Paper Chromatography of Steroids A Systematic Approach Using a Generalized
RA
Function
PETER KABASAKALIAN and ALVIN BASCH Chemical Research and Development Division, Schering Corp., Bloomfield, N. 1.
b A series of Zaffaroni-type solvent systems has been chosen which encompasses the entire range of steroidal polarities commonly encountered. Based on calculations from values given in this paper, the practical chromatographic range of each solvent system can be quantitatively described. The mobility of a steroid in each system can be estimated by inspection of its structure, thus permitting the choice of a suitable solvent system without literature references or experimentation. The method is developed in detail for the pregnane series and illustrated by application to a related series of compounds. The method is general and can be extended to include other solvent systems and steroidal types.
ment of the chromatogram. The Bush systems require careful and sometimes prolonged equilibration of solvents and paper. Consequently, they are difficult to maintain (with regard to tank equilibrium), are strongly temperature-dependent, and often show strong adsorption (tailing) ( 7 ) . For these reasons, they were considered unsuited. The third requisite necessitates the use of some function which will predict the mobility of the compound in question. Martin (8) has shown that the change in free energy accompanying the transfer of a mole of substance A from a mobile phase, M, to a stationary phase, S, under equilibrium conditions is: A ~ = A RT In ( N Y I N : ) (1) where
P
chromatography has been extensively used to study steroid biochemistry and steroidal transformations induced microbiologically. I n these cases the prime problem is the identification of unknown products. The prevalence of large numbers of solvent systems to chromatograph the same compound is very helpful for identification purposes. I n contrast, this laboratory has been interested in chromatographing known compounds to determine as simply as possible whether chemically induced steroidal transformations resulted in particular products, and their degree of purity. Besides giving good resolution, the series of solvent systems used should be easily maintained, require short development times (2 to 6 hours), and be interrelated. so that a method could be developed which would enable a choice of the appropriate solvent system without resort to literature references. The first two requisites are amply fulfilled by the Zaffaroni type (3, 14) of system, which involves the use of a stationary and relatively nonvolatile polar solvent with a volatile nonpolar solvent as the mobile phase. The alternatively available Bush systems (4) combine relatively volatile polar and nonpolar solvents, with the polar portion being preferentially absorbed onto the paper before or during the developAPER
458
ANALYTICAL CHEMISTRY
N A
is the mole fraction of A
( N Y / N j ) is the partition coefficient a
Therefore In CY = A ~ A / R T (2) Martin further showed that A p A may be regarded as being made up of the sum of the free energies of the various groups (X, Y, Z) of which molecule A is composed.
4- ~ A P Yi- CAW 4- . . . . (3) Thus, the addition of a group X
in steroid paper chromatography arid have derived results useful in conformational analysis of sapogenins. Instead of restricting the RMfunction to one solvent system a t a time, it has been extended to cover other solvent systems simultaneously by a normalization technique. The extrapolation of the R M function of a standard solvent system beyond its normally practical range without the use of reference compounds has been attempted, fol] loned by the assignment of [ R M ranges ( [ R M ] = R M normalized to a standard system) for other solvent systems based on the standard solvent system. This would enable the direct calculation of the R F value of a steroid in a number of solvent systems from its structure. To determine the utility of this approach-i.e., the quantitative interrelation of chromatographic solvent systems and the prediction of chromatographic mobilities of steroids-both the calculated and experimental [ E M ] values in suitable solvent systems for a related series of compounds were obtained. The series used was the laboratory synthesis of dexamethasone, 9afluoro-llp,17a,21-trihydroxy- 1 6 a - m e -
thyl-l,4-pregnadiene-3,20-dione.
