Swedish Medical Research Council, Research Unit and the Paediatric Endocrinology Unit, Karolinska sjukhuset, 10401 Stockholm 60
Reproductive Endocrinology
A SENSITIVE AND SPECIFIC IN VITRO BIOASSAY METHOD FOR THE MEASUREMENT OF FOLLICLE-STIMULATING HORMONE ACTIVITY
By M.-P. Van Damme, D. M. Robertson, R. Marana, E. M. Ritz\l=e'\nand E. Diczfalusy ABSTRACT An in vitro bioassay method for hFSH is presented. The method is based on the principles previously described by Dorrington et al. (1976b) and involves the assay of oestradiol produced from 19-hydroxyandrostenedione by dispersed Sertoli cells of 10-day old rats when cultured in the presence of graded doses of FSH. Using the 1st International Reference Preparation for human pituitary gonadotrophins (FSH and LH/ICSH) for bioassay (code no. 69/104) as standard, the useful range of the method is from 0.5 to 32 mIU/chamber (2 to 128 mIU/ml). The sensitivity of the method is 0.5 mIU/chamber. The mean index of precision obtained from 16 multiple assays over 2 or 3 dose levels was 0.084. Parallelism was obtained between the 69/104 preparation and all preparations under study. The practicability of the proposed assay method is such that 15 preparations at 3 dose levels can be assayed by one person in 3 days. The specificity of the assay was investigated by determining the FSH activity in the following preparations: hFSHa-\g=a\and \g=b\-subunits, hLH, hCG, hTSH, ACTH, human growth hormone (hGH) human prolactin (hPRL) and luteinizing hormone-releasing hormone (LH-RH). The ACTH, hGH, hPRL and LH-RH preparations studied showed no detectable FSH activity in the assay. In the remaining preparations very low levels of FSH activity were found, corresponding to 0.004 to 0.6 % of the weight of ') Ford Foundation Fellow in Reproductive Endocrinology 2) To whom reprint requests should be directed
these preparations when compared with a highly purified hFSH preparation, suggesting that the method is specific for FSH. The possible synergistic or antagonistic influence of the above preparations when assayed in the presence of the 69/104 preparation was also assessed. No evidence of a synergistic or antogonistic effect was found. The assay of the hFSH potencies of a limited number of hFSH preparations of varying purity by the proposed in vitro bioassay, an hFSH radioreceptor method and an hFSH specific radioimmunoassay technique revealed that \p=n-\although the relationship of the various potencies obtained with each method showed a close agreement \p=n-\ the bioassays yielded the highest potency estimates, and the radioimmunoassays the lowest ones. Since the proposed bioassay method is sensitive and considered to be specific for hFSH activity, it provides a suitable basis for the assessment of the specificity of other in vitro methods (radioreceptor and radioimmunoassay) currently used for detecting low levels of FSH activity.
In contrast to the assay of LH, there is no in vitro bioassay of FSH currently available which is sufficiently sensitive and precise to detect the low levels of hFSH present in plasma of men and women during the fertile age. The im¬ portance of such an assay can be gauged by the differences noted between various studies on the profile ot immunologically active hFSH in plasma during the menstrual cycle (for reviews see Taymor 8c Miyata 1969; Franchimont 8c Burger 1975; Diczfalusy 8c Landgren 1977) in which the magnitude of the midcycle hFSH peak can differ markedly. In addition, the correlation between hFSH in serum and urine is not as satisfactory as that of hLH (cf. Wide et al. 1973) and major discrepancies exist in the profile of hFSH in urine during the normal menstrual cycle, when assayed by biological and immunological methods (cf. Diczfalusy 8c Landgren 1977). Many of the above problems could be re¬ solved if a sensitive and specific bioassay method would be available for hFSH. Hence, the purpose of this study was to develop a sensitive in vitro bioassay method for the quantitation of hFSH. which fulfils the recognized criteria of reliability. The Sertoli cell is a well established target for FSH but not for other known peptide hormones (Hansson et al. 1973; Means 1977). Therefore, the FSH-dependent secretion of the testicular androgen binding protein (ABP) specific for the Sertoli cell was first evaluated as an endpoint of FSH action. However, due to high baseline secretion of ABP and poor sensitivity, this ap¬ proach had to be abandoned. The method described below is based on the initial observations of Dorrington 8c Armstrong (1975) and Dorrington et al. (19766) that a dose-response relationship exists between the amount of oestradiol pro¬ duced when FSH but not LH is added to cultured Sertoli cells obtained from immature rats. The experimental approach employed in the present study was similar to that previously used to develop a reliable in vitro bioassay method for LH activity (Van Damme et al. 1974).
MATERIALS
Abbreviations EM-I: chicken
minimum essential medium containing L-glutamine (2 ihm), 2 °/o penicillin (100 IU/ml), streptomycin (50 /ig/ml) and fungizone (0.63
Eagle's serum,
Mg/m\).
