| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Obstetrics and Gynecology, Northwestern University Medical School, Chicago, Illinois 60611
b Department of Neurobiology and Physiology, Northwestern University, Evanston, Illinois 60208
c Division of Endocrinology, Metabolism, and Molecular Medicine, Northwestern University Medical School, Chicago, Illinois 60611
d Reproductive Endocrine Unit, Massachusetts General Hospital, Boston, Massachusetts 02114
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
subunit and a ß subunit, while activins are dimers of ß subunits. Numerous studies of these proteins have centered on their roles in the reproductive axis, primarily in the pituitary and in the ovary—notably inhibin's inhibitory effect upon pituitary FSH secretion and activin's opposing effect [6]. In addition to the ovary and pituitary, there is evidence that nongonadal tissues, such as endothelial, neural, muscular, and exocrine glandular tissues, produce FS and that FS mediates activin activity in these tissues [1, 7–9]. Follistatin proteins have apparent molecular sizes of 32 and 35 kDa [10, 11]. These molecular size forms arise from alternatively spliced mRNAs between the fifth and sixth exons resulting in 288-amino acid (FS288) and 315 (FS315)-amino acid forms [8]. The truncated FS288 protein binds cell surfaces through an exposed proteoglycan-binding domain, while the longer FS315 form does not seem to bind cell surface proteoglycans [12]. The activin-binding site of FS is in the N-terminus, and all forms of FS bind activin with a similar affinity [12]. FS proteins of 300 amino acids and 303 amino acids also exist and are the products of proteolytic processing [12].
Follistatin mRNA localization and accumulation have been examined in the ovary and pituitary throughout the reproductive cycle; however, the various FS proteins and protein complexes have not been well characterized in these or other tissues [3, 13]. The molecular weight of recombinant hormone complexes can be directly compared to that of detected FS forms from biologic specimens. In the present study we specifically investigated FS and FS complexes in the ovary and pituitary during the normal rat estrous cycle. In addition, FS protein was measured in the rat kidney, a tissue replete with FS [8, 9, 14], and in urine from pregnant women and rats. Identification of FS and FS complexes will allow further exploration of their function in these activin/inhibin-regulated tissues.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Recombinant human (rh) activin A and rh-FS288 were provided through the National Pituitary and Hormone Distribution Program (Rockville, MD) of the NIH. Recombinant human inhibin A was generated in the laboratory. Antibodies that recognize FS, activin, and inhibin used in this study have been previously described [15, 16]. Normally cycling and timed-pregnant Sprague-Dawley-derived female rats were obtained from Harlan BioSciences (Indianapolis, IN). The animals were maintained on a 14L:10D schedule with constant access to food and water prior to and during the study. All animal experimentation was conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals. The collection of urine from pregnant women was approved by the Northwestern Medical School Institutional Review Board review committee. Third-trimester urine samples were collected from healthy pregnant women at their routine prenatal visits. Nonpregnant female urine and male urine samples were collected from volunteers. All samples were centrifuged to remove sediment and stored at -80°C. Separate aliquots were used in each assay. Urine samples were concentrated using Microsep-10 microconcentrators (Filtron Technology [Pall Gelman Sciences, Ann Arbor, MI]).
Analysis of Protein Complexes
Recombinant human FS, rh-inhibin A, and rh-activin A standards, tissue homogenates, and urine samples were subjected to SDS-PAGE in a 4–20% or 12% Tris-glycine gel. Tissues were homogenized in a Tris buffer containing 1 mM EDTA, 10 µg/ml pepstatin A, 0.25 M sucrose, and 17.4 µg/ml PMSF, and total protein amounts were equivalent per gel lane loaded. Samples were reduced in 10% SDS sample buffer with 2.5% ß-mercaptoethanol and then heated at 90°C for 2 min. The gels were fixed in a 10% acetic acid/40% ethanol solution before staining with silver nitrate reagents (serially in 1.7 mM sodium dithionite for sensitization, 12 mM silver nitrate/0.025% formaldehyde solution, then 40 mM sodium thiosulfate pentahydrate solution for development) or were transferred to polyvinylidene fluoride membranes for immunoblot analysis. The monoclonal antibodies used to detect FS via immunoblot (Western) analysis were 7FS30 and 6FS6, either alone or in combination, and immunoreactivity was determined using ECL (enhanced chemiluminescence) detection (Amersham, Arlington Heights, IL). The primary antibodies used in the experiment were biotinylated, and incubation of the blot with horseradish peroxidase (HRP)-streptavidin by itself did not result in the detection of any protein bands. A secondary rabbit anti-mouse antibody conjugated with HRP was used in the pituitary and ovarian studies. Cross-linking of proteins was done as previously described [17].