A ~ = A aApx
changes the partition coefficient by a given factor depending on the nature of the group and on the pair of phases employed, but not on the rest of the molecule. He also derived (6) the relationship between the partition coefficient, CY, and R F as a =
K(~/RF 1)
(4)
K being the ratio of the volumes of mobile and stationary phase in the chromatogram. Based on these concepts Bate-Smith and Westall ( 1 ) defined a new function, RM,as RM = log ( ~ / R F 1)
(5)
and showed that ARM values for the components of a molecule contribute additively to its chromatographic mobility (the RM values being a function of the free energy of the compound). Brooks, Hunt, Long, and Mooney (2) have shown the validity of this concept
EXPERIMENTAL
Seven solvent systems previously reported (9) were chosen (Table 111) and run in the descending manner on Whatman No. 1 paper in a constant temperature room (22' C.) Kith the following modifications. I n the propylene glycol systems, %yo propylene glycol in methanol was used as the stationary phase. Methyl Cellosolve was used undiluted, while phenyl Cellosolve was used as an 18y0v./v. solution in acetone. The solvents used were standard reagent grade with the exception of formamide, where only the stabilized reagent grade of the Fisher Scientific Go. proved consistently suitable. Toluene-propylene glycol was chosen as the reference solvent system. A suitable reference standard was run with each chromatogram to ensure that the system was functioning properly. Using Equation 5, ARM values for the common steroidal substituents were determined directly by chromatographing compounds with and without the desired substituent in the reference
solvent system. ARM values obtained for hydroxyls and ketones in different positions were separated into more polar and less polar groupings and each grouping was averaged independently. Using these values, ZAR, values were calculated for 15 steroids having a pregnane nucleus. RM values for these compounds in the reference system were experimentally obtained. The ARtf value for the pregnane nucleus was obtained from the intercept of a plot of ZARMus. RMof the above-mentioned steroids. Using the ARw values of substituents and the pregnane nucleus, the [R.u]values of a wide range of calibrating steroids were calculated using Equation 6; then their experimental R,vIvalues in the remaining six solvent systems were determined.
The following order of decreasing polarity was found (Table I) for the functional groups studied : hydroxyl > conjugated carbonyl > carbonyl > acetate > 16-methyl. This order is similar to that reported by previous workers (9),who have explained these differences in terms of hydrogen bonding ability, steric factors, and intramolecular interactions. The AR, values of substituents determined in the reference toluenepropylene glycol system (Table I) are in reasonable agreement with the ARM values of Brooks et al. (2) as calculated from Savard's data (12, IS) and a-ith values calculated from the data of Reineke (11) and support the
[RM= ] ZARM (6) A plot of the calculated [R.w]us. RM experimentally determined was made for each of the six solvent systems from which the constants a and 6 indicated in Equation 7 were determined for each solvent system.
Table 1. ARM Values of Substituents Determined in Reference ToluenePropylene Glycol System
+
[ R M ]= 0: RM b (7) The useful chromatographic [ R Mrange ] for each solvent system was calculated using Equation 7 and R,u limits corresponding to 0.05 and 0.60 R Funit. The RF values of the evaluation steroids (Table IV) were calculated using Equations 5 , 6, and 7. RESULTS
Substituent Hvdroxvl " "
