Effect of Time on Fluorescing Power of Estrogenic Steroids HERBERT S. STRICKLER, ROBERT C. GRAUER, and M A R Y R U T H C A U G H E Y Department o f Research in Endocrinology and Metabolism, W m . Allegheny General Hospital, Pittsburgh, Pa.
The fluorescing power of four estrogenic steroids has been found to fall off with time, if certain concentrations of sulfuric acid are used for sensitizing and dilution. Because under these conditions the background fluorescence of the urinary extract is constant, correction can be made for impurities.
I
Memorial Research Laboratory,
Studies by the authors of Goldzieher’s (IO) conditions of assay indicate, in accordance with his report, that the fluorescing power of pure estrogen is indeed constant with time. There is, however, no opportunity by his method to correct for the presence of impurities. Cognizance of the falling off of the fluorescence power with time lends additional precision to the determination of these substances. EXPERIMENTAL
I i STUDYIXG the ability to fluoresce, induced by sulfuric
acid treatment of various steroids, it was noted that under the conditions used in this laboratory the fluorescing power of estrone, 17 @-estradiol,estriol, and 17 or-ethinylestradiol decreased with the progress of time after the sensitizing reaction. Diethyldid not exhibit this stilbestrol (aJ~’-diethyl-4,4’-stilbenediol) phenomenon. Other investigators ( I , IO) report constancy of fluorescence LTith time, but they differ in their procedure, in that they dilute less after the sensitizing reaction and use stronger acid. It is believed that the decay observed may be due to the lower viscosity of the diluted solutions. Diczfalusy (4) observed such a decrease and therefore chose other conditions for his assay. However, it was noted that purified urinary extracts have a background fluorescence due to residual impurities, which under the conditions of analysis remain essentially constant with time. Hence by the measurement of the fluorescence of the solutions at different times the apparent level of estrogen content can be corrected for impurities. Should this “background’, fluorescence turn out to be indicative of the presence of other compounds related to estrogen metabolism, new methods based on further refinement in fractionation, together with a combination of fluorometry and colorimetry, would need to be evolved.
H. Singer
A direct-reading Farrand Fluorometer (Farrand Optical Co., K’ew York, N. Y.) with quartz optics was used. Improved stability was attained by the use of an electronic voltage regulator, Stabiline Model IE5 1005 (Superior Electric Co.). The smallest apertures ( S o . 5 or 6 ) were chosen in order to limit the action of the ultraviolet light on the solutions. Uranyl acetate, 0.1% in 5% acetic acid, was used as the reference standard for readings with the 490-mp secondary filter, and uranyl acetate, 0.04% in water, was used with the 524mp eecondary filter. Solutions of sodium fluorescein, 0.0005% in 0.0001N sodium hydroxide, and quinine sulfate, 0.01% in 0.1N sulfuric acid, were usually read before each assay in order to check for any deterioration of the ultraviolet lamp or phototube. The primary (source) filters that were used were 435-mp interference type (Farrand) and 365mp glass (Corning No. 5860). The secondar filters were 490mp interference (Baird) backed with glass ( L r n i n g So. 4303), and 524m.u interference type (Farrand). These choices of filters were based on recommendations of the manufacturers of the instrument and of Engel and his associates ( 6 , I I ) . The countercurrent distribution machine used was the glass type, 10-ml. lower phase, two banks of 50 tubes each (H. 0. Post Co.). The contents of the countercurrent tubes after partition were evaporated a t 0.4 to 0.8 mm. of mercury with controlled cooling. This procedure appears to be desirable because of removal of the easily volatile phenols. METHODS
Table I.
Comparison of Sensitizing Conditions
Strickler and Coworkers Goldzieher (Bates-Cohen) (Engel) 0.1 ml. 1 t o 19 alcohol-toluene 0.2 ml. 1 t o 20 alcohol-toluene 0.5 ml. (1 ml.) 90 wt. % sulfuric 1 ml. “90%” sulfuric acid (90 ml. c.P., 10 ml. water) acid 10 min., water bath, 90°, 100’ SOo C., 10 minutes (ovens 100-103° 10-12 min.) Water b a t h Dilute with 3.5 ml. (7 ml.) 65 wt. % Dilute with 5 ml. “65%” acid (65 ml. c.P.. 35 ml. water) acid Fluorescence readinns fall off with (Bates-Cohen use 6 ml.) ’ time Fluorescence stable 3 24 hr. Final mixture Final mixture -68 wt. % sulfuric acid -76 wt. % sulfuric acid 15 Viscosity, 9 N 9 Viscosity I) Sulfuric &id. c.P., taken as 96 wt. %. sp. gr. 1.836
-
Table 11.