Immunoassays
Unbound FS was measured in 96-well plates that were coated with 4 µg/ml of the monoclonal antibody 6FS7. Standards and samples were added coincident with a biotinylated secondary antibody, monoclonal 7FS30. The antibody-antigen mixture was allowed to incubate overnight on a shaker at room temperature. After an overnight incubation, the plate was washed and incubated with HRP-streptavidin. The enzyme was incubated for 2 h and the plate was then washed; substrate (orthophenylene diamine) was added, and the samples were read at 490 nm. Recombinant human FS was recovered (>95%) in buffer or urine. Total FS was measured using an assay previously described [15]. Both assays used rh-FS288 as standard.
Inhibin A and inhibin B assay kits were purchased from Serotec (Oxford, England); the characteristics of these dimer-specific assays have been previously described [18]. The assays were validated within the laboratory using rh-inhibin A, rh-inhibin B, rh-activin A, and rh-activin B. The inhibin A standard was the World Health Organization-National Institute for Biological Standards and Control (91/624). The inhibin B standard was prepared by immunoaffinity purification from human follicular fluid and was provided by Dr. Nigel Groome. No cross-reactivity between heterologous ligands was found for any of the assays using purified recombinant proteins. Recombinant human ligands were added at two dilutions into human urine from a pool of men. Inhibin A recovery was 100% and 110% for high and low standards. Inhibin B recovery was 90% and 92% for high and low standards. The assays were sensitive (11.7 pg/ml inhibin A; 31.1 pg/ml inhibin B) and precise. The intraassay variation was 4.8% and 4.2% for inhibin A and inhibin B assays, respectively. The interassay variation was 8.2% and 9.8% for inhibin A and inhibin B assays, respectively.
The activin A concentration was measured using a one-step (simultaneous) monoclonal antibody-based ELISA (2F8:6H5) as described elsewhere [16]. Briefly, Nunc (Nalge Nunc International, Rochester, NY) Immunlon Maxisorp microtiter plates were coated overnight (2F8, 4 µg/ml). Standards, controls, or diluted samples (1:5 or 1:10) and freshly diluted HRP-conjugated 6H5 were added and incubated for 2 h at room temperature. The bound conjugated antibody was measured at 490-nm absorption after addition of orthophenylene diamine and hydrogen peroxide. The assay limit of detection was 100 pg/ml. The ELISA had inter- and intraassay coefficients of variation of less than 11% and 10%, respectively. Activin A was recovered (>95%) in urine samples.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The apparent molecular masses of rh-FS288 analyzed by SDS-PAGE were 39 kDa and 42 kDa under reducing conditions and 32 kDa and 35 kDa under nonreducing conditions (Fig. 1A). The two molecular size forms are accounted for by glycosylation. The increase in apparent mobility when the molecule was reduced may be because the reduced protein occupies more space, which would decrease mobility through the gel. FS contains 36 sulfhyrides that form intramolecular bonds resulting in a compact molecule when intact (i.e., nonreduced). FS-inhibin complexes of 66-kDa, 97-kDa, 133-kDa, and > 220-kDa molecular masses were clearly detected when rh-FS was cross-linked to radiolabeled rh-inhibin A (Fig. 1B). When inhibin A was cross-linked to itself, a faint band at 66 kDa was seen, likely secondary to a small amount of aggregation of this 32-kDa protein. The increased intensity of the 66-kDa band after inhibin-FS cross-linking is likely due to a 1:1 ratio of the two proteins, although experiments aimed at analyzing the ratio of inhibin to FS on a molar basis were unsuccessful (with SDS-PAGE and silver staining). The larger molecular mass proteins may represent a 2:1 combination of two inhibin molecules to one FS. Proteins of 133 kDa and > 220 kDa may be multimers of the 66-kDa or 97-kDa complexes. When radiolabeled activin was cross-linked to FS, 66-kDa, 97-kDa, and larger molecular mass bands were detected by autoradiography. When rh-activin A and FS were combined and cross-linked in molar combinations, a 97-kDa product was detected predominantly when the ratio of FS to activin was 2:1 (Fig. 1C) after SDS-PAGE and silver staining. This result is consistent with an association of two FS molecules with one activin A protein (one FS per ßA subunit). Lastly, FS and activin or inhibin complexes were not retained in complex form when the proteins were analyzed