a ring. The repulsions between two hydroxyl groups should be quite strong, although not as great as that between -OH and -CH20H, if size of group is the most important factor involved. Information about the conformation of methyl a-D-talopyranoside (or, better, 1 ,5-anhydro-~talitol), in which the two effects are in opposition, might settle this point. Reeves assigned a heavy unfavorable weighting to the close approach of three oxygen atoms (&‘A2effect”)‘O, l1 encountered in methyl 0-D-mannopyranoside. His weighting may be excessive, when it is considered that no
(0
(74) R. I<. Summerbell and J. R . Stephens, J. Ant. Chem. Soc., 76, 731 (1954). (75) R. C. Cookson, Chemistry & Industry, 337 (1953). (76) Asima Chatterjee, A. K. Bose and S. Pakrashi, Chemistry & Industry, 491 (1954). (77) J. McKenna, Chemistry & Industry, 406 (1954). (78) K. Alder and H. A. Dortmann, Chem. Ber., 86, 1544 (1953). (79) I. G . M. Campbell, Ann. Repts. on Progr. Chem. (Chem. SOC.London), 60, 164 (1953). (80)H. Voetter and F. Wollrab, Acta Phys. Austriaca, 6, 529 (1952):
CYCLIC DERIVATIVES OF CARBOHYDRATES
17
marked instability seems to be associated with comparably close approaches of tjhree or even four oxygen atoms in other compounds, such as ort,hoesters and the 2,4:3,5-diacetals of iditol (p. 39). It must be emphasized that the exact shape of these heterocyclic chairform rings is not known. The different interatomic distances (C-C, 1.54; C-0, 1.43; C-S, 1.8 K.) in the rings, and uncertainty about valence angles, may cause some departure from the nice parallelism of bonds found in cyclohexane, but probably not enough to render invalid arguments developed for cyclohexane. It may, howeve?, be more difficult to draw fine distinctions between conformations. Replacement of carbon atoms by oxygen decreases the possibilities for eclipsed C-H bonds (to zero for s-trioxane), and the persistence of the chair ring as the favored conformation is good evidence for the postulated conformational effect8of unshared pairs of electrons. Replacement of carbon by oxygen also reduces the possibilities for interactions between substituents and axial hydrogen atoms; this is discussed in detail in the Section on cyclic acetals (p. 29). c. Other rings.-Three- and four-membered rings are so strained that they may be considered to be planar for all purposes. The largest ring of importance in carbohydrate chemistry is seven-membered, and little is known of the fine structure of the alicyclic analog, cycloheptane; most probably it resembles cyclohexane in general shape, but is more flexible and therefore imposes less restraint on reactions requiring a specific alignment of groups. 3. Structures Containing Fused Rings a. Two Five-membered Rings.-The cis (XIII) and trans forms of bicyclo[S.S.O]octane are known, but combustion data8' show that the trans isomer is the less stable. If the ring junction is labilized by an adjacent carbonyl group, derivatives of the cis form only are obtainable.82Evidently the trans ring junction requires a distortion of the rings beyond that normally attainable. Any factors tending to increase the planarity of the rings in a pair of fused five-membered rings should further increase the stability of the cis isomer relative to the trans. In agreement with this, systems of two fpgmembered rings in which tetrahydrofuran or 1,S-dioxolane is one of the components have so far been encountered only as cis isomers. This should not be taken as proof that the trans forms are incapable of existence; it might be possible to prepare, for example, 1 , 4 : 3,G-dianhydrodulcitol, if a suitable method were chosen. All molecules (in this class) which have cis ring junctions may be re(81) J. W . Barrett and R. P. Linstead, J . Chem. Soc., 611 (1936). (82) A. H. Cook and R. P. Linstead, J . Chem. Soc., 946 (1934).
IS
J. A. MILLS
garded as being composed of t,wo planar rings meeting a t an angle of about 120", in a broad V-shape. This causes an important type of isomerism: A
H
H
0 H
H
B
XIV
XI11
substituents a t other than angular positions may be either inside the T' (endo) or outside (em). An ezo substituent is necessarily far apart from any substituent in the other ring, but an endo substituent may be fairly close to an endo substituent in the other ring. The closest approach is found with two endo substituents located at the points X and Y in XIV, but the endo substituents A and B are also fairly close to each other. (A model of XIV requires that the V be below the plane of the paper, rather than above the plane, in a projection onto paper.) The lesser interactions with other atoms or groups should render compounds with e m substituents relatively more stable than their endo isomers. Furthermore, the faces of the rings inside the V are less accessible than the faces outside, so ex0 and endo isomers should exhibit differences in reactivity in a direction determined by the steric requirements of the reaction in question. 6. TWOSix-membered Rings.-The isomeric decalins had been studied for many yearszgbefore final details of their stereochemistry were elucidated. I t is now known4' 6 9 f 8* that the component rings of both trans-decalin (XV) and cis-decalin (XVI) exist preferentially in the chair form, as in
XV
XVI
cyclohexane itself. trans-Decalin is rigid and essentially flat, with the hydrogen atoms at the angular positions held in axial relationship to both rings. cis-Decalin is L-shaped and flexible; each angular hydrogen atom is axial to one ring and equatorial to the other, and, by a transformation that converts every axial bond to an equatorial bond and vice versa, there is (83) 0. Bastiansen and 0. Hassel, Nature, 167, 765 (1946). (84) R.B.Turner, J . A m . Ghem. SOC.,74, 2118 (1952).
19
CYCLIC DERIVATIVES OF CARBOHYDRATES
derivablea3an alternative stable conformation which is the mirror image of XVI but is ,otherwise stereochemically equivalent to it. In substituted trans-decalins, one of a pair of epimeric substitution products is necessarily axial and the other equatorial. In cis-decalins, the possibility of inversion of the ring conformation permits an equatorial arrangement for each of a pair of epimers.48, 6 0 , a5 If two or more substituents are present in cis-decalin, there will be a t least one axial group in some of the stereoisomers, and it then is necessary, when selecting the most stable conformation, to avoid conformations that show the effect seen in XVII, where an axial group is inside the L-shape and very close to three axial hydrogen atoms. That trans-decalin is more stable than cis-decalin has been shown in various ways,2g a7 the reason for this beings4repulsions between the four hydrogen atoms inside the L-shape (shown in XVI). In trans-decalin, the axial H-H distances are the same as in cyclohexane. Fusions of two m-dioxane rings are found in the 1 , 3 :2,4- and 2,4:3,5diacetals of glycitols, and fusions of a m-diosane and a pyranose ring in the 4,6-acetals of glycopyranosides. If the constituent rings retain the chair forms, the bicyclic systems with trans junctions will be rigid and completely analogous to trans-decalin. However, an important new feature arises in those with cis junctions. Formulas XVIII and X I X show two possible I
XVII
XVIII
x IX
conformations of 1,3:2,4-di-O-methylene-~-threitol,the simplest of the acetals containing two m-dioxane rings with a cis junction. They are interconvertible by the operation described for cis-decalin, but are not stereochemically equivalent: XIX, which will be referred to as “H-inside,” has a close approach of four axial hydrogen atoms, as for cis-decalin; whereas in XVIII, “0-inside,” axial hydrogen atoms do not interfere, but there is a fairly close approach of four oxygen atoms. “H-Inside” and “0-inside” conformations are also possible for fusions of a m-dioxane ring and a pyranose ring with a cis junction, but only three oxygen atoms are then involved. The available evidence strongly indicates that “0-inside” is t,he (85) I<. I,. Elirl, E.ipericnfLn,9, 91 (1953). (S6) C;. F. Davies and 15. C. Gilbert, J . *tm. (‘hem. Soc., 63, 1585 (1941). (87) N. D. Zelinsky and M . B. Turowa-Pollak, Ber., 66,1299 (1932).
20
J. A. MILLS
more stable arrangement. This might have been expected: the unshared electrons on the oxygen atoms may have an effects in promoting chair conformations for the rings, but their close approach is not likely to be as unfavorable as the close approach of the van der Waals barriers of four hydrogen atoms. c. One Five-membered and One Six-rnembered Ring.-For the hydrindans, the heats of combustion29 and the epimerizations effected by aluminum bromide**point to the trans form (XX) as the more stable of the hydrocarbons, but epimerization of 1-hydrindanone by base89 shows that the cis form is the more stable. With cyclic acetals containing an analogous fusion of rings, only cis isomers are stable: this is well known for cyclohexanol, the cyclitols, and H pyranose rings, which never afford acetals engaging trans glycol groupings unless6*the six-membered ring is considerably sx distorted by other fused rings. These results are attributable to the differing shapes of the five-membered ring component. I n cyclopentane, the C-H bond interactions force the ring out of a plane; therefore, the distortions required for the formation of trans-hydrindan tend to stabilize the cyclopentane component, although they destabilize the cyclohesane component by an unfavorable crowding of axial hydrogen 68 On balance, these opposed effects seem to confer greater stability on the trans isomer than on the cis, in which iiiteractions from eclipsed C-H bonds are important. However, if the five-membered component becomes more nearly planar, as in ryclopentanone, tetrahydrofuran, or I ,3-dioxolane, the greater crowding of axial hydrogen atoms in the six-membered ring (required for a 2rans junction) becomes the dominant factor, and the cis junction is the more stable arrangement. d. Other Rings.-A cis junction, only, is possible when three-membered or four-membered rings are fused to other small or medium-sized rings. and 1,3-anhydro-2,43,5-Anhydro-l,2-O-isopropylidene-~-xylofuranose~~ of fusion of a four-membered @ m e t hyl e ne - ~ ~ -x y lito are l~ representatives ~ ring to a five-membered ring and to a six-membered ring, respectively, and other combinations of four-membered rings undoubtedly could be prepared. Anhydro sugars containing an ethylene oxide ring fused to a pyranose ring are an important class. Their conformations are probably the same as
03
(88) N . D. Zelinskg and M. B Turowa-Pollak, Rer , 62, 1658 (1929). (89) W. Huckel, Margarethe Sachs, J. Yantschulewitscli and F. Nerdel, Ann., 618, 155 (1935). (90) 1’ A. 1,evene and A I,. R a y m n n d , J . Bzol. (‘hen, , 102, 331 (1933). (91) R. M IInnn, N Ii Richtmycr, H. W. Dkhl and C. S. Hudson, d A m . Chem. Soc., 72, 561 (1950).
CYCLIC DERIVATIVES OF CARBOHYDRATE8
21
those of cyclohexene oxide and cyclohexene, which contain a chair ring somewhat modifiedg2by the requirement that four atoms in the sis-membered ring be coplanar. Both isomers of 1,2-~ycloheptanediolaffordg3 cyclic 0-isopropylidene derivatives; therefore, both cis and trans forms should be stable in any fusions of seven-membered rings to five-membered or larger rings.
4. Brzdged Rings Bridging of six-membered rings, only, will be considered, as other types do not a t present seem to be important. Alicyclic compounds related to OicycZo[d.d.llheptane (XXI) arc very numerous, but no carbohydrate analogs of proved structure seem to be known. They will of necessity be confined to 1,4-aiihydroglycopyranoses (1,5-anhydroglycofurarioses)and 2,5-anhydroglycofuranosides.Some rep95 but the structures resentatives of the former may have been prepared,34* await confirmation. I n X X I and its analogs, the six-membered ring is necessarily in a boat form, with eclipsed bonds arid also some valence-angle strain. The 1,5-anhydroglycofuranoses are, therefore, likely to be fairly reactive. Dimler has suggestedI2 that they could profitably be examined, in vieu. of the peculiar properties of 1,G-anhydroglycofuranoses, but the latter probably have an entirely different conformation (see below). BicycZo[2.2.d]octane(XXII) also is a rigid structure containing a boat
XXI
XXII
ring, but differs from XXI in being free of valence-angle strain, The numerous eclipsed bonds nevertheless make it rather unstable (see below). Methyl 2,6-anhydro-~~-~-altropyranoside~~ is the sole carbohydrate representative; it is hydrolyzed by acids much more readily than are normal pyranosides. BicycZo[3.2.l]octane(XXIII) has not been studied in detail, but the prop7 8 , 7 9 are entirely consistent with the conerties of the tropane alkaloids75* formation XXIV for nortropane. The 1 ,6-anhydroglycopyranoses,such as (92) D . H. R. Barton, R. C. Cookson, W. Klyne arid C. W. Shoppee, Chemistry & Indiistry, 21 (1954). (93) J. Boeseken and H. G. Derx, Rec. trav. chim., 40, 529 (1921). (94) K. Freudenberg and E. Braun, Ann., 460, 288 (1928); Ber., 66, 780 (1933). (95) P. A. Levene and E. T. Stiller, J. Biol. Chewi., 102, 187 (1933). (96) D. A. Rosenfeld, N. K. Richtmyer and C. S. Hudson, J . A m . Chem. Soc., 70, 2201 (1948).
22
J. A. MILLS
I
SSIII
OH
0 I3
XSIV
OH XXVI
XSV
GH Ho
2
OH
H' XXVII
1,6-anhydro-P-~-glucopyranose (XXV), are knownlO* I1 to contain the chair ring, and presence of a chair ring (XXVI) is consistent'O with some propIt is therefore reasonable erties of methyl 3,6-anhydro-a-~-glucopyranoside. to expect that X XIII and other heterocyclic analogs of i t have chair rings. This will include the 1,6-anhydroglycofuranoses, such as 1 ,6-anhydro-P-~glucofuranose (XXVII), and two other carbohydrate groups, the 2,6anhydroglycofuranosides and 2,5-aiihydroglycoseptanosides, of which no representatives are yet known. Bridging a chair ring in this way causes some distortion of valence angles but, in compensation, non-bonded interactions are small. Evidence t,hat reduction in non-bonded interactions is more important than the introduction of some valence-angle strain is found in the rearrangement of bicyclo[S.S.O]octune (XIII) to X X I I I on treatment with aluminum chloride,*l and in rearrangementsg7~ of derivatives of bicgcZo[2.d.b]octune to derivatives of XX I I I during displacement reactions. A difference between alicyclic and carbohydrate chemistry is found in the rearrangement of methyl 3,6anhydroglycopyranosides of suitable configuration to methyl 3,6-anhydroglycofuranosides under non-hydrolytic conditions of acid catalysid6; the change in ring structure is in the opposite sense to the change of XI11 to XX I I I . The smaller interaction between eclipsed bonds in the tetrahydrofuran rings is the main factor favoring the system of two fused five-membered rings in the carbohydrate derivatives, but the difference in bond lengths in the two classes of compound cannot entirely be neglected. The action of bases on 6-0-tosyl derivatives of methyl a-D-mannopyrano(97) W. von E. Doering and M. Farber, J . Am. Chem. Soc., 71, 1514 (1949). (98) W. C . Wildman and D. R. Saunders, J. Am. Chem. Soe., 76, 946 (1954).
CYCLIC DERIVATIVES OF CARBOHYDRATES
23
affords in each case the 3 ,&anhydro sideg9and of 1,5-anhydro-~-mannitol~~~ derivative rather than the 2,6-anhydro derivative. This selectivity is undoubtedly a conformational effect, but the reactions are irreversible, and probably are not to be taken as conclusive evidence for the relative stability of the two types of anhydro compound. A chair ring may be bridged unsymmetrically by three atoms, and one conformation of the structure has two interlocking chair rings (XXVIII); it is free of valence-angle strain, but there is a close approach of the two hydrogen atoms shown. Some heterocyclic compounds containing this fused-ring system are kno~vn,7~*lot and it is possible that cis-l,3-glycol groupings in chair rings may, under some circumstances, give acetals with this ring structure.lol* Cuprammoiiium complexes of certain cis-l,3-glycols seem to have been obXXVIII tained." 78v
5. The Course of Irreversible Ring Closures Many examples are known of ring closures that afford a five-membered ring when a six-membered ring could be formed and would be the more stable structure. The action of mercuric chloride on thioacetals in neutral 16, ~ 38 ~and 1 ,4-anhydro-hexitols result solution affords g ly co fu ran ~ sid es,~ Other examples from the deamination of I-amino-l-de~xy-hexitols.~~~ may be found in the reviews by PeatI6 and Wiggins.I7It has also been found that, whereas heating ~-mannitolwith hydrochloric acid for a short time affords 1,4-anhydro-~-mannitol,longer heating affords 1 ,5-anhydro-~mannitol.102-*03 Furthermore, PeatI6 has pointed out that formation of a11 ethylene oxide ring seems to take precedence over that of a tetrahydrofuran ring, but his example may not be strictly comparable to the above, as it concerned anhydride formation from 1 ,2-O-isopropylidene-6-O-tosyl-~g l u c o f u r a n ~ s eNevertheless, .~~~ it might plausibly be assumed that, in some at least of the other examples quoted above, the formation of the fivemembered ring was preceded by formation of an epoxide that subsequently (99) W.T.Haskins, R. M. Hann and C. S. Hudson, J . Am. Chem. Soc., 68, 628 (1946). (100) R . C. Hockett and Elizabeth L. Sheffield, J. A m . Chem. Soc., 68, 937 (1946). (101) G.A. Haggis and L . N . Owen, J . Chem. SOC.,399 (1953). [S.B. Baker, Can. J . Chem., (10la) 1,5:3,6-Dianhydro-2,4-0-methylene-~-glucitol 32, 628 (1954))is an example. (102) A. B. Foster and W. G. Overend, J . Cherri. SOC.,680 (1951). (103) H. G.Fletcher, Jr., and H. W. Diehl, J. Am. Chem. Soc., 74, 3175 (1952). (104) H. Ohle and L. von Vargha, Ber., 62, 2435 (1929).
24
J. A . MILLS
suffered intramolecular attack by the hydroxyl group found to be engaged in the product. The above results are understandable when the influence of changes in entropy on rates of reactionlos is taken into account. If the preferred conformation of a glycitol is that8with the planar, zig-zag arrangement of the carbon backbone,58the probability of finding the chain in the configuration favorable for ring closure becomes progressively smaller as the size of the ring is increased; work is required in order to rotate the chain to the favorable configuration, and appears in the rate equation as a large entropy of activation. It has been found106 that the rate of formation of ethylene oxide is over six times that of tetrahydrofuran when the appropriate chlorohydrins are treated with a base, yet the relative heats of activation are strongly in favor of tetrahydrofuran. The controlling factor is the very much greater change in entropy required in the formation of tetrahydrofuran, which quite overshadows the effect of the heat of activation and leads to a slower rate of formation for the less-strained ring. If entropy differences, as well as the inherent strain in the rings, are taken into account, it is reasonable to assume that the ease of formation of anhydro rings will decrease in the order three- > five- > six-membered rings. The position of four-membered rings in the series is difficult to assess; in practice, they are rarely found. If an ethylene oxide ring is formed as an intermediate stage, its enlargement to a four-membered ring through intramolecular displacement mould appear to be highly improbable. The relative ease of scission of anhydro rings, three- >> four- > five- > six-membered, is more nearly related to the degree of strain of the rings. The carbohydrate field is unique in the opportutiities it offers for the study of factors controlling the formation of rings. For example, on treatment of a 1-0-tosylhexitol with base it is possible to observe, in competitive reactions, the relative rates of formation of anhydrides with from three to six atoms in the ring, all resulting from closure a t a secondary hydroxyl group. Derivatives of glycitols up to a t least the octitol stage are accessible,Ia7and the configuration of the carbon chain and the position of the reactive group may be varied. Techniques such as filter-paper chromat ~ g r a p h y or ’ ~ ionophoresis ~~ now make possible the detailed examination of the mixtures of products likely to be formed. Digressing, we may note an interesting application of the postulate58 of a planar, zig-zag arrangement of the carbon chain in polyhydric alcohols. I n such chains, the 0-0 distances have been calculated as 2.51 A. for pC(105) A. E. Remick, “Electronic Interpretations of Organic Chemistry,” John Wiley and Sons, Inc., New York, N. Y., 1943, pp. 196,206. (106) H. W. Heine and W. Siegfried, J . Am. Chem. Soc., 76, 489 (1954). (107) C. S. Hudson, Advances in Carbohydrate Chem., 1, 1 (1945). (107a) G. N. Kowkabany, Advances in Carbohydrate Cham., 9, 303 (1954).
CYCLIC DERIVATIVES OF CARBOHYDRATES
25
groups and 3.43 A. for /?T-g ro ~ p s.~ Now, * in the synthesis of higher aldoses by the cyanohydrin method, an interesting regularity has beeii notedlo*;the predominant product almost, always is the aldonic acid in which the hydroxyl group a t the new asymmetric cetiter is in the pT-relationship to the hydroxyl group that had been located a t C3 in the original aldose, aiid this is not affected by the steric arrangement a t C2 in the aldose. This means that the preferred epimer is the one with the more favorable 0-0 separation, if the aldonic acids also have the planar, zig-zag arrangement). Some examples are known where the formation of alternative, anhydroring structures is possible, and the deformation of the molecule required in the alternative reactions is approximately the same, so that entropy changes may be largely neglected. In such cases the ethylene oxide ring is not favored. The action of bases on methyl 3,4-di-O-acety1-2,6-di-O-tosyl-p-Dglucopyranoside10g and on methyl 2,6-di-O-mesyl-a-~-galactopyranoside~~~ affords predominantly the 3,6-anhydro-2-0-sulfonyl derivative in each case. The above arguments do not apply to the formation of cyclic acetals. In the more stable classes of six-membered acetal rings, the shape of the carbon chain is not greatly distorted from that of the free glycitols,68aiid it is not necessary to postulate that formation of a five-membered acetal ring is intermediate to the formation of six-membered acetals. V. CYCLICACETALS 1. General Remarks
The properties of ring systems discussed above may be used to answer two questions raised by the use of cyclic-acetal groupings for the selective masking of hydroxyl groups in carbohydrates. What is the most, probable ring structure of the acetals obtainable from a particular polyhydroxy compound ? For a given acetal, how many diastereoisomers are likely to be stable ? The first question has been largely answered for acetalation of alditols by the development of empirical rules by H a m , Ness and Hudson,2** and by the extension of these rules by Barker and Bourne.16' However, the empirical approach becomes very cumbersome when attempts are made to cover all the complexities of multi-ring compounds, and its success is entirely dependent on the range and quality of published work available. (108) See ref. 107, p. 26, for a summary of the evidence, and also C. S.Hudson, J . Am. Chenz. SOC.,73,4498 (1951), for variations in the ratio of epimers under certain conditions. (109) E. Hardegger, R. M. Montavon and 0. Jucker, Helu. Chim. Acta, 31, 2247 (1948). (110) A . B. Foster, W. G. Overend, M. Stacey and 1,. F. Wiggins, J. Chem. SOC., 2542 (1949). (111) A. T. Ness, R. M. Hann and C. S.Hudson, J. Am. Chern. Soc., 70, 765 (1948).
26
9. A . MILLS
There is now enough evideuce available to permit an analysis of' the coilformational factors concerned in the formation of cyclic acetals from all classes of polyhydrosy compounds likely to be encountered, and to allow the recognition of structural types for which more evidence is needed. I t will be found that the empirical rules are all sound in principle and may be used to assess quickly the types of acetals likely to be encountered in a given situation, but that a study of conformations is necessary where alternative structures seem to be possible from the empirical rules.21 Barker, Bourne and WhiffenS8concluded that the empirical rules were soundly based by examining the consequences of a planar, zig-zag conformation of the carbon chain in glycitols. The writer feels that a study of the end-products of reaction is the safer approach to the problem of acetal formation, and of other reversible reactions, because deductions based on the conformations of the reactants will only be sound if these conformations, and the mechanism of the reactions, are well established. If all factors concerned could be accurately assessed, the two approaches would give identical answers. Formation of cyclic acetals seems to be the only instance in which both approaches to the problem of preferred ring structure are possible. The answer to the second question above is of considerable practical importance: if several of the possible stereoisomers of an acetal are present in the crude product, the separation of a pure compound may be very difficult. The analysis of structure of acetal rings shows the few cases where mixtures of stereoisomers of comparable stability are likely to be encountered; measures to avoid these mixtures may then be taken. When the analysis indicates that only one stereoisomer will be stable, it defines the configuration at the new asymmetric centers in the acetal rings. The analysis presented below relates, in the main, to acetal formation in the approximate temperature range of 15" to 100". There is evidence that relative orders of stability may sometimes be altered at higher temperatures, five-membered rings becoming increasingly favored.112The change from six-membered to five-membered rings may be accompanied by a disIi4 ruption of fused-ring 2. Monocyclic Acetals
a. Five-membered Rings.-The possibilities for isomerism are generalized in formulas X X I X and XXX. Type XXIX, with an aT-ring, is knownI6' 18 (112) S. M. Trister and H. Hibbert, Can. . I Research, . 14B. 415 (1936). (113) P. A. Levene and A. L. Raymond, Ber., 66, 384 (1933). (114) C. I,. Mehltretter, R. L. Mellies and C. E. Rist, J. Am. Chem. Soc., 70, 1064 (1948).