8. Y
I n aliquot
c
(calcd.)
Example of Time Study Calculations
Estrone 0 24 hr. hr. 152.80 73.0 56.3 42.3 0.21
36.0
- Estriol 17 @-Estradiol 24 48 0 24 48 0 hr. hr. hr. hr. hr. hr. hr. 52.0 141.3 111.5 98.5 93.3 48.3 31.0 39.0 50.3 46.5 44.5 33.3 21.3 17.5 0.21 0.15 0.16 0.36 0.34 48
36.5
12.2
Specimen 3-55-138purified b y ccd. a Galvanometer readings are referred t o a reading of 35 for the standard. Where solutions were too concentrated, the sensitivity controls were adjusted
to eome smaller value and the observed readings were calculated t o the bases of 35.
The initial preparation of urinary extracts was essentially that of Engel ( 7 ) . However, hydrochloric acid was used for hydrolysis, and, to minimize losses of estriol, the final extraction of estrogens with ether was made from the solution saturated with sodium bicarbonate rather than with the carbonate-bicarbonate buffer. At this point, countercurrent distribution for exploration of a sample was sometimes carried out with 65 transfere and 50 tubes, using the system 70% methan01-30% water/40% chloroform-60% carbon tetrachloride (8). Just before countercurrent distribution the sample Fas usually partitioned between 50 ml. of benzene and 50 ml. of 0.3X sodium carbonate (singlestage Mather). The estrone and estradiol remain primarily in the benzene. The carbonate phase, after acidification and saturation with sodium bicarbonate, was re-extracted with ether. The countercurrent distribution then proceeded in one bank, but in two parts. Various procedures were tried, which -xi11 be discussed elsewhere (19). In recent work done at this laboratory, the estradiol and estrone were separated by a simple countercurrent distribution procedure using only five tubes and the same system of solvents. The estradiol is removed to the extent of 81% (at K I = 1) in eight w i t h d r a d s , two more withdrawals are discarded, and the estrone is then present in tubes I, 2, 3, and 4 to the extent of 85% ( K , = 0.2). ( K is the distribution coefficient.) Tube 0, which contains much pigment, is discarded. The estriol was likewise removed in the same type of distribution, but the solvents used mere those employed by Bauld (2) in his Celite system (70% methan01-30% water/ethylene chloride) and 2 =
1240
1241
V O L U M E 28, NO. 8, A U G U S T 1 9 5 6 0.5-i.e., only 5 ml. of upper phase are used in each transfer. Samples may be subjected to purification with a Bauld column, but the authors have found that recovery is lower. Further purification of the urinary extracts is desirable. Estrone and estradiol were separately purified on Celite-sodium hydroxide columns, essentially according to Bitman and Sykes ( 3 ) . Columns in this laboratory were 10 mm. in inside diameter. The first 5 ml. of eluate benzene (total 80 ml.) were discarded in the case of estrone. For estradiol the first 55 ml. of eluate were discarded (total 170 ml.). At assay, a single distribution of the estriol fraction between 40% ethyl acetate-60% cyclohexane/ 31.6% ethan01-68.4% water (mixture TG) removed sufficient impurity into the lover layer so that aliquots of the upper phase could be used. K for estriol was found to be 0.83. Such separations are good starting points for bioassay as well as for fluorometry, since the impurity level is considerably reduced. In bioassay, the separation of the three estrogens is desirable. This is of importance because, as shown in this and other laboratories, varying proportions of the three biologically active estrogens will exhibit potentiation or depression, depending upon the mixtures of the estrogens employed ( 5 , 1 2 ) .
a galvanometer deflection each day. The decrease of fluorescence observed was therefore not due to the lability of the sensitized steroids to the exciting radiation. Further heat treatment (50') was applied in order to clarify the solutions. This did not affect the previously observed decrease in fluorescence. It was not due to oxygen diffusion into the samples. The fluorescence decrease indicated instead that the sensitized estrogens shifted structure in some manner, so that the population of sensitized molecules fell off in a manner analogous to decay of radioactive substances. Hence, the differential equation dG/dt = -A'G was applied to the data. Here G is the galvanometer reading corrected for blank and/or impurity, t is the time, and A' is a constant. Integration of this equation yields In Gt = In GO
- d't:
log,, GI = log,~Go- At
(1)
Consequently, the data should lie on straight lines on a semilog plot (cf. Figure 1). Numerous other experiments, including those in which every precaution was taken to avoid traces of turbidity (filtration, heating at 50") indicate that this equation is rigidly applicable only with some sets of data. The decay in many cases follows a more complex law.
