Analyst, July, 1968, Vol. 93, Pfi. 453455 453 Determination of the Number of Oxygen Substituents of Steroids by Chromatography BY D. J. H. TRAFFORD AND R. W. H. EDWARDS (Institute of Child Health, University of London, 30 Guilford Street, London, W.C. 1 ) ARm values obtained by difference of Rm values in the presence and absence of formaldehyde are shown to group in a manner determined by the number of polar functional groups and, to a lesser degree, by the nature of the steroid skeleton. It is proposed that determination of the AR, value provides a means of characterising steroids from natural sources. PARTITION chromatography gives ARm values that are constant for each substituting func- tional group in a particular family of solvent mixtures (reviewed by Bush1 and Edwards2). The present observations stem from the use of formaldehyde in the stationary phase in an attempt to cause strong association with hydroxylic and ketonic substituents, and thus to alter the AR, values.Paper chromatography of steroids with mixtures of aqueous formaldehyde, light petro- leum and benzene gave excessive tailing of the chromatographed spots. Incorporation of methanol eliminated tailing, and the mixture effectively became a modification of the Bush3 system. EXPERIMENTAL The general apparatus, methods, reagents and details of steroid chromatography are described elsewhere2 and only special points will be elaborated. REFERENCE STEROIDS- These were generally purchased from Steraloids Ltd., Croydon, Surrey, and were checked for correctness of melting-point, chromatographic properties and chemical reactions.Other steroids were kindly provided by Professor W. Klyne from the Medical Research Council reference steroid collection. CONDITIONS OF CHROMATOGRAPHY- Whatman No. 2 paper was used at 37" C, with 3 hours' equilibration. The solvent front was allowed to proceed 40 cm by descent, and steroids were located by chemical and physical procedures. SOLVENT MIXTURES- The stationary phase consisted of a mixture of methanol and saturated aqueous formalde- hyde (1 + 1). Tailing occurred at lower concentrations of methanol and higher concen- trations lead to monophasic conditions. The mobile phase consisted of light petroleum and benzene mixtures. As the ARm values for replacement of light petroleum by benzene were proportional to the benzene concentration, as reported for the Zaffaroni systems4 and Bush systemsI3 it is not necessary to define precise solvent mixtures.Rmo for the family is defined as the Rm value determined for the mobile phase of light petroleum alone. For precision, observations were made within the limits +O-6 to -0.6 Rm units. In the Bush systems3 the benzene concentration factor was 0-0195 Rm units. It was found that R,, = observed Rm + 0.026 (benzene concentration, per cent. v/v, when preparing), 0 SAC and the authors.454 TRAFFORD AND EDWARDS: DETERMINATION OF THE NUMBER OF [ArtahySt, VOl. 93 RESULTS The Rmo values obtained by chromatography of ninety-eight steroids are summarised in Tables I and 11. TABLE I KEY STEROID Rmo VALUES Steroid* P4-3 : 20-one 5 /3P-3a-ol-20-one P5-3 /3-01-20-0ne 5aA-3a-ol-17-one 5 /3A-3a-ol-l 7-one As-3 8-01-1 7-one 5PP-3a : 17a : 2Oa-01 P4-3 : 11 : 20-one-17a: 21-01 P4-3 : 20-one-l1/3 : 17a: 21-01 Trivial name Progesterone Pregnanolone Pregnenolone Androsterone Aetiocholanolone Dehydro-epi-androsterone Pregnanetriol Cortisone Cortisol Rmo t - 0.45 - 0.37 -0.18 + 0-02 + 0.27 + 0.37 + 1-34 + 2.25 -j- 2-54 * Bush1 abbreviated nomenclature.t R,, = Rm of substance in methanol - aqueous formaldehyde - light petroleum mixture (1 + 1 + 2), obtained by calculation if necessary; see text. TABLE I1 MEAN AR, VALUES IN FORMALDEHYDE SYSTEMS Substituent ARm 1/3-01 . . .. . . 1.20 la-ol .. .. . . 1-40 2a-01 .. .. . . 