Purification and Properties of Follicle-stimulating Hormone from Sheep Pituitary Glands*

SUMMARY A highly purified follicle-stimulating hormone (FSH) preparation has been obtained from sheep pituitary glands by extraction with ethanol followed by fractionation with metaphosphoric acid and ammonium sulfate, gel filtration on Sephadex ion exchange chromatography on carboxymethyl hydroxylapatite chromatography, preparative acrylamide disc gel electrophoresis, and gel filtration The purified FSH contained 133 units of ovine follicle-stimulating hormone standard (NIH-FSH-Sl) per mg of dry weight as judged by augmentation bioassays. This preparation also contained standard per of dry weight determined by ovarian ascorbate depletion and more than 1 unit of NM-LH-Sl per mg as determined by the hypophysectomized male rat ventral prostate bioassay. Radioimmunoassay indicated that the purified FSH contained only 3% as much immtmoreactive luteinizing hormone as found in the National

This preparation also contained 0.04 unit of the National Institutes of Health ovine luteinizing hormone standard (NIH-LH-Sl) per mg of dry weight as determined by ovarian ascorbate depletion bioassays and more than 1 unit of NM-LH-Sl per mg as determined by the hypophysectomized male rat ventral prostate bioassay.
Radioimmunoassay indicated that the purified FSH contained only 3% as much immtmoreactive luteinizing hormone as found in the National Institutes of Health ovine FSH standard (NIH-FSH-S4). Administration of low doses of the purified FSH to hypophysectomized female and male rats not only resulted in increased weights of the ovaries and testes but also of the uteri, ventral prostates, and seminal vesicles.
Analytical acrylamide disc gel electrophoresis of the purified FSH at pH 8.9 displayed a single broad zone; however, immunodiffusion in agar showed a diffuse precipitin line plus two fine precipitin lines. The molecular weight of sheep FSH was approximately 33,000 as determined by ultracentrifugation. Electrofocusing in carrier ampholytes indicated an isoelectric point of pH 4.6. Amino acid analysis showed that the purified FSH contained a higher content of threonine and half-cystine and a lower content of phenylalanine than sheep FSH preparations previously described.
During the last 5 years, several methods have been described for obtaining purified follicle-stimulating hormone from sheep * This investigation was supported by National Institutes of Health Research Grant GM 02154, Public Health Service Training  and Ford Foundation Grant 63-505. $ Present address, CIBA Corporation, 556 Morris Avenue, Summit, New Jersey 07901.
These FSH' preparations were reported to contain from 30 to 58 units of NIH-FSH-Sl per mg (6) with little (l- 3,5) or no (4) evidence of contamination with luteinizing hormone.
These authors reported yields ranging from 3.2 to 8.0 mg of purified FSH per kg of fresh tissue, with the exception of Cahill et al. (4) who reported 25 mg per kg.
The present report describes a procedure used to obtain a preparation of sheep FSH which contained more than 2 times as much FSH activity per mg of dry weight as preparations previously described.
Emphasis was placed on the development of a method which separated FSH from LH to a high degree early in the purification procedure.
This was done with the point of view that subsequent purification methods, which resolved the two hormones, would further reduce LH contamination.
Results are presented from three different assay techniques used to determine the level of LH contamination in the purified FSHnamely, ovarian ascorbate depletion bioassay (7), ventral prostate bioassay (8), and radioimmunoassay (9). Data in this report show that the injection of low doses of highly purified sheep FSH into hypophysectomized male and female rats not only increased the weights of the testes and ovaries, but also the seminal vesicles, ventral prostates, and uteri.
Information is also given on the molecular weight, isoelectric point, and amino acid content of sheep FSH. 0. D. Sherwood, H. J. Grimek, and W. H. McXhan 2329 aerylamide disc gel electrophoresis was conducted in a mode1 PD 2/70 column obtained from Canalco, Rockville, Maryland. Electrofocusing was done in a model 8101 ampholine electrofocusing column obtained from LKB.
A Beckman model E ultracentrifuge with a model Bn-D rotor (Spinco Division, Palo Alto, California) was used for molecular weight determinations.