by SDS-PAGE under reducing or nonreducing conditions if the complexes were not first cross-linked (not pictured). To confirm this assignment, a cocktail of monoclonal antibodies (6FS7, 7FS30) specific for FS detected 66-kDa bands and 97-kDa bands in FS-activin samples (Fig. 1C). Analysis of the recombinant complexes using antibodies against the inhibin or activin subunits was not very successful, suggesting that epitopes for these molecules are partially or completely masked when the proteins are associated with FS. By using three different techniques—i.e., silver-stained SDS gel analysis, Western analysis, and gel-shift analysis with iodinated inhibin and activin that were then cross-linked to FS—it was demonstrated that inhibin and activin form complexes with FS with apparent molecular sizes of 66 kDa and 97 kDa. Protein complexes of 133 kDa and > 220 kDa were also formed.
|
Detection of FS Complexes in the Ovary and Pituitary
Protein homogenates from the ovary were analyzed for the presence of FS-containing protein complexes by SDS-PAGE and Western blotting. Protein complexes of 64 kDa, 66 kDa, and 133 kDa were detected. No significant changes in the abundance of these FS-containing complexes were detected during the reproductive cycle—from proestrus 1500 h to proestrus 2400 h to estrus 1500 h of the rat reproductive cycle (Fig. 2 and data not shown). Similarly, proteins of 66 kDa and 133 kDa were detected in homogenates of the pituitary and were invariant through proestrus and estrus (Fig. 2). To determine whether the FS-containing complexes were capable of binding activin (free FS rather than bound FS), pituitary blots were incubated with iodinated activin and exposed to x-ray film. Neither the 66-kDa nor the 133-kDa proteins were detected by ligand blotting, indicating that the FS contained in these proteins was fully occupied (data not shown).
|
Detection of FS Complexes in the Kidney
Follistatin mRNA is present in the kidney; therefore, homogenates from rat kidneys were generated and examined by Western blot analysis using anti-FS antibodies. Identical amounts of total protein were loaded per well. In nonpregnant kidney protein homogenates, FS proteins were detected at 14–20 kDa, 25 kDa, 30 kDa, and 35–40 kDa (Fig. 3). In the pregnant kidney, immunoreactive FS bands were detected at 30 kDa, 35–40 kDa, 97 kDa, and 133 kDa, with evidence of larger molecular size bands as well (Fig. 3). The smaller molecular size proteins may represent proteolytic cleavage products of the FS monomer.
|
FS and FS Complexes in Urine
Because FS mRNA and protein are present in the kidney, it was hypothesized that urine may also contain FS or FS complexes. Using Western blot analysis, FS was not detected in nonpregnant female (human) urine. However, urine collected from pregnant women was replete with the proteins previously documented as FS complexes of 66 kDa, 97 kDa, 133 kDa, and 220 kDa. To further clarify the nature of the urine-related FS, total activin A (free activin and activin complexed to binding proteins), total inhibin A (inhibin A or inhibin A complexes), total inhibin B (inhibin B or inhibin B complexes), free FS, and total FS enzyme-linked immunosorbent assays were established to study the urine samples. Inhibin A was detected in urine from pregnant women (62.19 ± 17.6 pg/ml; n = 8) and was not detected in urine from nonpregnant women or from men. The urinary inhibin A diluted linearly and in parallel to the rh-inhibin A standard curve. Neither inhibin B nor activin A was detected in any of the urine samples studied. An FS ELISA was developed that did not detect FS upon addition of activin. Therefore, the assay was designated a free FS assay. FS was not detected by this ELISA using urine from pregnant or nonpregnant women at various dilutions. This implies that there is no free FS in the urine, consistent with the Western blot analysis. However, in an assay that detects FS when it is associated with inhibin or activin (i.e., "bound" FS), FS was detected in the pregnant human urine samples, suggesting that urine contains complexes of FS and inhibin A (7.1 ± 3.2 ng/ml; n = 5).