CYCLIC DERIVATIVES OF CARBOHYDRATES
XXIX
27
xxx
to be favored over type XXX, with an &-ring, as would be expected from interaction between groups R1 and R2.The same order follows from the rotations needed68in order to form the rings from a zig-zag chain. This order of stability for aT- and aC-rings will hold for any carbonyl component. If no other type of ring is possible, formation of aC-rings may occur, even in the decidedly unfavorable case of erythro-1 2-diphenyl-l}2-ethanedi01.~ From theoretical considerations,68 as well as from the general experience that a more symmetrical substitution of a ring tends to increase stability, the aT-ring should be more stable than the terminal a-ring, in which R1 or R2is hydrogen. The best experimental evidence for this is the rearrangeto 2,3:4,5-di-Oment115 of 1, 2:4 , 5 -di-O-isopropylidene-~~,-galactitol isopropylidenedulcitol, catalyzed by pyridinium chloride or quinolinium chloride. Stereoisomerism a t the acetal carbon atom is, of course, impossible if the I n an aT-ring, it is impossible carbonyl component is symmetrical (R3= R4). when R' = RZ,whatever the nature of R3 and R4.In all other instances, it is possible; and no information is available to show whether one of the stereoisomers will be markedly more stable than the other. If R' and R2 are - closely similar, the two isomers of XXIX will almost certainly be obtained in comparable amounts. To avoid the formation of mixtures, aTacetal rings should therefore be prepared from symmetrical carbonyl compounds. Two isomers are known1I6of' 1 ,6-di-O-benzoy1-2,3 )4,5-di-O-benzylidenedulcitol, and, since only the 2,3:4,5-type of acetal, with two aT-rings, is likely t o be formed, the known compounds are two of the three stereoisomers possible for the 2 , 3 :4,5-di-O-benxylidene derivative. The possibility of stereoisomerism a t the acetal carbon atom in a-rings is of considerable practical importance when they are prepared from aldehydes other than formaldehyde. For the cis and trans isomers XXXI and XXXII i t is impossible to predict whether, in any given case, one will be (115) R . M. H a n n , W. D. Mnclay and C. S. Hudson, J . A m . Cheni. SOC.,61, 2432 (1939). (116) W. T.Haskins, R. M. Hitnn and C. S. Hudson, J . A n ! . Cheni. Soc., 64, 136 (1942).
28
J. A. MILLS
Xi11
XXXII
so much more stable than the other as to be virtually the sole product.
Stereoisomerish in the five-membered ring probably occasions the difficulty encountered117-11gin preparing crystalline 1, 3 :2,4:5,6-tri-O-ethylidene-~glucitol, and accounts for the variable melting point quoted120,lz1 for 1,3:2,4:5,6-tri-O-benzylidene-~-glucitol; it will be shown that stereoisomerism in the 1,3:2,4-portion of these acetals is unlikely. The pendent rings of 1,2:5 , 6 and 2 , 3 :5 , 6 diacetals of furanose sugars may also provide diastereoisomers, unless symmetrical carbonyl compounds are used in acetal formation. Two stereoisomers of 2,3-S-benzylidene-2,3-dideoxy-2,3dithio-DL-glyceritol have been obtained.122 b. Six-membered Rings.-Consideration of the generalized, six-membered, cyclic-acetal structures X X X I I I and XXXIV shows that, if the ring is a chair form, the possibilit,ies for stereoisomerism will in practice be more restricted than for five-membered rings. In the pC-ring X X X III, R1and R2 may both be equatorial, whereas in the pT-ring XXXIV one must be axial and one equatorial. This is the reason for the observed precedence16,l8 of PC- over pT-rings. Furt,her, once R1 and R2are assigned equatorial positions in XXXIII, it is clear t>hatthe most stable arrangement a t the acetal carbon atom is that which has the larger group in the equatorial position R3. Since formation of six-membered acetal rings is practically confined to aldehydes, the stable forms of the acetals will be represented by X X X I I I with R4= H. The rule for the selection of the most stable configuration of a six-membered, cyclic-acetal ring is, therefore, that the two residues and the substituent at the acetal carbon atom will all be on the same side of the ring. However, if the groups R3and R4in a DT-ring (XXXIV) are different, it will in general be impossible to decide which will be t,he more stable arrangement a t the acetal carbon atom, unless, for example, Rl is far more bulky than R2, and R3 than R4,which would then require an equatorial arrangement. of Rl and R3.Therefore, if a pT-ring is to be established, formaldehyde is the reagent of choice. (117) (118) (119) (120) (121) (122)
H . Appel, J . Chewi. Soc., 425 (1935). W. R. Sullivan, .J. Awl. (.'hem. Soc., 67, 837 (1945). E. J. Bourne and L. F. Wiggins, J . Cheitr. SOC.,1933 (1948). S. J . Arlgyal and J. V. I,awler, J . A I N .V / m r l . Soc., 66, 837 (1944). 1'. Iiarrer and J. Riichi, H e h . C/iini. .Ictn, 20, 86 (1937).
I,. W . C. Miles and L. K. Owen, J . C h e n i . Soc., 2938 (1950).
CYCLIC DERIVATIVES OF CARBOHYDRATES
20
xxsv A more detailed analysis of the chair-form pC-ring yields some interesting results. The method followed is that used by Barton65in assessing the relative stability of substituents a t various positions in highly substituted derivatives of cyclohexane. In generalized formula XXXV, R5and R6 represent the hydrogen atom and hydroxyl group that will be present a t this point. Consider first the acetal carbon atom. Group R4 (axial) is close to the two axial hydrogen atoms, but group R3 (equatorial) has no hydrogen atoms close enough for interaction. On the reasonable assumption that the repulsive forces from unshared electrons on the ring oxygen atoms are considerably less than those from a hydrogen atom, the difference in stability, in favor of the equatorial group, is much greater than for two similar groups attached to a cyclohexane ring. Therefore, it is highly unlikely that the isomer of a PC-acetal having the substituent a t the acetal carbon atom in axial arrangement will ever be stable enough to be isolated. It may likewise be shown that groups R1 and R2 are strongly constrained to equatorial positions, and that PT-rings are, in consequence, particularly unfavorable structures. The preferred configuration is very clearly defined, even for terminal acetal rings. The situation of the groups R5 and R6 is even more interesting. R6will suffer repulsions from R', R2, and the two axial hydrogen atoms, but R6, although subjected to the same repulsion from R1 and R2as for RE,has no adjacent hydrogen atoms. The repulsion between R6and the oxygen atoms cannot be accurately assessed, but it may well be that the axial group R6 is more stable than the equatorial group R5. There is, therefore, no barrier to the occurrence of axial hydroxyl groups in BC-acetnls of the general formula and XXXIII. 2,4-O-Methylene-~-glucitol,1, 3 :4,6-di-O-mcthylenedulcitol, the related et,hylidene and benzylidene derivativesI5 are stable acetals with axial, ring-hydroxyl groups. The benzylidenation of arabitol has so
30
J. A. MILLS
far afforded only one 0-benzylidenearabitol (D and L forms k n o ~ n ) , ~ * ~ 9 proved124lZ6to be the l13-acetal; this must have the ring-hydroxyl group in axial position. The other possible six-membered acetal of arabitol, the 3 ,5-acetal, would have an equatorial hydroxyl group in the ring, and it is most interesting that formation of the isomer with the axial hydroxyl group seems t o be favored. 1, 3:4,g-Diacetals of mannitol are readily obtainable,16 so an equatorial position for the hydroxyl group is quite permissible. An attempt t o introduce one tosyl group into 1,3-0-benzylidene-~arabitol'26showed that the ring hydroxyl group is unusually reactive, and attempted unimolecular tosylation of 1-deoxy-2,4-O-methylene-~-xylitol~~ afforded a t least 14% of the 3,5-di-O-tosyl derivative. Such ready esterification of axial hydroxyl groups is attributable to the absence of axial hydrogen atoms, which normally hinder the reactions of axial groups. This should be kept in mind in future attempts at selective esterification of hydroxyl groups in compounds containing six-membered acetal rings. Two isomers of 1,3-O-benzylideneglyceritolhave been prepared, but the configurations are not yet known.lZ8The conformations will be represented by XXXV, with R1 = R2 = R4 = H, R3 = Ph, and with the free hydroxyl group at R6in one isomer and a t RRin the other. It will be seen that no sound conclusions could be drawn from the results of any experiments designed to establish the relative reactivity of the hydroxyl groups in the epimers by esterification, or the relative stability by epimerization reactions, and other methods will have to be used for the configurational assignments. The writer knows of no other simple, cyclic, secondary alcohol for which these established methods of configurational determination would be in-. effective. According to the above analysis, the isomerism of the 1,3-O-benzylideneglyceritols is not to be regarded as an example of isomerism a t the acetal carbon atom, as the phenyl group will be equatorial in each isomer. Indeed, true isomerism at the acetal carbon atom in simple, six-membered rings [other than the rare pT-rings) is most improbable. It has been claimed that 1,3 :5,7-di-O-benzylidene-~and -L-perseitol are obtainable in stereoisomeric forms,ll1 but the two forms had identical rotations, and it is difficult to see how the reported interconversion of the isomers could be effected in the basic solvent used. The two substances are probably dimorphic modificaa
(123) E. Fischer, Ber., 27, 1524 (1894); C. A. Lobry de Bruyn and W. Alberda van Ekenstein, Rec. trav. chim., 18, 150 (1899). (124) Marguerite Steiger and T. Reichstein, Helv. C'him. Acta, 19, 1016 (1936). (125) W. T. Hnskins, R. hl. Hanil a n d C. 8. Hudson, J . Am. ('hem. Soc , 66, 1663 (1943). (126) R Grewe and €1. Pnchaly, ('heui. Ber., 87, 46 (1954). (127) E. Zissis and N . K. Richtmyer, J . Am. Cheni. Soc., 76, 129 (1953). (128) P. E. Verknde and J. D. van Roon. Rec. trav. chim., 61, 831 (1942).
CYCLIC DERIVATIVES OF CARBOHYDRATES
31
ttione, and the extremely large difference in melting poiiits (153-155" and 280 f 2 O ) becomes less surprising on noting that the molecule consists of several planar groups separated by bonds about which free rotation is possible, and thereby has several possible modes of packing into a crystal lattice. The demon~tration6~ that the two known isomers of 2,4,6-trimethyl-strithiane have the chair conformation, and that one therefore has an axial methyl group, might be thought to disprove the reasoning used above to show that substituents at acetal carbon atoms in PC-rings are strongly constrained t o an equatorial position. However, in the s-trithiane ring the distance from an axial group t)o adjacent, axial, hydrogen atoms is about 25 % greater than in m-dioxane (if the axial bonds are truly parallel in each), and the repulsive interactions will consequently be considerably reduced. A reason for the marked tendency of acetone and other ketones to afford five-membered rather than six-membered acetal rings is not readily apparent when the process of acetal formation is examined.68It has now been suggested129that formation of a six-membered 0-isopropylidene derivative is not favored because it would require one of the methyl groups to be axial. This view is supported by the demonstration above that such a group would be conformationally unstable, but it is possible that slight alterations in bond lengths or valence angles may be part of the reason for the marked preference for five-membered rings with the gem-dimethyl group. More information about the relative stability of five- and six-membered isopropylidene acetals might be obtained by acetonation of derivatives of allitol or ribitol in which the terminal hydroxyl groups are masked; the possibilities are the formation of the unfavorable &-ring, or the pC-ring with an axial methyl group.21 3. Acetals Containing Fused Rings a. Fused Five-membered Rings.-The only members of this class requiring detailed discussion are the acetals resulting from reaction at adjacent hydroxyl groups in reducing sugars. The course of acetal formation from these sugars is harder to classify than that from glycitols, and, although sufficient evidence is already available to define some of the main features of the reactions, much more is needed. The commonest type of acetal in this class has the I ,3-dioxolane ring fused cis to a furanose ring. Formation of 1,2-acetals of a-D-glucofuranose has been proved for reactions of D-glucose with acetone,130cycl~hexanone,'~~ (129) See ref. 8, footnote 46. (130) C. G. Anderson, W. Charlton and W. N. Haworth, J. Chem. SOC.,1329 (1929). (131) R. C. Hockett, R. E. Miller and A. Scattergood, J. Am. Chem. SOC.,71, 3072 (1949).
32
J. A. MILLS
be n~a l de hyde ,'~ ~ and f0rma1dehyde.l~~ Other examples of acetals with two fused five-membered rings are 2,3: 5,6-di-O-isopropylidene-Dm a n n o f u r a n ~ s e , ' 2,3-0-isopropylidene ~~'~~~ derivatives of D-ribofuranose,9 5 ' 138 and 1 , 2 :3 ,5-di-O-isopropylidene-~-xylofuranose.~~~-~~~ The second acetal grouping in the diacetals may be combined in different ways, but on partial hydrolysis of several diacetals of glucose,'31 137 mansystem of two five-membered rings nose,137and x y l ~ s e141, ~the ~ ~bicyclic ~ always is the more stable portion. The stable, bicyclic monoacetals are of two kinds, containing one or other of the isomeric structures XXXVI (glucose, xylose) and XXXVTI (mannose, ribose) ;since both structures are 1
reasonably stable, the stability is due to steric rather than electronic factors, the most important being the decrease in interaction of non-bonded groups compared to the alicyclic analogs. An entirely different arrangement, in which two five-membered acetal rings are fused to a pyranose ring, is found in the 1,2:3,4-di-O-isopropylidene derivatives of 143 and a r a b i n 0 ~ e . l14b ~ ~Altrose . is reported146t o afford a mixture of 1 ,2 :3,4-and I ,2 :5,6-di-O-isopropylidene derivatives. The one formula XXXVIII shows the stereochemistry of the 142q
(132) J. C. Sowden and Dorothy J. Kuenne, J . A m . Chem. Soc., 74, 686 (1952); B. Helferich and A. Porck, Ann., 682, 233 (1953). (133) Anne White and R. M. Hixon, J . Am. Chem. SOC.,56, 2438 (1933). (134) 0. T. Schmidt, A . Distelmaier and H . Reinhard, Chem. Ber., 86, 741 (1953). (135) K. Freudenberg and A. Wolf, Ber., 68, 300 (1925). (136) H. Ohle and Gertrud Bercnd, Ber., 68, 2590 (1925). (137) K . Freudenberg, W Durr and H. von Hochstetter, Ber., 61, 1735 (1928). (138) P. A. Levene and E. T. Stiller, J . B i d . Chem., 104, 299 (1934). (139) 0. Svanberg and R.Sjoberg, Be?-., 56, 863 (1923). (140) W. N . Haworth and C. R. Porter, J . Chem. SOC.,611 (1928). (141) P. A. Levene and A. 1,. Raymond, J . Biol. Chem., 102, 317 (1933). (142) K. Freudenberg, A. Noe and E. Knopf, Be?-., 60, 238 (1927). (143) P. A. Levene and G. M. Meyer, J . B i d . Chem., 64,473 (1925); 92, 257 (1931). (144) K. Freudenberg and 0. Svanberg, Ber., 65, 3239 (1922). (145) P. A. Levene and J. Compton, J. Biol. Chenz., 116, 189 (1936). (146) F. H . Newth and L. F. Wiggins, J . Chem. SOC.,1734 (1950).
33
CYCLIC DERIVATIVES OF CARBOHYDRATES
Go Ra ,R1
O
3
R2,R1
2
M e l k O HH XXXVIII
)Me*
M e . L O HH 0
Me2
XXXIX
1,2:3,4-di-O-isopropylideiiederivatives of L-arabinose (Rl = R2 = H), D-galactose (R' = €1, R2 = CH20H) and L-altrose (R' = CH20H, R2 = H). The significant feature is the fusion of the two five-membered rings on opposite sides of the pyranose rings: in the nomericlature normally used for alicyclic compoundsg'") the steric arrangement is cis-anti-cis. This involves much smaller repulsions between substituents than the isomeric cis-syn-cis arrangement XXXIX. It could not have been predicted that, for galactose and arabinose, the acetals with ring structure XXXVIII would be more stable than the alternatives containing the structure XXXVI, but it is clear that ribose should show no tendency to form a 1,2:3,4-di-O-isopropylidene derivative of structure X X X I X (R' = R2 = H). From the evidence available147148 the most stable isopropylideiie derivative of fructose seems to be the 2 , 3 :4,5-diacetal, which also has a cis-anti-cis arrangement of rings, similar to XXXVIII. The preference for the formation of the structure XXXVIII in the examples quoted could possibly be due to the use of acetone as the carbonyl reagent : in I ,2-O-isopropylidene-a-D-galactofuranose (XL) one methyl group and the two-carbon "tail" would be in endo positions a t the points where maximum interference between endo groups is possible (p. 18) ; the same would be true of analogous derivatives of arabinose, altrose, and fructose. This explanation probably has to be rejected, because 2,343benzylidene- and 2 , 3 :4 ,5-di-O-benzylidene-~-fructopyranoseare readily obtainable directly from fructose,149 rather than 2,3-0-benzylidene-~fructofuranose, although the last has no endo substituent a t the acetal carbon atom. D-Galactose affords 1 , 2 :3,4-di-O-methylene-~-gaIactopyran0se,lS0although in rather poor yield. The present indications are that the cis-anii-cis arrangement XXXVIII is produced because it is inherently a stable structdre, but more work is desirable on the four sugars concerned. The reason for the formation of two different diacetals from D-altrose is not apparent. 9
(147) H. Ohle and Ilse Koller, Ber., 67, 1566 (1924); M. L. Wolfrom, W. L. Shilling and W. W. Binkley, J. Am. ('hen]. SOC., 72, 4544 (1950). (148) W. L. Glen, G. S. Myers and G . A. Grant, J . Chem. Soc., 2568 (1951). (149) P. Brig1 and 0. Widmaier, Ber., 69, 1219 (1936). (150) L. Hough, J. K. N. Jones and M. S. Magson, J . Chem. Soc., 1525 (1952).
34
J. -4. MILLS
a0)m XL
H O C H ~ . ~ ~ . ; H
HOCHi
9
2
HO XLI
0
HocH H
O V 0
O V 0
Me XLII
XLIII
hle2
The relative stability of exo and endo substituents in two fused fivemembered rings decides the course of acetal formation from aldoses that could afford either 1 , 2 or 2 , 3 acetals of the furanose form.z0D-Ribose is an example. Acetals of the pyranose form may be disregarded for the reasons given above, but acetonation could still afford three different acetals : 1,20-isopropylidene-a-D-ribofuranose (XLI) and the CY and 0 agomers (XLII the latter pair being and XLIII) of 2,3-O-isopropylidene-~-ribofuranose, interconvertible under conditions permitting mutarotation. The 1,2-acetal (XLI) has two endo substituents (a methyl group, and the hydroxyl group a t C3), but the 2,3-acetal has only one endo substituent (methyl) in the &form XLIII. The relatively greater stability of compounds with ex0 substituents therefore*Ois the factor determining the observed formation of the 2 , 3 - a ~ e t a lSimilar .~~ arguments explain the observed formation of the 2 , 3 :5 , 6 - , not the 1,2: 5,6-,di-O-isopropylidene derivative from mannose, and it may be predictedz0 that allose, gulose, and talose will be found to afford 2,3:5,6-diacetals. An interesting problem in the chemistry of ribose may now be solved. Levene and Stillerg6attempted a monotosylation of their preparation of 2,3-O-isopropylidene-~-ribofuranose, but instead obtained a small yield of a di-0-tosyl derivative in which only one tosyloxy group was replaceable by iodine; the resulting monoiodo-mono-0-tosyl derivative was resistant to alkaline hydrolysis. They tentatively suggested that the tosylation had given the 1,5-di-0-tosyl derivative XLIV, and that the inert tosyloxy group was that in position 1. This opinion has been endorsed,16' but is inadmissible, since glycosyl p-toluenesulfonates are highly r e a ~ tiv e ,3and ~ it is doubtful whether a compound of structure XLIV would have survived the conditions (151) R. W. Jeanloz asd H, G. Fletcher, Jr., Advances in Carbohydrate Chem., 6, 435 (1951).
CYCLIC DERIVATIVES OF CARBOHYDRATES
35
of isolation. Levene and Stiller's '2 3,0-isopropylidene-ribose had not been purified through a crystalline derivative, and the presence in it of a small amount of the Il2-acetal XLI might reasonably be expected, and would not have been detected by the methods usedg6for structural assignment. The subsequent isolation of the 3,5-di-O-tosyl derivative of this minor isomer accountsz0for all of the facts mentioned above. Isomerism a t the acetal carbon atom is theoretically possible in fivemembered acetal rings formed when pyranose or furanose sugars are condensed with aldehydes. For furanose derivatives, the isomerism is exo-endo, and the superior stability of exo isomers ensures that one isomer only will be obtainable, and this of the exo configuration.20 1 ,2-0-Benzylidene-a~-glucofuranose'~~ therefore has the complete configuration XLV. If the
H Tso"';Q~Ts
HOCHt-CH OHy' $4 3o y b
O V 0
Mel XLIV
KO XLV
five-membered acetal ring is fused to a pyranose ring, no convincing arguments in favor of a preferred configuration a t the acetal carbon atom can be devised, and a mixture of diastereoisomeric acetals may be encountered. Configurations differing only at C* in XLVI have been assigned2' to the isomeric di-0-benzylidene derivatives of methyl a-D-mannopyranoside,lb2 and the pair of isomers of 1 ,5-anhydro-di-0-benzylidene-~-mannit0~~~~ is of the same type. A compound believed to be 1 6-di-0-benzoyl-3 4-0-benzylidene-2 5-0metlhylene-D-mannitol (XLVII) has been prepared from 1 6-di-0-benzoyl-
I P11 XLVI
XLVII
2 5-0-methylene- and 1 6-di-O-benzoyl-S,4-0-benzylidene-~-mannito1,'6~ (152) G.J. Robertson, J . Chem. SOC.,330 (1934). (153) Y.Asahina and H. Takimoto, Ber., 64, 1803 (1931). (154) A.T. Ness, R . M. Hann and C. S.Hudson, J . Am. Chem. Soc., 66,2215 (1943).
36
J. A. MILLS
but the structure is not rigidly proved, as a rearrangement of acetal groups could have occurred during the preparations. Such a fusion of five- and seven-membered rings with a h m s junction is stereochemically feasible.5s b. Fused Sixmembered Rings.-The fusion (cis junction) of a six-membered acetal ring to a furanose ring is known in several diacetals, but is in general less stable than the five-membered ring also present in the diacetals. A t low temperatures, 1,2-0-isopropyhdene-a-~-g~ucofura1ioseaffords a 3 ,5-0-benzylidene deri~ative,"~and 1 , 2 :3 ,5-di-O-methylene-a-~-gluco15" A st,ructural examination of other difuranose is readily 0btainab1e.I~~) ~~ be made, to determine whether 1,2:3,5acetals of ~ - g l u c o s e 'should diacetals are the favored type in condensations of D-glucose with aldehydes. It will be noted that the arrangement of rings is cis-anti-cis. Fusions of the six-membered 0-isopropylidene ring to the furanose ring are known : D-xylose affords 1,2: 3 5-di-O-isopropylidene-a-~-xylofuran0se139-14i and, if position 6 is masked, derivatives of 1,2:3,5-di-O-iso157~are 6 ~ obtainable from D-glucose. propylidene LY ~-g lu co fu ran o ~ e* Interesting conformational effects are possible with the six-membered 0-isopropylidene ring. It has been found158that formation of 6-deoxy1,2 :3 ,5-d~-~-~sopropyl~dene-~-n~tro-a-~-g~ucofuranose (XLVIII) occurs much much less readily than formation of the corresponding derivative (XLIX) of L-idofuranose when the 1 2-0-isopropylidene derivatives are )
)
6
CHzNOz
CHzIiOz
Me2
hIez
Me
Me XLVIII
XLIX
condensed with acetone; the 1 ,Zmonoacetals may be separated by preferential acetonation of the idose derivative. The inferencelK9that this selectivity is due to a difference in distance between the relevant hydroxyl groups in the monoacetals is unsound; the distance is the same, but there is a conformational difference. The 3,5-acetal rings may be somewhat distorted from the true chair form, but probably not enough to prevent identification of equatorial and axial groups; therefore, in XLVIII, there must be (155) R. L. Mellies, C . L. Mehltretter and C. E. Rist, J . Am. Chem. Soc., 73, 294 (1951). (156) H. Ohle and L. von Vnrghe, Ber., 61, 1203 (1928). (157) K. Freudenberg, H. Toepffer and C. C . Andersen, Ber., 61, 1750 (1928). (158) J. M. Grosheinta and H. 0. L. Fischer, J . Am. Chem. SOC.,70, 1476 (1948). (159) H. G. Fletcher, Jr., Aduances in Carbohydrate Chem., 3, 45 (1948).