Methods
Routine Bioassays-Intact 21-day-old n-ale and female rats were used to monitor FSH and LH activities during the development of the purification procedure. Subcutaneous injections were administered the afternoon that the rats were obtained (Day 21) and twice daily the next 4 days. The animals were autopsied on the morning of Day 26. FSH activity was based on an increase in ovarian weight (10). LII activity was based on an increase in seminal vesicle weight (10,11).
Specific FSH activity was determined for the principal FSH fractions by Steelman-Pohley augmentation bioassays (6). Three doses of the FSH reference standard NIH-FSH-S3 and each unknown were administered to immature female rats. Four animals were used for each dose. The results were evaluated by statistical methods for parallel line assays as described by Bliss (12).
Specific LH activity was determined by a modification of the Parlo\\-ovarian ascorbate depletion bioassay (7). Two doses of the NIH-LH reference standard and unknown were administered with each bioassay.
Four rats were used for each dose. Both ovaries were removed 4 hours f 10 min after the hormone was injected into the tail vein. The ovarian ascorbic acid content was determined by the procedure of Schaffert and Kingsley (13)  Disc Gel Electrophoresisnalytical acrylamide disc gel electrophoresis at pH 8.9 (16) was used routinely to follow the progress of chemical purification.
Dry Weight Determinations-The weights of purified FSH fractions given in this report (yields, doses for bioassays, etc.) are based 011 protein determinations.
This was done to avoid the inconvenience and time routinely required to dry and weigh each fraction.
Since FSH is a glycoprotein, the dry weight of the hormone is greater than the weight suggested by protein determinations.
Therefore, the ratio of the weight suggested by protein determinations to the actual dry weight was determined with the highly purified FSH preparation.
This enabled an accurate determination of the yield and relative potency of this purified FSH.
Protein Determinations-The protein content of the extract and subsequent fractions was determined by the biuret reaction (17)  fication of the procedure of Ellis (21) was used to precipitate inert protein with metaphosphoric acid. The ethanol extract was diluted with sufficient pH 7.3 0.025 M sodium phosphate buffer to produce a protein solution consisting of 2 g/100 ml. The pI1 of this solution was adjusted to 4.2 by the gradual addition of freshly prepared 0.035 M metaphosphoric acid. The inert precipitate which formed was removed by centrifugation at 16,300 X g for 20 min. The supernatant fluid, metaphosphoric acid supernatant, was readjusted to pH 7.3 by the addition of 1 N NaOH and then dialyzed for 48 hours against 10 liters of 0.5 saturated (NH4)2S04 buffered at pH 7.3 with 0.025 M sodium phosphate.
The resulting precipitate which contained the bulk of the LH activity was removed by centrifugation at 16,300 x g for 20 min. The supernatant fluid was dialyzed for 48 hours against a dilute solution of CH&OONHk (0.25 g per liter) and dried by lyophilization.
This supernatant obtained by ammonium sulfate fractionation which contained nearly all of the FS H activity was designated the (NH&S04 FSH fraction. This fraction contained 922 mg of protein per kg of fresh tissue.
Gel Filtration on Sephadex G-150-An 8-to g-fold purification of the (NH&SO4 FSH fraction was accomplished by gel filtration with Sephadex G-150 as described in the legend to Fig. 1.
The contents of tubes containing FSH activity were pooled to form the Sephadex G-150 fraction. Ion Exchange Chromatography on Cm-Xephadex-The FSH in the Sephadex G-150 fraction was further concentrated by ion exchange chromatography on Cm-Sephadex C-50 as described in the legend to Fig. 2. The contents of tubes containing FSH activity were pooled to form the Cm-Sephadex fraction.
Chromatography on Hydroxylapatite-Hydroxylapatite chromatography was used for further purification of the Cm-Sephadex fraction as described in the legend to Fig. 3 5. Gel filtration of the preparative electrophoresis fraction (2 mg of protein) on a column, 1.4 X 65 cm, of Sephadex G-75 at 4".
The column was eauilibrated and run with 0.05 M. nH 6.8. CH&OONH4 buffer. Fiactions were collected every 12'min at a flow rate of 8 ml per hour.
The tubes which contained molecules small enough to penetrate Sephadex G-75 were pooled to form the' fraction purified FSH.
The mean protein yield of purified FSH was 1.9 mg per kg eq of fresh tissue.