statement of Bate-Smith and Westall (1) that R.u values and ratios are more nearly constant than RF values. The effects of positional isomerism and neighboring interactions of the more polar substituents were treated by calculating ARM values for each substituent in different positions, separating the values into more polar and less polar groupings, and averaging the values for each grouping independently. The average ARM values so obtained seldom differed by more than 15% from the absolute values. This approximation is adequate for this purpose. These values are obtained from the simplest available model compounds and may be somewhat modified in complicated molecules. An illustration of this may be found in Table I, where the ARM values of 16methyl substituents vary with the nature of the substituent a t position 17. When unusually strong interactions appear, the results are even more dramatic. For example, the 16a-hydroxyl is a very polar group, yet with a 17a-hydroxyl, its ARM value is reduced to +0.66 (from +1.56). Similar observations have been made for the 14a,l7a-dihydroxy grouping (6). These exceptions are infrequent and when observed may provide a clue to structure. The RM of a steroid of the pregnane series could be calculated when the ARM value of the pregnane nucleus was determined from the linearly ob-
Positions ARM 3a. 38.6a. 6 8 . 1 1 ~ ~ . i5& 16; ' $1.56 llp, 170, 21 $1.04 3 f1.36 3 fl.06 3 +O. 76 11,20 +O .55 3 $0.66 21 $0.44 "
Hydroxyl A1v4-3-one A4-3-one Carbonyl Carbonyl Acetate Acetate Methyl (l7a-H) Methyl ( 17a-OH ) Methyl (17a-OH)
16a, 166
-0.21
16R
-0.25
16a
-0.42
'
The AR,w values obtained for the various functional groups commonly found in steroids as determined in the reference toluene-propylene glycol system are listed in Table I. The compounds used to determine the ARM of the pregnane nucleus experimentally are listed in Table 11, which also contains ZARM values calculated from the data in Table I and the experimentally determined R.w valum. The ARMvalue for the pregnane nucleus was found to be -3.00. The slopes and intercepts obtained as a result of plotting the calculated [R,tf]us. the experimental RMvalues of the calibration compounds in the seven solvent systems studied are listed in Table I11 together with their useful [R.tf] ranges (Figure 1). The predicted and experimental [ E M ] values for the chemically related series of compounds together with their calculated and experimental Rp values are listed in Table IV. DISCUSSION
The contribution of any given functional group to the chromatographic mobility of a molecule will depend mainly upon three factors: the nature of the substituent (usually the predominant factor), the position of the substituent with respect to the rest of the molecule, and the nature of the chromatographic solvent system.
-
Table
II.
Calibration Data for ARM of Pregnane Nucleus in Reference ToluenePropylene Glycol System
Compound 38,17a, 21-Trihydroxy-5a,6~-dichloro-l6~-methylallopregnan-20one 21-acetate Qa,11~-Dichloro-17~,21-dihydroxy-1,4-pregnadiene-3,2Odione Qa,118-Dichloro-l7a, 21-dihydroxy-1,4-pregnadiene-3~2O-dione 21-acetate 17a-Hydroxy-4,Q ( 1l)-pregnadiene-3,2O-dione 17~~,21-Dihydroxy-4,9( 1l)-pregnadiene-3,20-dione 9a-Fluoro-llp,21-dihydroxy-4,16-pregnadiene-3,20-dione21-acetate 16p-Methy1-17a,21-dihydroxypregnane-3,11,20-trione 17a-Hydroxy-1,4,9 ( 1l)-pregnatriene-3,20-dione 17a,21-Dihydroxy-1,4,9 ( 1l)-pregnatriene-3,2O-dione2 1-acetate
2 l-Hydroxy-4-pregnene-3,20-dione 17a,21-Dihydroxy-4-pregnene-3,20-dione 16~-Methyl-17a,2l-dihydroxy-4pregnene-3,2O-dione 16a-Methyl-17a,21-dihydroxy-4-pregnene-3,2O-dione 17a,2l-Dihydroxy-4-pregnene-3,11,20-trione 3&21-Dihydroxy-5-pregnen-20-one 4
ZARx"
RM
3.17 3.99
0.14 0.95 .. .. 0.05
3.39
2.65 3.69 3.09 3.79 2.95 3.39 2.65 3.69 3.44 3.27 4.24 3.15
-0.23
0.79
0.07 0.87
0.12 0.18 -0.25 0.72 0.45
0.29 1.20 0.33
ZAR, doea not include pregnane nucleus. Table 111.
Useful [ R x ] Ranges for Steroidal Solvent Systems
Solvent System 1. Chloroform-formamide 2. Benzene-formamide 3. Toluene-propylene glycol 4. Cyclohexene-formamide 5. Ligroine-propylene glycol 6. Heptane-methyl Cellosolve 7. Heptanephenyl Cellosolve a* [ R M ]= a RM b. 0 Based on 0.05 to 0.60 Rp range.