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bilezikjian et al. [2] detected protein bands at greater than 200 kDa, 90–100 kDa, 50–55 kDa, and 42–44 kDa from anterior pituitary lysates [2]. When rh-FS288 was added to the lysate samples, the antibodies preferentially bound the recombinant ligand, and subsequently the higher molecular weight proteins were not detected. While the lower molecular size proteins were suggested to be variants of FS processing, no explanation was given for the larger proteins. Similarly, Michel et al. [19] demonstrated the presence of different FS forms seen in human cerebrospinal fluid (CSF) in comparison to FS from porcine endothelial cells. The FS proteins in CSF included 65-kDa and 78-kDa sizes. These authors hypothesize that the higher molecular weight proteins may be due to brain-specific posttranslational modifications or to an unknown substance that can cross-react with the anti-FS antibody. In developing bone, Funaba and colleagues [20] identified a 55-kDa as well as a 50-kDa protein (under reduced conditions) for FS. When the same samples were studied for the presence of inhibin/activin ßA subunit, no 30-kDa or smaller band, namely a 14-kDa band consistent with the ßA subunit alone, was identified (implying no activin or inhibin), only a 55- to 60-kDa band. No subunit was detected with antibodies. Therefore, the authors concluded that the bands represent precursor to activin or activin/FS complexes that do not dissociate even after SDS-PAGE.
Given our findings of FS complexes in light of these other data, it is conceivable that, in vivo, after FS binds the ligand (either inhibin or activin), there is a conformational change in the structure of the proteins involved such that prolonged or aggressive reduction is necessary to break the complex into its components. This seems plausible for FS because of structural aspects of the protein: the presence of numerous cysteines that could form cysteine knots [21]; the presence of splice variants, FS288 or FS315, differing by a 27-amino acid sequence that is 44% acidic in humans; and the ability to bind to proteoglycans, which may allow FS to be membrane bound [10]. The different FS isoforms—FS288, FS303 (by proteolytic cleavage), and FS315—may also vary in functional capacity. It has been suggested that the presence of the 27-amino acid tail appears to mask a heparan sulfate proteoglycan-binding site, such that the three different isoforms have different affinities to a cell surface [11] (i.e., the surface of a granulosa cell). FS288 binds with greatest affinity (ED50 = 2 ng/ml), while FS303 has less affinity (ED50 = 10 ng/ml) and FS315 does not associate [11]. Another consideration regarding the difficulty in reducing FS may be that the reduced FS has a less negative charge because its structure or hydration shell does not allow for accessibility to a reductant, thus resulting in a larger protein size upon reduction. Although the activin-binding activity is fairly similar for the three isoforms (Kd = 540–680 pM), it is possible that the difference in the isoform size or in cellular binding ability could alter FS interactions with inhibin. The particular affinity of FS for activin and inhibins may also vary on the basis of specific conditions, such as differing pH milieu or differing amounts of 2 macroglobulin (a low-affinity, large-capacity activin-binding protein) in different organ systems, thereby altering the balance of activin/inhibin action as well as the actions of other related proteins such as TGF-ß, which has been reported to oppose the actions of activin in human trophoblast cells, possibly via alterations in FS [22].
Significant changes in serum activin and FS occur during pregnancy. Specifically, levels of activin A increase in serum from both the pregnant human and rat during the third trimester or its equivalent. [23, 24]. Total FS increases during gestation but free FS remains unchanged [23]. These observations led to the study of FS complexes in the pregnant kidney and urine. The presence of these complexes in third-trimester pregnant urine and its absence in the nonpregnant urine are consistent with the evidence that there are significantly more FS complexes in the pregnant rat kidney in comparison to the nonpregnant kidney. The inability to clearly detect an FS-inhibin complex in nonpregnant human or rat urine suggests two possibilities: the amounts present are below the limits of the Western or ELISA assays, or the protein is not present in the nonpregnant state. While the complete physiology of activin, FS, and inhibin is not yet understood, there is now clear evidence that alterations in serum activin and inhibin levels are associated with abnormal pregnancy states, such as preeclampsia [25]; therefore, changes in urine FS during pregnancy may reflect pathological alterations in normal activin/inhibin function.