CYCLIC DERIVATIVES OF CARBOHYDRATES
37
two axial groups on one side of the ring in either of the possible conformations (in one, the groups are the terminal carbon atom and the methyl group below the plane of the rings, and in the other, C 2 and the other methyl group), whereas the idose derivative XLIX can exist in a conformation with no interfering axial groups. Examples of six-membered acetal rings fused to another six-membered ring are very numerous, and discussion will be restricted to the two cases in which the second ring is a pyranose ring or another acetal ring. Compounds of this type with a trans ring junction present no problems. The bicyclic system is comparable to trans-decalin, and is therefore rigid, with sharply defined equatorial and axial positions a t every point of substitution. The substituents at acetal carbon atoms will be equatorial. Apparently the first definite assignment of a chair form to a six-membered, cyclic-acetal ring was made5 in order to explain the occurrence of only one stable isomer of methyl 4,6-0-benzylidenc-~-~-glucopyranoside, which will have the conformation L.15Qa The stable forms of 4,6-O-benzylidene derivatives of methyl P-D-mannopyranoside, methyl P-D-allopyranoside, and methyl p-D-altropyranoside will differ from L only in the configurations a t C2 and C3.30 The pattern of formation of fused-ring diacetals of glycitols having a trans junction of the rings conforms to expectations,21the known stable acetalsi5 of this type all having the residues in equatorial positions. 2,4:3,5-Di-O-methyleneallit01~~~ (LI, R = H) is a representative of the most stable of all such acetals, those with two equatorial residues. Substituents a t acetal carbon atoms will have the configurations shown. The trans bicyclic system with one equatorial residue (LII) will occur in 1 ,3:2,4-
Ph L
LI
LII
(159:~)I n Ref. 5 , the mirror image of this was inadvertently reproduced. (160) M. I,. Wolfrom, B, W. Lew and R. h1. Goepp, J r . , J . A m . ('hem. Soc., 68,1443 (1946).
38
J. A. MILLS
diacetals of ribitol, 1,3:2,4-diacetals and 1,3:2,4 :5,6-triacetals of allitol, and 1,3:2,4-diacetals and 1,3:2,4:5,6-triacetals of D-talitol. No triacetals of allitol have been described, and this may be taken as evidence that the symmetrical substitution in LI is more stable than the unsymmetrical substitution in LII. The precise reason for this difference is not known; the argument adduced for the superior ease of formation of $3-rings relative to P-ringsKsmay be relevant, but is based on rate of reaction rather than stability. The acetalation of allitol under forcing conditions would be of has been prepared,’e’ but interest. 1,3 :2,4: 5,6-Tri-O-methylene-~-talitol in poor yield, and the 1,3:2,4-diacetal was not detected. A tri-0-benzylidene-n-talito1 has, however, been made by several workers,1Kand undoubtedly is the 1,3 :2,4: 5,6-triacetal. A more exhaustive study of acetals of talitol would be useful, because another type of acetal theoretically obtainable from it is the 3 , 5 :4,6-diacetal, which would be of the “H-inside” type discussed below. The reaction of 1,6-disubstituted allitols with acetone would be of interest, as there is a possibility of obtaining the 2,4:3,5-di-O-isopropylidene derivative with two axial methyl groups instead of the 2 , 3 :4,5-diacetal with two crC-rings.21 Bicyclic trans diacetals with one axial residue will be considerably less stable than the acetals with equatorial residues; this is shown by the absence of 3 , 5:4,6-diacetals of glucitol, 1 ,3 :2,4-diacetals of mannitol and 2,4:3,5-diacetals of arabitol from the products of acetalation of the free glycitols. However, by suitable masking of hydroxyl groups, some trans diacetals with axial substituents should be obtainable. The most interesting will be 2,4: 3,5-di-O-methylene-~-talitol(LIII), in which one terminal group is axial and the other equatorial (see p. 43). It would seem impossible to prepare trans bicyclic acetals with both residues axial, as this is a very unfavorable conformation, and reaction under any conditions could afford other, more stable types of acetal. For example, in attempts to prepare 2,4:3,5-diacetals of dulcitol, which would have two axial groups, there would be no means of preventing the formation of the stable 2 , 3:4,5-diacetals with two a?’-rings. Discussion of the conformation and relative stabilities of bicyclic diacetals with cis ring junctions requires consideration of the possibilities of “Hinside” and “0-inside” conformations of the flexible ring system (XVIII and XIX) as well as of axial and equatorial positions of residues. One possibility that may be excluded is an “inside” axial position for a substituent of the type XVII shown for cis-decalin, as this would be extremely unstable compared to other ronformations. (161) It. M. IIanri, W . T.Haskiris arid C. 6 . IIudson, J . h i . Chew. S o c . , 69, 624 (1947).
CYCLIC DERIVATIVES OF CARBOHYDRATES
30
Diacetals of iditol therefore present no problem. The general formula for 2,4:3 ,j-diacetals of L-iditol is LIV, with “0-inside” ring system and
LIII
LIV
equatorial positions for residues and substituents a t the acetal carbon atoms. The conformation is relatively rigid, and retains the planar zig-sag arrangement6s of the carbon chain of iditol. The formation of the 2,4:3,5diacetals occurs very readily.68The same ring conformation LIV, but with only one substituent (equatorial) will be contained in the 1,3:2 ,bdiacetals of iditol, glucitol, and xylitol, and in 1,3:2,4:5,6-triacetals of iditol and glucitol. The triacetals of glucitol are readily obtainable,I6 but an attempt to prepare a tri-0-methyleneiditol failed.’@ The occurrence of a tri-0benzylidene-L-iditol has been proved,163 and it undoubtedly is the 1,3:2,4: 5,6-triacetal, for which the complete configuration may be defined a t all points except the acetal carbon atom in the 5,B-ring. The 2,4:3,5-diacetals of glucitol and its derivatives are of great interest and importance, and masking of the hydroxyl group a t C1 is sufficient to ensure the synthesis of these diacetals. 2,4:3 ,ti-Di-O-rnethylene deriva166 D-gluconic 167 and D-glucaric tives are known for ~-‘glucitol,’~~1 6 5 * 16’, In these acetals, one residue must be axial and one equatorial for any combination of chair rings, and it is easy to see that the preferred conformation is LV, with “0-inside” rings and the terminal group at C6 axial; the alternative “H-inside” arrangement is excluded because itt would have the terminal group at C1 in the “inside” axial position, analogous t o XVII. The formation of the 2,4:3,5- rather than the 3,5:4,6diacetal from D-glUCOniC acid shows that “0-inside” diacetals with cis junction and one axial group are more stable than diacetals with a trans junction and one axial group. Chemical evidence supporting the conforma(162) R. M. Hann and C. S. Hudson, J . Am. Chem. SOC., 67, 602 (1945). (163) J . SicB, J . Am. Chem. SOC.,76, 1661 (1954). (164) R. M. Hann, J . K. Wolfe and C. S.Hudson, J . Am. Chem. Soc., 66, 1898 (1944). (165) W. N. Haworth and L. F. Wiggins, J . Chem. SOC.,58 (1944). (166) 0. T. Schmidt and H. Heiss, Chem. Ber., 82, 7 (1949). (167) C. L. Mehltretter, R. L. Mellies, C. E. Rist and G. E. Hilbert, J . Am. Chem. SOC.,69, 2130 (1947). (168) W. G. M. Jones and L. F. Wiggins, J . Chem. SOC., 364 (1944).
40
J. A. MILLS
tion LV for 2,4:3,5-diacetals with the D-glucitol configuration is found in the acetolysis of 2 , 4 :3, 5-di-0-methylene-D-glucitol; the 3 5-ring, with the axial group, is cleaved preferentially.”j4 Diacetals of maniiitol with the groups disposed 2 , 3 , 4 , 5 present several problems. The structure of only one has been established beyond doubt: 2,4:3 ,5-di-O-methylene-~-mannitoP”168 has been converted to 2 ,4 :3,5-di0-methylene-L-idaric acid by oxidation and epimerization. Two conformations for this diacetal are possible, the “H-inside” with two equatorial, terminal groups (LVI) or “0-inside” with two axial groups (LVII). It is )
0
1
R LV
LVI
LVII
reasonable to select LVI as the more stablelZ1but this is by no means proved, and it is possible that the diacetal is a typical example of conformational instability, showing some properties of each of the two conformations LVI and 1,VII. There is evidence from other sources suggesting that t,he “H-inside” conformation with equatorial substituents is not very stable. Dulcitol could afford 1, 3 :2 4-diacetals with this conformation, but only 1 , 3 :4 ,6diacetals have been obtained from condensations of dulcitol with aldehydes; consequently, the 1 3 :2 ,4-structure is less st,able than an isolated @-ring. The @ring is less stable than the pC-ring,‘5 and the pC-ring is less stable than the “0-inside” conformation with two equatorial substituents (shown by the formation of diacetals, not monoacetals, from idito1162),therefore the “H-inside” conformation with one equatorial residue is considerably less stable than the “0-inside” conformation with two equatorial residues. Arabitol seems to show considerable reluctance to give 1, 3 :2,4-diacetals, which are conformationally analogous to the 2,4:3,5-diacetals of mannitol, the possibilities being “0-inside” with one axial residue or “H-inside” with )
)
CYCLIC DERIVATIVES O F CARBOHYDRATES
41
one equatorial residue. However, 1,3 :2 ,4-di-O-methylene-~-arabitol has recently been obtained,168athereby fulfilling the predictionz1that it would be more stable than the 2 ,4 :3 ,5-diacetal. The conformational flexibility postulated for the 2 ,4: 3 ,5-diacetals of mannitol would disappear in acetals prepared from higher aldehydes and therefore possessing substituents at the acetal carbon atoms, but the status of such acetals is a t present obscure. -Two isomers of 1,6-dichloro-l ,6dideoxy-2,3 ,4 ,5-di-O-ethylidene-~-mannitolare known,lBgbut the structures have not been determined. Two stereoisomers with the 2,4:3 ,5structure and related t,o the “H-inside” and “0-insidel’ conformations LVI and LVII, respectively, are clearly possible, but the structures 2,3:4,5 and 2 ,5 :3 ,4 must also be considered. There is considerable scope for more work 011 2, 3 ,4,5-diacetals of mannitol, and confirmation of the structure XLVII advanced by Ness, Hann and Hudson164for their mono-0-benzylidenemono-0-methylene derivative is a necessary step in the assessment of the relative stability of the different structural types. The situation becomes even more complex on noting that two isomeric di-0-isopropylidene derivatives have been obtained from 1 ,6-dichloro-1 ,6dideoxy-~-mannitol.‘~~ These must be structural isomers, and of the three possibilities, 2,4:3,5, 2,5:3,4 and 2,3:4,5, the first is found to be conformationally very unfavorable, although none of the three may be excluded from consideration. A diacetal could not be obtained”O from the reaction of acetone with 1,6-di-0-benzoyl-D-mannitol. Two other approaches to the problem presented by the diacetals of mannitol are available. The first is through the base-catalyzed epimerizaacids. It is known3z.1 6 6 - 168 that the tion of 2,4:3,5-di-O-methylenehexaric derivatives with L-idaric, D-glucaric, and D-mannaric configurations (derived by oxidation of LIV, LV and LVI, respectively) are interconvertible in basic solutions, and, although this is not proved, an equilibrium probably is established. The composition a t equilibrium is not accurately known, but the available evidence shows that the L-idaric derivative is the most stable, and the D-mannaric derivative by far the least stable. At present it may only be said that the two possible conformations for 2,4:3 ,5-di-0methylene-D-mannaric acid (corresponding to LVI and LVII) are both unfavorable, but if similar epimerizations could be achieved with ethylidene and benzylidene derivatives it might be possible to assigri structures and configurations to some 2 ,4: 3,5-diacetals of mannitol. The second possible approach is an examination of the reactivity of sub(168a) E. Zissis and N. I<. Richtmyer, J . A m . Chenb. Soc., 76, 5515 (1954); Dr. Richtmyer kindly provided this information in advance of publication. (169) L. F. Wiggins, J. Chem. Soc., 384 (1946). (170) A. Muller, B e y . , 66, 1055 (1932).
42
J. A . MILLS
stituents at the terminal carbon atoms: axial and equatorial groups differ in degree of accessibility, and therefore in reactivity, and this difference is a recognized aid to conformational studies. Of the acetals under consideration, 2)4:3,5-diacetals of glucitol will provide the basic information required, because it is known (p. 39) that C1 is equatorial and C6 axial. A difference in reactivity a t the two centers has been observed several times, but in most cases it is not definitely known whether C1 or C6 was the more reactive center. A di-0-methylenehexonic acid was isolated166as an intermediate in the oxidation of 2 4 :3 ,5-di-O-methylene-~-glucitolto 2 , 4 :3 ,5-di-o-methyleneD-glucaric acid, but its structure is not known; if oxidation occurs preferentially a t the equatorial group in LV, the acid will be the D-gluconic 3,5-di-0derivative. One chlorine atom in 1,6-dichloro-l,6-dideoxy-2,4: methylene-D-glucitol is more reactive than the other in replacement by iodine or elimination by treatment with metallic sodium,171and one bromine atom in 2 4: 3,5-di-O-benzylidene-l 6-dibromo-l,6-dideoxy-~-glucito1 is replaceable by iodine more readily than the 0ther.1~~ The precise course of these reactions is unknown, and information on this point would be of great importance in establishing the mechanism of the reactions. The sugges172 that the (terminal) group a t C6 is the more reactive site is based ti0nl~~0 only on a comparison of the reactivities of terminal groups in the 2,4:3 )5diacetals of mannitol and iditol, without any knowledge of the conformations of the acetals. In fact, in the series of 2,4:3,5-di-O-methylene derivatives of iditol, mannitol, and glucitol, the (terminal) group a t C6 in the glucitol derivative is different from all the other terminal groups in being strongly constrained to an axial position. I n one instance, the course of a selective reaction a t one terminal group has been established, with a very surprising result. The tosyloxy group a t C6 in 2 , 4 :3 5-di-O-methylene-l , 6-di-O-tosyl-~-glucitol is replaced by iodine more readily than that a t C1 when the di-0-tosyl derivative is heated with sodium iodide in acet0ne.17~Preferential reaction a t the axial group would not have been expected in a reaction that seems to be39 rather strongly retarded by increase in the size of substituents in the vicinity of tosyloxy groups. The observed result is, however, in agreement with other results obtained during replacements of terminal tosyloxy groups in bicyclic diacetals with cis ring junctions. Replacement by iodine of the tosyloxy groups in 2 , 4 : 3,5-di-O-methylene-l,6-di-O-tosyl-~-iditol,which are both equatorial, is very slow in acetone at 100°,162 whereas both groups in the )
(171) I,. F. Wiggins and D. J. C. Wood, J . C h e m . Soc., 1180 (1951). (172) W. G. Overend, R . Montgomery and L. F. Wiggins, J . Chem. SOC.,2201 (1948). (173) A. T.Ness, R. M, Hann and C. S.Hudson, J . Am. Chem. Soc., 86,1901 (1944).
CYCLIC DERIVATIVES OF CARBOHYDRATES
43
corresponding derivative of D-mannitol, which may he axial (see LVII), are readily replaceable under the same condition^.'^^ Replacement by iodine is also slow for the equatorial tosyloxy groups in 2 ,4 : 3 ,5-di-O-methylene-lO- t os yl - ~~- xy litoand l’~ ~ 2,4: 3 ,5-di-0-henzylidene-l-0-tosyl-n~-xylitol.~7~ The extension of the reaction with sodium iodide tlo tosyl derivatives of other acetals of iditol, glucitol, and mannitol could provide evidence for the conformation of 2,4:3,5-diacetals of mannitol and, more important, lead to a clarification of the steric requirements of this reaction. The importance of 2,4:3,5-diacetals of talitol, a t present unknown, is now clear: they will have one axial and one equatorial terminal group (see LIII), and the trans ring junction imposes a rigidity of conformation greater than that of any of the diaeetals with cis junctions. Attention may be drawn to a new reagent, chromium trioxide in pyridine, that is said177to oxidize primary alcohols to aldehydes. The potential importance of its application to bicyclic diacetals with free terminal groups is clear, both as a preparative method for aldoses and as a means of assigning conformations. The above evidence seems to show that the “H-inside” conformation is not favored for bicyclic diacetals, and it might therefore be expected that the “0-inside” conformation will be the more stable in fused ring systems made up from a pyranose ring and a six-membered acetal ring, although only three oxygen atoms are present in the rings. Methyl 4,6-O-benzylidene-a-D-galactopyranoside will then have the conformation LVIII, rather than the alternative “H-inside” conformation, and the ,f3 anomer will also have the same conformation and the same configuration a t the acetal carbon atom. The behavior of the two glycosides in cuprammonium solution178 shows that the assigned conformations are correct, a t least for the pyranose ring portion. The position is not so clear for methyl 4,6-O-benzylidene-a-~-idopyranoside. This was made”9 from the galactoside LVIII under conditions not likely to invert the configuration a t the acetal carbon atom, and therefore should also have the “0-inside” conformation (LIX), yet the behavior in cuprammonium solution suggests1O that the pyranose ring has the conformation found in the “H-inside” conformation L X for the bicyclic system. (174) W. T. Haskins, R. M. Hann and C . S. Hudson, J . A m . Chem. SOC.,66, 67 (1943). (175) R . M. Hann, A. T. Ness and C. S. Hudson, J. A m . Chem. SOC.,66,670 (1944) (176) M. L. Wolfrom, W. J. Burke and E. A. Metcalf, J . A m . Chem. SOC.,69, 1667 (1947); A. T. Ness, R. M. Hann and C. S. Hudson, ibid., 76, 132 (1953). (177) G. I. Poos, G. E. Arth, R. E. Beyler and I,. H. Sarett, J . A m . Chem. SOC., 76,422 (1953). (178) R . E. Reeves, J . A m . Chem. SOC.,71, 1737 (1949). (179) E. Sorkin and T. Reichstein, Helv. Chim. Acta, 28, 1 (1945).
44
J. A . MILLS
LIX
LVIII
LX
Thc conformation LIX has a highly unfavorable arrangement of three axial groups in the pyranose ring, and it is possible that the acetal is an example of conformational instability, showing some properties of each of the conformations LIX and LX, or boat forms related to them; the configuration a t the acetal carbon atom is probably that shown in LIX. A preparation of a 4,6-O-benzylidene derivative directly from methyl CFDidopyranoside would be helpful in settling the problem of the configuration and conformation of the stable form. The question is of considerable importance, because this is the first example of conflict between information (on conformation) derived by the use of cuprammonium solution and that derived by other means. The possibility that different workers may not have used the same stereoisomer of methyl 4,6-O-benzylidene-a-D-galactopyrano~ide"~" adds to the difficulty. Parallel arguments apply to methyl 4,6-O-benzylidene-P-~-idopyranoside: the conformation of the pyranose ring observedLoin cuprammonium solution is in conflict with that deduced from the observed conformation of methyl 4,6-O-benzylidene-~-~-galactopyranoside.~~~ Six-membered acetal rings fused to seven-membered rings are found in acetals formed directly from mannitol and certain of its derivatives. The 1 , 3 :2 , 5 :4,6-structure has been proved for the tri-0-methylene164and triO-ethylidenelso derivatives of mannitol, and the 1 ,3:2,5-structure for the di-0-methylene derivative of 6-deoxy-~-mannitol.~~~ These acetals are markedly more stable than the 2,4:3,5-diacetals discussed above. Their , ~ ~is stability has been related to the probable mechanism of f ~ r m a t i o nbut also understandable on conformational grounds.21 The ring junctions are trans, and the triacetals LXI have the trans-anti-trans configuration, which (179a) H. G. Fletcher, Jr., H. W. Diehl and R. K. New, J . A m . Chenz. Soc., 76. 3029 (1954). (180) E. J. Bourne, G. T. Bruce and L. F. Wiggins, J . Chetn. SOC.,2708 (1951). (181) W. T. Haskins, R. M. Hann and C. S. Hudson, J . Am. Chem. SOC.,67, 1800 (1945).
45
CYCLIC DERIVATIVES OF CARBOHYDRATES
is kiiuwns(a) to confer marked stability on the analogous arrangement of three cyclohexane rings. Only glycitols with the nianno configuration can afford triacetals of the trans-anti-trans type; this is a consequence of the “all-trans” disposition of substituents in 2,5-acetals (LXII) of maunitol.
CHzOH
4-J: LXI
CH,OH LXII
2,5-O-Methylene-~-mannitol (LXII, R = H) is the only acetal containing a single seven-membered ring to be isolated so far, and it probably is the most stable of acetals of its class. In the acetals LXI and LXII the sevenmembered acetal ring seems to be conformationally equivalent to a cyclohexane ring. Axial and equatorial positions a t the acetal carbon atoms in the six-membered rings of LXI are clearly defined because of the trans ring junctions; therefore, when R = Me, LXI represents the stable configuration of I ,3 :2 ,5 :4 ,B-tri-O-ethylidene-~-mannitol,of which only one diastereoisomer is known, in agreement with theory. Triacetals of mannitol prepared from other aldehydes will probably also have the configuration LXI. It will be interesting to see whether a similar combination of seven- and six-membered acetal rings is obtainable from other glycitols for which they are theoretically possible. 1,4:3 ,5-Di-O-methylene-~-arabitolwill be conformationally analogous to 6-deoxy-l,3: 2 ,5-di-O-methylene-~-rnannitol, but it lacks the terminal equatorial group; the present indications are that the absence of this group renders the 1,4:3 ,5-diacetal less stable than the 1,3 :2 ,4-diacetal of arabitol. D-ghJcero-D-talo-Heptitol (volemitol) could afford the 2 , 4 :3 , 6 :5 ,7-tri-0-methylene derivative (LXIII) in which the free terminal group is and this might be expected to exceed the triacetal of mannitol (LXI, R = H) H°CHz in stability. An alternative reaction of volemitol would be formation of the 1 , 3 :2,4-di-O-methylene derivative, with two six-membered rings, trans junction, and one 0 ,, 5 equatorial substituent; a careful study of the acetalation of volemitol may therefore establish the true order of stability of the tricyclic system LXI relative to the more 0 common bicyclic acetals. LXIII (182) W. S.Johnson, J . Am. Chem. Soe., 76, 1498 (1953).
c20
46
J. A . MIIALS
The above discussioii of acetals of glycitols shows that the present empirical rules relating to acetal formatioiiIb have a sound theorctic~1.1basis, aiid that, by an esamination of ring conformations, it is possible to detect further subtle differences in stability of acetals. However, even for glycitols, it is impossible at present to draw up a complete order of stahility of the different classes of cyclic acetal, as some of the experimentztl tests of stability will have to be carried out with heptitols, octitols, and iionitols. For esample, it would be possible to compare the relative stability of bicyelic diacetals having a trans ring junction arid one equatorial residue with that, of those having an “0inside” cis ring junction and one equatorial residue by examining the products from D-ylgcero-D-yluco-heptitol, the two structures named being contained in 4 , G :5,7- and 1 , 3 :2,4-diacetals of this heptitol. It must be pointed out once more that the discussion is based on thermodynamic stability, not rate of formation, of ring structures. It is quite conceivable that the most stable acetals will not be those formed most rapidly in a reaction misture, and adequate time for equilibrium to be established must be allowed when attempting to assess small differences in stability.