+
Slope4 1.0 1.0 1.0 1.0
0.76 1.0 1.0
Interceptb 1.26 0.47 0.00 -0.60 -1.28 -1.62 -2.16
[ R M ]Rangec +2.54 to f 1 . 0 8 f1.75 to +0.29 +1.28 t o -0.18 +0.68 -0.78 -0.31 to -1.42 -0.34 to -1.80 -0.88 to -2.34
VOL. 32, NO. 4, APRIL 1960
0
459
tained plot of the ZAR, (which did not include the pregnane nucleus) us. the experimentally determined R,w values of steroids in the reference toluenepropylene glycol system. Thus the chromatographic mobility of steroids in the reference toluene-propylene glycol system could be predicted using Equation 6. The R.lf function of the reference toluene-propylene glycol system is symbolized by [R.w]. The graphical plots of predicted [R.$f] us. experimental RM values of the calibration compounds were linear in all seven solvent systems. Only in the ligroine-propylene glycol system was the slope different from 1.0 (Table 111). This linear relationship enables one to extend the prediction of chromatographic mobility to all seven solvent systems using Equation 7. The relative polarities of the solvent systems in Table 111 are listed in decreasing sequence as indicated by their intercepts. There was some scatter in the points of the calibration plots. The scatter of points increased the more removed polaritywise the solvent system was from the reference solvent system. The scatter of points in these plots may be due to a number of simplifications which were introduced as a result of the primary purpose of this study-i.e., to be able t o choose suitable solvent systems rather than predict exact R F values for steroids under study. Substituents or spatial configurations which contribute relatively little to mobilities were neglected; for although they often provide the basis for chromatographic separation, they will seldom influence the choice
Table IV.
of the system. Thus isolated double in these systems causes little difficulty, as in every case the stationary phase is bonds, epoxides, halogens, and Cg isomerism have been omitted from virtually insoluble in the mobile phase. The variations that do occur will not these calculations. Of course, the less affect direct AR.w comparisons between substituted a molecule is, the greater the effect of these groupings, as will two compounds on the same chromabe seen from Table IV [Compare 1613togram, nor will they invalidate the choice of the system. The use of methyl-l7(20)-allopregnene - 3p,20p- diol 3,20-diacetate, 16a-methyl-licr,20craverage AR.M values and the unaccountability of all possible interactions epoxyallopregnane-3/3,2O,B diol 3,20-diacetate, and 16a-methyl-lia.2la-dialso contribute to the loss of accuracy. hydroxy-1,4-pregnadiene-3,20-dione 21Thus, although the chromatographic acetate, 9~,1lp-epoxy-16a-mcthyl-l7~~mobility of a steroid in the seven solvent systems could be calculated, these 21 - dihydroxy - 1,4 - pregnadienr - 3.20 cannot be used to confirm the structure dione 2 1-acet a t e]. Another limitation to the accuracJof a compound uniquely. In practice, achievable lies in the difficulty of reif the R p value of a steroid nith a proposed structure differs by more producing Rp values. By norking in a constant temperature area, vigorously than 0.2 Rr unit from its calculated standardizing the procedure for inivalue in a specific solvent systcm, its structure should be considercd suspect. pregnating the paper, and using ne11 equilibrated tanks, Rr values constant The practical [ R I f ]ranges listed in within 0.05 RF unit have been obtained. Table 111 for the seven solvent systems The composition of the mobile pha>e werr limited by R w values corre-