EM-II: EM-I bulfer containing 0.1 % bovine serum albumin instead of chicken serum. 19-OH-A: 19-hydroxy-androstenedionc (19-hydroxy-4-androstene-3,17-dione), LHRH: luteinizing hormone-releasing hormone, hGH: human growth hormone, hPRL: human prolactin, RIA: radioimmunoassay, RRA: radioreceptor assay, / index of precision of Gaddum (1933), measured as the ratio of the standard deviation of the response metameter to the slope of the dose-response line. =
Preparations International Reference Preparation. Human pituitary gonadotrophins (1st IRP of human pituitary gonadotrophins (FSH and LH/ICSH) for bioassay (code number 69/104)) with a stated content of 10 IU FSH activity per ampoule was provided by the National Institute of Biological Standards and Control (NIBSC), London. -
Hormone preparations. Highly purified preparations of hFSH (Lot No. 19344), (Batch 1974, 7089 (6770:7408)IU/mg; in vitro bioassay standard 69/104), hTSH (Lot No. 19365, 10.1 IU/mg; RIA standard 68/38 (NIBSC, London) data provided by Kabi Diagnostica) and hGH (Lot No. 19407, 2 IU/mg) were obtained from Kabi Diagnostica, Stockholm. Highly purified preparations of hFSH (LER 1801-3), and of hFSHa- and /i-subunits (hFSH«, N-594-A and hFSH/?, N-596-C) and ACTH (syn¬ thetic) were provided by the National Pituitary Agency, NIH. Bethesda, Md., USA. Synthetic LH-RH was obtained from Hoechst A.G., Frankfurt, Germany, and a highly purified hPRL preparation (Lot No. 389521) was purchased from Calbiochem, San Diego, California, USA. A highly purified urinary hCG preparation (12 800 IU/mg, Lot No. E 477 ter-2) was purchased from Istituto Farmacologico Serono, Rome, Italy. A saline extract was prepared from a pool of seven pituitaries obtained post mortem from post-menopausal women and used without further purification. hLH
-
Reagents. 19-Hydroxyandrostenedione was obtained from Sigma, St. Louis, Mis¬ souri, USA. Fungizone, chicken serum, HEPES and Earle's buffers and Eagle's medium were obtained from Flow Laboratories, Irvine, Scotland, UK. The tissue culture plates. -
containing 96 chambers of 300 a\ capacity, were obtained from Falcon Plastics (Type 3040), Oxnard, California, USA. [2,4,6,73H] Oestradiol (85 Ci/mM) was pur¬ chased from the Radiochemical Centre, Amersham, England. The anti-oestradiol serum used was a gift from Dr. J.-P. Raynaud, Roussel UCLAF Research Centre, France. This antiserum was raised against bovine serum albumin conjugated with 3,17/?dihydroxy-l,3,5(10)-oestratrien-7-one (O-carboxymethyl) oxime. The antiserum showed low cross-reaction with oestrone (3.6%), oestriol (3.6%) testosterone (< 0.01 °/o) and 19-OH-A (< 0.01 %). Animals. Sprague-Dawley rats (aged 5,10,12,14 and 20 days) were purchased from each
AB Anticimex. Stockholm. -
METHODS
method
Proposed
Assay design. A parallel line assay design (Finney 1978) was employed in each experiment with the amount of oestradiol produced as the response. Each dose level consisted of 4 replicates with a ratio of two between successive doses. This design was modelled on that used for an in vitro bioassay of LH activity (Van Damme et al. 1974). -
Preparation of ication testes
to
cells.
The
are
Dorrington el al. (1975) without modif¬ Sprague-Dawley strain. Decapsulated trypsin and deoxyribonuclease digestion in order
procedure
used, employing 10-day old -
was
minced
prior
to
a
combined
of
rats of the
separate the interstitial cells from the seminiferous tubules. After the addition of
trypsin inhibitor, the tubules are digested with collagenase. The resulting suspension is then centrifuged and re-suspended in EM-I medium. Under the experimental conditions employed, Sertoli cells from the testes of 36 (10-day old) rats
soya bean
cell are
sufficient for 220 chambers.
Initial culture of cells. Aliquote (200 µ ) of the cell suspension are dispensed into tissue culture chambers (300 ,ul capacity) and incubated at 37°C in an 0¿/COL) (93.5/6.5) atmosphere. After 21-24 h, the attached cells are washed twice with EM-II buffer. -
To each chamber 19-OH-A in EM-II (50 µ , Incubation conditions with hFSH. 10-('m) (Dorrington et al. 1976 ) is added, followed by a dilution (200 µ\ in EM-II) ol the standard or unknown sample. The chambers are gassed with 93.5 : 6.5 Oo/COo and incubated in a gassed container for 24-26 h at 37°C.