While significant differences exist in the FS complexes present in the nonpregnant and pregnant biological samples, our data did not demonstrate any significant change in overall FS levels in the pituitary or ovarian homogenates across the estrous cycle. In a previous study, Besecke et al. [3] demonstrated that pituitary FS levels decreased and then peaked in proestrus. The discrepancy in results may be attributable to limited time points in our sampling such that the peaks and valleys are not detected, or perhaps to a decreased sensitivity of the Western technique when compared to an RIA. The other possibility is that since the prior study was based upon pituitary extracts from tissues that had been perfused, while our data are derived from tissue homogenates, the discrepancy may be related to the presence of cellular-bound forms of FS.
Follistatin has been considered primarily as the binding protein for activin; however, there is evidence that FS may serve other functions that do not relate to its ability to bind activin. FS-deficient mice demonstrate a variety of structural anomalies (shiny taut skin, thirteenth pair of ribs, whisker and tooth irregularities) and a failure to survive after birth, while activin-deficient mutant mice do not show such a widespread pattern of anomalies [26]. This discordance of effects suggests that the impact of FS in development surpasses the range of effects with activin, implying that FS is more than just an activin bioneutralizer. This, in addition to recent data providing evidence that FS can inhibit the TGF-ß effect on cytotrophoblast outgrowth, suggests that FS may also modulate other members of the TGF-ß superfamily or may function independently of these ligands [22]. The demonstration of the FS-inhibin complexes in the present study is intriguing; these complexes may in themselves serve a biological role as a ligand, as a possible method to sequester free FS from activin; but at present it is unclear what functional role, if any, exists for these complexes.
Our data illustrate the elaborate ability of FS to form complexes with activin and inhibin. Significantly, the free form of FS is not detected in biological tissues. While the bioneutralizing actions of FS on activin at the level of pituitary FSH secretion and erythroid cell differentiation are understood, FS regulation of inhibin, the presence of cell-bound FS, and potential conformational changes of FS-inhibin/activin complex have not been correlated with functional roles in the kidney and other organs. The physiology of these alternative FS forms warrants investigation; in the meantime, the ability to detect changes in the levels of FS and inhibin in the urine in pregnancy should be further explored under the conditions of normal and abnormal pregnancy.
|
FOOTNOTES |
|---|
2 Correspondence: Teresa K. Woodruff, Northwestern University, Department of Neurobiology and Physiology, O.T. Hogan 4–150, 2153 Sheridan Road, Evanston, IL 60208. FAX: 847 491 2224; tkw{at}nwu.edu
Accepted: August 17, 1998.
Received: April 17, 1998.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Schwartz Intercellular Communication in the Anterior Pituitary Endocr. Rev., October 1, 2000; 21(5): 488 - 513. [Abstract] [Full Text]
|
|
|
|
 |
|
 Q. Wang, H. T. Keutmann, A. L. Schneyer, and P. M. Sluss Analysis of Human Follistatin Structure: Identification of Two Discontinuous N-Terminal Sequences Coding for Activin A Binding and Structural Consequences of Activin Binding to Native Proteins Endocrinology, September 1, 2000; 141(9): 3183 - 3193. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. C. Cipriano, L. Chen, T. R. Kumar, and M. M. Matzuk Follistatin Is a Modulator of Gonadal Tumor Progression and the Activin-Induced Wasting Syndrome in Inhibin-Deficient Mice Endocrinology, July 1, 2000; 141(7): 2319 - 2327. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Wu, H. Patel, S. Mukai, C. Melino, R. Garg, X. Ni, J. Chang, and C. Peng Activin, Inhibin, and Follistatin in Zebrafish Ovary: Expression and Role in Oocyte Maturation Biol Reprod, June 1, 2000; 62(6): 1585 - 1592. [Abstract] [Full Text]
|
|
|
|
 |
|
 D. C. Danila, X. Zhang, Y. Zhou, G. R. Dickersin, J. A. Fletcher, E. T. Hedley-Whyte, M. K. Selig, S. R. Johnson, and A. Klibanski A Human Pituitary Tumor-Derived Folliculostellate Cell Line J. Clin. Endocrinol. Metab., March 1, 2000; 85(3): 1180 - 1187. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