1. Practical ranges of the seven Zaffaroni-type chromatographic solvent systems listed in Table 111 Figure
[RM]
4
[RM]
2.5
2.0
1.5
1.0
0.5
0
0.5
and
RF
Solvent Calcd. System [R.v] 2.13 lla,17~,21-Trihydroxy-16~-methy1-1,4-pregnadiene-3,20-dione 1 2 1.53 lla,17~,21-Trihydroxy-l6~-methyl-1,4-pregnadiene-3,2O-dione 21-acetate 1.61 1 9a-Fluoro-llp,17oc,21-trihydroxy-l6~-methyl-1,4-pregnadiene-3,2O-dione 9a-Bromo-ll0,l ia,21-t rihydroxy-16a-methyl1,4-pregnadiene-3,20-dione 2 1.01 21-acetate 9~-Fluoro-l18,17~,21-trihydroxy-l6a-methyl-1,4-pregnadiene-3,2O-dione 2 1.01 21-acetate 0.17 3 3p. 17~,21-Trihydroxy-l6a-methylallopregnan-2O-one 21-acetate -0.03 3 16~-Rlethyl-17~,21-dihydroxy-l,4-pregnadiene-3,20-dione 21-acetate -0.03 16a-Riethyl-17oc,21-dihydroxy-1,4,9( 1 l)-pregnatriene-3,2O-dione21-acetate 3 9~,11p-Epoxy-16~-methyl-l7~,21-dih~droxy-l,~-pregnadiene-3,2O-dione 3 -0.03 21-acetate 5 -0.89 3p-Hydroxy-5,16-pregnadien-20-one 6 -0.89 3p-Hydroxy-5,16-pregnadien-20-one -0.27 5 3~,17~-Dihydrox~-16~-methyl-2l-bromoallopregnan-2O-one -0.63 5 16a-RIeth~l-17~,21-dihydroxyallopregnane-3,2O-dione 2 1-acetate 2~,4e-Dibromo-l6~-methyl-l7a,2 l-dihydroxyallopregnane-3,20-dione 5 -0.63 21-acetate 6 -1.89 22a-5-Spirosten-3p-01 (Diosgenin) -1.89 7 22a-5-Spirosten-3p-01 (Diosgenin) 6 -1.79 3p-Hydroxy-5,16-pregnadiene-20-one 3-acetate -1.31 6 3pHydroxy-16a-methyl-5-pregnen-20-one 7 -1.31 3~-Hydroxy-l6~-methyl-5-pregnen-20-one -1.31 6 3p-Hydroxy-16a-methylallopregnan-2O-one c -1.31 3~-H~droxy-16a-methylallopregnan-2O-one -2.22 I 16p-Methyl-17(20)-allopregnene-38,2O~-diol3,2O-diacetate r -2.22 16a-hIethyl-17a,20au-epoxya1lopregnane-3~~ 200-diol3,20-diacetate r
ANALYTICAL CHEMISTRY
1.15
2.0
Values of Steroids Used in Evaluation Study
Compound
460
1.0
[R.d
Exptl. Calcd. [ R M ] [R.vlc - [RML RF 1.79 0.34 0.12 0.27 0.08 1.26 0.31 1.96 -0.35
Exptl. RF 0.23 0.14 0.17
0.92
0.09
0.22
0.27
1.01 0.16 -0.39 -0.27
0.00 0.01 0.36 0.24
0.22 0.40 0.52 0.52
0.22 0.41 0.71 0.65
-0.21 -0.78 -0.96 -0.42 -0.72
0.18 -0.11 0.07 0.15 0.09
0.52 0.24 0.16 0.05 0.13
0.62 0.18 0.18 0.07 0.15
-0.52 -1.65 -1.61 -1.71 -1.12 -1.47 -1.12 -1.50 -2.76 -1.58
-0.11 -0.24 -0.28 -0.08 -0.19 0.16 -0.19 0.19 0.54 -0.64
0.13 0.65 0.35 0.60 0.33 0.13 0.33 0.13 0.54 0.54
0.09 0.52 0.24 0.55 0.24 0.17 0.24 0.18 0.80 0.21
sponding t o 0.05 and 0.60 RF unit using Equation 7 . The range of RF values (0.05 to 0.60) 15 as chosen because greater resolution is generally achieved in this range than a t higher RF values. where diffusion and zone spreading become pronounced. Figure 1 shons the results in bar graph form and shons that nithin these ranges there is considerable overlap among the various solvent systems. Thus a quantitativr interrelation of seven chromatographic solvent systems has been obtained which can be used to cover the [ R u ] range continuously from t 2 . 