2.5
—
Radioimmunoassay of oestradiol. The RIA procedure employed is based on the method of Van Damme et al. (1973). In brief: 200 µ\ sample, 100 «1 [3H]oestradiol (18 000 dpm) and anti-oestradiol serum (öxlG1-12 moles oestradiol binding sites/tube) 10 min followed by 30°Cx30 min. Charcoal (500 µ , 0.5 °/o are incubated at 60°C Norit A) is added and allowed to stand for 15 min at 4°C. After centrifugaron, the radioactivity in the supernatant is counted as previously described (Van Damme et al. 1974). The levels of oestradiol in the samples are calculated from a simultaneously prepared oestradiol standard curve after a logit-log dose transformation. The sensitivity of the RIA (logit + 2) is approximately 5 pg/tube. That the product of incubation was indeed oestradiol was ascertained by crystal¬ lization to constant specific activity of the -5H-labelled oestradiol formed in the assay system following incubation with 3H-labelled testosterone. The details of these spe¬ cificity studies will be published elsewhere. -
Melamelric transformations Several transformations of the response (pg of oestradiol produced) were examined in order to extend the linear region of the log dose response line and to reduce the heterogeneity of variances. As in a previous study on the development of an in vitro bioassay for LH (Van Damme et al. 1974), the untransformed response (pg) was com¬ pared with various transformations (pg1/2 or log pg). In contrast to the untransformed response line (Fig. 1), it was possible to linearize the response over 4 to 5 doses after either the square root or logarithmic transformation, although the former was less effective below 2 mlU/chamber and the latter above 8 mlU/chamber. Heterogeneity of variances as assessed by the test of Barllett (1937) was observed over a 16-fold dilution in all experiments examined (n=10), using the untransformed data; in 6/10
with the square root transformation and only in 1/10 when the logarithmic transformation was used. Based on these findings, the logarithmic transformation was used in the 0.5-8 mlU region of the response line. Above this dose level, the use of the square root transformation gave a straight line relationship, with no evidence of heterogeneity of variances over a maximum of three doses. Thus, in this region of the dose response line, the square root transformation can be used.
experiments
Standard
preparation
In all studies, unless otherwise stated, the 69/104 preparation
was
used
as
standard.
Variables
affecting the assay animals. The influence of the age of the rats on the various characteristics of the FSH dose response curve was investigated. Sertoli cell preparations at concen¬ trations proportional to the weight of the original testes were obtained from animals aged 5,10,12,14 and 20 days. The responsiveness of the Sertoli cells was assessed by calculating the amount of oestradiol produced at two dose levels of hFSH (0.5 and 8 mlU). Similar responses were observed with Sertoli cell preparations obtained from 5- and 10-day old animals: however, with older animals (12 to 20 days) the responsiveness decreased, with a cor¬ responding decrease in sensitivity, so that by 20 days the amount of oestradiol produced over control values was non-detectable. Since testes from 5-day old animals are in practice more difficult to handle, 10-day old animals were employed routinely in all Age of
-
bioassays.
Choice of cell concentration. Since cell quantitation is complicated by clumping, the cell concentration was expressed in terms of the equivalent weight or number of the original testes used in each chamber, assuming constant cell losses during pre¬ paration. Using cell preparations obtained from 10-day old animals, final Sertoli cell concentrations corresponding to 0.3 to 1 testis/chamber gave a response curve in which the oestradiol produced could be detected at the lower dose levels of F'SH (0.5-2 mlU). On the other hand, at cell concentrations lower than 0.3 testes/tube, a sharp decrease in the response to hFSH was observed, which was much greater than that which could be expected on the basis of simple cell dilution. Microscopic examination indicated that, under such experimental conditions, there was a disproportionate reduction in the number of cells cultured. The influence of cell concentration on the relative potency of a highly purified hFSH preparation (hF'SH-Kabi), when assayed against the relatively impure reference pre¬ paration (69/104), showed no significant differences over a 6-fold dilution of cells (1.8 to 0.3 testes/chamber), indicating that the cell concentration within this range is not a critical factor. The final cell suspension that was obtained after dispersion of the testicular cells contained aggregates of Sertoli cells and germ cells. During the initial 21-24 h of culture, the Sertoli cells adhered to the bottom of the chamber and spread out to form a monolayer. The germ cells did not adhere, and were therefore mostly rinsed away before the FSH-containing sample was applied. However, degenerating germ cells might constitute up to 5-10% of the total cell number even after rinsing, and fibroblasts up to 1 %. A relative estimate of cell density in the chambers was obtained by microfluorometric measurements of acridine orange binding to the cells after the incubation was com¬ pleted. The cells were first fixed in ethanol and then air dried. Fifty µ of acridine -
(4 /(g/ml in 0.145 M NaCl, 6.5 mM NaHoPOj, 185 added, and after an equilibration period of > 15 min, the 2525 nm fluorescence emitted from a large field (2.43 mm diameter, including sg 30 000 cells) was recorded by a microfluorimeter adapted to an inverted microscope orange in saline
mM
phosphate
Na.jHP04, pH 7.4)
(Cohen (1971).