VI. ANHYDRO COMPOUNDS 1. The 1 ,4 :3,6-Dianhydrohexitols
The very interesting proper tie^'^, of I , 4 :3,G-dianhydro-D-mannitol (LXIV), 1 ,4 :3 ,6-dianhydro-D-glucitol (LXV), and 1 ,4 :3,G-dianhydro-Liditol (LXVI) have iiot yet, been fully explained, but an explanatioii of most,
LXIV
LXV
LXVI
of them may be deduced from the stereochemistry of the three anhydrides. In the iditol derivative, LXVI, both hvdroxyl groups are e m ; in the mannito1 derivative LXIV, both are endo; aiid in the glucitol derivative LXV, that a t C2 is exo and that a t C5 is endo. The configurations LXIV and LXVI are interconvertible by simultaneous inversion of configuration a t both free hydroxyl groups, whereas such aii operation on 1,XV reproduces the original configuration. The order of stahility of the three dianhydrides, when deduced from ease
CYCLIC DERIVATIVES OF CARBOHYDRATES
47
of formation'sY.la4, and e p i m e r i ~ a t i o nunder ~ ~ ~ conditions of catalytic hydrogenation-dehydrogenation, is LXVI' > LXV > 1,XIV. In LXV, t,he 3 ,6-anhydro ring is established less readily,186 and opened by acids more readily,Is7 than is t.he I ,4-anhydro ring. These differences show that the presence of an exo substit>uentincreases the stability of the ring to which 3,is attached, by virtue of the diminished opportunities for interaction between non-bonded subst,ituent,s. The two endo hydroxyl groups in LXIV are sufficiently close together to afford cyclic acetals, and 1 , 4:3 ,6-dianhydro-2, 5-O-metfhylene-D-mannitol is readily obtainable.188~ lag The formation of cyclic acetals from LXV and LXVI appears to be impossible because of the wide separation of t'he hydroxyl groups, mid the substai1c.e reported as 1 , 4 : 3 ,&dianhydro-2,5-0methyleae-L-iditol'aa must have some other structure. The most interesting behavior of the three diaiihydrides is found in the displacement reactions of t>heir0-sulfonyl derivat,ives, and at present it seems that the report>ed results may be covered by three general rules. (I) Displacement of sulfonyloxy groups from the 0-sulfonates may be achieved under certain conditions, and occurs with inversion of configuratJion a t t,he point, of displacement. (2) Ammonia and halide ions displace sulfonyloxy groups, but alkali hydroxides (and perhaps alkoxides) merely hydrolyze the su1fonat)eesters by 0-sulfonyl fission, with ret,ention of configuration. (3) Displacements by halide ions are possible only for endo sulfoiiyloxy groups. These rules are only tent,ative, and may be modified when more sulfonyloxy compounds, and t,heir behavior to a wider variety of anions, have been studied. Rules (1) and (3) are extensions of the arguments used by Matheson and Angyallgoto explain the behavior of the di-0-tosyl derivatives of the three dianhydrides t,oward sodium iodide. These authors found that, in the Dmannitol derivative, both tosyloxy groups are readily replaceable by iodine; but in the D-glucitol derivative only one, and in the L-iditol derivative neither, tosyloxy group was replaceable. Wiggiiis and Wood171had obtained a parallel series of results from react,ions of the di-0-tosyl and di-0-mesyl (183) It. Montgomery and L. F. Wiggins, J. Chem. Sac., 390 (1946). (184) R. Montgomery and 1,. F. Wiggins, J. Cherri. Sac., 433 (1947). (185) H. G. Fletcher, J r . , and It. M. Goepp, J r . , J . A m . Chem. Sac., 67, 1042 (1945); 68, 939 (1046). (186) S.Solt,zberg, R . M. Goepp, J r . , and W. Freudenherg, J . A n / . Cherri. Soc., 68, 919 (1946). (187) R . Montgomery and I,. F. Wiggins, .I. ( ' h e m . Sac., 237 (1938). (188) S. B. Baker, C a n . J . Phent., 31, 821 (1953). (159) J. A. Mills, unpublished observation. (190) N . K. Matheson and S. J. Angyal, J . C h e m . SOC.,1133 (1952).
48
J. A. MILLS
derivatives with sodium iodide and lithium chloride, respectively. The resistance to displacement by halide ions is very marked for the ex0 sulfonyloxy groups.171These results show1g0that the approach of large anions to the faces of the rings on the inside of the V-shaped molecules, required to displace ezo groups with inversion, is strongly hindered. It follows that the product of a successful displacement of sulfonyloxy groups by halide ions will always have the configuratioii of 1,4:3 ,G-dianhydro-~-iditol;for example, a disulfonate of 1 , 4 :3,G-dianhydro-D-glucitol will react with sodium iodide to afford the sulfonate of I ,4:3,6-diankydrod-deoxy-5-iodoL-iditol. The esception noted for hydroxyl ions in rule ( 2 ) is well established for other sulfoiiic esters of carbohydrate^,^^ which therefore differ from simple of hydrolysis of an ezo sulfonyloxy group is alkyl s u l f o n a t e ~An . ~ ~example ~ provided by the compound described as “methanesulfonyl-monochloro1 ,4:3 ,6-dianhydrosorbit 01’’ 171 n-hich must now be designated 1 ,4 :3,6dianhydro- 5 - chloro - 5 - deoxy -2- 0-mesyl- L-iditol: reaction with sodium hydroxide affords the free, hydroxy compound of the same c~nfiguration.~~’ Examples of the hydrolysis of endo sulfonyloxy groups are not available in the dianhydrohexitols, but the endo tosyloxy group in the stereochemically equivalent 3,G-anhydro-1 ,2-O-isopropylidene-5-O-tosyl-~-glucofuraiiose (LXVII) is hydrolyzed hy sodium hydroxide to the corresponding hydroxyl
LXVII
H LXVIII
compound (VII) without any evidence of inversion a t C5.42This is a surprising result, as the conditions are seemingly ideal for displacement of the tosyloxy group with inversion of configuration to afford an exo substituent. The reactions of tosyl derivatives of the three dianhydrohexitols with ammonia1g”* l g 3 are explicable if it is assumed that ammonia, by virtue of its small size, is able to approach either inside or outside faces of the rings and effect displacement with inversion. The product from 1 , 4 : 3 ,G-dianhydro2,5-di-0-tosyl-~-mannitol~~~ will then be 2,5-diamino-l ,4: 3 ,G-dianhydro2,5-dideoxy-~-iditol,and that from 1 ,4:3 ,G-dianhydro-2,5-di-O-tosyl-Dg l ~ c i t o lwill ~ ~ be ~ 2,5-diamino-1 ,4:3.G-dianhydro-2, 5-dideosy-D-glucito1, inversion at C2 and C5 occurring in each case. The formation of 1,4:3,G(LXVIII) from 1,4: 3 , Gdianhydro-2 ,5-dideoxy-2,5-imino-~-mannitol (191) See ref. 26(b)! p. 341. (192) R. Montgomery and L. F. Wiggins, J . Chcni. SOC.,303 (1946). (193) V. G. Bashford and L. F. Wiggins, J . Cheni. SOC.,371 (1050).
CYCLIC DERIVATIVES OF CARBOHYDRATES
49
dianliydro-2 ;5-di-O-tosyl-~-idit,01~93 conforms to the same pattern: initial displacement of one of the e m tosyloxy groups, with inversion of configuration, is followed by attack of the resulting endo amino group on the other tosyloxy group, with a second inversion of configuration. The deamination of the 2,5-diamino-l 4: 3 6-dianhydro-2,5-dideoxy derivatives of L-iditol and of D-glucitol has been observed'g3to yield some 1 4 :3 6-dianhydro-L-iditol in each case. The reactions evidently follow the s ~ mechanism 1 often observed in deaminations of simple aliphatic amines,46 and afford the product of t,he most stable type, with two P X O hydroxyl groups. Analogies drawn between the behavior of the dianhydrohexitols and the 2,4:3,5-diacetals of hexaric acids17'171 may be somewhat misleading, because the behavior of the latter is determined solely by thermodynamic stability, whereas in the former the steric requirements of reactions become very prominent. The two classes are in some ways complementary, the dianhydrohexitols affording opportunities for studying the relations between steric arrangement and course of reaction at secondary carbon atoms, and the diacetals providing similar opportunities for studying these relations a t terminal groups. 2. 1 6-Anhydro Derivatives of Pyranose Sugars The most interesting feature of the chemistry of these anhydro compounds is the spontaneous formation of certain isomers from the free sugars under conditions that completely hydrolyze other isomers back to the free sugars. The present position is as follows. Spontaneous conversion to a 1 ,6-anhydroaldopyrariose (or equivalent derivative of a ketose) occurs in l g 4 . lg6 acidic solutions of several aldoses and ketoses having the a l t r ~ , 'lg7 ~ ~or . g ~ configurations. l ~ ~ The ~anhydrides ~ are present in equi.librium with the free sugar, and the reported conversions to anhydride range from 43 to 85 %. On the other hand,. it is well known that the 1,6anhydro derivatives of glucopyranose, mannopyranose, and galactopyranose are apparently almost. completely hydrolyzed to tlhe free aldoses in dilute acid solutions. (However, levoglucosan has been isolated'98a as a (194) L.F . Wiggins, J . Chem. SOC.,1590 (1949). 74, 2210 (195) J. W. Pratt, N. K. Richtmyer and C. S. Hudson, J . Am. Chem. SOC., (1952) ; 76, 4503 (1953). 62, 961 (1940). (196) N. K. Richtmyer and C. S.Hudson, J . .4m.Chem. SOC., (197) J. W. Pratt, N. K. Richtmyer and C. S.Hudson, J . Am. Chem. Sac., 74,2200 (1952). (198) Laura C. Stewart, N. K. Richtmyer and C. S. Hudson, J . Am. Chem. SOC., 74, 2206 (1952). (198a) A. Thompson, Kimiko Anno, M. L. Wolfrom and M. Inatome, J. Am. Chem. SOC.,76, 1309 (1954).
50
J. A . MILLS
component of the products formed by the action of dilute hydrochloric acid 011 D-glucose.) Reeves’O has suggested that the spontaneous formation of 1 ,G-anhydro derivatives of idoppanose and altropyranose may he due to the conformational behavior of the aldoses: his “instability factors” show that p-Didose and P-D-altrose will exist partly in the conformations favorable for closure of the 1,g-anhydro ring. This reasoning could lead to the further conclusions that D-talOSe, but probably not D-gulose, should fairly readily afford a 1,G-anhydro derivative of the pyranose form. Since the formation of the anhydrides in acid solution iiivolves an equilibrium, it is more correct to examine the conformations of the end-products of the reaction. An explanation is then apparent. I n 1 ,G-anhydro-P-Didopyranose (LXIX), the three free hydroxyl groups are all equatorial in the chair ring’” known to be present in these anhydrides, and two equatorial (LXX) and hydroxyl groups are present in I ,6-anhydro-P-D-altropyranose (LXXI). The known stable types of in 1,6-anhydro-p-~-gulopyranose
HO
OH
OH LSS
LXIS
LXXI
1,6-anhydropyranoses are therefore relatively free of interactions between non-bonded substituents. Of the three known unstable anhydrides, 1 ,Ganhydro-P-D-glucopyranose(XXV) has all three hydroxyl groups axial, and it will be seen that the derivatives of mannose and galactose each have two axial hydroxyl groups, one group being on the same side of the ring as the bridge. Non-bonded interactions in these are therefore very considerable. If’ steric strain due to repulsions between substituents in the chair ring is the chief factor deciding the stability of 1 ,G-anhydroaldopyranoses, it is possible to calculate that, as the configuration of the nldose is varied, the amount of anhydride present at, eyuilihrium in dilute acidic solutions should decrease in the following or er of configurations,
4
ido
> altro, gulo > talo > a110 > galacto, manno > gluco.
The latest work, lgaa. 19*b however, appears to place allo before talo, and shows that the glum configuration permits some formation of anhydride. (198b) J. W. P r a t t and N. K. Richtmyer, Abstracts Papers Am . Chem. Soc., 126, 22D (1954); L. D. Ough and R. G. Rohwer, ibid., 126, 16D (1954).
CYCLIC DERIVATIVES OF CARBOHYDRATES
51
It is possible that, in t,he allo and gluco anhydrides, hydrogen bonding between a xkl hydroxyl groups may promote stability. A similar order should hold for ketoses, adthough the additional equatorial group (at the anomeric center) may increase the stability of individual isomers, and the diketose dianhydrides are actually always formed. 3. The Scission of Epoxide Rings Since t.his topic was last reviewed in this serieslL6much additional information about the scission of epoxy derivatives of pyranose sugars has become pvailable. Only this t)ype of epoxide will be discussed here. Several w o r k e r ~ ' ~ have.cornpiled 9-~~~ lists of t,he available evidence. There is general agreement that scission of the epoxide ring to provide two free substituenh always occurs with inversion of configuration16a t the point of attack by the reagent, and the chief problem is the predic~tioiiof t,he direct,ion of opening of the ring when two alternatives are possible, which is always the case for 2,3- and 3,4-anhydro derivatives of pyranose rings. For example, a 2,3-anhydride with the D-manno configuration may afford products with the D-gkKO configuration (attack a t C2), or with the D-altro configurat'ion (attack a t C3); products with the D-gluco and D-altro configurat'ions may also result from a 2,3-anhydride with the D-a,llo configuration, but in this case by attack a t C3 and C2, respectively. Some interesting regularities were noted. Derivatives with the D-altro configuration predominate in the products of scission, by bases, of methyl and methyl 2,3-an2,3-anhydro-4,G-0-benzylidene-a-D-allopyranoside hydro-4,6-O-benzylidene-a-~-mannopyranoside.~~~~ 204 Scission of methyl 2,3-anhydro-4,6-O-benzylidene-a(and P)-D-talopyranoside and methyl 2,3-anhydro-4,6-O-benzylidene-a(and @)-D-gulopyranosideaffords, almost ~ ~The ~ * action of ammoniazo6or sodium entirely, derivatives of ~ - i d o s e .*06 gives derivatives of methoxidez06 on 1 ,6: 2,3-dianhydro-p-~-talopyranose 1 ,6-anhydro-p-~-galactose,and t,he same reagentszo6with 1 ,6:3,4-dianhydro-P-D-talopyranose yield derivatives of I ,B-anhydro-P-~-manqo(199) F. H. Newth, W. C. Overend and L. F. Wiggins, J. C h e w . Soc., 10 (1917). (200) S. Mukherjee and H. C. Srivastilva, Proc. I n d i a n A c a d . Sci.,36A, 178 (1952). (201) A. I(.Bose, D. K . It. Chaudhuri and A. K. Bhattacharyya, Chemistry & Industry, 869 (1953). (202) F. H. Newtjh, Chenristry & I n d u s t r y , 1257 (1953). 1193 (1935); G. J. Robert(203) G . ,J. Robertson and C. F. Griffith, J . Chem. SOC., son and W. Whitehead, ibitl., 319 (1940); W. €I. Myers and G. J. Robertson, J . A m . ('hem. Soc., 66, 8 (1943). (204) N. K. Richtmyer and C . S. Hudson, J . A m . Chevr. Soc., 63, 1727 (1941); N . K. Richt,myer, ,~dva7iccsZ'TL ('arbohydrafe ('/ienr,, 1, 37 (1945). (205) L. F. Wiggins, J . f'hem. Soc., 522 (1944). (206) Syllil 1'. .James, F. Sinit!, M. Stacey and I,. F. Wiggins, J . <:hem. Soc., 625 (1946).
52
J. A . MILLS
pyranose; very little of the alternative product with the D - i d 0 configuration is formed in any of these reactions. Other epoxy derivatives of pyranose sugars do not show any regularity in the direction of scission. Methyl 2,3-anhydro-4,6-di-O-methyl-@-~mannopyranoside reacts with ammonia to afford predominantly the product with the D-altro configuration, but with sodium methoxide equal amounts of D-altro and D-gluco isomers are obtained.*07No useful rules could be formulated to predict the direction of opening of epoxides by acidic reagents.Ig9.2oo The writer pointed out20,208 that, in cases where the conformations of the products of scission of epoxides are readily deducible, the major product of scission is that with the two resulting groups in axial position; such carbohydrate epoxides therefore appear to conform to the rule formulated in the steroid field,209namely, that scission of epoxide rings occurs in the direction affording two axial groups in the products. The conformation of methyl 4,6-O-benzylidene-a-D-altropyranoside is rigidly fixed as LXXII, with two axial hydroxyl groups, by virtue of the trans ring junction ; therefore the predominance of products with the altro configuration in the scission of 2,3-anhydro-4,6-O-benzyliPI. OMe dene derivatives of the met,hyl mannopyOH ranosides and methyl a l l o p y r a n o ~ i d e s204~ ~ ~ ~ LXXII conforms to the rule. The preferred direction of scission for 1,6:2,3- and 1,6:3,4dianhydro-/3-D-talose206also provides two axial groups in the products. The preponderance of D-id0 derivatives in the products of scission of the and -Depimeric methyl 2,3-anhydro-4,6-O-benzylidene-~-talopyranosides gulopyran~sides'~~* *05 will also conform to the rule if methyl 4,6-0-benzylidene-a-D-idopyranoside has the "0-inside" conformation L I X discussed on p. 44. The above rule is empirical, and was intended to be used only with epoxides that will afford products having well defined conformations. It was restricted to reactions with basic reagents, because of the likelihood that acids would destroy the rigidity of the conformations by removal of henzylidene groups or cleavage of stabilizing anhydro rings before effecting 208 scission of the epoxide C o o k ~ o n has ' ~ examined the mechanism of the scission of rarbohydrate epoxides in more detail. He considers that the initial product of scission by
k : ?
(207) W. IT. G. Lake and S. Peat, J. Cheni. SOC.,1417 (1938); W. H G. Lake, W. N. Haworth and S. Peat, ibid., 271 (1939). (208) See F. H . N m t h and R. F. Homer, J . ('hem. SOC.,989 (1953). (209) A. Fiirst and P. A. Plattner, 12th Intern. Congr. Pure and A p p l . ( ' h e m . , New I-ork, 1951; Abstracts, p. 409.
CYCLIC DERIVATIVES OF CARBOHYDRATES
53
acids or bases mill always have the two groups in axial positions (antiparallel arrangement), as would be anticipated from the transition state of the reaction (p. S), and that the direction of scission will depend largely on the preferred conformation of the epoxide, which may be deduced with some confidence in many cases.13~92 In rigid ring systems, the epoxide will have a rigid conformation, arid will open in one direction; the initial arrangement will be retained] and the major product has two axial groups. For epoxides of monocyclic pyranosides, the conformation will not be so rigidly fixed, and there is more possibility for scission in each of the two possible directions. The initial product of opening, in either direction, will have axial groups, but may then pass into a more stable conformation with two equatorial groups. Cookson has also postulatedL3a directing electronic effect from the ring-oxygen atom in the opening of non-rigid epoxides by acidic reagents] but more work on this type of scission would be desirable. At present, it may safely be stated that scission, by basic reagents, of epoxide rings in rigid ring structures will afford predominantly the product with two axial groups. This is likely to be true for acidic reagents, too, if the conditions of reaction are such that benzylidene or other stabilizing groups are not removed prior to scission of the epoxide.208The scission of epoxides by Grignard reagents does not conform to the rule.210An attempt to apply conformational rules to the opening of 1 ,2-anhydro derivatives of pyranose sugarszoLis premature, as it has never been shown that epoxides of the glycosan type can suffer scission a t a point other than the anomeric center. (210) A. B. Foster, W. G. Overend, M. Stacey and G. Vaughan,
3308 (1953).
J. Chem. Soc.,
This Page Intentionally Left Blank
COLUMN CHROMATOGRAPHY OF SUGARS AND THEIR DERIVATIVES
BY W. W. BINKLEY Depnrlwient [$ Chemistry, T h e Ohio State University,(hlunibus, Oh,io
CONTESTS 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Adsorbents for the Column Chromatography of Sugars, Sugar Alcohols, and Sugar Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Carbon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Fuller’s Earth Clay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Carbohydrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Ion-exchange Resins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Calcium Acid Silicate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Other Adsorbents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Column Chromatography of Sugar Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . 1. p-Phenylazobenzoyl Esters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Acetate Esters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Methyl Ethers.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Other Derivatives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55 56 56 64 70 75 77 79 79 79 83 88 92
I. INTRODUCTION The stunning complexity of some of the sugar mixtures from natural sources and from synthetic studies presents a formidable deterrent to many investigators. New hope now exists. The serious application, beginning scarcely more than a decade ago, of column chromatography to carbohydrate investigations has produced resolutions rivaling in magic that of Aladdin’s lamp. Column Chromatography is a specialized manipulation of the general technique of selective adsorption with powdered materials.’ These adsorbents are molded vertically in the shape of cylindrical columns. Three techniques are usually employed in the formation of such chromatographic columns: dusting dry powder into an empty dry tube, slowly adding dry adsorbent to a tube containing liquid, and pouring a thin slurry of adsorbent and developer into an empty tube. Other useful techniques closely or remotely related to column chromatography involve use of the (1) Column chromatography is the separation, by selective adsorption, of a mixture of substances into zones or bands, on a column of powdered materials, after the passage of proper solvents.
55
56
w. w.
BINKLEY
chromatopile,? the ~hromatopack,~ and blocks of cellulo~e.~ The removal, with a powdered adsorbent, of some of the components of a solution of a mixture of sugars has been achieved5’6 ; but this important process is not chromatography. The recent flood of reviews and books 011 chromatography refleck in part the iiicreasing popularity of this versatile tool.?
11. ADSO~~BENTS FOR THE COLUMN CHROMATOGRAPHY OF SUGARS, SUGARALCOHOLS,AND SUGAEACIDS 1. Carbon
The use of carbou columns or beds for t,he purification of sugar mixtures by selective adsorption* preceded the advent of modern chromatography by nearly a c e i ~ t u r yOnly . ~ little more than a decade ago, the potent)ialities of this adsorbeut for the resolution of sugar mixtures were recognized.1° A (2) H. I<. Mitchell and F. A. Haskins, Science, 110, 278 (1949). (3) W. 1,. Porter, A n a l . Chwn., 23, 412 (1951). (4) L. S. Cuendet, R. Montgomery and F. Smith, J . Atti. Che.nz. Soc., 76, 2764 (1953). ( 5 ) F. Hnyashi, J . Biocheni. (Japan), 16, 1 (1932); Chem. Abstracts, 27, 8 (1933). (6) Edna M. Montgomery, H. F. Conway and F. B. Weakley, U. S. P a t . 2,610,931 (1952); Chcni. Abstracts, 47, 2523 (1953). (7) (a) H . G . Cassidy and others, “Conference on Chromatography,” Ann. N . Y . Acad. Sci.,49, 141-325 (1948); (b) E. L. Hirst and J . K. N . Jones, Discussions FaraXo. 7, 268 (1949); (c) H. H. Strain, A n a l . Chein., 22, 41 (1950); (d) T. I. d a y SOC., Williams, Discovery, 396 (1950) ; (e) H. H . Strain, “Chromatographic Adsorption Analysis,” Interscience Publishers, Inc., New York, N . Y ., 1945 (revised reprint) ; ( f ) T . I. Williams, “An Introduction t o Chromatography,” Blackie & Son, Ltd., London, 1946; (g) W. W. Binkley and M . L. Wolfrom, “Chromatography of Sugars and Related Substances,” Sugar Research Foundation, N . Y . , Sci.Rept. S e r . , No. 10 (1948); (h) E. Lederer, “ProgrBs de la Chromatographie, 1. Chimie Organique e t Biologiqae,” Hermann & Cie., Paris, 1949; (i) L. Zechmeister, “Progress in Chromatography, 1938-1947,” John Wiley and Sons, Inc., New York, N. Y., 1950; (j) H. G . Cassidy, “Adsorption and Chromatography,” Vol. V of “Technique of Organic Chemistry,” A. Weissberger, ed., Interscience Publishers, Inc., New York, N . Y., 1951; (k) 6 . N . Balstori and B. E. Talbot, “A Guide to Filter Paper and Cellulose Powder Chromatography,” H. Reeve Angel and Co., Ltd., London, 1952; (1) R . C. Brimley and F. C. Barrett, “Practical Chromatography,” Reinhold Publishing Carp., New York, N . Y., 1953; (m) L. Hough, “Analysis of Mixtures of Sugars by Paper and Cellulose Column Chromatography,” Vol. I of “Methods of Biochemical Analysis,” D. Glick, ed., Interscience Publishers, Inc., New York, N . Y., 1954; (11) R . J . Dimler, T r a n s . A7n. Assoc. Cereal Chemists, 12, 1 (1954); ( 0 ) T. I. Williams and H. Weil, A r k i v Kemi, 6, 283 (1953). (8) E. P. Barrett, Advar~cesin Curbohydrafe Chew., 6, 205 (1951). (9) NI. Tswett, Ber. deut. botan. Ges., 24, 316 (1906). (10) 4 . Tiselius, A r k i v Kemi, Mineral. Geol., 14B, No. 22, 4 (1940).
57
COLUMN CHROMATOGRAPHY OF SUGARS
versatile mcthod for the chromatography of colorless substances 11 as developed.ll Sugar solutions and developers, under pressures of up to one atmosphere, were forced u p through an adsorbent bed packed in a metal tube (5-20 mm. high by 20 mm. inside diameter). This tube was connected to the bottom of a rectangular glass (or Lucite) tube (a cuvette, 19 by 94 mm. cross-section by 150 mm. height) which facilitated observation of the refractive index and photographing of the interfaces12 in the effluent from the chromatogram. The entire chromatographic unit was immersed in a thermostatically controlled, clear bath. The utility of carbon for the separation of mono- and di-saccharides was apparent a t the outset (see Fig. (a)
-
L
Meniscus
Sodium chloride
:I 3
v
.Q .4-
Lactose
U
I
c 0
EL
/
increasing volume of developer
FIG. 1.-(a) Sketch of Interfaces Formed in the Effluent from the Chromatogmm, on Carbon (Kahlbaum Active Carbon), of Sodium Chloride, D-Glucose and Lactose; (b) Graph of the Refractive Index u s . Volume of the Effluent of the Same Chromatogram.ll
and separations were extended quickly t o a trisaccharide (raffi(11) A. Tiselius, Arkiv Kemi,Mineral. Geol., 14B. No. 32 (1941); Science, 94, 145 (1941); R. J. P. Williams, L. Hagdahl and A. Tiselius, Arkiv Kemi, 7, 1 (1954).