5 4 to -2.34 units. K h e n the calculated [ R u ]valuc, of a compound lies nithin the ranges of tmo solvent systems. it has generally been profitable t o use the less polar system for optimum resolution. The application of Equation 7 implies that the ilR,w values of individual substituents in the extrapolated solvent systems are directlj proportional to the ARII values of the substituents in the reference toluene-propylene glycol system. This direct relationship of ARlf values in the various solvent systems illustrates hy reversals in order of polarity are usually not observed upon changing solvent systems, thus eliminating the possibility of twodimensional paper chromatography for steroids. To extend these relationships to the androstane series, a n assumption was made that the 17-keto and 178-hydroxy1 groups had 4R.,f values approximately equal t o their polar counterparts in the pregnane series, and accordingly 4Rtf values of 0.76 and 1.56 were assigned t o the 17-keto and 178hydroxyl, respectively. Using these
values, a calibration curve similar to that used to derive a value for the pregnane nucleus (-3.00) yielded a 4R.v value of -2.45 for the androstane nucleus. This higher (less negative) value for the androstane nucleus explains the greater polarity of androstane derivatives as compared to similar pregnane derivatives. Use of these values has resulted in satisfactory agreement betneen calculated and e\perimental mobilities in thcse solvent systems, but with somen hat higher average deviations. To evaluate the usefulness of the systematic approach to steroid paper chromatography, chemically related series of compounds (10) were used instead of fortuitously chosen unrelated steroids. The compounds listed in Table IV comprise steroids involved in the laboratory synthesis of dexamethasone. The accuracy of the method in predicting desired solvent systems and approximate chromatographic mobilities can be judged by examining the data listed in Table IT'. Although the agreement betneen the calculated and experimental [ R Y ]values is not as good as one would like, nevertheless it is fair enough t o permit the choosing of the proper paper chromatographic solvent system, as evidenced by a comparison of the calculated and experimental RF values. The worst deviation encountered corresponds t o an error of only 0.64 [R,tf]or 0.33 Rp unit in the corresponding solvent system, while the average RF deviation is only 0.09 RF unit. This systematic approach to steroid paper chromatography makes possible the prediction of the approximate
mobility of a steroid of the pregnane and androstane series in all of the seven solvent' systems by inspection of its struct,ure and has thus greatly simplified routine paper chromatographic analyses. It is capable of extension to other steroid families and other solvent systems and makes steroid paper chromatography a little lrss of a n art. LITERATURE CITED
(1) Bate-Smith, E. C., Westall! R. G., Biochim. Biophys. -4cta 4 , 127 (1950).
(2) Brooks, S. G., Hunt, J. S..Long, A. G., Nooney, B., J . C h e m . SOC.1957, 1173. (3) Burt,on, R . B., Zaffaroni, h.,Keutman, E. H., J . Biol. Chem. 188, 763 (1931). 14) Bush. I. E.. Biochem. J . 50.370 11952).
Chem. SOC.80,3382 (1958) ( 7 ) Heftmanri, E., C h e m Xez1.s. 55, 679 i1%5 i. -- , \
(8) lfartin, .A. J. P.,Biochein. SOC.Symposia 3 , 6 (1950). ( 9 ) Seher, R., J . Chromatog. 1, 205 (1958). 110) Oliveto. E. P.. Rausser. R.. Weber. L.. Nussbaum. L.. kebbrt. W.1 Coniglio, C. T., H&hberg, E. B.: Tolksdorf, S., Eider, hI., Perlman, P. L., Pechet, 11. hI. J . Am. Chem. Sac. 80, 4431 (1958) i l l i Reineke. L. AI., AAsa~. C H m r . 28. 1853 (1956) (12) Savard, I