buffer
was
al. 1974). The acridine orange staining is described by Adams 8c Kamentsky The intra-assay coefficient of variation of the fluorescence recordings was 17.4% (n 220). The coefficient of variation for oestradiol values within triplicates or quadruplicates of identical samples was smaller than that for the ratio oestradiol/ fluorescence, indicating that the former variation was not due to uneven distribution of cells between the chambers. Also, the presence or absence of FSH did not influence the cell density at the end of the culture, as revealed by acridine orange binding. Thus, determination of cell density did not add to the precision of the assay. et
=
Choice of incubation buffer. The influence of the use of various buffers during the incubation of cells was examined. EM-II and Earle's balanced salt solution in the presence and absence of HEPES buffer (0.025 °/o) were investigated in the in¬ cubation of cells with hFSH. The buffer with the highest response as assessed by the amount of oestradiol produced with 8 mlU hF'SH was EM-II (250 pg), followed by Earle's (180 pg), EM-II + HEPES (180 pg) and Earle's + HEPES (49 pg). As a conse¬ quence, EM-II without HEPES was used as the bioassay buffer. -
Use
of 19-OH-A as steroid substrate in the assay The oestradiol produced by Sertoli cells in the presence of hFSH (hFSH-Kabi) was significantly increased (2-3 times) when incubated with 19-OH-A (0.5x10-°m) rather than with the same amount of testosterone as a precursor. These findings confirm the previous observation of Dorrington et al. (1976(7). Since the various characteristics of the dose-response curve (sensitivity, slope) were not adversely influenced by this mod¬ ification, 19-OH-A was used as the steroid substrate in the assay. Choice of incubation conditions. The influence of the time of incubation on both the initial culturing of the cells and the subsequent incubation with hFSH was asses-
Table 1. The influence of the time of incubation on the relative potency of two highly purified hFSH preparations when assayed against the first IRP for human pituitary gonadotrophins (FSH and LH/ICSH) for bioassay (code No. 69/104). The incubation times stated (days) refer to the time employed for the initial culturing of the cells (first), followed by the time of incubation with hFSH (second). The results obtained with the 1 day + 1 day incubation conditions were considered as unity. Potency estimates with 95 °/o fiducial limits from one experiment are presented.
Incubation time (days) Second First
hFSH-Kabi Relative potency 1.00 1.08 1.18 1.05
(0.87 (0.88 (0.99 (0.90
1.15)
1.33)
1.40) 1.21)
hFSH-NIH Relative potency 1.00
0.95 1.04 1.04
(0.73 (0.77 (0.84 (0.92
1.37) 1.18)
1.22) 1.14)
sed; firstly, from the changes in the various characteristics of the dose-response line, and secondly, from the changes in the relative potency of two highly purified hFSH when assayed against a less purified standard preparation (69/104). A small decrease (25 °/o) in the amount ol oestradiol produced was found by extending the initial culturing time from 1 to 2 days. Following a 2-day incubation with hFSH, a higher production of oestradiol (50%) was observed than with a 1-day incubation, with a small increase only (30 %) in the sensitivity of the assay. No significant change in the relative potency of the hFSH preparations was observed over a wide range of incubation conditions (Table 1). Based on these findings, a one-day culturing of the cells, followed by a one-day incubation with hFSH was employed in the assay.
preparations
Temperature of
incubation. A slight increase in the amount of oestradiol produced observed in the lower dose range (1-2 mlU) when incubating at 37°C in com¬ parison with 34°C. Hence 37°C was chosen as the incubation temperature in the assay. was
-
Radioreceptor and radioimmunoassay methods The radioreceptor assay method (RRA) of Cheng (1975), utilizing membrane pre¬ parations from bovine testes, and an hF'SH radioimmunoassay (RIA) method, using an anti-hFSH serum of high specificity (cf. Table 2), were employed. In the RIA the antiserum added to each assay tube was absorbed with 2 IU hCG to increase the spe¬ cificity of the assay. Equilibrium assay conditions (3 days at 4°C) were employed and the incubation was terminated by a second antibody procedure. The [12'I]hFSH pre¬ paration used in both assay methods was prepared by a lactoperoxidase technique and fractionated by a high resolution gel filtration method (97 1.2 cm, Ultrogel AcA 54, LKB Produkter, Bromma, Sweden). This fractionation technique removes the iodinated subunits, which represent a potential source of error in the RIA. The peak tubes cor¬ responding to ['-'IjhFSH were pooled and used as tracer in the assays. Multiple dose, parallel line assay designs with randomization were used in both methods (Finney 1978). A detailed description of these techniques is presented elsewhere (Maraña et al,, submitted for publication).
RESULTS
Reliability of the in vitro bioassay The useful range of doses of the 69/104 standard preparation giving a linear log dose response line when plotted against various transformations (pg1/2 and log pg) of the amount of oestradiol produced was 0.5 to 32 mlU/chamber (or 2 to 128 mIU/ml). The sensitivity of the assay calculated as the smallest re¬ sponse giving a significant difference from control values was 0.5 mlU/chamber. The precision of the assay as characterized by the mean index of precision ( ) was 0.079 in 6 multiple assays at 2 dose levels, and 0.087 in 10 multiple assays at 3 dose levels. The between-assay variation as assessed from the reproducibility of the potency of 6 human gonadotrophin preparations with varying FSH activity when assayed on 2 to 4 occasions was 8.4 °/o. No significant deviation from parallelism with the 69/104 preparation was observed over a 2- to 16-fold dilution of the various
preparations.