The mean relative potency of the purified FSH was 188 units of NIH-FSH-Sl per mg of protein.
In agreement with Braselton (see Footnote 2), the material excluded from the Sephadex G-75 was biologically inert and contained a low protein content (21 pg).
of tubes containing FSH activity were pooled to form the hydroxylapatite fraction. Preparative Electrophoresis and Gel Filtration on Sephadex G-76 -The removal of inert proteins from the highly purified hydroxylapatite fraction was accomplished by preparative acrylamide disc gel electrophoresis as described in the legend to Fig. 4. The contents of tubes containing FSH activity, designated preparative electrophoresis fraction, were concentrated in a Diaflo model 50 ultrafiltration cell. This fraction was then passed through a column of Sephadex G-75 as described in the legend to Fig. 5. This was done in order to remove a contaminating substance apparently acquired during preparative acrylamide disc gel electrophoresis. 2 Molecules which were not excluded by the Sephadex G-75 were pooled to form the FSH preparation purified FSH. The procedure used to obtain each fraction is described under "Experimental Procedure." Steelman-Pohley bioassays (6) were conducted as described under "Experimental Procedure" with doses of 30, 60, and 120 pg of NIH-FSH-S3 or NIH-FSH-S6 augmented with 50 i.u. of human chorionic gonadotropin.
The protein yields and total units recovered are based on 1 kg of frozen sheep pituitary glands as starting material.
The geometric mean relative potencies and their 95$& confidence limits were determined by the procedure of Sheps and Moore (29) and adjusted to NIH-FSH-Sl.
One unit is equivalent to 1 mg of NIH-FSH-Sl. The percentage of FSH activitv recovered in each fraction was calculated with reference to the original ethanol extract which is expressed as lOOoi,. The indices of precision for t,hese bioassays ranged from 0.09 to 0.21.
27 . (6) and ovarian ascorbate depletion bioassays for LH (7), hypophysectomized rats were used to examine the effects of purified FSH on the gonads and sexual accessory organs. Female and male rats were hypophysectomized and shipped at 22 days of age (nay 22). Injections were administered twice daily for 4 days starting on Day 25. Autopsy was performed on the morning of Day 29. The ovaries, uteri, testes, ventral prostates, and seminal vesicles were removed and weighed.
LPI activity was determined by the ventral prostate bioassay (8). The ovaries and testes were also prepared for histological examination.
The level of immunoreactive LH in purified FSH was determined by radioimmunoassay.
The double diffusion method in agar gel as described by Ouchterlony (23) was used for immunological examination of purified FSH. Sntibodies to sheep FSH were developed in rabbits with the Cm-Sephadex fraction.
Antibodies to sheep LH were developed against a highly purified sheep LH fraction prepared by the authors in this laboratory.4 Antiserum to both FSH and LH were absorbed with normal sheep serum before use in double diffusion studies.
Molecular Weight Determination-High speed sedimentation equilibrium studies were performed with a Spinco model E ultracentrifuge.
The apparent weight-average molecular weight was calculated according to the method described by Yphantis (24). 3 The radioimmunoassay (9) involved the use of the National Institutes of Health ovine LH preparation LER1056X2 as standard.
The authors are grateful to Dr. K. W. Thompson for performing this assay.
A partial specific volume of 0.72 ml per g (2) was used in the calculations. Two hundred micrograms of the hydroxylapatite fraction were dissolved in 1 ml of sodium phosphate at pH 5.0 and ionic strength 0.1 (4) and dialyzed for 5 days aglanst this buffer. Ultracentrifugation was conducted at 40,000 andi 48,000 rpm at 4".
Isoelectric Point Determination-The isoelectric point of sheep FSH was determined by electrofocusing (25) as described in the legend to Fig. 9.

Amino
Acid Analysis-Samples of purified FSH were hydrolyzed in constant boiling 6 N HCl (500 pg of protein per m of acid) in sealed evacuated tubes at 110" for 22 hours. The hydrolyzed samples were dried in a vacuum desiccator and then analyzed with the Beckman Spinco model 120C amino acid analyzer according to the method of Spackman (26) and Benson and Patterson (27). Tryptophan was determined according to the procedure of Bencze and Schmid (28). Table I summarizes the protein yields, relative potencies and confidence limits, total units of FSH activity, and percentages of FSH activity recovered in the principal FSH fractions obtained throughout purification. Table II summarizes the results of ovarian ascorbate depletion bioassays on the initial and final FSH fractions.