(12) Method reviewed by H. Schardin, “Das Toeplersche Schlierenverfahren,” VDI-Verlag, Berlin, 1934.
58
W. W . BINKLEY
iiose).ll, l3 The sensitivity and flexibility of this chromatographic method were improved markedly by allowing t)he sugars and developers to pass, under pressure, down through the adsorbent into a very small cuvette (80.0 mm. in length by 1.4 mm. inside diameter, volume 0.13 ml.; see Fig. 2 ) , Pressure applied
Refractive index observed
~Light source
Windows
FIG.2.-Sketch of a Chromatographic Assembly for Interferometric Measurements 011 the Column Effluent.'*
which was connected to a frartion collector outside the temperaturecontrolled bath.I4 The cuvette was aligned with a sensitive interfer~meter'~ which permitted rapid rec3gnition of composition changes in the effluent. The effluent turnover was quick in this small cell; thus, the opportunity for the intermixing of zones was greatly reduced. The combination of this (13) A. Tiselius, Advances in Colloid Sci.,1, 89 (1942); Kahlbaum active carbon, Carboraffin, Suprsnorit, Merck animal carbon, and Merck blood charcoal were found useful in these separations. Norit P-3, Carboraffin Supra, and Darco G-60 were used subsequently. (14) A. Tiselius and S. Claesson, A r k i v K e m i , Mineral. Geol., 16B,No. 18,l (1942); Claesson, Ann. N . Y . Acad. Sci., 49, 183 (1948). T) F. Haber and F. Liiwe, Z . angew. Cheni.,23, 1393 (1910).
COLUMN CHROMATOGRAPHY O F SUGARS
59
excellent method together with an established scale of certain sugars in the order of their increasing affinity to carbon (see Table 1)16 provided a powerful tool for the resolution of sugar mixtures. Three general analytical procedures have employed this method ; they are identified by the mode of development used, namely, frontal, elution, and displacement analysis.l6,l 7 An aqueous solution containing D-glucose, sucrose, and raffinose is allowed to pass through a carbon column; no developer is used. Pure D-glucose (the least tightly held, see Table I) appears first in the effluent, followed by a mixture of D-glucose and sucrose, and finally by all the components of the original solution. Each change in effluent composition produces a new vertical front or step in the graph TABLEI T h e Order o j Increasing A f i n i t y to Carbon (Norit P - 3 ) of Certain Sugars, Phenol, and Ephedrine f r o m Their Aqurous Solictions'O L-Rhamnose L-Arabinose o-Fructose D-Xylose D-Glucose D-Galactose D-Mannose Sucrose Lactose Maltose Raffinose Phenol Ephedrine
showing the refractive index as a function of effluent volume (see Fig. 3).18 This has been designated a fronta>lanalysis; t,he yuant,itative value of such a diagram is limited. An aqueous solution of D-glucose, sucrose, and raffinose can be adsorbed a t the top of a carbon column and developed with wat)er and ethanol-water mixtures until all of t,he sugars have passed through the column. Separately, D-glucose, sucrose, and finally raffinose will be recovered, and each component will produce a peak in the interferometric change of the column effluent with volume as portrayed in Fig. 4. This is (16) A . Tiselius, Kolloid-Z., 106, 101 (1943). (17) A . Tiselius, Arkiv K e m i , Mineral. Geol., 16A, No. 18, 1 (1943). (18) T,. Hagdahl, Acta C h e w . S c a d . , 2. 574 (1948). The definition of the diagrams of the chromatograms, by t'rontrtl analysis arid displacement development, was improved markedly by the use of n. series of t.hree small columns connected t o a micromiser (a capillary filled with qunrte sand) which led directly t o the cuvette for the interferometric readings.
60
W . W. BINKLEY
an elution analysis. The amount of each component in the mixture can be estimated from the area under its respective peak (see Fig. 4). Complete recovery of strongly held substances by elution chromatography is slow
I
I I Raffinose, I
X '0 QI
.-E 5 .-a
c 0
2
L
a
E
Ic
Increasing effluent volume
FIG.3.-Frontal Analysis on Carbon (Carboraffin Supra) of a Mixture of D - G ~ u cose, Sucrose, and Raffinose (Refractive Index us. Volume of EBuent). Developer
-+ Water Jr.-
5%Ethanol
+-15%Ethanol-
Increasing effluent volume
FIG.4.--Elution Analysis, on Carbon, of a Mixture of o-Glucose, Sucrose, and Raffinose; Developed with Water a.nd Ethanol-Water Mixtures (Refractive Index us. Volume of Effluent).
and often difficult. These obstarles are largely eliminated by addition to the developer of a substance more strongly adsorbed than any of the other adsorbates. Soon, the adsorbent is saturated with this substance which has displaced all of the less strongly adsorbed materials; the order of their entry into the effluent is not altered (see Fig. 5). This has been designated a displacement analysis. The concentration of an individual component in a mixture can be estimated from the length of its respective plateau (with
61
COLUMN CHROMATOGRAPHY OF SUGARS
the aid of proper calibration diagrams). The applicationig of this displacement technique to a corn-starch, limit dextrin fractionz0of molecular weight, 975 showed the presence of tri-, tetra-, penta-, and hexa-saccharide comI I I
I
I
Increasing effluent volume
FIG.5.-Chromatography on Carbon (Carboraffin Supra) of a Mixture of D-Glucose, Sucrose, and Raffinose by Displacement Development with 0.5% Ephedrine in Water (Refractive Index us. Volume of Developer) .I *
I
I I
Saccharides
I
I
I
1
I
I
I
I
L
Increasing effluent volume
FIG.6.-Chromatogram on Carbon (Carboraffin Pretreated with 0.15% Ephedrine) of a Corn-starch Limit Dextrin (MW 975), by Displacement Development with 0.5% Ephedrine (Refractive Index us. Volume of Effluent). 1 9
ponents (see Fig. G). The resolution of gram quantities of sugars was found to be sharper with a series of successively smaller carbon rolumns (see Fig. 7)'l than with a single large o m . (19) C. Weibull and A. Tiselius, A i k ~ Kemz, v M i n e r a l . G ~ o l .19A. , NO.19,22 (1945). (20) B. Ortenblad and K. Myrback, HLochCwL. Z , 303, 335 (1940). (21) S. Claesson, A r k z v Kemz, Mzneral. Geol., 24A, No. 16, 3 (1947). The displace-
ment developer was 4% phenol in water.
62
W. W. BINKLEY
Developer
columns
Effluent
FIG.7.-A
Series of Carbon Columns Arr:mged for Chromatography of Fairly Large Quantities of Sugars.18, 21
Recently, the versatility of carbon (Darco G-GO) as an adsorbcnt for the segregation of sugars into molecular-weight groups was emphasizedz2using the simple chromatographic apparatus of Tswet.t.9In a model experiment, a mixt,ure of a mono-, a di-, and a tri-saccharide and limit dextrin was adsorbed on t8hecarbon column ; the monosaccharide was developed off the column with water, and the disaccharide with 5 % ethanol. The operation of this flowiiig chromatogram was facilitated by use of an automatic supplier of developer and a large, laboratory-scale, automat,ic effluent>collector.23n'umerous effluent fractionators have been proposed for column c h r o m a t ~ g r a p h y .24~ ~The . successful application of these devices was (22) R . I,. Whistler and D. F. Durso, J . Anz. Chem. SOC., 72, 677 (1950); 74, 5140 (1952). (23) D. F. Durso, E. D. Schall and R. L. Whistler, Anal. Chem., 23, 425 (1951); J . L. Hickson and R. I,. Whistler, ibid., 26, 1425 (1953). (24) I,. Hough, J. K. N . Jones a n d W. H. Wadinan, . J . Chwm. SOC.,2511 (1949); D. M. P. Phillips, Natctrc, 164, 545 (1949); S. 9. Itnndall arid A . J. 1'. Martin, Biochem. d . (I,ondon), 44, €'roc.. ii (1941)); A . R. Gilson, C h e n i i s f r g Le. I r ~ d z r s l l y29, , 155 (1951); R . A . Grcirit a n d S. R.. Stitch, ibitl., 230 (19.51); J . 0 , Harris, ibid., 255 (1951); I,. A. Boggs, I,. S. Caendct, M. Dulmis and F. Sniitli, Annl. Chem., 24, 1148 (1952); J.Etlelnian and It. V. Martin, Bioch,ern. J . (London), 60, Proc. xxi (1952); A . T.Carlander and S. Gardell, Arkiv Kemi, 4, 461 (1952); J . E. Varner and W. A . Bulen, J . Chem.
63
COLUMN CHROMATOGRAPHY O F SUGARS
exemplilied with a chromatogram of an acid hydrolyzate of an alpha Schardinger cleutrin, using gradient development with ethanol in water (the concentration of the ethanol in water was changed gradually and continuously from 0 to 20 % during the course of the development,; see Fig. 8).2b
0.4-
0
50
10
80
Fraction no.
FIG.8.-Chromatography of a Mixture of Mono- t o Hexa-saccharides from an Acid Hydrolyzate of a Schardinger Dextrin. [The Adsorbent was Carbon (Darco G-60) Pretreated with 1% Ethanolic Stearic Acid; the Developer was Gradually Changed from 0% to 15% Ethanol.]
The unusual utility of this adsorbent has been demonstrated by its role in the preliminary isolation of isomaltose,26(a)maltotriose,26(a)maltotetraose,26(b)and panose26(n) from the hydrolytic products of starch; in the preparation of D-xylo-biose, -triose, -tetraose, -pentaose, -hexaose, and -heptaose from a partially hydrolyzed xylanZ7;in the resolution of bimolecular dianhydrides of L-sorbose28(a) and D-fructose2s(b) ; in an improved procedure for the preparation of s t a c h y ~ s e and ~ ~ ; in the preparation of a series of maltodextrins (as high as maltoheptaose) from the partial, acid Educ., 29,625 (1952) ; H. J. Huisman and S. A. Krans, Chent. Weekblad, 48,1007 (1952) ; R. J . Dimler, J. W. Van Cleve, Edna M. Montgomery, L. R. Bair, F. J. Castle and J. A. Whithead, Anal. Chenz., 26, 1428 (1953); E. S. Sandcrson, ibid., 26, 944 (1954); D. Fraser, ibid., 26, 1858 (1954). (25) A. Tiselius, Endeavour, 11, No. 41, 13 (1952); R. S. Alm, R. J. P. Williams and A. Tiselius, Acta Chem. Scand., 6, 826 (1952); R . S. Alm, i b i d . , 6, 1186 (1952). (26) (a) M. L. Wolfrom, A. Thompson, A . N . O’Neill and T . T . Galltowski, J . Am. Chem. SOC.,74, 1062 (1952); A. Thompson and M. L. Wolfrorn, i b i d . , 74, 3612 (1952); 73, 5849 (1951); (b) R. L . Whistler and J. L . Hickson, zbid., 76, 1671 (1954); Edna M. Montgomery and F. B. Weakley, J . Assoc. O j i c . Agr. Chemists, 36, 1096 (1953). (27) R. L . Whistler and C.-C. T u , J . A m . Chem. SOC.,73, 1389 (1951); 74, 3609 (1952); 76,645 (1953); R. L. Whistler, J. Bachrach and C.-C. Tu, i b i d . , 74,3059 (1952). (28) (a) M. 1,. Wolfrom and H. W. Hilton, J . A m . Chem. Soc., 74, 5334 (1952); (b) B. Wickberg, A d a Chem. Scand., 6, 961 (1952). (29) M. L. Wolfrom, R C. Burrell, A. Thompson and S. S. Furst, J . A m . Chern. Soc., 74, 6299 (1952); R. A. Laidlaw and Clare B. Wylam, J . Chem. SOC., 567 (1953).
64
W. W. RINKLEY
hydrolysis of pot,ato amylase."" Carbon-column chromatograms of the sugars of Cmbilicaria pustulnta, when developed wit>h 1, 15, and 50% aqueous ethanol, yielded D-arabitol, D-marinitol, umbilicin [3-0-(@-~galactopyranosy1)-o-arabit,ol],a ,a-trehalose, and sucrose.31 The carboncolumn analysis of the acid hydrolyzates of the hemicelluloses of Scots pine and black spruce showed the presence of D-xylose, ~-arabinose,D-mannose, D-galactose, L-rhamnose, L-fucose, D-galacturonic acid, ~-xylose-(2-+ I ) 4-~-methyl-~-g~ucosiduronic acid, and 4-O-methyl-~-g~ucuronic acid.32 A crystalline nitrogen-containing tetrasaccharide was obtained from human milk by carbon c h r ~ m a t o g r a p h yUse . ~ ~ of this adsorbent made possible the isolation of certain oligosaccharides from enzymic transglycosidations using sucrose in the substrate; some examples are k e ~ t o s earid ~ ~n e o k e ~ t o s ewith ~~ yeast invertase, a new trisaccharide with T a k a d i a ~ t a s e and , ~ ~ a reducing O-(D-fructosyl)-D-glucoseamong the products from sucrose, D-ghcose, and yeast invertase.37 The salient chromatographic advantages of carbon are general availability, low cost, keen selectivity, and good capacity. Less attractive features of this adsorbent are that it is disagreeable to handle and that it usually requires some pretreatment before use. Furthermore, the rate of solvent flow through columns of finely divided carbon is too low unless up to an equal weight of an inert diluent38is added; a corresponding decrease in column capacity results from this addition. The last objection is met by employing columns composed only of carbon capable of passing through a 40- or 60-mesh screen but retained by ail 80-mesh screen.39 2. Fuller's Earth Clay
Zones formed in the chromatography of colorless substances on columns of colorless or light-colored adsorbents may be detected by streaking the (30) W. J. Whelan, J . M. Bailey and P. J. P. Roberts, J . Chem. SOC.,1293 (1953). (31) B. Lindberg and B. Wickberg, Acla Chem. Scand., 7 , 140 (1953). (32) A. R . N. Gorrod and J . K . N . Jones, J . Chem. Soc., 2522 (1954). The acidic disaccharide was also resolved chromatographically from an acid hydrolyzate of hemicellulose-B from corn cob; R. I,. Whistler and L. Hough, 6 .A m . Chem. Soc., 76, 4918 (1953); R. L . Whistler, H . E. Conrad and L. Hough, ibid., 76, 1668 (1954). (33) R . IZuhn, Adeline Gauhe and H . H. Baer, Chem. BeT., 86, 827 (1953). (34) N . Albon, D. J . Bell, P. H. Blanchard, D. Gross and J. T . Rundell, J . Chem. SOC.,24 (1953). (35) D. Gross, P. H. Blanchard and D. J . Bell, J . Chem. Soc., 1727 (1954). (36) J . S. D. Bacon and D. J . Bell, J . Chem. SOC.,2528 (1953). (37) D. J. Bell and J. Edelman, J . Chem. S O C .4652 , (1954). (38) Celite No. 535, a diatomaceous filter aid, is a product of the Johns-Manviile Co., New York, N. Y . , and is often used for this purpose. (39) R . L. Whistler, Science, 26, 899 (1954).
COLUMN CHROMATOGRAPHY OF SUGAHS
65
extruded columii with a suitable indicator (streak reagent).40The potent'ialities of this method are great. The rapid evaluation and selection of adsorbeiit,s and developers are made possible, using only milligram quantities of adsorbate. Aqueous, alkaline, potassium permanganate4' is often employed as a reagent for the detection of sugars and their colorless derivatives. The st,reaked adsorbent column is cut into sections as indicated by the reagent. The streak is scraped off the column section, and the zone is then eluted with a solvent mixture containing 10 to 20% more water than t.he developer employed in the chromatogram. The recovered adsorbate is oft,en obtained in crysballiiie form. The suitability of fuller's eart>h for the chromatography of carbohydrates was demonstrated by the establishment of an adsorpt,ioii series which included nearly 100 sugar subst,ances (see Table 11).43 The developers mere aqueous 2-propanol and aqueous et,hanol; the water content of these mixtures was gradually increased in order to develop more st'rongly adsorbed carbohydrates, particularly sugar Here, the developing effect of aqueous ethanol was the reverse of its role in the chromatography of sugars 011 carbon.45 In general, the adsorption of sugars on clay did not present a definite pattern. The adsorptive bond of the members of the D-glucose-sucrose-raffinose series did become stronger with an increase of molecular weight ; glycerol, erythritol, ribitol, and allit,ol behaved similarly. Glycosides are held more (30) 12. Zechmeister, L. de Cholnoky and (Mlle.) E. Ujhelyi, Bicll. soc. chim. bid., 18, 1885 (1936). (41) W. H. McNeely, W. W. Binkley and M. L. Wolfrom, J . A m . C'hem. Suc., 67, 527 (1945). (42) These clays are hydrated magnesium aluminum silicates containing some iron and calcium. (43) B. W. Lew, M. L. Wolfrom and R . M. Goepp, Jr., J . A m . Chem. SOC.,68,1449 (1946). A suitable fuller's earth clay was Flores XXX; it was produced by the Floridin Co., Warren, Pa. Other selective clays were Floridin XXX (Floridin Co., Warren, Pa.), Types A and AA Attapulgus Clays (Attapulgus Clay Co., Attapulgus, Ga.), Bleaching Clay 260 (Industrial Minerals and Chemical Co., Berkeley, Calif.), and J. Neutrol (Filtrol Corp., Los Angeles, Calif.). (44) Often, sugar acids can be detected with Congo Red indicator, and ascorbic acid with 2,6-dichlorophenolindophenol. (45) The detection of sugar alcohols in chromatograms developed with aqueous ethanol was difficult because the streak reagent, alkaline potassium permanganate, was rapidly decolorized. This difficulty was largely overcome in small exploratory chromatograms by the use of aqueous tert-butyl alcohol as the developer. The rate of flow of these solvents through fuller's earth clay was very slow and this developer was unsatisfactory for larger columns of this adsorbent. These columns were developed with aqueous ethanol, and extruded. They were wrapped with aluminum foil, leaving exposed t o the air an area, 1 cm. wide, from the top t o the bottom of the adsorbent column. Sufficient solvent had evaporated from the exposed surface in 18 hours to permit the use of alkaline permanganate.
66
W . W . BINKLEY
TABLE I1 Chroinatoyraphzc Adsoi ptzon Series,u on Fzrllci 's E a r t h Clay, oJ Curbohydrulcs and Some Related Substances43 Class I (10 cc. of 70% ethyl alcoholb) Potassium acid D-glwarate, D-glucosamine hydrochloride, 1)-galactosamine (chondrosaniine) hydrochloride (dertro)-Tartaric acid D-Gluconic acid, ammonium wgluconnte, sodium D-gluconate D-Arabonic acid, L-arabonic acid, potassium D-arabonate, citric acid D-Mannonic acid a-D-Galacturonic acid Class ZI {lo cc. of 90% ethyl alcohol) Lactitol, melibiitol L-Iditol, inyo-inositol (m. p. 225"),stitchyose, Schardinger a-dextrin Lactose D-Glucitol, dulcitol, L-perseulose, D-perseulose, Schardinger 8-dextrin, (Zeao)malic acid Raffinose Xylitol D-Mannitol, u-talitol, gentiobiose Class IZI (5 cc. of 90% ethyl alcohol) D-Gulose Melibiose, D-mannose, wribose, 3,6-anhydro-~-glucitol,u-psicose Cellobiose l-Deoxp-o-glucitol Ascorbic acid D-ArabitoI Class Ik' (4 cc of 95% ethyl alcohol) Maltose, u-galactose, D-fructose, melezitose, u-)r2unno-heptrIlose, trehalose, D glycero-D-yulo-heptose L-Fucitol, turanose, 1,5-anhydro-u-niannitol, 1,4-anhydro-~-mannito1, D-lyxose D-Rhamnitol, L-rhamnitol, 2-deoxy-~-"glucitol,"D-gluco-heptdose L-Arabinose, D-arabinose, ribitol, quercitol, diethylene glycol (2,2'-oxydiethanol), dipentaerythritol Sucrose, erythritol, L-altrose ~-G~ucurono-2,6-lactone L-Sorbose, succinic acid Anhydroenneahepti to1 n-Glucose, glycerol Class 1' (5 cc. of 90% isopropyl alcohol) L-Allose 1,4:3,6-Dianhydro-1)-mannitol, amygdalin ~-D-Ga~ncto-metasitccharinic acid 3-Hydroxypropyl 8-o-glucopyranoside rJ-Fucose
BT
COLUMN CHROMATOGRAPHS O F SUGARS
TABLEI1 (Contznued) - ______-____. ~-
_
_
~
Class VZ (6 cc of 97% isopropyl nlcoliol)
Methyl cu-i,-glucopyranosidc, methyl ~-u-glucopyranoside Methyl a-D-mannopyranoside, D-xylose, D-glucono-l,5-lactone 1,4-Anhydro-o-gluc~itol,dipropylene glycol, pentaerythritol, 1,5:3,6-dianhytIrou-mannitol, 1,4:3,6-dianhydro-~-glucitol D-Clucono-l , 4-lactone Class V I I (4 c c . of 97% isopropyl alcohol) L-Rliamnose, salicin Ethylene glycol (1,2-ethancdiol) Propylene glycol (I ,2-propnnediol) 1’11lori x i n 1 , 4 : 3 ,A-Dianhydro-~-idi to1 a Arranged in decreasing order of adsorptive strength. Adsorbent: 1.68 g. of Florex XXX43/Cclite38 (5/1). Dimensions of adsorbent column: 0.9 X 6 cm. of adsorbent. Adsorbate solution: I nig. in 0.5 cc. of developer shown. Developer: noted following the class heading. Prepared by adding 30 cc. of water t o 70 cc. of absolute ethanol. Other dilutions were made similarly, except, t h a t 95% ethyl alcohol was the axeotrope.
loosely than sugars, sugar acids more tightly. Columns of d a y are difficult, to extrude from the usual, straight, glass tubes; they are removed more readily from metal tubes or from glass tubes tapered uniformly and gradually outward from the bottom to the top.46The rate of developer flow through clay columns was slow; the rate was improved by the admixing of a filter aid.38The flow rate can be increased &fold when the clay is air-blown (to remove the finer particles) and the developer is subjected to a pressure of 120 Ibs. per sq. in.47The keen selectivity of clay was exemplified by the separation and isolation in gram quantities (by the extrusion method) of two new di-D-fructose dianhydrides, D-fructopyranose-Dfructofuranose 1,2’:2, l’-diaiihydride4* and a second of unestablished ~ t r u c t u r e The . ~ ~ interconversion, by heat, of D-fructose tlo D-glucose was discovered with the aid of extrusion, clay chromatography. bo 2,3-0-Isopropylideue-D-fructose was isolat,ed chromatographically from partially (46) Ref. 7(g), p. 7 . Tapered glass tubes are produced by the Scientific Class Co., Bloomfield, N . J. (47) D. F. Mowery, Jr., J . Am. C h ~ m SOC.,73, 5047 (1951). Heavy-walled Pyrex tubes were needed for these chromatograms. (48) M. L. Wolfrom and M. Grace Blair, J . Am. Chem. SOC.,70, 2046 (1948); M. I,. Wolfrom, W. W. Binkley, W. I,. Shilling and H . W. Hilton, ibid., 73, 3553 (1951). (49) M. L. Wolfrom, H . W. Hilton and W. W. Binkley, d Am. Chern. SOC.,74, 2867 (1952). (50) M. 1,. Wolfrom and W. L. Shilling, J . Am. Chem. SOC.,73, 3557 (1951).
-
68
W. W. BINKLEY
hydrolyzed “ij-diacetoiie-D-fru~tose,” showing the structure of thc latter to be 2,3:4,5-di-~-isopropylidene-~-fructopyraiiose.~~ Clay has been used effectively in the flowing or elutJioiichromatography of sugars. The residual sucrose was removed almost quantitatively from beet molasses, arid in 73 % yield from cane, blackstrap molasses.52Subscucose
0.19
0.16
u) u)
-
2
.+.
0
@h
2 L
g o.oe u
0.04
0.01
t
I
I
15
39
54
89
Fraction no.
FIG.9.-Chromatography on Fuller’s Earth Clay of a Fraction from Cuban, Blackstrap Molasses. [The Developer was Azeotropic Ethanol from X t o Y , and Ethanol/Water (90/10) from Y t o Z.]
quently, crystalline D-glucose and myo-inositol, and zones containing Dfructose and D-mannitol, were isolated from cane molasses (see Fig. 9); a similar chromatogram of its yeast-fermentation residue revealed the presence of D-glucose, sucrose, erythritol, D-fructose, D-arabitol, and (51) M. L. Wolfrom, W. L. Shilling and W. W. Binkley, J . Am. Chem. Soc., 74, 4544 (1950). (52) W. W Binkley and M. L. Wolfrom, J. Am. Chem. Soc., 69, 664 (1947); M. L. Wolfrom and W. W. Binkley, U. S. Pat. 2,504,169 (1950); Chem. Abstracts, 44, 6180 (1950).