200
log pg
o D
Reproductive Endocrinology
A SENSITIVE AND SPECIFIC IN VITRO BIOASSAY METHOD FOR THE MEASUREMENT OF FOLLICLE-STIMULATING HORMONE ACTIVITY
By M.-P. Van Damme, D. M. Robertson, R. Marana, E. M. Ritz\l=e'\nand E. Diczfalusy ABSTRACT An in vitro bioassay method for hFSH is presented. The method is based on the principles previously described by Dorrington et al. (1976b) and involves the assay of oestradiol produced from 19-hydroxyandrostenedione by dispersed Sertoli cells of 10-day old rats when cultured in the presence of graded doses of FSH. Using the 1st International Reference Preparation for human pituitary gonadotrophins (FSH and LH/ICSH) for bioassay (code no. 69/104) as standard, the useful range of the method is from 0.5 to 32 mIU/chamber (2 to 128 mIU/ml). The sensitivity of the method is 0.5 mIU/chamber. The mean index of precision obtained from 16 multiple assays over 2 or 3 dose levels was 0.084. Parallelism was obtained between the 69/104 preparation and all preparations under study. The practicability of the proposed assay method is such that 15 preparations at 3 dose levels can be assayed by one person in 3 days. The specificity of the assay was investigated by determining the FSH activity in the following preparations: hFSHa-\g=a\and \g=b\-subunits, hLH, hCG, hTSH, ACTH, human growth hormone (hGH) human prolactin (hPRL) and luteinizing hormone-releasing hormone (LH-RH). The ACTH, hGH, hPRL and LH-RH preparations studied showed no detectable FSH activity in the assay. In the remaining preparations very low levels of FSH activity were found, corresponding to 0.004 to 0.6 % of the weight of ') Ford Foundation Fellow in Reproductive Endocrinology 2) To whom reprint requests should be directed
these preparations when compared with a highly purified hFSH preparation, suggesting that the method is specific for FSH. The possible synergistic or antagonistic influence of the above preparations when assayed in the presence of the 69/104 preparation was also assessed. No evidence of a synergistic or antogonistic effect was found. The assay of the hFSH potencies of a limited number of hFSH preparations of varying purity by the proposed in vitro bioassay, an hFSH radioreceptor method and an hFSH specific radioimmunoassay technique revealed that \p=n-\although the relationship of the various potencies obtained with each method showed a close agreement \p=n-\ the bioassays yielded the highest potency estimates, and the radioimmunoassays the lowest ones. Since the proposed bioassay method is sensitive and considered to be specific for hFSH activity, it provides a suitable basis for the assessment of the specificity of other in vitro methods (radioreceptor and radioimmunoassay) currently used for detecting low levels of FSH activity.
In contrast to the assay of LH, there is no in vitro bioassay of FSH currently available which is sufficiently sensitive and precise to detect the low levels of hFSH present in plasma of men and women during the fertile age. The im¬ portance of such an assay can be gauged by the differences noted between various studies on the profile ot immunologically active hFSH in plasma during the menstrual cycle (for reviews see Taymor 8c Miyata 1969; Franchimont 8c Burger 1975; Diczfalusy 8c Landgren 1977) in which the magnitude of the midcycle hFSH peak can differ markedly. In addition, the correlation between hFSH in serum and urine is not as satisfactory as that of hLH (cf. Wide et al. 1973) and major discrepancies exist in the profile of hFSH in urine during the normal menstrual cycle, when assayed by biological and immunological methods (cf. Diczfalusy 8c Landgren 1977). Many of the above problems could be re¬ solved if a sensitive and specific bioassay method would be available for hFSH. Hence, the purpose of this study was to develop a sensitive in vitro bioassay method for the quantitation of hFSH. which fulfils the recognized criteria of reliability. The Sertoli cell is a well established target for FSH but not for other known peptide hormones (Hansson et al. 1973; Means 1977). Therefore, the FSH-dependent secretion of the testicular androgen binding protein (ABP) specific for the Sertoli cell was first evaluated as an endpoint of FSH action. However, due to high baseline secretion of ABP and poor sensitivity, this ap¬ proach had to be abandoned. The method described below is based on the initial observations of Dorrington 8c Armstrong (1975) and Dorrington et al. (19766) that a dose-response relationship exists between the amount of oestradiol pro¬ duced when FSH but not LH is added to cultured Sertoli cells obtained from immature rats. The experimental approach employed in the present study was similar to that previously used to develop a reliable in vitro bioassay method for LH activity (Van Damme et al. 1974).
MATERIALS
Abbreviations EM-I: chicken
minimum essential medium containing L-glutamine (2 ihm), 2 °/o penicillin (100 IU/ml), streptomycin (50 /ig/ml) and fungizone (0.63
Eagle's serum,
Mg/m\).