PurQication
This table clearly shows that fractionation of the crude ethanol extract with (NH4)$04 essentially separated the LH from the FSH. Fig. 6 shows the results obtained when each of the principal fractions was analyzed by analytical disc gel electrophoresis at pH 8.9. In general, it was observed that fewer stained zones were distinguished with the more highly purified FSH fractions. Although the quality of the reproduction of Gels e and f is poor, the point is clear that no fraction showed a single discrete narrow zone. The ovarian responses brought about by the injection of purified FSH into intact (6) and hypophysectomized female rats are presented in Table III.
Examination of these data shows that   very low doses of purified FSH relative to the ovine NIH-FSH standard were required for equivalent stimulation of ovarian weights with both bioassay methods. The ovaries obtained from hypophysectomized rats injected with purified FSH consisted of clear follicles with no evidence of luteinization as judged by gross and histological examination (30). Low doses of purified FSH also increased uterine weights in hypophysectomized female rats.
The level of LH activity measured in purified FSH varied with each bioassay method used for its detection. Ovarian ascorbate depletion bioassays suggested that purified FSH contained 0.05 unit of NIH-LH-Sl per mg of protein or 0.04 unit per mg of dry weight (Table II) rats, the ventral prostates, as well as the testes of the rats which received purified FSH, were stimulated to a greater degree than the rats which received NIH-LHSll at both doses (Table IV). Moreover, the seminal vesicles of the animals which received the high dose of purified FSH were heavier than the seminal vesicles of the animals which received the high dose of NIH-LH-Sll. Although the mean ventral prostate weights obtained with the low doses of both preparations (in both bioassays) were significantly above the 0.9% NaCl solution-treated controls, neither bioassay provided a sufficient dose response with the NIH-LH standard to allow a statistically valid calculation of the relative potency. Nevertheless, consideration of the magnitude of the responses obtained with purified FSR relative to those obtained with similar doses of NIH-LH clearly suggested that. purified FSH contained more than 1 unit of NIH-I&S11 per mg of dry weight as determined by the ventral prostate bioassay (8).
Mensuremcnt of immunoreactive LH by means of radioimmunoassay indicated 1.2 ng of the KIH-LH standard (LER 10%X2) were present in 1 pg of purified FSH, while 40 ng of LH were detected per pg of NIH-FSIJ-S4.
This indicated that the level of immunoreactive LH in purified FSH was only 3% the level of immunoreactive LH in the ovine NIH-FSH standard. An,alyses of Physicochemical Homogeneiky of Purified FSH-Analytical acrylamide disc gel electrophoresis of purified FSH at pH S.9 (16) displayed a single broad protein zone near the center of the gel (Fig. 6g). When analytical electrophoresis was conducted at pH 4.3 (22), the protein failed to migrate sufficient distance for a critical estimation of purity.
Repeated double diffusion in agar gel with absorbed anti-FSH (A-FXH) and absorbed anti-LIT (A-IX) against purified FSH (FSH) displayed a diffuse precipitin line (a) near the anti-FSH well after 48 hours (Fig. 7a). Two additional thin precipitin lines (b, c) were apparent after 96 hours (Fig. 7b). No precipitin line was seen between anti-LH and purified FSH. The halos surrounding the antiserum wells which could not be removed with repeated 0.9% NaCl solution and distilled water washings were thought to be artifacts. Molecular Weight Determination (Fig. 8)-The hydroxylapatite fraction was centrifuged at two speeds to enable duplicate calculations of the molecular weight.
The apparent weightaverage molecular weights at 40,000 and 48,000 rpm were 33,800 and 32,700, respectively.
Isoelectric Point Determination (Fig. Q)-Electrofocusing of the hydroxylapatite fraction resulted in the concentration of most of the FSH activity around pH 4.6. The nature of the absorbing peaks at the extremes of the pI1 gradient associated with the anode and cathode buffer solutions is not known.
Protein determinations of these peaks have produced essentially negative results.