0.81 6a-01 .. .. . . 1.81 6/3-01 . . .. . . 1.44 7-one .. .. . . 1.19 701-01 .... . . 1.84 ll-one . . .. . . 0.91 lla-ol . . .. . . 1-52 1201-01 . . .. . . 1-49 16/3-01 . . .. . . 2.03 17-one . . .. . . 1.65 17a-01 (P) . . .. . . 1.03 20-one . . . . . . 0.89 21-01 .. . I . . 1.04 11p-01 . . .. . . 1.09 CH, . . .. .. . . 0.22 5a-+5/3 . . .. . . 0.12 3a-ol+ 3/3-01 . . . . 0.29 %one + 3/3-01 . . . . 0.58 4-ene-3-one + 3a-ol-5p . . 0.24 1 l-one +- 11 /?-01 . . . . 0.26 20-one + 2OP-01 . . . . 0.33 20-one +- 20a-01 . . . . 0.57 Number of examples 3 1 1 1 2 1 1 8 3 12 1 3 1 9 1 6 4 9 2 2 10 7 5 5 &Standard deviation 0.15 - - 0.17 0.16 0.15 - - 0.07 0.07 0.14 0.15 DISCUSSION On plotting the steroid Rmo values obtained with and without formaldehyde, the points were found to group about three parallel lines. Each group shared a common number of functional groups, being di-, tri- and tetra-, or more, oxygenated.An alternative method of plotting is summarised in Fig. 1 , where the plot of ARm value caused by the formaldehyde is plotted against the number of polar functional groups. In this figure the observations are plotted as the mean value of each group and the standard error. It is clear that the observed values cluster in a manner useful in characterising steroids from natural sources into one of the following groups: tetra-, or more, oxygenated; tri-oxy- genated, with indications at the upper and lower ends of androstane or pregnane skeletons; di-oxygenated androstane or pregnane derivatives, with no ambiguity.July, 1968] OXYGEN SUBSTITUENTS OF STEROIDS BY CHROMATOGRAPHY 455 Q Number of polar functional groups Fig.1. Correlation of ARm when formaldehyde is incor- porated with several polar functional groups. The mean, number of steroids in each group, standard error and the prob- abilities of significance of differences are indicated. Groups marked A are androstane derivatives and P, pregnane The above characterisation of steroids by number of functional groups was carried out initially on simple hydroxylic and ketonic steroids without vicinal effects. On extension to vicinally substituted steroids no marked divergence was found for 2,3-, 16,17-, 17,220- and 20,21-di-substitution, provided that the substances were not penta-substituted. Double bonds in isolation, or in conjugation with carbonyl groups, had no effect, and the acetoxy group behaved as &-functional.A further examination of the results was carried out separately in the androstane and pregnane groups by multiple regression to distinguish between the effects of hydroxyl groups and ketones and assess each standard error. The pairs of values were not significantly different, providing justification for the empirical procedure used above, in which the A& effects of hydroxyl and ketone groups were assumed to be equal. The present procedure has been applied in the characterisation of substances from the urine of a new-born child5 and indicated that certain oxidation products were tri-oxygenated androstane derivatives. This was in agreement with other observations on these substances. Although the present study is limited to steroids it may be presumed that similar results would be obtained with other groups of substances that do not react chemically with formaldehyde. REFERENCES 1. 2. 3. 4. 5 . Bush, I. E., “The Chromatography of Steroids,” Pergamon Press Ltd., Oxford, London, New York Edwards, R. W. H., in Smith, I., Editor, “Chromatographic and Electrophoretic Techniques,” Bush, I. E., Biochem. J., 1952, 50, 370. Kasabakalian, P., and Basch, A., Analyt. Chem., 1960, 32, 459. Edwards, R. W. H., and Trafford, D. J. H., Biochem. J., 1968, 108, 185. and Paris, 1961. Third Edition, Volume 1, William Heinemann Ltd., Publishers, London, 1968. Received October 31sl, 1967