EM-II: EM-I bulfer containing 0.1 % bovine serum albumin instead of chicken serum. 19-OH-A: 19-hydroxy-androstenedionc (19-hydroxy-4-androstene-3,17-dione), LHRH: luteinizing hormone-releasing hormone, hGH: human growth hormone, hPRL: human prolactin, RIA: radioimmunoassay, RRA: radioreceptor assay, / index of precision of Gaddum (1933), measured as the ratio of the standard deviation of the response metameter to the slope of the dose-response line. =
Preparations International Reference Preparation. Human pituitary gonadotrophins (1st IRP of human pituitary gonadotrophins (FSH and LH/ICSH) for bioassay (code number 69/104)) with a stated content of 10 IU FSH activity per ampoule was provided by the National Institute of Biological Standards and Control (NIBSC), London. -
Hormone preparations. Highly purified preparations of hFSH (Lot No. 19344), (Batch 1974, 7089 (6770:7408)IU/mg; in vitro bioassay standard 69/104), hTSH (Lot No. 19365, 10.1 IU/mg; RIA standard 68/38 (NIBSC, London) data provided by Kabi Diagnostica) and hGH (Lot No. 19407, 2 IU/mg) were obtained from Kabi Diagnostica, Stockholm. Highly purified preparations of hFSH (LER 1801-3), and of hFSHa- and /i-subunits (hFSH«, N-594-A and hFSH/?, N-596-C) and ACTH (syn¬ thetic) were provided by the National Pituitary Agency, NIH. Bethesda, Md., USA. Synthetic LH-RH was obtained from Hoechst A.G., Frankfurt, Germany, and a highly purified hPRL preparation (Lot No. 389521) was purchased from Calbiochem, San Diego, California, USA. A highly purified urinary hCG preparation (12 800 IU/mg, Lot No. E 477 ter-2) was purchased from Istituto Farmacologico Serono, Rome, Italy. A saline extract was prepared from a pool of seven pituitaries obtained post mortem from post-menopausal women and used without further purification. hLH
-
Reagents. 19-Hydroxyandrostenedione was obtained from Sigma, St. Louis, Mis¬ souri, USA. Fungizone, chicken serum, HEPES and Earle's buffers and Eagle's medium were obtained from Flow Laboratories, Irvine, Scotland, UK. The tissue culture plates. -
containing 96 chambers of 300 a\ capacity, were obtained from Falcon Plastics (Type 3040), Oxnard, California, USA. [2,4,6,73H] Oestradiol (85 Ci/mM) was pur¬ chased from the Radiochemical Centre, Amersham, England. The anti-oestradiol serum used was a gift from Dr. J.-P. Raynaud, Roussel UCLAF Research Centre, France. This antiserum was raised against bovine serum albumin conjugated with 3,17/?dihydroxy-l,3,5(10)-oestratrien-7-one (O-carboxymethyl) oxime. The antiserum showed low cross-reaction with oestrone (3.6%), oestriol (3.6%) testosterone (< 0.01 °/o) and 19-OH-A (< 0.01 %). Animals. Sprague-Dawley rats (aged 5,10,12,14 and 20 days) were purchased from each
AB Anticimex. Stockholm. -
METHODS
method
Proposed
Assay design. A parallel line assay design (Finney 1978) was employed in each experiment with the amount of oestradiol produced as the response. Each dose level consisted of 4 replicates with a ratio of two between successive doses. This design was modelled on that used for an in vitro bioassay of LH activity (Van Damme et al. 1974). -
Preparation of ication testes
to
cells.
The
are
Dorrington el al. (1975) without modif¬ Sprague-Dawley strain. Decapsulated trypsin and deoxyribonuclease digestion in order
procedure
used, employing 10-day old -
was
minced
prior
to
a
combined
of
rats of the
separate the interstitial cells from the seminiferous tubules. After the addition of
trypsin inhibitor, the tubules are digested with collagenase. The resulting suspension is then centrifuged and re-suspended in EM-I medium. Under the experimental conditions employed, Sertoli cells from the testes of 36 (10-day old) rats
soya bean
cell are
sufficient for 220 chambers.
Initial culture of cells. Aliquote (200 µ ) of the cell suspension are dispensed into tissue culture chambers (300 ,ul capacity) and incubated at 37°C in an 0¿/COL) (93.5/6.5) atmosphere. After 21-24 h, the attached cells are washed twice with EM-II buffer. -
To each chamber 19-OH-A in EM-II (50 µ , Incubation conditions with hFSH. 10-('m) (Dorrington et al. 1976 ) is added, followed by a dilution (200 µ\ in EM-II) ol the standard or unknown sample. The chambers are gassed with 93.5 : 6.5 Oo/COo and incubated in a gassed container for 24-26 h at 37°C.