69
COLUMN CHROMATOGRAPHY OF SUGSRS
D-mannitol (see Fig. 10).53 Crystalline D-fructose was prepared chromatographically from invert sugar (acid-hydrolyzed sucrose) .64 Fuller's earth clay is capable of some unusual and difficult separations of sugars, and because of its comparatively light color this adsorbent is adaptable to extrusion, streak-reagent chromatography; furthermore, it is cheap. Most organic solvents pass rapidly through columns of clay. The
D-Frucrose D-Arabitot
o . 0 6 ~, x Erythritol
5
10 13
D-Mannitol
19
29
35
49
Fraction no.
FIG.10.-Chromatography, on Fuller's Earth Clay, of a Fraction from the NonFermented (by Yeast) Residue of Cuban, Blackstrap Molasses. [The Developer was Aeeotropic Pzthanol from X t o Y, and EthanollWater (90/10) from Y to Z.]
addition of wat,er to these solvents markedly decreases tjhe rate of their percolation, and so the admixture of 15 t,o 20 % of filt,eraid t30t,he adsorbent is desirable. The capacity of fuller's eartjh clay is low, and some pretreat'ment is required when t>heclay is t,o be used for isolation work. Whereas the adsorptive propert,ies of clays from different areas are variable, this adsorbent from any particular geographical locatJion is except.ionally uniform. The poor extrusion propert.ies of the makrial are partially overcome by the (53) W. W .Binkley and hl. L. Wolfroni, J . A m . Chevr. Soc., 72, 4775 (1950). (54) D. F. Mowery, Jr., J . h ~ Cheni. . SOC.,73, 5049 (1951); D. F. Mowery, Jr., and G. R . Ferrante, i b i d . , 76, 4103 (1954), for the separation of methyl u-galactoside anomers.
70
W. W. BINKLEY
use of metal, or tapered-glass, chromatographic tubes. Certain inorganic substances are leached from the adsorbent during the elution process, particularly when water-rich solvents are employed, and these form troublesome suspensions. They are removed from an aqueous solution of the adsorbate by treatment with decolorizing carbon and passage through acid-washed asbestos under diminished pressure, followed by gravity filtrat,ion t,hrough hardened filter paper.49
3. Carbohydrates T s v ~ e t t himself, ,~~ recognized the potentialities of powdered sucrose as a chromatographic adsorbent, and employed t,his sugar in his investigations on chlorophyll. Soon after,56the branched form (smylopectin) and the h e a r form (amylose) of potato starch were separated on fibrous celliilose (washed cotton) ; corn starch has also been so f r a c t i ~ n a t e d Interestjingly, .~~ the linear form is the more st,roiigly adsorbed although its molecular weight is lower. Amylose may be able to form with t,he cellulose, a t best,, some t3ype of weak inclusion complex.58 Thus, the removal of amylose from amylopectin is not complete,59and some modifications of this adsorption procedure likewise were unsuccessful.60 The behavior of cellulose nitrate and of non-carbohydrate polymers, on carbon, is similar.6‘ The advent of modern, filt>er-paperchromatography,62and its extension to ~ u g a r s , 6have ~ produced a powerful, qualitative, analytical method. Often, only microamounts of adsorbat,es are handled by this procedure; but larger yuantities of resolved materials were needed for some quant,itative separations, isolations, arid proper identifications, and columns of tightly packed, powdered cellulose were found to be 64 A mixture containing 50 mg. each of L-rhamnose, D-ribose, L-arabinose, and D-galaCtOSe was adsorbed on a 12 by 1.25 in. (diam.) column of powdered cellulose7~ prewetted with t,he (55) RI. Tswett, Y i p o g r . Warshawskago iitsch,ebnago Okrwgn, Warsaw, 1910 (in Russian). (56) C. Tariret,, C o w p t . rend., 168, 1353 (1914). (57) F:.Pacsu :tnd J. W. Mullen, 11, J . A m . Cheni. Soc., 63, 1168 (1941). stsJlizat,ion,” i i i “Technique of Organic Chemistry,” A. (58) R. S. Tipson, ‘‘ Weissherger, ed., Interscience Publishers, I n c . , New York, N. Y., Vol. 111,2nd Edition, i n press. (59) T. J . Schoch, Arlaances in Corhohydrate Chenf,.,1, 247 (1945). (60) G . A. Gilbert, C. T. Greenwood and F. J. H y h r t , J. Chew. Soc., 4454 (1954). (61) 8. Claesson, A r h i v K e m i , Jlineral. Geol., 26A, No. 24 (1948). (62) R . Corisden, A . H. Gordon nnd A. J. P. Martin, Bioche,ni. .I. (London), 38, 224 (1048); A . J . P. Martin, P r i x N o b r l , 110 (1952). (63) S. M . Part,ritlge, .Vatitre, 168, 270 (1946). See G. 1;.Komkahany, Adzinnces in Cnrhoh!/drnta C h e n ~ .9, , 303 (19.54). (64) I,. Hough, J . K . N. Jones iind W. I€.Wildman, ;Vnfur.e,162,448 (1048). (65) A commodity of H. Reeve Angel and Co., Inc., 52 I3uane St., New Yorlr 7 , N.T.
COLUMN CHROMATOGRAPHY O F SUGARS
71
developer, 1-butaiiol saturated with water coiitaiiiiiig 1 % of ammoilia. The flowing chromatogram was employed, and the resolved, cystalline sugars were recovered in 95 % yield from the column effluent. Columns of wet, powdered cellulose are extruded readily i t ith compressed air, hut no suitable streaking reagent is available at present. Fortunately, the chromatographic behavior of powdered cellulose is identical with that of f lterpaper strips and sheets. The flowing chromatogram of this adsorbent, using an automatic, effluent-fraction collectorz4 coordinated with paper-strip analysis, has produced a strong tool. ITsually, the rate of' solveiit percolation through this adsorbent is sufficiently rapid that neither the application of pressure (or vacuum) nor the addition of an inert dilueut (filter aid) is required. The method is also very effective in the resolution of TABLE111 Flowing Chromatogram, on Powdered Cellulose, or 1.6 g. of the Sugars from the Incomplete Hydrolysis or the Gum of the Sterculia setigera Trees@(*) Cofiiponrnls
Yield, mg
I I1 111 IV V
15 413 145 373 640
a ketose L-rhamnose wtagatose and an aldose wtagatose, wgalactose, and an aldose 11-galactose
Total, 1,586 a
The developer was 1-butanol, saturated with water.
mixtures of methylated sugars (which is discussed in detail in a later Section). The broad applicability of this method is already apparent in the successful elucidation of the structure of certain naturally occurring, plant polysaccharides. Powdered-cellulose chromatography of the sugars from the incomplete, acid hydrolysis of t,he gum from the Sterculia setigera tree yielded pure, crystalline L-rhamnose hydrate and D-galactose, as well as D-tagatose (also crystalline, see Table III)66(a); other methods showed the presence of D-galacturonic acid residues. Pure crystals of' L-rhamnose and D-xylose were obtained by powdered-cellulose chromatography of a hydro(66) (a) E. L. Hirst, L. Hough and J. K. N. Jones, Nature, 163, 177 (1949); J . Chem. Soc., 3145 (1949); (b) P. Andrews, D. H. Ball and J . K. N . Jones, ibid., 4090 (1953),
for peach and cherry gums; (c) P. Andrews and J. K. N . Jones, ibid.,1724,4134 (1954), for lemon and golden apple gums; (d) E. L. Hirst and A . S. Perlin, ibid., 2622 (1954), fov the gum of Acacia pycn.antha.
72
W. W. BINKLEY
lyzed mucilage from flax seed; L-arabinose was also 67 The acidhydrolyzed mucilage of the slippery elm (Ulmus ,fuZva,) was found to contain ~-rhamnose-(2+ 1) D-galactosiduronic acid and a mixture of sugars, resolved by this chromatographic method int>o it,s components, D-galactose, L-rhamnose, and 3-O-methyl-~-galactose.~~ The application of the method, using a hydrocellulose, to the acid hydrolyzates of several galactomannans (from clover,'j9 and fenugreek seed70) gave crystalline D-mannose and D-galactose; n-glucose was obtained, in addition to these sugars, from certain Iris The residues of D-galactose, Larabinose, and 4-O-methyl-~-glucuronic acid (isolated by another procedure) were shown, using powdered-cellulose columns, to be present in gum myrrh72;and D-xylose and D-galactose residues were found to be components of the polysaccharide of tJhefresh water alga, N o s l o ~An . ~ ~L-arab. inopyranose disaccharide [3-O-(P-~-arabinopyranosyl)-~-arabinose] was separated from the hydrolytic products of larch g gal act an.^^ Chromatography, on powdered hydrocellulose, of the sugars from the hydrolyzed, animal polysaccharide of frog-spawn mucin led to the identification of L-fucose, D-mannose, D-galactose, D-glucosamine, and D-galactosamine among the components (see Table IV).75 Further evidence of the diversity of chromatographic applicability of hydrocellulose was shown in its ability t o separate a hexitol from a hexose; crystalline D-arabitol and D-galactose were isolated from acid-hydrolyzed umbilicin of Umbilicaria p ~ s f u l a t a . ~ ~ The powdered-cellulose column has performed valuable service in the study of sugar synthesis. DL-Xylose was resolved from the products of the action of heat and alkali on paraformaldehyde; DL-xylose and DL-arabinose were isolated after the action of alkali a t 20" on a mixture of glyceraldehyde and gly~olaldehyde~';subsequently, DL-ribose (identified as the p-tolylsulfonylhydrazone7*) was obtained79 using a column operating a t 60". (67) D. G. Easterby and J . K. N. Jones, Nature, 166, 614 (1950). (68) L. Hough, J. K. N . Jones and E . L. Hirst, Nature, 166, 34 (1950); E.L. Hirst, I,. Hough and J . K. W. Jones, J . Cheni. SOC.,323 (1951); for okra mucilage, R . L. Whistler and H. E. Conrad, J . A m . Chem. SOC.,76, 1673, 3544 (1954). (69) P. Andrews, L:Hough and J. K. N. Jones, J . A m . Chem. SOC.,74, 4029 (1952). (70) P. Andrews, L. Hough and J. K. N. Jones, J . Chem. SOC.,2744 (1952). (71) P. Andrews, L . Hough and J. K . N . Jones, J . Chem. SOC.,1186 (1953). (72) L. Hough, J . I(. N. Jones and W. H. Wadman, J . Chhem. SOC.,796 (1952). (73) L. Hough, J . I<. N. Jones and W. H. Wadman, J . Cham. Soc., 3393 (1952). (74) J. K. N. Jones, J . Chem. S O C . 1672 , (1953); Chemistry & Industry, 954 (1952). 2136 (1950). (75) B. F. Folkes, R. A . Grant and J. K. N. Jones, J . Chem. SO C., (76) B. Lindberg, C. A . Wachmeister and B. Wickberg, Acta Chem. Scand., 6, 1052 (1952). (77) L. Hough and J . K. N . Jones, J . Chem. SOC.,1122 (1951). (78) L. Hough and J . K. N. Jones, J . Chem. Soc., 3191 (1951). (79) J. N . Counsell, L. Hough and W. H. Wadmzn, Reseaich (London), 4,143 (1951).
COLUMN CHROMATOGRAPHY OF SUGARY
73
n-threo-Pentulose (D-xylulose, tentatively identified as the osazone) was found chromatographically to be among the products of the action of a pea enzyme on glycolaldehyde and triose phosphate (prepared from D-fructose 1 ,G-diphosphate)*O;similarly, 6-deoxy-D-fructose and ti-deoxyL-sorbose resulted from m-lactaldehyde,BOsedoheptulose from D-eryt>hrose,B" 5-deosy-~-threo-pentulosefrom acetaldehyde and a triose phosphate,81 and D-ido-heptulosan from D-xylose.82 The powdered-c*ellulose, flowing chromatogram of an unfermentable residue from the mixture obtained by the action of heat, on D-fructose yielded a sirup containing a ketose having a mobility greater than that of D-fructose; i t may be ~-psicose.*~ D-Psicose (identified as the phenylosazone TABLEIV 17lowing Chrornntograrn on Powdered Cellulose of the Sugars from the Hydrolysis of 1.082 y. of Frog-spawn Mucin7$ Ej4uent fractiona
Yield, mg.
I I1
59 16
111 IV V VI
15 148 108 -
Component
L-fucose o-mannose (identified as the phenylhydrazone) amino sugar o-galactose D-glucosaminc o-galactosamine (chondrosamine)
1
Physicel form o//raclion
crystalline sirup sirup crystalline crystalline hydrochloride crystalline hydrochloride
~
Total, 346 0
The developer was 1-butanol, half saturated with water.
and as 1,2 : 3 ,4-di-O-isopropylide11e-~-psicofuranose)was found among the products of the action, a t 37") of ammonia on ~ - g l u c o s e . ~Chromato~(~) graphic assay of the products formed by the action of this base on maltose and lactose revealed their partial conversion to m a l t u l o ~ e ~and ~ ( ~lactu) 1 0 ~ e , ~respectively; ~(~) ammoniacal melibiose yielded melibiulose, ti-O-(c~-~galactopyranosy1)-p-D-mannose, D-tagatose, and ~ - g a l a c t o s e . ~ ~ ( ~ ) A fine example of the rapacity and selectivity of powdered cellulose is (80) L. Houyh and J . K. N Jones, J . Chenr. Sac., 4047 (1952); 4052 (1952); 342 (1953). (81) P. A ,J. Gorin, L. Ho~ighand J . K. N . Jones, d . Chem. Sac ,2140 (1953). (82) J. K. N. Jones, J . Chern. Sac., 3643 (1954). (83) L. Sattler, F. W. Zerhan, G . I,. Clark, C.-C. Chu, N . Albon, D. Gross and H. C. S. de Whalley, Ind. Eng. Chem., 44, 1127 (1952). (84) (a) L . Hough, J . K. K.Jones and E. L. Richards, J . Chem. Sac., 2005 (1953); (b) 295 (1954).
74
W. W. BINIiLEY
the quantitative determination of raffinose in raw-beet sugars.85h chromatogram of 20 g. of the raw sugar, on a 30 by 7.5 cm. (diam.) column of adsorbent, sharply separated 0.5 % of raffinose (as the pentahydrate) from the very large amount of sucrose and extremely small quantity of stachyose; the developer was 2-propanol/l-butanol/water(7/1 /2).a6 Starch columns were found suitable for the separation of certain monos a c c h a r i d e ~These . ~ ~ columns, pretreated with aqueous I -butanol and the developer 1-butanol/l-propanol/water (4/1/1), and developed under
2oot
a
c
c
3
L
150-
L L
f
-
E
e
l
100P aJ
: 5 Fn
x
50 -
f.n
01
I
I
I
20
50
100
Effluent volume, mi.
FIG.11.-Chromatogram, on a 20 x 0.9 cm. Potato-Starch Column, of (a) L-Rhamnose, (b) L-Fucose, (c) D-Ribose, (d) D-Xylose, (e) D-Mannose, ( f ) D-Glucose, and (g) o-Galactose. [The Developer was 1-Butanol/l-Propanol/Water(4/1/1).]
pressure, \wre capable of sharp separations (see Fig. 11). The capacity of starch was found to be lorn; the maximum weight-ratio of adsorbate to adsorbent was ca. 1:1000. The coordinated use of powdered-cellulose and starch chromatography has far-reaching potentialities. Cellulose powder has moderately good capacity, and the rate of flow of developer through its columns is good if slight pressure is applied. Filterpaper strips and the cellulose powder have identical adsorptive properties, (85) D. Gross and N . Albon, Arinlyst, 78, 191 (1953). (86) Unless otherwise noted, solvent ratios will be expressed on a volume basis. (87) S. Gardell, Acta Chem. Scand., 7, 201 (1953). Potato starch was conditioned by suspension in a mixture of 1-butanol and water; the total moisture in the system was in the weight ratio of waterlanhydrous starch:3/10.
COLUMN CHROMATOGRAPHY OF RTJGARS
75
and they form a useful chromatographic partnership. The progress of colored and of fluorescent (iii ultraviolet light) suhstaiices, moving on cellulose, is readily followed. Chromatographic cellulose is expensive, and it usually requires some treatment before use. Starch is capable of separating closely related sugars. I t is inexpensive, but t,he rate of flow of developer through starch columns under pressure is low, and the adsorbate capacity of starch is very low. This carbohydrate must be conditioned and prewashed before it is suitable for chromatography. 4. Ion-Pxchange Resins
Serious thought and a considerable amount of effort have been devoted to the utilization of ion-exchange resins in the sugar industry.88Some success has been achieved in the beet industry, but the situation is much less certain in cane-sugar production. The principal objectives are the retention by the resins of the ionic (organic and inorganic) impurities and the maximum recovery of sucrose. Quite different is their use in the separation of sugars. A column of an anion-exchange resin (Amberlite IR-400,s9 regenerated with sodium hydroxide) retained, in general, reducing sugars when developed with water. Sugar alcohols and methyl a-D-glucopyranoside were not adsorbed,g0 and sucrose was only partially retained. The sugars were completely eluted by 10 % sodium chloride. The chromatographic separation of sugar mixtures was not attempted. The polyhydroxy functions of sugars and related substances permit the formation of complexes with the borate ion of boric acid and borate salts.g1 These complexes are readily formed in low concentrations of borate and display different adsorptive strengths when added to a column of an anionexchange resin (Dowex-la2) regenerated to the borate form.93 The chromatogram was developed first with 0.018 M sodium borate; D-fructose and D-galactose, in succession, appeared in the effluent, followed by D-glucose, which required 0.03 M sodium borate. The adsorptive strength of nonreducing sugars is dependent upon their molecular weights, stachyose being more strongly bonded than raffinose, and raffinose more tightly held than (88) R . Kunin and R . J. Myers, “Ion Exchange Resins,” John Wiley and Sons, Inc., New York, N . Y., 1950; R . Kunin, Chem. Eng. News, 32, 3046 (1954). (89) A product of the Resinous Products Division of Rohm and Haas Co., Philadelphia, Pa. (90) S. Roseman, R . H. Abeles and A . Dorfman, Arch. Biochem. and Biophys., 36, 232 (1952). (91) J. Boeseken, Advances i n Carbohydrate Chem., 4, 189 (1949). (92) A product of the Dow Chemical Co., Midland, Mich. (93) J. X. Khym and L. P. Zill, J. A m . Chem. Soc., 73, 2399 (1951).
76
W. W. RINKLET
The effect of a functioiial group was even greater, the adsorptive strength of D-fructose being higher than that of the oligosaccharides. A mixture of the pentoses, D-ribose, D-arabiiiose, and D-xylose, was resolved .~~ by Dowex-1 when developed with 0.015 M potassium t e t r a b ~ r a t e The fine selectivity of this method was established by its successful application to a series of disaccharides (see Table V). The progress of these chromatograms was followed quantitatively (for total sugar) with anthrone963 y7 and orcinol r e a ge nt ~ .~The 8 individual sugars were identified by paper chromatography after the removal of potassium ions with a cation-exchange resin. The isolation of pure sugars was completed after the formation, and removal under diminished pressure, of the boric acid as its methyl ester.99 This method is also applicable to sugar alcohols, as exemplified by the TABLE V Chromatography; on an Anion-exchange Resin ( D o w e x - I ) , of Certain DisaccharidesQs Disaccharide (order of appearance in efluenl)
Volume required lo produce maximum concentralion in eBuent, 1nl.b
Sucrose Trehalose Cellobiose Maltose Lactose
175 370 660 750 X1@ ~
The developer was 0.005 M K2B,07 . 6 An amount of 10-25 mg. of each sugar waa added t o an 11 by 1 cm. (diam.) column of resin. a
separation of D-glucitol, galactitol and D-mannitol; to sugar acids in the resolution of D-galacturonic and D-glucuronic acids on Dowex-1,92 acetate or formate form, with aqueous acetic or formic acid as the developerloo; and to amino sugars in the chromatography of D-glucosamine and D-galactosSedoheptulosan is readily sepamine (as hydrochlorides) lol on D0wex-50.~~ arated from the more strongly adsorbed s e d o h e p t u l o ~ e . ~ ~ The chief disadvantage of the method is the unusually large volume of (94) G. R. Noggle and L. P. Zill, Arch. Biochem. and Biophys., 41, 21 (1952). (95) J. X. Khym and L. P. Zill, J . Am. Chem. Soc., 74, 2090 (1952). (96) R. Dreywood, Ind. Eng. Chem., Anal. Ed., 18, 499 (1946). (97) D. L. Morris, Science, 107, 254 (1948). (98) A. H. Brown, Arch. Biochem., 11, 269 (1946). (99) L. P. Zill, J. X. Khym and G. M. Cheniae, d . A m . Chenz. Soc., 76, 1339 (1953). (100) J. X. Khym and D . G. Doherty, J . A m . Chem. SOC.,74, 3199 (1952); J. X. Khym and W. E. Cohn, ibid., 76, 1818 (1954); J. X. Khym, D. G. Doherty and W. E. Cohn, ibid., 76, 5523 (1954) for the chromatography of o-ribose phosphates. (101) S. Gardell, Acta Chem. Scand., 7,207 (1953); S . Gardell and S. Rastgeldi, ibid., 8, 362 (1954); E. Drake and S. Gardell, Arkiw Kemi, 4 , 469 (1952).
COLUMN CHROMATOGRAPHY OF S U G A R S
77
developer required in order to achieve the desired separations. The durutiori of the experiments is long, and the recovery of small amounts of separated sugars from large volumes of solution is tedious. TABLEV I Chromatographic Adsorption Series,= on Calcium Acid Silicutc, of Some Sugars and Related S r ~ b s t u n c e s ~ ~ ~ CInss Z (10 cc. of 90% dioxane) a-D-Galacturonic acid Lactose monohydrate 1,actitol Drilcitol Melezitose, raffinose pentahydrate, gentiobiose, D-glyccro-D-gubheptose Sucrose, maltose monohydrate, cellobiose, D-glucitol (sorbitol) D-Galactose, D-mannitol D -Glucose, o -fructose, I) -mannose, L-sorbose L-Fucitol L-Arabinose Class IZ (5 cc. of 90% dioxane) L - Fu cose o-Xylose, L-rhamnose monohydrate Methyl a-D-ghcopyranoside Class I11 (10 cc. of 90% diosane) D-glycero-D-gulo-Heptonamide o-Galactonamide, D-glnconamide Class ZV (5 cc. of 90% dioxane) D-Lyxonamide D-Ribonamide L-Fuconamide a Arranged in decreasing order of adsorptive strength. Adsorbent: 1.8 g . of Silene EF10Z/Celite38 (5/1) (wt. ratio). Dimensions of adsorbent column: 0.9 X 11 cm. Adsorbate solution: 2 mg. in 0.2 cc. of 90% dioxane. Developer: noted following the class heading.
5 . Calcium Acid Silicate In addition to fuller's earth clays,43certain synthetic silicates are useful in the column chromatography of sugars. Calcium acid silicate (Silene EF102) was, in its original e v a l ~ a t i o n , found ' ~ ~ to be suitable for the separation of such groups of sugars as pentoses from hexoses, and mono- from oligo-saccharides, following the general pattern of increased adsorptive strength with molecular weight (see Table VI) . The influence of molecular (102) A product of the Columbia Chemical Division, Pittsburgh Plate Glass Co., Barberton, Ohio. (103) I,. W. Georges, R . S. Bower and M. L. Wolfrom, J . Am. Chem. Soc., 68, 2169 (1946).
