Steroid Estrogen Glycosides FORMATION OF GLUCOSIDES AND GALACTOSIDES

Abstract Homogenates of human liver can transfer both glucose and galactose from their respective uridine nucleotides to the 17α-hydroxyl of 17α[6,7-3H]estradiol. Glucose was also transferred, but less readily, to the 16α-hydroxyl of [6,7-3H]-estriol. Kidney homogenates effect these same transfers. The radioactive steroid glycosides formed, namely 17α-estradiol-17-β-d-glucopyranoside, 17α-estradiol-17-β-d-galactopyranoside, and estriol-16α-β-d-glucopyranoside, were rigorously identified by crystallization to constant specific activity with authentic glycosides. The chemical preparation of the latter two glycosides is described for the first time. The results are the first demonstration of steroid glucoside formation by human kidney, and of steroid galactoside formation by any human tissue.

The chemical preparation of the latter two glycosides is described for the first time. The results are the first demonstration of steroid glucoside formation by human kidney, and of steroid galactoside formation by any human tissue.
Homogenates of human liver were recently shown to effect the transfer of glucose from TTDP-glucose to the 17.hydroxyl group of 17oc-eetradiol (1). This was the first demonstration, with human tissue, of the formation of a steroid glucoside. In these experiments no transfer of glucose to the 17.hydroxyl of 17/%estradiol, or to the phenolic 3-hydroxyls of estrone, or of either epimer of estradiol could be shown. This suggested that the microsomal steroid glucosyltransferase had a high specificity for the a-oriented 17.hydroxyl group. The present paper describes the finding that human kidney, as lee11 as liver, exhibits this glucosyltransferase activity, and that both liver and kidney can also effect the in vitro formation of 17a-estradiol-17.P-u-galactoside.
The specificity of these glycosyl transfer reactions has been further investigated with a limited series of steroids as substrates. Kidney tissue for preliminary experiments was obtained by excising random areas, containing both cortex and medulla, from the kidney of a 56.year-old man, which had been removed because of hypertensive renal failure. This tissue was obtained at surgery and homogenized at once. Incubat'ions were carried out immediately with portions of the homogenate, and the remainder was stored at -15".
Another small sample of kidney tissue was obtained from the macroscopically normal area, containing both cortex and medulla, of the kidney of a 37-year-old man who was undergoing surgery for removal of a renal calculus.
This tissue was homogenized and incubations were carried out immediately after surgery.
These constantls remained unchanged by further c*rystallizat,ion.
These spectra are in agreement with the assigned structure of 17a-estradiol-17.B-D-galactopyranoside.
As a final check on the synthesis, 500 pg of the product were hydrolyzed with hydrochloric acid and the sugar residue was examined on thin layer chromatography on Silica Gel H by the procedure described in detail by Williamson et al. (4). The systems used were n-propyl alcohol-water (7 : 1) and methyl acetate-isopropyl alcohol-water (I 8 : 1: 1).
In both systems the sugar residue had the same RF as reference galactose, and was separated from reference glucose and mannose. The crystallization of steroid glycosides with fractional molecular equivalents of water, which is not removed by prolonged drying under vacuum, has been well documented (4-6).
Preparation of ~striol-26cu-P-D-glucopyranoside-A sample of estriol glucuronide donated by Dr. Saul Cohen was used as starting material.
This material was isolated in crystalline form from human pregnancy urine (7), and its structure has been established (8, 9) as estriol-16oc-B-n-glucopyranosiduronic acid. The compound was methylated and reduced to the corresponding glucoside (10, 11). The procedure followed was exactly as described by Collins et al. (2,. The final product melted sharply at 260" and gave an infrared spectrum entirely reconcilable with the structure of estriol-16a-fi-n-glucopyranoside; v::: 3350 (OH), 1600 and 1495 (aromatic), 1220 to 1280 (ester). Hydrolysis of the product with almond emulsin and with hydrochloric acid yielded products which were identified by thin layer chromatography, by the procedures de-  (4)) as estriol and n-glucose. a Ul)l'(:, IJI)P-glucose; IJl)P(;al, UI)P-g&ctose Xethods-The procedures for the preparation of homogenates h N.T., not, tested. the presence of free steroid and of glucoside by thin layer chromatography (2, 3).