PREPARATION FOR FLUOROMETRY
The estrogens were sensitized by heating the residues which were obtained on evaporation with 0.1 ml. of toluene-absolute alcohol (19 to 1) and 0.5 ml. of 90 weight % sulfuric acid for 10 minutes in a boiling water bath ( 7 ) . Other temperatures and concentrations of acid have been tried. On cooling, the solutions were diluted with 3.5 ml. of 65 weight % acid. In the case of urinary extracts, chilling, followed by filtration through sintered glass and/or centrifugation, seemed desirable in order to avoid any traces of turbidity. The clear solutions were then transferred to cuvettes and heated at 50" for 45 minutes prior to being read immediately after their removal from the oven. The fluorometric readings were completed in a few seconds and the tubes were then capped and stored in a cupboard until the next day. The amounts of acid employed are compared in Table I to those of two other investigators (1, 10), who indicated that a constancy of fluorescence for over 24 hours i s obtained. Evidently the difference lies in the fact that the present authors used dilution to a greater extent and with weaker acid. The conditions followed were those published by Engels' group ( 7 ) . The falling off of the fluorescence with time is considered to be dependent on the concentration of the acid solutions. The method of making up the solutions is therefore detailed here. Du Pont C.P. concentrated sulfuric acid, specific gravity 1.8411.844, is used. The 90 Keight % acid is made up with 400 ml. of the concentrated acid and 46 ml. of water. The diluent acid (65 weight %) is made up from 231 ml. of concentrated acid and 200 ml. of water. These are cooled in an ice bath when being prepared. The differences in viscosity (IS) listed in Table I may afford an explanation of the observed differences. In the less viscous solutions there is a fall-off of sensitized molecules, which is perhaps due to an increased frequency of collisions. There is, consequently, an opportunity to distinguish betn-een the stable background (impurity) fluorescence and that due to the estrogen under the conditions of the investigation. This is not true when the higher acid concentrations are used. TIME STUDY OF FLUORESCENT (SULFURIC ACID-SENSITIZED) ESTROGENS
The procedure followed in this laboratory involves measuring the time decrease of the fluorescence in material obtained from a given sample, and noting the degree of fluorescence daily for 3 or 4 days (Table 11). The samples were purposely exposed to the ultraviolet excitation for only the hrief period necessary to obtain
0.01
I 06
E,
:
x
E,
0
E,
A
04
02
1 20
60
40
80
100
HOURS
Figure 1.
Semilog plot of galvanometer readings 2;s. time for pure estrogens
Upper line of each pair, 1.0 y i lower line, 0.5 y . Samples heated a t 50' just before reading. Filters P435/5490. EL = estrone; E1 = 118 estradiol; Es = estriol
The filter system used here was P435/5490. The relation is stoichiometric, for 0.5-7 lines parallel 1.0-7 lines. It was observed in this laboratory that 17 a-ethinylestradiol also gives fluorescence which falls off with time. Slaunwhite, Engel, and coworkers (16) observed that 17 a-ethinylestradiol did not give a linear response with concentration and the authors have confirmed this. The compound gives a pink color n-ith an absorption curve peaking at 530 mfi (18). Appreciable self-absorption would therefore be expected a t either 490 or 524 mp. On examining some samples with the filter system P435/S524 it was noted that the fluorescence readings also fell off with time, as did the values obtained on correcting for fluorescence (at equal blank readings) at P365/5524 according to Finkelstein (9). It was found, for estrogenic urinary extracts prepared rssentially according to Engel ( 7 ) ,that the impurity contribution to the
1242
ANALYTICAL CHEMISTRY
Table 111. Approximate Constancy of Background Fluorescence Contrasted with Fall Off of That of Internal Standard Corrected Galvanometer Peading Specimen and Time, Time, Time, Fraction (Aliquots) Reading hr. Reading hr. Reading hr. E-55-427; eetradiol 23.5 2.5 23.3 20.1 22 3 44.2 1 y 17p-estra153.7 2.5 115.3 20.1 105.0 44.2 diol E-55-408; estriol0 32.: 2,5 33.0 20.1 32 0 44.2 + 1 y estriol 87.t 2,s 47.2 20.1 39.5 44.2 Pooled male urine; estrone 19.5 2.0 17.0 22.0 i5,3 n o 1 y estrone 184.5 2.0 93.5 22.0 34.8 70.0 Gonadectomized true hermavhrodite E-55-376 b ; estrone 5.8 2.3 5.2 20.0 5.5 44.0 1 yestrone 121.5 2.3 39.2 20.0 18.7 44.0 E-55-327 6 : estriol !?.3 2.3 18.7 20.0 21.2 44.0 1 y estriol 00.3 2.3 37.8 20.0 30.7 44.0 a Finally cleaned up on Celite-sodium bicarbonate-benzene column. 6 See Table IV.