2.5
—
Radioimmunoassay of oestradiol. The RIA procedure employed is based on the method of Van Damme et al. (1973). In brief: 200 µ\ sample, 100 «1 [3H]oestradiol (18 000 dpm) and anti-oestradiol serum (öxlG1-12 moles oestradiol binding sites/tube) 10 min followed by 30°Cx30 min. Charcoal (500 µ , 0.5 °/o are incubated at 60°C Norit A) is added and allowed to stand for 15 min at 4°C. After centrifugaron, the radioactivity in the supernatant is counted as previously described (Van Damme et al. 1974). The levels of oestradiol in the samples are calculated from a simultaneously prepared oestradiol standard curve after a logit-log dose transformation. The sensitivity of the RIA (logit + 2) is approximately 5 pg/tube. That the product of incubation was indeed oestradiol was ascertained by crystal¬ lization to constant specific activity of the -5H-labelled oestradiol formed in the assay system following incubation with 3H-labelled testosterone. The details of these spe¬ cificity studies will be published elsewhere. -
Melamelric transformations Several transformations of the response (pg of oestradiol produced) were examined in order to extend the linear region of the log dose response line and to reduce the heterogeneity of variances. As in a previous study on the development of an in vitro bioassay for LH (Van Damme et al. 1974), the untransformed response (pg) was com¬ pared with various transformations (pg1/2 or log pg). In contrast to the untransformed response line (Fig. 1), it was possible to linearize the response over 4 to 5 doses after either the square root or logarithmic transformation, although the former was less effective below 2 mlU/chamber and the latter above 8 mlU/chamber. Heterogeneity of variances as assessed by the test of Barllett (1937) was observed over a 16-fold dilution in all experiments examined (n=10), using the untransformed data; in 6/10
with the square root transformation and only in 1/10 when the logarithmic transformation was used. Based on these findings, the logarithmic transformation was used in the 0.5-8 mlU region of the response line. Above this dose level, the use of the square root transformation gave a straight line relationship, with no evidence of heterogeneity of variances over a maximum of three doses. Thus, in this region of the dose response line, the square root transformation can be used.
experiments
Standard
preparation
In all studies, unless otherwise stated, the 69/104 preparation
was
used
as
standard.
Variables
affecting the assay animals. The influence of the age of the rats on the various characteristics of the FSH dose response curve was investigated. Sertoli cell preparations at concen¬ trations proportional to the weight of the original testes were obtained from animals aged 5,10,12,14 and 20 days. The responsiveness of the Sertoli cells was assessed by calculating the amount of oestradiol produced at two dose levels of hFSH (0.5 and 8 mlU). Similar responses were observed with Sertoli cell preparations obtained from 5- and 10-day old animals: however, with older animals (12 to 20 days) the responsiveness decreased, with a cor¬ responding decrease in sensitivity, so that by 20 days the amount of oestradiol produced over control values was non-detectable. Since testes from 5-day old animals are in practice more difficult to handle, 10-day old animals were employed routinely in all Age of
-
bioassays.
Choice of cell concentration. Since cell quantitation is complicated by clumping, the cell concentration was expressed in terms of the equivalent weight or number of the original testes used in each chamber, assuming constant cell losses during pre¬ paration. Using cell preparations obtained from 10-day old animals, final Sertoli cell concentrations corresponding to 0.3 to 1 testis/chamber gave a response curve in which the oestradiol produced could be detected at the lower dose levels of F'SH (0.5-2 mlU). On the other hand, at cell concentrations lower than 0.3 testes/tube, a sharp decrease in the response to hFSH was observed, which was much greater than that which could be expected on the basis of simple cell dilution. Microscopic examination indicated that, under such experimental conditions, there was a disproportionate reduction in the number of cells cultured. The influence of cell concentration on the relative potency of a highly purified hFSH preparation (hF'SH-Kabi), when assayed against the relatively impure reference pre¬ paration (69/104), showed no significant differences over a 6-fold dilution of cells (1.8 to 0.3 testes/chamber), indicating that the cell concentration within this range is not a critical factor. The final cell suspension that was obtained after dispersion of the testicular cells contained aggregates of Sertoli cells and germ cells. During the initial 21-24 h of culture, the Sertoli cells adhered to the bottom of the chamber and spread out to form a monolayer. The germ cells did not adhere, and were therefore mostly rinsed away before the FSH-containing sample was applied. However, degenerating germ cells might constitute up to 5-10% of the total cell number even after rinsing, and fibroblasts up to 1 %. A relative estimate of cell density in the chambers was obtained by microfluorometric measurements of acridine orange binding to the cells after the incubation was com¬ pleted. The cells were first fixed in ethanol and then air dried. Fifty µ of acridine -
(4 /(g/ml in 0.145 M NaCl, 6.5 mM NaHoPOj, 185 added, and after an equilibration period of > 15 min, the 2525 nm fluorescence emitted from a large field (2.43 mm diameter, including sg 30 000 cells) was recorded by a microfluorimeter adapted to an inverted microscope orange in saline
mM
phosphate
Na.jHP04, pH 7.4)
(Cohen (1971).