Amino Acid Analysis-The results of amino acid analysis of purified FSH are presented in Table V. In general, very good agreement was obtained between the two determinations made with this preparation.
Threonine and half-cystine were the most abundant amino acids followed by glutamic acid, lysine, and aspartic acid, respectively.
The least abundant amino acids were methionine, tryptophan, and phenylalanine.

DISCUSSION
In response to reports by Ellis (21) and Papkoff et al. (3,31), metaphosphoric acid was used to precipitate inert proteins from the Koenig and King extract.
During the course of early experi ments, it was discovered that the use of dilute phosphate buffe by guest on March 23, 2020 http://www.jbc.org/ as solvent prevented a drastic drop in pH with the addition of metaphosphoric acid. Firm control of the pH around ~114.0 was considered desirable since the incubation of sheep LH at pH 4.0 (32), PI-I 2.0 (33), and pI1 1. 3 (31) was reported to cause a loss of LH activity.
With this technique, inert proteins which comprised about 5Ooi, of the extract were precipitated.
It was considered unlikely that sheep FSH and LH could be completely separated on the basis of one purification step. Therefore, a technique which separated these hormones to a high degree was sought early in the purification procedure with the point of view that subsequent purification steps which resolve FSH and LH would further reduce LH contamination.
Following the pioneering efforts of Evans,Simpson,and Pencharz (34) and Jensen et al. (35) most investigators (3-5, 21, 36-38) have used the differential solubility of sheep gonadotropins in 0.5 saturated (NH&SO4 to separate FSH from LH. In agreement with these authors, the supernatant (NH&SO4 FSH fraction obtained by salt fractionation contained very little LH (Table II). Two subsequent purification steps, hydroxylapatite chromatography and preparative acrylamide disc gel electrophoresis, were used to remove contaminating LH as well as inert protein based on the observations that these two procedures served to resolve sheep FSH from LH (30). The level of LH in purified FSH was low as determined by ovarian ascorbate depletion bioassay (Table II), radioimmunoassay, and double diffusion in agar gel (Fig. 7). Although radioimmunoassay indicated that purified FSH contained only 3% as much immunoreactive LH as the NIH sheep FSH standard, NIH-FSH-S4, ovarian ascorbate depletion bioassays suggested that purified FSH contained more ovarian ascorbate depletion activity than NIH-FSH-S4 (0.04 unit and 0.017 unit of NIH-LII-Sl per mg, respectively). Moreover, the level of LH in purified FSH indicated by the ventral prostate bioassay (8) was several-fold (at least 20-fold) greater than the level indicated by the ovarian ascorbate depletion bioassay. The wide range of results obtained with these assay methods suggest that the three methods are not specifically measuring the same contaminant, namely LH, in the purified FSH preparation.
The specificity of the ovarian ascorbate depletion bioassay for LH (7) has been challenged. Gibson et al. (39) have reported ovarian ascorbate depletion activity in homogenized starch gel, extracts from human cerebral cortex, plasma of hypophysectomized cockerels, and propylene glycol.
DeGroot (40) has found ovarian ascorbate depletion activity in serum from hypophysectomized male rats. Roed and Hamburger (41) recently observed ovarian ascorbate depletion activity in the urine of hypophysectomiaed women as well as children between the ages of 5 and 9 and concluded that the depletion of ovarian ascorbate was probably due to nonspecific proteins in the urine.
The significance of the discrepancy between the results of the ovarian ascorbate depletion bioassay and radioimmunoassay cannot be established at this time; however, when one considers the following-(a) the failure of preparative disc electrophoresis to reduce the ovarian ascorbate depletion activity of purified FSH over the previous purification step (Table II), (b) the low level of immunoreactive LH in purified FSH detected by radioimmunoassay, (c) the high specific FSII activity of purified FSH (Table I), and (d) the previous reports which challenge the specificity of the ovarian ascorbate depletion bioassay (39-41)-one can speculate that a highly purified FSH preparation might contain slight intrinsic ovarian ascorbate depletion activity.
This could account for the higher level of LH contamination suggested by the ovarian ascorbate depletion bioassay than by the radioimmunoassay.