+ +
+ +
fluorescence may be considered to be a constant, C, invariable ivith time (cf. Table 111). This is deduced from the readings listed and many other observations on estracts of different purity, and representing different physiological states. -4s purification proceeds, the value of C decreases, so that the values of y from time studies and from the initial readings (“apparent” estrogen) converge. Thus, some of the values in Table IV are nearly the same. E-55-376 gave a large apparent peak of estrone and estriol on exploratory countercurrent distribution but none by time study (19). This was before intensive purification. The eqnation based on this assumption is:
G = - r+EC
Q x-here G is the galvanometer reading, y the number of micrograms of a particular estrogen in the aliquot, and E the reading of 1 7 of pure estrogen, Q is the quench coefficient as defined by
Some correlation with structure is evident from time study measurement. Dehydration at positions 16 and 17 in the steroid ring would convert estriol to estrone. In the sensitization with sulfuric acid, dehydration probably occurs at these positions. Consequently, the fluorescent complex obtained from estriol should be similar to that obtained from estrone. Indeed, the time decay slopes of the two are similar. 17 @-Estradioland 17 a-ethinybstradiol, which differ from the other two at the fivemembered ring, have different slopes. 17 p-Estradiol shows less fluorescent decay than the other three estrogens. Diethylstilbestrol, another synthetic estrogen but not of the steroid family, gives very mild fluorescence and does not decrease with time. It may be concluded that the presence of the phenolic structure alone is insufficient to give strong fluorescence under these conditions (cf. 16). Recently it has been found in this laboratory that the methyl ethers of estrone, 17 @-estradiol,and estriol show fluorescence and time decay comparable to the free phenols. So far, no decrease of fluorescing power with time comparable to those of the estrogens has been observed with other steroids Dehydroisoandrosterone, 3 a-, 20 a-pregmediol, 17-hydroxypregnanolone, desoxycorticosterone, and compound F do n o t show the decrease in fluorescence with the same technique. The conditions of Linford (14) were also used with slight modification (0.1 ml. of alcohol Kith concentrated sulfuric acid at 60” for 1 hour). No marked fall in fluorescence was noted using P435/S524 filters. Countercurrent distribution, Celite chromatography, and fluorometry were used to assay a number of urines. In Table IV apparent estrogen is referred t o as that estrogen computed from the ratio of the galvanometer reading of the unknown and the standard, computed without time studies-Le., at or near zero hours. These values are similar in magnitude to those of hligeon (Is),using different countercurrent systems and fluorometry.
Table IV.
fE
Q = wherej is the number of micrograms of pure estroG , - G’ gen added to an equal aliquot, and G, is the resulting galvanometer reading (19). All galvanometer readings are corrected for reagent blank. If Equation 2 is written at two different timese.g., 0 and t-then on solving for y, C drops out and ~
(3) and Here Q has been replaced by its definition at each time. Data illustrating the application of this principle are shown in Tables I1 and 111. DISCUSSION
The use of conventional acid hydrolysis throughout this work is justified on the basis that only by restricting the type of processing could one become at all familiar with the types of impurities likely to be encountered. The countercurrent distribution procedures were designed to deal with observed interference. Despite these and supplementary procedures, it would appear that the use of time studies yields the lowest and most convincing values. Thus, in the case of male urine, the Engel group (17) has never demonstrated any estrogen by redistribution in different solvents. The time study values from the authors’ preliminary experiments indicate the presence of either none or low levels of estrogen in male urine. The materials used by Engel may have been subject to deterioration, because he employed repeated redistribution and quench corrections may not have been applied.
Specimen 3-55-346 E-55-553 E-55-327 E-55-376 E-55-319 E-55-291 E-55-285 3-55-243 3-55-218 E-55-233 E-56-367 3-55-644
Comparison of Apparent Estrogen Values with Those from Time Studies Estrone ApTime parent study 6.0 6.7 0.3 0.1 0.3 go.09 go.09 0.4 Q0.l 0.2 0 9 0.6 1.3 1.0 1.4 2.2 1.3 1.4 2.7 3.1 1.6 2.0 9.6 9.4
Estradiol ApTime parent study 1.0 dO.09 (0 2 0.4 2o:z 0.6 0.4