buffer
was
al. 1974). The acridine orange staining is described by Adams 8c Kamentsky The intra-assay coefficient of variation of the fluorescence recordings was 17.4% (n 220). The coefficient of variation for oestradiol values within triplicates or quadruplicates of identical samples was smaller than that for the ratio oestradiol/ fluorescence, indicating that the former variation was not due to uneven distribution of cells between the chambers. Also, the presence or absence of FSH did not influence the cell density at the end of the culture, as revealed by acridine orange binding. Thus, determination of cell density did not add to the precision of the assay. et
=
Choice of incubation buffer. The influence of the use of various buffers during the incubation of cells was examined. EM-II and Earle's balanced salt solution in the presence and absence of HEPES buffer (0.025 °/o) were investigated in the in¬ cubation of cells with hFSH. The buffer with the highest response as assessed by the amount of oestradiol produced with 8 mlU hF'SH was EM-II (250 pg), followed by Earle's (180 pg), EM-II + HEPES (180 pg) and Earle's + HEPES (49 pg). As a conse¬ quence, EM-II without HEPES was used as the bioassay buffer. -
Use
of 19-OH-A as steroid substrate in the assay The oestradiol produced by Sertoli cells in the presence of hFSH (hFSH-Kabi) was significantly increased (2-3 times) when incubated with 19-OH-A (0.5x10-°m) rather than with the same amount of testosterone as a precursor. These findings confirm the previous observation of Dorrington et al. (1976(7). Since the various characteristics of the dose-response curve (sensitivity, slope) were not adversely influenced by this mod¬ ification, 19-OH-A was used as the steroid substrate in the assay. Choice of incubation conditions. The influence of the time of incubation on both the initial culturing of the cells and the subsequent incubation with hFSH was asses-
Table 1. The influence of the time of incubation on the relative potency of two highly purified hFSH preparations when assayed against the first IRP for human pituitary gonadotrophins (FSH and LH/ICSH) for bioassay (code No. 69/104). The incubation times stated (days) refer to the time employed for the initial culturing of the cells (first), followed by the time of incubation with hFSH (second). The results obtained with the 1 day + 1 day incubation conditions were considered as unity. Potency estimates with 95 °/o fiducial limits from one experiment are presented.
Incubation time (days) Second First
hFSH-Kabi Relative potency 1.00 1.08 1.18 1.05
(0.87 (0.88 (0.99 (0.90
1.15)
1.33)
1.40) 1.21)
hFSH-NIH Relative potency 1.00
0.95 1.04 1.04
(0.73 (0.77 (0.84 (0.92
1.37) 1.18)
1.22) 1.14)
sed; firstly, from the changes in the various characteristics of the dose-response line, and secondly, from the changes in the relative potency of two highly purified hFSH when assayed against a less purified standard preparation (69/104). A small decrease (25 °/o) in the amount ol oestradiol produced was found by extending the initial culturing time from 1 to 2 days. Following a 2-day incubation with hFSH, a higher production of oestradiol (50%) was observed than with a 1-day incubation, with a small increase only (30 %) in the sensitivity of the assay. No significant change in the relative potency of the hFSH preparations was observed over a wide range of incubation conditions (Table 1). Based on these findings, a one-day culturing of the cells, followed by a one-day incubation with hFSH was employed in the assay.
preparations
Temperature of
incubation. A slight increase in the amount of oestradiol produced observed in the lower dose range (1-2 mlU) when incubating at 37°C in com¬ parison with 34°C. Hence 37°C was chosen as the incubation temperature in the assay. was
-
Radioreceptor and radioimmunoassay methods The radioreceptor assay method (RRA) of Cheng (1975), utilizing membrane pre¬ parations from bovine testes, and an hF'SH radioimmunoassay (RIA) method, using an anti-hFSH serum of high specificity (cf. Table 2), were employed. In the RIA the antiserum added to each assay tube was absorbed with 2 IU hCG to increase the spe¬ cificity of the assay. Equilibrium assay conditions (3 days at 4°C) were employed and the incubation was terminated by a second antibody procedure. The [12'I]hFSH pre¬ paration used in both assay methods was prepared by a lactoperoxidase technique and fractionated by a high resolution gel filtration method (97 1.2 cm, Ultrogel AcA 54, LKB Produkter, Bromma, Sweden). This fractionation technique removes the iodinated subunits, which represent a potential source of error in the RIA. The peak tubes cor¬ responding to ['-'IjhFSH were pooled and used as tracer in the assays. Multiple dose, parallel line assay designs with randomization were used in both methods (Finney 1978). A detailed description of these techniques is presented elsewhere (Maraña et al,, submitted for publication).
RESULTS
Reliability of the in vitro bioassay The useful range of doses of the 69/104 standard preparation giving a linear log dose response line when plotted against various transformations (pg1/2 and log pg) of the amount of oestradiol produced was 0.5 to 32 mlU/chamber (or 2 to 128 mIU/ml). The sensitivity of the assay calculated as the smallest re¬ sponse giving a significant difference from control values was 0.5 mlU/chamber. The precision of the assay as characterized by the mean index of precision ( ) was 0.079 in 6 multiple assays at 2 dose levels, and 0.087 in 10 multiple assays at 3 dose levels. The between-assay variation as assessed from the reproducibility of the potency of 6 human gonadotrophin preparations with varying FSH activity when assayed on 2 to 4 occasions was 8.4 °/o. No significant deviation from parallelism with the 69/104 preparation was observed over a 2- to 16-fold dilution of the various
preparations.
200
log pg
o D