78
W. W. BINKLEY
weight was as usual overshadowed by that of the carhosyl group of the sugar arid ; in the series studied, tlic 1,oud het~\veeua-D-galacturouic acid and this silicate adsorbent was the st,rongest. C'ertain properties of this adsorbent, preseiit, a st,roiig rase for its c-hromatographic~use. ITnlikcfuller's eart'h clays, t,his syiit,hetic: silicate is readily estjrricled from st,rsight, glass tubes. The adsorlieiit is white, arid if it is used with t,he proper dcvelopers (aqueous dioxilrie or aqueous lert-butyl alcohol), dctectioii of slowly reduring sugar substances is possible with the permangallate-streak reagent,. Separate lot,s of t,his silicate varied greatly in adsorbent properties when test'ed under st,aiidard conditions with il mixture of D-glucaose and maltose."'" The adsorbents were gradually deact,ivated by continuous, water washing; they were reactivated by sodium hydroxide and certain salts. The pH of an aqueous susperisioii of the alkali-reactivat,ed adsorbent is 10-1 2; and so TABLE VII T h e pH of th,e Supernatant Liquors froni the M i x t u r e of 1 9. of Various Silicate Adsorbents with 16 ml. of Water'"" Adsorbent*
- ~ _ _ ~ _ _ _ _
I
__-_~
Hgdrnt,etl c:ilciiim :wid silic:ite, lot. K O . 1" Hytlr:Lt,ed c:ilciuu x i d silicate, lot, N o . 2" Hydrat.erI calcirini acid silicate, lot, N o . 3" Hydrated magnesium acid silicate,, Fuller's earth clay"
PfJ
10.3 10.4 10.6 8.3 8.6
a The adsorbents were mixed with :t filter aid in the weight-rtit,ios of 5/1. EF.'a2 c Magnesol.'o5 d Florex XSX.43
* Silene
the adsorbent is then unsuitable for the resolution of alkali-sensitive sugars. The alka1iiiit.y of untreated calcium acid si1icat.e was greater t,hsn that, of some other si1icat.e adsorbents (see Table 1'11). Sufficient. colloidal and soluble iiiorganir substances are roiitributed to the eluat,e by the adsorbent, during zoiie elution, t,hat a tedious procedure must, be followecl hi order t o recover the resolved adsorbate. These difficulties are not. encountered in t,he chromat,ography of sugar derivatives which require much less polar developers. The principal advantages of t,his adsorbentJ are it,s low cost, its \vhite color, the fairly high rate of solveiit percolation through it, and t,he ease of its ext,rusioii. (104) D . 0 . IIotTrnaii (with h l . I,. Wolftom), P1i.D. Ilissertation, The Ohio St,:it,e Universit,y, 1048. A n mioiirit of 0.2 ml. of :L solut>ioriof r)-glucosc (100 mg.) anti malt,ose (100 ing.) in a Inixcture of I nil. of w:iter :mtl $1 ml. of dioxane was added a t the t q of a 120 X 0 mm. (tlinm.) colu~niiof Silenc E;F/filter aid (.5/1, \ v t . rat,ios). The clirornnt,ograin was developed wit,h 8 ml. of :L soliition prel)*.red by mixing 1 i d . of vi:tt,er wit,h 10 ml. of diossrie.
COLUMN CHROMATOGRAPHY O F SITGARS
79
6. Other Adsorbents Certain adsorbents have been selected for specific, sugar separations, without consideration of t>heirpossible general applicability t,o carbohydrates. Mixt'ures of C14-labeledD-glucose and D-fructose were resolved on a column of hydrated magnesium acid silicate,lo6employing 05 % ethanol as the developer."J6The acid-treated silicate (essentially silicic acid) was useful as an adsorbent in the investigation of the structure of sapote gum.107 Alumina,'O* one of the most popular chromatographic adsorbents, has rarely been used for uncombined sugars. A sharp separation of the branched (upper zone) from t,he linear (lower zone) constituent of starch was stated to be effected on a column in which the upper section of the alumina was adjusted to pII 4.5 and t,he lower section tJopH 8.2; t'he developer was 0.0025 N iodine in pot)assium iodide.logA multiplicity of zoiies resulted on chromatographing animal glycogens on freshly precipitated calcium carbonate; aqueous iodine was employed as the developer and zone indicator.l10 The adsorption on silica gel of the polysaccharides of Mycobacterium tuberculosis and successive development with hexane, chloroform, methanol, and water produced 21 distinct fractions.11*
111. COLUMN CHROMATOGHAPHY O F SUGAR DEarvATIVEs I . p-Phenylaxobenxoyl Esters
The epic of the column chromatography of sugars began a little more than a decade The sugars D-glucose and D-fructose, being colorless, were esterified with a colored acyl halide, to yield colored sugar derivat,ives. The acylating agent chosen was p-phenylazobenzoyl ("azoyl") chloride, and red-colored, crystalline pentaazoates were obtained. A mixture of these esters was adsorbed on a column of silica, and developed under pressure with a 1/1 mixture of benzene/petroleum ether (b. p. SO-SO"). The progress of the resolution was followed visually; a typical Tswett chromatogram was (105) hlagnesol, it hydrated magnesium acid silicate produced by the Food Machinery and Chemical Corporation, Westvaco Chemical Division, South Charleston, W. Virginia. This adsorbent has been useful in the chromatography of several types of acyl sugar derivative. (106) G. R. Noggle and R. A . Bolomey, Plant Physiol., 26, 174 (1951). (107) E.V. White, J . A m . C h m . SOC.,76,4906 (1954). (108) H. Brockmann and H. Schodder, Ber., 74, 73 (1941). (109) pvl. Ulmnnn, Makrorrrol. Chem., 9,76,97 (1952); 10, 147,221 (1953); Die Sturke, 4, 73 (1952); Pharmazie, 7, 787 (1952). (110) V . V. Koval'skil, Doklady A k a d . Nauk S. S . S. R., 68, 1083 (1947); Chem. Abstracts, 44, 7902 (1950). (111) B. Siege], G. A . Candela and R. M. Howard, J . Am. Ch.e,m. SOC.,76, 1311 (1954). (112) W. S. Reich, Compt. rend., 208, 589, 748 (1939); Biochem. J . (London), 33, 1000 (1939).
80
W . W. BINKLEY
produced. A red-colored zone containing the more strongly adsorbed D-fructose ester remained near the top of the column; below it was an orange-colored interzone, and lower still on the column was a second redcolored zone containing the D-glucose ester. This classical demonstration initiated the modern, column chromatography of sugars and led to the preparation of a new and interesting series of sugar esters (see Table VIII) In addition to alumina and silica, &lagneso1106 and silicic were found to be useful adsorbents for the separation of sugar azoates; the TABLE VIII Sugar Azoates Suitable f o r Clhromatograph?/‘13 Azoafe o/
a-D-Glucose 0-D-Glucose a-D-Galactose p-D-Galactose 0-D-Fructose a-D-XylOSe (3-D-Arabinose p - L - Arabinose Sucrose a ,a-Trehalose (3,fl-Trehalose a-Lactose @-Lactose a-Gentiobiose p-Maltose &Cellobiose p-Melibiose Meleeitose
Sinfers at, “C.
265
268
.Welling poinf, “C. (mi-.)
265-266: 252-253 275-276 255-255.5 124.5-125.5 15C-157 261.5-262 262-262.5 125-125.5 134-134.5 328-329 287-288 199-204 232-233 274275 272-273 279.5-280
127-130 143-145
223
-50
436 170 -440 244 -755 755 35 210 17 320 167 62 2 105 172 188 146
selectioii of the proper developers is critical in the evaluation of these adsorbeiits (see Table IX) . l I 3 “Azoyl chromatography” of sugars, employing Magnesol, separated mono- from di-saccharides, such as D-galactose from lactose, as well as the two disaccharides, sucrose from lactose.115Application of this method, with silicic acid as the adsorbent, resolved these binary mixtures of sugar azoates: D-galactose from D-fructose, D-glucose from melezitose, and a- from P-D-galactose; and was extended successfully to ternary and quaternary (113) G. H. Coleman end C. M. McCloskey, J . Am. Chern. Soc., 66, 1588 (1943). (114) Reagent grade, Merck and Go., Rahway, N. J. (115) G. H. Coleman, 8.G. Farnhani and A . Miller, J . Am. Chem. SOC., 64, 1501 (1942).
COLUMN CHROMATOGRAPHY OF SUGARS
81
Several factors have limited the general acceptability of this method. Completely azoylated sugars may contain up to 85% of the azoyl group. In addition to the tedious procedure for the preparation of these esters, coiisiderahle quantities of the comparatively rare axoyl chloride are (YJI~sumed. Reconversion of the axoates isolated t80 the free sugars is uiiTABLE IX InJ?iiPnce of the Dciirloprr on the Bvuluatzon of Adsodents for the Separation p-o-Gltirose Penfaazonfe and p-Cellohzosr Clctanzoatc'13 Adsnrbenl
I
/)eveloper
:
(ij
Re.solulioir
Magnesol and Dicalitee Silicic acid
I,igroine/benzene/chloroforni 1/1/1 (Developer A)
Magnesol and Dicalite Silicic acid Magnesol and Dicalite Silicic acid Silicic acid
A containing 0.4% of ethanol"
Good; two zones; cellobiose zone remained a t top of column Good; two zones
A Containing 0.4% of ethanol A containing ]-lo% of ethanol
Fair; two zones Poor t o none
A containing ]-lo% of cthanol Chloroform
Poor t o none Adequate; two zones; cellobiose zone remained a t top of column Adequate; two zones; cellobiose zone remained at top of column Adequate; two zones; cellobiose zone moved down somewhat from top of column Poor
A
Silicic acid
Chloroform with 0 1% of ethanol
Silicic acid
Chloroform with 0.2% of ethanol
Silicic acid
Chloroform with more than 0 2% of ethanol
None
A diatomaceous filter-aid produced by the Dicalite Co., New York, N . Y. cent,
* Volume per
satisfactory. However, the idea of combination of a colored group with sugars to be chromatographed was fuiidameritally sound, arid some of these objections have now been eliminated. Partially methylated D-glucoses were azoylated and then chromatographed 011 alumina, using chloroform as the developer,L16and 011 silica (prepared from sodium silicate) with chloroform/beneene/petroleum ether (b. p. 65-1 10"):l / l / l . I L 7 The color (116) K. Myrback and C. 0. Tsmm, Svensk Kern. l'idskr., 63,441 (1941). (117) J. K. Mertzweiller, D . M. Carney and F. F . Farley, J . Am. Chem. Soc., 66, 2367 (1943).
82
w.
U'. %INKLEY
iiitroduced into a hexose molecule I)y a single azoyl group is suficient to permit visual observat,ion of t.he progress of t,he adsorbat>edown the column. The adsorpt>ivebonds become weaker with increase in number of azoyl groups and with devrease in number of methoxyl groups. The yields in the formation of this type of compound were 95 ?4 or better, and the recovery of resolved substances was nearly quantitative. Hardly believable though it may seem, a mixture of methylated sugars had been separated, virtually quantitat,ively, at room temperature with very simple apparatus. A new and powerful tool for the elucidation of structure of polysaccharides had been forged; but, until recently, it was rarely used. This method was further refined"* before actual use in the structural studies of polysaccharides. Partially methylated D-glucoses were reduced t o their corresponding hexitols, which were converted to their azoyl esters, thus avoiding the complication of anomers. Markedly simplified chromatograms were obtained with alumina.118- The utility of this method was demonstrated by its successful application120to the hydrolyzate from fully methylated lichenin. Elution of the zones from the chromatogram on alumina led to the recovery of 68 f 4 % of 1,4,5-tri-0-azoyl-2,3,6-tri-O-methy~-D-glucitol (denoting 1 -+ 4 links) and of 32 f 4 % of 1,3,5-tri-O-azoyl-2,4,6-tri-O-rnethyl-~glucitol (denoting 1 -+ 3 links), which confirmed the results obt,ained by a different chromatographic method.121 Another interesting variation of the chromatography of sugar derivatives containing both colorless and colored substit,uent groups is the use of the acetates of p-phenylazophenyl glycosides.lz2These compounds are prepared from interaction of poly-0-acetylglycosyl chlorides (formed more easily than methylated sugars) with p-hydroxyazobenzene (more readily obtained than azoyl chloride). They were adsorbed on silica gel'23and then developed with chloroform or benzene, or with a mixture of these solvents with petroleum ether. Typical separations were D-xylose and D-glucose; (118) R. A. Boissonnas, Erperienlia, 3,238 (1947) ;Helv. Chim. Acta, 30, 1689 (1947). (119) The alumina must be exactly neutralized with acid prior t o use as an adsorbent for sugar esters since these are decomposed by alkali. R. W . Jeanloz, H . G. Fletcher, Jr., and C. S. Hudson, J . Ant. Chern. SOC.,70,4054 (1948). H . E. Stavely and J. Fried, ibid., 71, 137 (1949). (120) R. A. Boissonnas. Helv. Chim. Acta, 30, 1703 (1947). (121) K. H . Meyer and P. Gurtler, Helv. Chim. Acta, 30, 751 (1947). (122) C. D. Hurd and R. P. Zelinski, J . Am. Chem. Soc., 69, 243 (1947). (123) The silica gel was prepared by slowly adding 200 ml. of 3 N hydrochloric acid t o a mechanically agitated solution of 100 g. of water-glass in 200 ml. of water. The gel was broken up; it was heated on a steam bath in a current of air t o remove most of the excess water and acid. The friable residue was crushed, washed with water until neutral t o litmus, and dried a t 70" for 3-4 days. It was ground t o 40-60 mesh before use.
COLUMN CHROMATOGRAPHY O F SUGARS
83
D-galactose, D-glucose, and maltose; lactose and maltose; and D-xylose, D-glucose, and lactose, all as their acetylated p-phenylazophenyl glycosides. 2. Acetate Esters
The introduction of the “brush” or “streak” technique40for the locating of zones of colorless substances added vital impetus t’o the progress of modern, column chromatography. Colorless compounds are adsorbed on columns of white or light-colored adsorbents; the chromatogram is developed, and the moist column of adsorbent is extruded. The column is then brushed or streakedI24lengthwise with a reagent sufficiently sensitive to undergo a color change in the presence of the adsorbates. The resolved substances are located; the column is cut into appropriate sect’ions,and the portion of the adsorbent wetted by the streaking reagent is removed by gentle scraping. These sections are then eluted, and the individual materials are recovered. This method has been extended to sugars and to several of their colorless derivatives. Its first and most extensive app1icat)ion has been to sugar acetates. Several factors influenced the selection of this group of compounds. They are a simple type of sugar derivative, are easily prepared, and are soluble in many organic solvents. Often, sugar acetates crystallize readily, and they are most extensively described in the sugar literature. Further, the individuality of the sugars is masked far less in the acetates than in their benzoate and azoate esters, and so sharper chromatographic separations can be expected. Directive acetylation was employed in order to encourage the formation of one anomer of each sugar; thus, each resolved sugar would be represented by a single zone. Hydrated magnesium and calcium acid silicates (MagnesolIo5and Silene EF,lo2respectively) are often suitable for sugar-acetate chromatography. The original extension of the brush technique to sugar acetates employed Magnesol,4* and this adsorbent has since been generally preferred. A major factor in the subsequent success of the method was t,he discovery, during its adaptat,ion to sugar acetates, of an unusually versatile txush or streak reagent., namely, alkaline ~erma1iganate.l~~ The chromatographic selectivity of Magnesol is notably dependent upon its moisture content.Iz6 The octaacetates of p-gentiobiose and p-maltose are st,rongly adsorbed and (124) This may he accomplished with a medicine dropper the tip of which is drawn out fine. (125) The streak reagent was prepared by dissolving 50 mg. of potassium permanganate in 5 ml. of 2.5 N sodium hydroxide. It was most, sensitive immediately after preparation. (126) M . L. Wolfrom, A . Thompson, T . T . Galkowski and E:. J. Quinn, Annl. Chcm., 24, 1670 (1952). This interesting eti’ect is not uniisual with adsorbeIit,s.
84
\V. W. BINKLEY
are only slightly separated at low moisture contents; zones of these substances are diffuse and widely separated when this adsorbent possesses over 20 % of moisture. The best results (based on the consequent separation of these acetates) mere obtained in the moisture-content range of 12-18 %. A benzeneltert-butyl alcohol (125/1) developer was used in these evaluations, and i t has frequently been preferred to benzene-ethanol mixtures because of its greater stability to alkaline ~ e r m a n g a n a t e . ~ ~
~ZZZZ-
a-Cellohexaose eicosaacetate
a-Celiotriose hendecaacetate
a-D-Glucopyranose pentaacetate FIG.12.-Chromatograms of Cellulose Acetolyzate on (a) Hydrated Magnesium Acid Silicate arid (b) Hydrated Calcium Acid Silicate. [The Chromatograms were Developed with Benzene/Ethanol (100/1) .] See Addendum, p. 94.
Mono-, di-, and tri-saccharide acet,ates were easily separated, using benzene -ethanol developer, into groups whose composition reflected the influence of molecular weight (see Fig. 12) .Iz7 The influence of t.he position of (see Fig. l3),lz6.128 and t,he loss of polarit,y of, a funct,ional group (see Fig. 14)Iz9on the adsorbabilit,y of acet,ates was interesting. Transformat,ion of t,he carbosyl group of pentn-O-acet,yl-D-gluconic acid t,o cert,ain nitro(127) E . E. Dickey and M. I,. Wolfrom, J . A m , Chptn. Soc., 71, 825 (1949); M. I,. Wolfrom and J . C . Dacons, ibl'tE.,'74, 5331 (1952); J. G . Leech, T a p p i , 36, 249 (1952). (128) W. W. Binkley and hf. 1,. Wolfrom, ./. Ant. Chevt.SOC.,68, 1720 (1946). (129) Ref. 7 ( g ) , p. 28.
85
COLUMN CHROMATOGRAPHS OF SUGARS
p-Gentiobiose octaacetate
@-Maltose octaacetate
-a-D-Glucopyranose pentaacetate a-D-Glucopyranose pentaacetate
~
~
FIG.13 -Chromatograms, on Hydrated Magnesium Acid Silicate, of (a) 01- and p-i>-Glucopyr:tnose Pentaaretntcs and of (b) p-Gentiobiose Octaacctate and p-Maltose 0 c t aa ce t at e .
pentaacetate
+ ’L’//,’///,/////:
fi-D.Glucopyranose __f
2,3,4,64etraacetate
OAC
H @.D.Glucopyranose
,’, / ,/
pentaacetate
A
c--- c‘’
’
‘OR‘
I
FIG.14.-Relationship Between the Adsorptive Strength and the Polarity of C1 of Certniri D-Glucose Acetates Chromatographed on Hydrated Magnesium Acid Silicate.
8G
W. W. BINRLET
genous and other derivatives produred, 011 a silicic acid adsorbeiit’”Owith henzenc-acetic acid developers, the following series (in order of derreasing adsorptive strength) : amides, phenylhydrazides, acids, lactones, and nitri1es.la1 Kumerous publit-atious i n the last decade have testified to the soundness of sugar-acetate chromatography on Magnesol as a separative procedure. The potentialities of the method are demonstrated in the elucidationz6, of the structure of the trisaccharide, p a n 0 ~ e . lThe ~ ~ acetate of its sugar alcohol (panitol dodecaacetate) was partially hydrolyzed and then chromatographed. The five components of the hydrolyzate were separated, recovered as crystalline materials, and properly identified. They were acetates of D-glucitol, p-D-glUCOpyranOSe, 0-isomaltose, maltitol, and unreacted panitol; thus, the structure of the trisaccharide was unequivocally established as 4-0-(cr-isomaltopyranosyl)-~-glucose.The presence in amylopectin of glycosidic 1 -+ 6 links (indicating branched structures) was proved by the chromatographic isolation of the acetates of p-isomaltose [~-O-(LY-Dglucopyranosyl)-p-~-glucose]~~~ 135 and panose.26(a) Maltotriose [4-0-(0(maltopyranosyl)-~-glucose]~~6 was found1” among the hydrolytic products of starch, and the structure of this triose was confirmed26(a)by “acetate chromatography.” Application of the method to glycogen revealed the .~~~ existence of glycosidic 1 + 6 links in this animal p o l y ~ a c c h a r i d eCrystalline aldehydo-D-galactose heptaacetate and aldehydo-D-xylose hexaacetate were separated chromatographically from the acetolyzates of guaran acetate and xylan acetate, respe~tive1y.l~~ 1
(130) A product of the Mallinckrodt Chemical Works, St. Louis, Mo. The adsorbent mixture was silicic acid/Celite No. 535 (3/1). (131) J. F. Haskins and M. J. Hogsed, J. Org. Chem., 16, 1275 (1950). (132) M. L. Wolfrom, A . Thompson and T . T . Galkowski, J . A m . Chem. SOC.,73, 4093 (1951); A . Thompson, M. L. Wolfrom and E. J . Quinn, ibid., 76, 3003 (1953). (133) S. C. Pan, A . A. Andreasen and P. Kolachov, Science, 112, 115 (1950); S. C. Pan, L. W. Nicholson and P. Kolachov, J . Am. Chem. SOC.,73,2547 (1951). (134) M. L . Wolfrom, J. T . Tyree, T. T . Galkowski and A . N . O’Neill, J . Am. Chem. Soc.,73,4927 (1951). (135) The first isolation of crystalline p-isomaltose octaacetate was accomplished by means of Magnesol chromatography; I,. W. Georges, I . L. Miller and M. L. Wolfrom, J . Am. Chem. SOC.,69, 473 (1947); M. L. Wolfrom, L. W. Georges and I. L. Miller, ibid., 71, 125 11949). (136) J. M. Sugihara and M. L. Wolfrom, J . A m . Chem. Soc., 71,3357 (1949). (137) M . L. Wolfrom, L. W. Georges, A . Thompson and I. L. Miller, J . Am. Chem. Soc .,71,2873 (1949). (138) M. L. Wolfrom, E. N. Lassettre and A. N . O’Neill, J . Am. Chem. Soc., 74, 3162 (1952). (139) R . 1,. Whistler, Eileen Heyne and J. Bachrach, J . A m . Chem. Soc., 71, 1476 (1949).
COLUMN CHROMATOGRAPHY OF SUGARS
87
In addition to its use in carbohydrate structural studies, sugar-acetate chromatography has been employed as an analytical tool. The first isolation of D-glucose and D-fructose (as their crystalline acetates) from cane juice was achieved by its aid.128D-Glucose, n-fructose, D-mannitol, and myoinositol were found in cane, blackstrap molasses (see Fig. 9),53.140 and, in addition to these, D-arabitol and erythritol in its fermentation residue (see Fig. The interconversion, by heat, of D-fructose to n-glucose was column chroreadily predicted on the basis of enolizatiori matography supplied the proof.50hiialysis of the acid reversion products of D-glucose by a combination of carbon chromatography and magnesium acid silicate chromatography led to the isolat,iori of gent.iobiose, isomaltose, maltose, cellobiose, sophorose, P ,P-trehalose, and 1evoglClcosan (all as acetates), and of the acetate of an unknown disaccharide, temporarily designated y - a ~ e t a t e . ' ~ ~ Sugar-acetate chromatography was instrumental in the preparation of crystalline, key-intermediate compounds in some intricate syntheses of sugars, such as keto-D-glycero-D-gulo-octulose heptaacetate in the format,ion of ~-glycero-~-gulo-octulose.~~~ Sugar diazo compounds have often been purified by this technique; some derivatives responding were l-deoxy-ldiazo-keto-L-galacto-heptulose p e n t a a ~ e t a t e ,1-deoxy-1 '~~ -diazo-keto-D-glyceroD-galacto-octulose h e ~ a a c e t a t e , ' ~and ~ 1-deoxy-1-diazo-keto-L-manno-heptulose p e n t a a ~ e t a t e . The ' ~ ~ anomeric a- and @-D-glucosaminepentaacetates A chromatogram of an acet'ylated, residual were resolved on Magnes01.l~~ sirup from tjhe preparation of N-methyl-L-glucosaminic acid nitrile separated the acet,ate thereof from its epimer, penta-0-acetyl-N-methyl-Lmannosaminic a~id.1~8 Magnesol chromatography was one of the keystones in the actual isolation of sucrose (as its octaacetate) after its first chemical synthesis.149This (140) W. W. Binkley, NI. Grace Blair and M. L. Wolfrom, J . A,tn. Chem. SOC., 67, 1789 (1945). L:Wolfrom I. and W. L. Lewis, J . Am. Che,m. Suc., 60, 837 (l(328). (141) & (142) A. Thompson, Kiiniko Anno, M. 1,. Wolfrom and M. Inatome, J . Atti. Chma. SOC.,76, 1309 (1951) ; the y-acetate is now known t o be 3 - O - ( o l - ~ - g l u c o p y r a n o sy l ) - ~ - ~ glucopyranose octaacetate. (143) M. L. Wolfrom and A . Thompson, J . Am. Chem. SOC., 68, 1453 (1946). (144) M. L . Wolfrom, J. M. Berkebile and A . Thompson, J . Am. Chem. SOC., 71, 2360 (1949). (145) RI. I,. Wolfrom and 1'. W. Cooper, J . A m . Chein. SOC., 72, 1345 (1950). (146) M. I,. Wolfrom and H. B. Wood, J . A m . Chem. SOC.,73, 730 (1951). 71, 2870 (147) M. I,. Wolfrom, R . U. Lemieux and S. M. Olin, J . A m . Cheni. SOC., (1949). (148) M. L. Wolfrom and A. Thompson, J . h a . Chem. SOC.,69, 1847 (1947). 76, 4118 (1953). (149) R . U. Lemieux and G. Huber, J . Am. Chem. SOC.,
88
W. W. BINKLEY
adsorbent can distinguish the difference between the glycosidic bonds of sucrose octaacetate and isosucrose octaacet at.e.150 A chromatographic series showing t.he adsorptive strength between certaiii sugar acetates and a hydrated calcium acid silicate (Silene EF) has been prepared (see Table X) .lo3 151 Silene EF is a weaker acetate adsorbent than is Magnesol; thus, the former has greater utility in the separation of the acetates of t,he larger oligosaccharides, such as those cf a-cellotetraose, a-cellopentaose, a-cellohexaose, and a-cellohept)aose (see Fig. 12).lZ7 T h e 9
TABLE
Chroriialoqraphic Adsorption Series; on Calcium Acid Silicate, of Some Sugar Derivatives'~3 J
Class I[I5 cc. of benzenelethanol (250/l)b] Raffinose hendecaacctate 8-R.lelibiose octaacetate, sucrose octaacetate 8-Maltopyranose octaacetate Class ZZ 115 cc. of benzene/ethanol (500/1)] keto-wFructose pentnacetate D-Glucitol (sorbitol) hexaacetate, D-mannitol hexaacetate 6-D-Glueopyranose pentaacetate ~ - - ~ - A m b i r i o p y m n otetraacetate se CIoss I I Z [15 cc. of benzenc/ethanol (lOOO/l)] WL-FUCO t>et,raacetate S~ Class I T T (15 cc. of benzenelethanol ( l O O / l ) ] 2,3-Di-O-methyl-~-glucose 2 , 3 ,B-Tri-O-methyl-n-glucose Class V [12.5 cc. of benzene/ethanol (250/1)] 2,3,4,6-Tetra-O-methyl-~-glucose a Arranged in decreasing order of adsorptive strength. Adsorbent: 1.8 g. of Silene EF1°2/Celite" (5/1) ( a t . ratio). Dimensions of adsorbent column: 0.9 X 11 cm. Adsorbate solution: 2 mg. in 0.5 cc. of absolute benzene or chloroform. Developer: noted in class heading. b The benzene was thiophene-free, and the ethanol was absolute.
increased adsorptive strength of this series reflected the increase in their molecular weights. 3. Methyl Ethers Nearly all polysaccharides are st,ableto alkalis, aiid often they are met,hylat,ed almost completely by ail appropriate, single t>reatment with methyl su1fat)e aiid a strong hase.15' Vsually, succwsive t'reat,ments with this or (150) W . W . I h k l e y atid hI. I,. Wolfroni, J . A v i , C h n . Soc.., 68, 2171 (1946). (151) The variability in adsorptive strength of various lots of Silene EF for sugaracetate chromatography was small, and its relatively large content of alkali and other water-soluble material was not troublesome. (152) W. N. Haworth, J . Chenr. Soc., 107, 8 (1915).