Although Greep,Van Dyke,and Chow (8) proposed that the enlargement of the ventral prostate of hypophysectomized immature male rats was a specific and quantitative bioassay for LH, in recent years conflicting reports have appeared concerning the influence of highly purified sheep FSH on this gland. Cahill et al. (4) reported that their purified sheep FSH preparation did not increase ventral prostate weights, whereas Papkoff (42) and Jutisz et al. (1) increased the weight of this gland when they injected their purified FSH. Relatively low doses of the highly purified FSH obtained in this study stimulated growth of the ventral prostate to a high degree (Table IV).
Based on the many indications of low LH contamination in this preparation, we conclude that the increase in weight of the ventral prostate in hypophysectomized male rats injected with purified FSH cannot be explained solely on the basis of LH contamination.
The mean relative potency of 133 units of NIII-FSII-Sl per mg of dry weight obtained with purified FSH with eight Steelman-Pohley augmentation bioassays (6) indicated that this preparation was more than 2 times as pure as previous sheep FSH preparations (Table I). The low mean protein yield and relatively high recovery of original FSH activity are consistent with this conclusion.
The stimulation of uterine, ventral prostate, and seminal vesicle weights that was obtained in hypophysectomized rats lends support to an earlier report by Papkoff (42) which postulated that purified sheep FSH might possess steroidogenic activity.
A previous preparation of FSH obtained in this laboratory (2) displayed a rather narrow band when subjected to analytical acrylamide electrophoresis (16). The hydroxylapatite fraction (Fig. 6f) described in this study was approximately twice as active as the previously reported FSH; moreover, considerable inert protein was removed from this fraction to obtain purified FSH (Fig. 6g). Therefore, it is suggested that the previously reported FSH preparation contained considerable inert protein in spite of its apparent electrophoretic homogeneity. The broad zone obtained with analytical acrylamide disc gel electrophoresis at pH 8.9 (Fig. 6g) and the rather broad band of biological activity obtained with electrofocusing ( Fig. 9), as well as the diffuse precipitin line obtained after 48 hours with double diffusion in agar gel (Fig. 7a), suggested that the purified FSH might consist of a microheterogeneous population of molecules as defined by Colvin,Smith,and Cook (43) and first proposed for sheep LH by Jutisz and Squire (44). Sheep FSH is a glycoprotein containing neuraminic acid, hexose, and hexosamine (3,4). It appears quite possible that natural alterations in &JO in amino acids or variations in kinds and amounts of the carbohydrates mentioned, or both, may occur among FSH molecules.
Those changes which would not eliminate the biological activity of the molecule might alter its physicochemical properties to the degree necessary for resolution with analytical techniques such as electrophoresis and double diffusion.
The authors also recognize the possibility that alterations of FSH molecules during storage or purification could cause the diffuse bands observed with these analytical techniques.
Molecular weights of purified sheep FSH ranging from 30,000 to 33,900 have been determined by ultracentrifugation (1,2,4). The apparent weight-average molecular weights obtained by sedimentation equilibrium are in close agreement with these reports. Although the FSH fraction used was 50% as pure as Sheep Pituitary Follicle-stimulating Hormone Vol. 245, Nix 9 purified FSH, the plot of fringe displacement (In c) against the square of the radial distance (x2) was linear (Fig. 8). This suggests that the inert contaminating substances, which may include inactivated FSH, were physicochemically very similar to FSH. The isoelectric point of pH 4.6 obtained for sheep FSH is in close agreement with the earlier studies of Li and Pedersen (45), Raacke,Lostroh,and Li (46), and Jutisz et al. (1) who reported isoelectric points ranging from pH 4.4 to pH 4.6.
The amino acid content of purified FSH showed rather notable differences when compared to the amino acid contents of other sheep FSH preparations (Table V). Specifically, the threonine and half-cystine content of purified FSH was higher than previously reported values, whereas the phenylalanine content of purified FSH was lower than previously reported. The somewhat reduced serine, proline, glycine, and leucine content may also be significant.
In conclusion, a reproducible procedure has been developed for the preparation of highly purified FSH from sheep pituitary glands. The specific FSH activity of this preparation (188 times NIH-FSH-Sl per mg of protein or 133 times NIH-FSH-Sl per mg of dry weight) indicated that it was 2 to 3 times as active as sheep FSH preparations previously obtained.