RESULTS
Preliminary experiments indicated that the steroid glucosyl transferase of human liver was located in the microsomal fraction obtained at 105,000 x g. The activity in this fraction was, however, considerably enhanced by the addition of small amounts of the 105,000 x g supernatant, indicating the probable requirement for a soluble cofactor, which is a quite cornmon situation with other steroid glycosyltransferases (12). Because of this finding, further experiments were carried out with whole homogenates. Table I shows the results of qualitative experiments to test the respective ability of liver and of kidney homogenates to transfer glucose and galactose from their respective uridine nucleotides to various steroids. No appreciable qualitative or quantitative difference in transferase activity was detected between the kidney tissue from the two donors. Loss of transferase activity on storage was negligible after 48 hours, but was about 50% after 2 weeks.
The results in Table I indicated that both liver and kidney tissue could effect the formation of both a glucoside and a galactoside of 17cr-estradiol.
Evidence was also obtained for the formation of a glucoside of estriol by liver and by kidney /  Vol. 247,No. 10 centage of the radioactive steroid which was converted to glycoside.
The conversion of 17a-estradiol to glucoside and galactoside by liver tissue was 16% and 5%, respectively. The corresponding figures for kidney tissue were 837, and 137,. The formation of estriol glucoside amounted to 1.5% of the steroid incubated with liver and 6.6% of that incubated with kidney.
These figures can be converted to picomoles of conjugate formed per g of wet tissue, and are shown in this form in Table II. The identity of the 17cr-estradiol-17-P-n-glucopyranoside formed by human liver has been previously est'ablished (1) by crystallization to constant specific activity with material prepared synthetically (4). Similar recrystallizations were carried out in the present work to establish the identity of the glucoside of 17a-estradiol formed by kidney tissue and of the galactoside of this steroid formed both by liver and by kidney. As shown in Table III, the tritiated conjugate formed in the incubations crystallized in each case to constant specific activity with the corresponding 17oc-estradiol-17.P-n-glycopyranoside.
In addition, the identity of the tritiated glucoside of estriol formed by liver homogenates was established by crystallization as estriol-16oc-fl-n-glucopyranoside.
In these experiments the calculated specific activity (Table III) was obtained for each compound by dividing the number of disintegrations per minute of tritium in the presumptive glycoside by the number of milligrams of authentic compound which were mixed with it prior to crystallization.
The weight of the radioactive material was negligible.
The figures in the individual cases were, estriol-16or-glucoside, 135,390 dpm, 9.05 mg; 17or-estradiol-17.galactoside (liver) 195,300 dpm, 15.41 mg; 17a-estradiol-17.glucoside, 136,540 dpm, 8.8 mg; 17ar-estradiol-17.galactoside (kidney), 180,955 dpm, 15.44 mg. DISCUSSIOK Williamson et al. (13) have detected the formation of galactosides of some phenolic steroids by rabbit liver, but the present work is the first demonstration of the formation of a steroid galactoside by human tissue. The finding that human kidney homogenates as well as those of liver had steroid glucosyland galactosyltransferase activities contrasts with the results obtained in the rabbit, where these glycosyltransferase activities were detected only in liver, and were apparently specific for the phenolic 3-hydroxyl group.
Human kidney tissue was apparently more effective than liver in forming the steroid glycosides (Table II), but it must be borne in mind that there were both sex and age differences in the donors of the liver and the kidney tissues. Liver was obtained from a 26.year-old woman, while the kidney sample which was used in the experiment reported in Table II was obtained from a' 56-gear-old man, with a history of hypertension probably as a result of kidney disease. However, it is of interest that the results obtained with apparently normal kidney tissue from a 37. year-old man were comparable to those shown in Table II. In the present experiments, the amount of galactoside formed was lower than that of glucoside in incubations with either liver or kidney tissue. However, this quantitative difference should be interpreted with caution. It was observed that the homogenates contained an active UDP-galactose-4-epimerase, which converted some of the added UDP-galactose to UDP-glucose, and led to the formation of considerabIe amounts of steroid glucoside as well as steroid galactoside in experiments in which UDP-galactose was the only nucleotide added to the incubation medium.
The quantitative relatiorrship of glucosyl-and galactosyltransferase activities which might obtain in vivo could well differ, therefore, from that found in vitro.
No precautions were taken to exclude bacterial contamination of the homogenates used in this work.
However, it was possible with kidney homogenates to carry out incubations within 2 hours of excision of the tissue, and these incubations were of 30.min duration.
The possibility that the steroid glycosidations were caused by bacterial contamination is extremely remote, the more so because such contamination would have had to affect to the same extent three tissue samples obtained at different times from different hospitals. Prolonged storage of the homogenates at -15" showed no evidence of any effect other than a slow loss of transferase activity.
This would be expected in an endogenous enzyme activity, and militates against the possibility that the transferases were produced by contamination during storage. The significance of the formation of these novel glycosides of 17Lu-estradiol is as yet unknown.
This steroid, if it is formed in the human at all, is of minor quantitative importance (1). It is of int,erest that neither the 17P-hydrosyl of 170.estradiol nor the 3-hydroxyls of any of the phenolic steroids tested appear to be acceptors of glucose or galactose.
Estriol, which is a quantitatively major excretion 1)roduct in the human, did form a 16a-glucoside.
This reaction, however, took place in poor yield, which might explain the failure to detect the formation of the corresponding estriol galactoside.
The number of steroids tested as substrates (Table 1) is limited, but the glycosylt.ransferase or transferases, in addition to the requirements with regard to the hydroxyls in Ring D, appear to have a considerable specificity for the steroids with an aromatic A ring. That this may not be an absolute requirement is suggested by the evidence obtained in a single experiment for the formation of a glucoside by epitestosterone. The fact that the presence of glucuronic acid at position 3 of 17or-estradiol prevents the addit'ion of glucose to the 17a-position is at variance with the results obtained in rabbits (2,4) where the glucuronidation of the 3-hydroxyl is obligatory for the addition of glucose to the 1701 position.