COLUMN CHROMATOGRAPHY O F STJG.4 RR
89
other methylatiiig s g e ~ l t s produce l~~ fully methylated products which may he used a.s at.art,iiig materials for t.hc elucidation of the structure of the polysaccharides. Formerly, t,licse et,heritied subst,aiices wcrc hydrolyzed, and the rt:sulting partiadly rnethylated proclucts were subjected to frart>ional distillatiou under diminished pressure. Rarely is this rectification completely satisfactory because of the inaccuracies introduced by demethylation, pyrolysis, arid incomplet,e recovery of the methylated sugars. These dificult.ies have tjo a great extent been eliminat'ed by chromat,ography. The colored azoyl esters of partially methylat,cd sugars were prepared, and the conditions established for t.heir resolution cn and silica.123Separat.ion of the methyl glycosides of 2 , 3 ,6-tri-O-met.hyl- and tetra-0-methyl-D-glucopyranose was accomplished with a flowing rhromatogram on The completely methylated hesoside moved rapidly on development with an ether-petroleum ether mixture, and was recovered in 94% yield from a portion of the column effluent,. The more strongly adsorbed methyl tri-0-methyl-P-D-glucoside appeared next in the effluent; however, development n-ith methanol was necessary in order to release the a - anomer; ~ the total yield of the anomers was 88%. These recoveries were good, but not quantitative. The magnitude of t.he repeat,ing units in rice starch and banana starch was found by this method to be 33 and 26 D-glucose residues, respectively; these values confirmed the rcsult,s obtained by fractional-distillation 156 The ariomeric met,hyl glycosides of fully methylated D-glucose and D-galactose have been partially separated on alumina, with chloroform as the developer.157The formation of anomers during the preparation of these met,hyl glycosides is a complication; often, two zones represent a single component of the original mixture. Polymerization and demethylation may also occur during glycoside formation. These object,ioris have for the most part been met by using a combination of (a) extraction with part,ition solvents and (b) chromatography on silica-water columns, for the resolution of di-, t,ri-, and tetra-0-methyl-Dg l ~ c o s e s The .~~~ Molisch reagent was used for detecting the zones in the extruded chromatograms of these substances. EssentJially the same separa1
(153) T. Purdie and J. C. Irvine, 6.Chem. Soc., 83, 1021 (1903). (154) J. K. N. Jones, J . Chern. Soc., 333 (1944). (155) E. L. Hirst and G. T. Young, J . Chem. Soc., 1471 (1939). (156) E. G . E. Hawkins, J . I<.N. Jones and G. T. Young, J . Chem. Soc., 390 (1940). (157) F. Brown and J . K. N . Jones, J . Chem. Soc., 1344 (1947). (158) D. J. Bell, J . Chem. Soc., 473 (1944); the silica-water adsorbent was prepared exactly as described by A. H. Gordon, A . J . P. Martin and R. I,. M. Synge, Biochem. J . (London), 37, 79, 86, 92 (1943); this adsorbent was employed in the elucidation of the structure of a dextran produced by Betacoccus arabinosaceotcs with a sucrose-mineral substrate [S.A . Barker, E. J. Bourne, G. T. Bruce, W. B. Neely and M. Stacey, J . Chew,. SOC., 2395 (1954)] arid of certain bacterial levans [D. J. Bell and R. Dedonder, ibin!., 2866 (1954)l.
90
W. Mi. BINKLEY
tions occur on filwous alumina, developed witJh an acetone-benzene mixture, followed by rechromatography of the effluent sugar mixture using only benzene as t,he deve10per.I~~ The combinattion procedure (extraction and silica-water column chromatography) has been extended to partially met)hylated f ruc toses . I 6o The adsorptive strengths of powdered cellulose and filter paper for methylated sugars are closely parallel. Filter-paper strip and sheet chromatography provide an essential service in the evaluation of the effluent from powdered-cellulose chromatograms. This adsorbent has recently been popular in end-group assay st,udies, more as a device for the preparation and isolat,ion of crystalline methylated sugars t.han as an analytical tool.64 Suitable micro-methods applicable to filter-paper sheets have been developed for t,he quantitative estimation of t,he resolved substances with alkaline hypoioditel6I and, colorimetrically, with aniline phthalate in methanol.162The conversion (with 85 % phosphoric acid) of powdered cellulose to hydrocellulose produces an adsorbent with a greater selectivity and capacity for methylated sugars.163Examples of the utility of this method are numerous. Equimolar quantities of 2,3,5-tri-O-methyl-, 2,3-di-Omethyl-, and 2-O-methyl-~-arabinosecrystals were obtained from a chromatogram of hydrolyzed, methylated, sugar-beet araban.64 The method was employed successfully in structural studies on : the galactomaimam of lucerne and clover fenugreek seeds,'O and certain iris seeds71;the xylans of esparto grasP4(a)and of wheat s t r a ~ 1 6 ~alginic ( ~ ) ; acidIss; Karaya gum166;degraded cherry and wood saponinsI68; an animal polysaccharide (from the intestine of horse and the rumen of sheep) which re(159) Ethelda J . Norberg, I . Auerbach and R. M. Hixon, J . Am. Chem. Soc., 67, 342 (1945). (160) D . J. Bell and Anne Palmer, Nature, 163, 846 (1949); J . Chein. Soc., 2522 (1940) ; 3763 (1952). (161) E. L. Hirst, L. Hough and J . K. N . Jones, J . Chem. Soc., 829 (1949). (162) (Miss) J. K. Bartlett, L. Hough and J . K. N. Jones, Chwnistr2/ & Industry, 76 (1951). (163) J. D . Geerdes, Bertha A. Lewis, R . Montgomery and F. Smith, Anal. Chem., 29, 264 (1954); P. A. Rebers and F. Smith, J . ,4m. Chem. Soc., 76,6097 (1954). (164) (a) S. K. Chanda, E. I,. Hirst, J . K. K.Jones and E. G. V. Percival, J . C h e m . Soc., 1289 (1950); G. 0. Aspinall, E . L. Hirst, R . W. Moody and E. G. V. Percival, ibid., 1631 (1953); (b) G. 0. Aspinall and R. S. Mahomed, i b i d . , 1731 (1954); I . Ehrenthal, R. Montgomery and F. Smith, J . Am. Chem. Soc., 76, 5509 (1954). (165) S. K. Chanda, E. L. Hirst, E. G. V. Percival and A. G. Ross, J . Ch.eni. Soc., 1833 (1952). (166) E. L. Hirst and Sonia Dunstan, J . Chem. Soc., 2332 (1953). (167) J . K. N. Jones, J . Chem. Soc., 3141 (1949). (168) R . A. Laidlaw, J . Ch,eirr. Soc., 752 (1954).
COLUMN CHROMATOGRAPHY O F SUGARS
91
sembles amyl~pectinl~g; and the di- and tri-galacturonic acids formed during the enzymic degradation of pectic a ~ i d . 1 ~ ~ The powdered-cellulose, chromatographic assay of t’he hydrolyzate of methylated stachyose has furnished important structural data.2g1 , 3 ,4, 6Tetra-0-methyl-D-fructose, 2 , 3 , 4 ,6-tetra-O-methyl-~-galactose,2,3,4-tri0-methyl-D-galactose, and 2 , 3 ,4-tri-O-methyl-~-glucose were recovered in approximat,e!y equimolar quantities; the glycosidic 1 + 6 bond for the D-glucose residue and for one of the D-galactose residues in stachyose is clearly indicated. Powdered-cellulose, chromatographic, end-group assay has also been applied to fructans. The fructose polysaccharides from rye grass (LoZium perenne)”l and leafy cocksfoot grass (Dactylis g l o r n e r a t ~ ) ’were ~ ~ shown to possess a linear structure, whereas a fructan from the rhizomes of couch grass (Triticum repens I,.) was highly branched (see Table XI).17aThe D-glucose in inulin hydrolyzates was found to originate from, D-glucose residues bound in this fructan.”* The application of this end-group assay to a glucan from barley grain revealed that it consisted of unbranched chains of P-D-glucopyranose units linked about equally 1 3 and 1 4 4,176 whereas a polysaccharide (“nigeran”) synthesized from sucrose by Aspergillus niger was found to contain linkages in the same positions, but the units were a - ~-g lu co p y ran o se.~ ~ A ~tentative structure for beechwood hemicellulose-A, a straight chain of approximately 70 &D-xylopyrnose units (I + 4) with a 4-0-methyl-~-glucopyranuronicacid residue linked (1 + 2) to every tenth unit,’” was based on a similar analysis. The first crystals of 2,3-di-O-methy1-a-~-xylose ever isolated were j
(169) G. Forsyth, E. L. Hirst and A. E. Oxford, J . Chem. SOC., 2030 (1953); G. Forsyth and E. L. Hirst, ibid.,2132 (1953). (170) J. K . N . Jones and W. W. Reid, Chemistry & Industry, 303 (1953); J . Chem. SOC., 1361 (1954). (171) R. A. Laidlaw and S.G. Reid, J . Chem. SOC.,1830 (1951); V. D. Harwood, R. A. Laidlaw and R. G . J. Telfer, ibid., 2364 (1954). (172) G . 0. Aspinall, E. 1,. Hirst, E. G. V. Percival and R. G. J. Telfer, J . Chern. SOC.,337 (1953); G. 0 . Aspinall and It. G. J. Telfer, Chemistry & Tndustry, 1244 (1952). (173) P. C. Arni and E. G. V. Percival, J . Chem. SOC.,1822 (1951). 1297 (1950). (174) E. 1,. Hirst, n. I. McGilvray and E. G. V. Percival, J . Chem. SOC., (175) G. 0. Aspinall and R. G. J. Telfer, J . Chem. SOC.,3519 (1954). (176) S. A. Barker, E. J. Bourne and M. Stacey, J . Chem. SOC., 3084 (1953); S. A. Barker and T. R. Carrington, ibid., 3588 (1953); S.A. Barker, E. J. Bourne and T. R. Carrington, i b i d . , 2125 (1954), oligosaccharides from sucrose and Aspergillus niger. (177) G . 0. Aspinall, E. 1,. Hirst a n d Ti. 8. Mahorned, J . Chem, SOC.,1734 (1954).
92
W. W. BINKLEY
formed in a chromatographically homogeneous sirup of this sugar obtained from a powdered cellulose column.178
4.Other Derivatives Aluminai79is an excellent adsorbent for the chromatography of certain sugar derivatives. It sharply differentiates between anomers, yielding crystalline substances from both sirupy arid crystallirie mixtures. Frequently, the p-D aiiomer moves down the adsorbent column more rapidly than does the (Y-D form; 1,2,4-tri-o-acctyl-G-deoxy-3-O-methyl-P-~-gluTABLE XI Powdcrcd-Cellulose Chromnlography of the Hydrolyzutes of Some Methyluted Fru~tans'~~ Per cent
0.f
coniponenl in j r u i t a n
Coinponenl
Rhizomes of couch grass
R y e grass
Inulin
-I 4 93.3 0.7a
42 11 4 39
-
-
-
-
99.8
98.0
9G
98
97 3
4 1,3,4,6-Tetra-O-methyl-~-fructose 1,3,4-Tri-0-methpl-~-fructose 93 3,4,6-Tri-0-methyl-D-fructose 3,4-Di-O-metliyl-wfructose 2.8 2 , 3, ~ , 6 - T e t r a - O - m e t h ~ ~ ~ - ~ - g 1 ~ 1 c o s eMixture of tri-0-methyl-D-glucoses
Total
2 _
_
_
_
-
2.2 3.7 100.1
~
Identification was incomplete. fructan. a
3.2
91
Concentration in hydrolgzate from untreated
cose,I8O methyl 2,~i-d~deoxy-~i-~odo-3-~-mcthy~-4-0-tosy~-~-~-g~i~opy side,'*I and met hgl2,6-dideoxy-3,4-di-0-tosyl-~-~-zyZohesopyrano8ide~~~ a,ll precede their momers. The order of deparhre, from t,he adsorbent column, of the anomers of 1 ,2,4-tri-O-acet~.l-~-dig.italose ( I ,2,4-tri-O-acetyl-6(178) S. K . Chanda, Elizabeth E. Percival and E. G. V. Percival, J . Chenz. SOC., 260 (1952). (179) The developers for thesc chromatograms on nlumina vary from petrolcum ether containing small amouiits of beneciie up t o 100% benzene (for \venkly held substances) nnd from benzerie w i t h smitll t o large porportions of ethyl ether, ethanol, or acetour for modrrately and st,rongly held :tdsorbxtev. (180) E. Vischer and T . Iieiclistriii, Ele[ci. Chi,w. 4 c t u , 27, 1332 (1944). (181) €1. Haueristeiri and T. Reichsteiii, Hclzi. Chi,t>.Actu, 33, 446 (1050). (182) H. R. Bolliger and T. Reichstein, Helv. Chiin. A c f a , 36, 302 (1953).
COLUMN CHROMATOGRAPHY OF SUGARS
93
deoxy-3-O-methyl-~-galactose)is, however, the reverse.183 Alumina is capable of sorting out the derivatives of certain hexoses according to the number of acyl groups they possess. The order of appearance in the effluent of a flowing chromatogram was methyl 4,6-O-benzylidene-2,3-di-O-tosy1-cr-~-g1ucopyranoside followed by methyl 4 ,6-0-benzylidene-2-0-tosyl-a-~-glucopyranosideIs4;that is, the adsorpt’ive bond between the free hydroxyl group (at C3) and alumina was greater than that from its more massive replacement, the tosyl ester. This adsorbent was discerning with respect to t’he position of the tosyl group, setting apart the preceding 2-0-tosyl derivative from its isomer having the acyloxy group at C3.lS4Alumina differentiated between such anhydrohexosides as methyl 2,3-anhydro-4,6-0-benzylidenea-D-talopyranoside and methyl 2,3-anhydro-4,6-O-benzylidene-c~-~-gulop y r a n o ~ i d e . ’Anhydro ~~ and deoxy compounds are readily separated, from for example, methyl 2,3-anhydro-4,6-di-O-tosyl-a-~-allopyranoside Outstanding success methyl 2-deoxy-4,6-di-O-tosyl-a-~-allopyranoside.~~~ has been experienced in the isolation of crystalline sugar derivatives from crystalline and sirupy mixt,ures.I87 Naturally occurring flavonoid glycosides form anot’her interesting group of sugar derivatives responsive to column chromatography. These glycosides, dissolved in acetone, were adsorbed on Magnesolio6and were resolved by development with aqueous ethyl acetate. Five glycosylquercetins (3,3’,4’, 5,7-pentahydroxyflavone) have been isolated from huckleberry (Vuccinium my~tiZZus)leaves,’**and quercitriii was obtained from a crude extract of oak bark (“lemonflavin”) .189 Cert,ain prepared mixt’ures of flavonoid substances were separable on Magnesol; they included quercitrin(183) C. Tamm, Helv. Chim. Acta, 32, 163 (1949). (184) H . R. Bolliger and D. A. Prins, H e h . Chini. Acta, 28, 465 (1945); 29, 1116 (1046); F. Reber and T. Reichstein, ibid., 28, 1164 (1945); for benzoyl esters, M. Gyr and T. Reichstein, ibid., 28, 226 (1945). (185) H. Huber and T. Rcichstein, Helv. Chhinz. Arta, 31, 1645 (1948). (186) H. R . Bolligcr and M. Thiirkauf, Helv. Chim. Acta, 36, 1426 (1952). (187) C. A. Grob and D. A . Prins, Helv. C h h . Acta, 28, 840 (1945); D. A . Prins, ibid., 29, 1 (1046); A . S. Meyer and T. Reiehstein, ibid.,29,152 (1946); F. Reber and T . Reichstein, ibid., 29, 343 (1946); R.W. Jeanloz, D. A . Prins and T. Reichstein, i b i d . , 29, 371 (1946); A . C. Maehly and T . Reichstein, ibid., 30, 496 (1947); M. Gut, D. A . Prins and T . Reichstein, ibid., 30, 743 (1947) ; C. Tamm and T . Reichstein, ibid., 31, 1630 (1948); F. Blindenbacher and T. Reichstein, ibid., 31, 1669 (1948); employing silicic acid as the adsorbent, R . W. Jeanloz, J . A m . Chent. Soe., 76, 555, 558 (1954); 1’. J. Stoffjrn and R. W. Jeanloz, i b i d . , 76, 561,563 (1954). (188) C. H.Ice arid S. H. Wender, J . An,. C‘hvu. Soc., 76, 50 (1953). (189) C. H . Ice a r i d S . H. Wender, Arch. Biochcm.a n d Bioph?/s.,38, 185 (1952). This adsorbent was successfully employed in similar assays on grapes [B. I,. Wil1i:tms and S. H. Weiider, J . A m . Clte7tr. Sor., 74, 4372 (1952)], apricots [Arch. Biochwti. cvnd R i o p h y s . , 43,310 (1953)], and black currants [ U . I,. Williams, C. H. Ice and S. H. Wcader, J . Am. Chem. Xoc., 74, 4566 (1952)l.
94
W. W . BINKLEY
rutin-quercetrin, xanthorhamnin-quercitrin-quercetin, and xanthorhamnin-rutin-quercitrin.'90 Magnesol chromatography proved essential for successful isolation of the synthetic isoquercitrin from the reaction, in liquid ammonia, between 2,3,4,6-tetra-O-acetyl-~-ghcosyl bromide and the monopotassium salt of quercetrin.lgl This method was applied successfully in the preparation of the rutin, hesperidin, and naringin glucosides from their rhamnoglucosides.192A glucofrangulin (as an octaacetate) was obtained from an extract of Rhamnus frangula bark by flowing chromatographylg3on Floridin XXX43and silicic acid. A similar investigation of the glycosides in certain seeds has been made, employing aluminalg4 l g 6 and magnesium silicate/Celite (2/l).1g4 Amides are useful sugar derivatives, and some aldonamides have responded to chromatography on calcium acid silicate, using aqueous dioxane as the developer (Table VI).lg6 Some degree of resolution of uronic acids and of their low molecularweight polymers has been achieved.66,l o o 9
(190) C. H. Ice and S. H. Wender, Anal. Chem., 24, 1616 (1952). (191) C . H. Ice and S. H. Wender, J . A m . Chem. Soc., 74,4606 (1952). (192) D. W. Fox, W. L. Savage and S. H . Wender, J . A m . Chem. Soc., 76, 2504 (1953). (193) E. Seebeck and 0. Schindler, Helv. Chim. Acta, 29, 317 (1946). (194) J. C. Hess, A . Hunger and T. Reichstein, Helv. Chim.Acta, 36, 2202 (1952). (195) 0.Schindler and T. Reichstein, Helv. Chim. Acta, 36, 921,1007 (1953). (196) M. L. Wolfroni, R. S. Bower and C . G. Maher, J . A m . Chent. Sac., 73, 875 (1951).
ADDENDUM In Fig. 12 (p. 841, the middle zone of column a is an artifact and represents a non-cellulosic disaccharide from the wood cellulose (filter paper) employed. It was absent when cotton cellulose was used.
GLYCOSYLAMINES
BY G. P. ELLISAND JOHNHONEYMAN* Deparfnienl o j C h e n i i s f r y , King’s (’ollegr, I ’ n i z ~ s i f oy j London, EnglantI
CONTENTS I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Physical Properties.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Mutarotation and Hydrolysis.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Methylation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 . Acetylation.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Benzoylation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. T r i t y l a t i o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. Periodate Oxidation.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8. Conclusion.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Diglgcosylamiiies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Diamide Derivatives of Aldoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. TJses of Glycosylamines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX. Tables of Properties of Glycosylamines.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95 96 97 101 102 102 104 109 110 115 116 116 120 120 121 124 125
I . INTRODUCTION
A large number of important, naturally occurring, carbohydrate compounds contain nitrogen. These include nucleic acids, coenzymes, polysaccharides, some virus components, and some of the members of the vitamin B group. Furthermore, compounds made up by union of proteins with carbohydrates are of possible wide occurrence and major biological importance. Reducing sugars react readily, under mild conditions, with compounds containing the primary amino group, notably with ammonia, hydroxylaniine, and a wide variety of amines, hydrazines, and amino acids. A number of second’ary amines give similar products, although with many of them reaction is more sluggish. In the main, the present article will be limited to discussion of compounds resulting from the condensation of one molecule of a reducing sugar with one molecule of ammonia or of a primary or secondary aliphatic, alicylic. or aromatic amine. Although, in recent years, con-
* Harold Hibbert Memorial Fellow a t McGill University, Montreal, Canada, during the period when this article was completed. 95
96
G. P. E L L I S AND J O H N HONEYMAN
siderahle progress has been made in determining the structure, react ions, and importance of these compounds, much reinnins to he done and worb is in progress in many laboratories. For this reason, this review will I P a report on selected parts of thc wide field. Information about nucleosidcs, nucleotides, and nucleic acids is contained in other reviews in this Series.' Derivatives of amino acids will not be discussed. Compounds containing two amide groups attached through their respective nitrogen atoms to C 1 of a monosaccharide will be mentioned incidentally. Since the special nature and properties of the products obtained by reaction of the sugars with o-phenylenedisniine have been reviewed recently by IZicht,myer,? these, also, will not be considered here. In addition, one very important reaction, known as the Aniadori rearrangeinelit, is given separate treatment in an accompanying Chapter.2a The importance, in the non-enzymic browning of foods, of compounds of glycosylamine nature is also outside the scope of the present artirle, but readers interested in the topic are referred to a recent review hy Danehy and Pig~man.~
11. NOMENCLATURE Three representatives of the kind of compound here considered result from the coiidensation of D-glUCOSe with aniline, or ammonia, or piperidine in equimolar amounts. That obtained from D-glucose arid aniline, and assumed for the present to be I, has been called D-glucose anilide, aniline D-glucoside, aniline N-D-glucoside, and D-glucosidoatiilinc, but none of these names can be justified on systematic grounds. The nomenc~lature adopted here is that agreed on between British and American committee^.^ Compound 11, prepared from D-glucose and ammonia, is named D-glUCOSy1amine, because it is ammonia with one of its hydrogen atoms replaced by
I
CHNII,