| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Monash Institute of Reproduction and Development, Monash University (S.L.M., E.M.A.B., A.E.O., J.F.S., D.J.P., G.P.R.), Melbourne, Victoria 3168, Australia; Prince Henry’s Institute of Medical Research (J.-F.E.), Clayton, Victoria 3168, Australia; and School of Biological and Molecular Sciences, Oxford Brookes University (M.C., N.P.G.), Headington 0X3 0BP, Oxford, United Kingdom
Address all correspondence and requests for reprints to: Prof. Gail P. Risbridger, Centre for Urological Research, Monash Institute of Reproduction and Development, Monash Medical Center, 246 Clayton Road, Clayton, Victoria 3168, Australia. E-mail: gail.risbridger{at}med.monash.edu.au.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Activin signal transduction is initiated by ligand binding to induce the formation of a heteromeric receptor complex, thereby stimulating the phosphorylation of downstream Smad signaling proteins (6, 7, 8). The interaction of Smad proteins with transcription factors and specific DNA sequences in promoter regions regulate the expression of specific subsets of genes. For example, in response to activin, the Xenopus DNA-binding transcription factor, Forkhead activin signal transducer-1, binds to the complex of Smad2 and Smad4 to activate the activin response element (ARE) on the Mix.2 promoter (9, 10). Examples of mammalian genes regulated directly by activin signaling through the Smads include the GnRH receptor (11), junB (12), and Smad7 (13).
Growth of the human prostate is an androgen-driven process modulated by members of the TGFß superfamily. Activins inhibit proliferation and induce apoptosis in human prostate tumor cells (1, 14). We have previously shown that both benign and malignant human prostate tissues express activin ßA- and ßB-subunit mRNA and protein (15, 16). Exogenous activin A protein inhibited DNA synthesis of the human prostate tumor cell line, LNCaP. Activin A also had an inhibitory effect on the growth of PC3 cells once the activin-binding protein, follistatin 288 (FS288), had been neutralized with an anti-FS288 antibody (1).
The novel activin subunit, activin ßC, was cloned in 1995 (17). Based on amino acid sequence, the activin ßC-subunit has been grouped with the ßD- and ßE-subunits in a subgroup distinct from the activin ßA- and ßB-subunits (18). The human activin ßC-subunit is expressed in the prostate (19) and other tissues such as the liver (17, 20), ovary, and testis (21). It is not known whether dimers consisting of the activin ßC-subunit transduce signals through activin receptors or if they use different receptors. The lack of any functional data regarding the actions of these activin subunits has hampered analysis of potential signaling mechanisms.
Despite the deficiency of information regarding the role of the activin ßC-subunit, expression patterns imply that the activin ßC-subunit has a function in various tissues, including the human prostate gland. In the prostate, the ßC-subunit protein colocalizes with ßA- and ßB-subunit proteins in the benign basal epithelium and in tumor cells (19). Human prostate tumor cell lines LNCaP, DU145, and PC3 express ßA, ßB, and ßC mRNA (22), although measurable levels of activin A were detected only in PC3 cells (1). In contrast to activin A, activin C homodimeric protein has no effect on the proliferation of LNCaP cells (19).
In addition, activin ßC-subunit knockout mice do not show any major abnormal phenotype in the prostate or any other organ where the activin ßC-subunit is expressed (23). However, investigations into prostate phenotypes involved gross histological analysis, so it is not known whether subtle phenotypes exist in these mice (23). Therefore, despite ongoing investigation of the activin ßC-subunit for almost a decade, no biological roles for the activin C homodimeric protein or the individual ßC-subunits have been established in the prostate or other organs.
To ascertain the function of the activin ßC-subunit in the prostate, we have focused our attention on the role of the monomeric activin ßC-subunit rather than the homodimeric protein. In vitro investigations from this laboratory have demonstrated that the activin ßC-subunit can heterodimerize with the activin ßA- or ßB-subunits to form the putative ligands, activin AC and BC (19). Therefore, we postulated that one of the consequences of formation of heterodimers containing the activin ßC-subunit was the regulation of levels and/or bioactivity of activin A, AB, and B dimers in prostate tumor cells by decreasing their intracellular formation (19, 24). However, without the development of a specific assay to differentiate between homodimers and heterodimers containing the activin ßC-subunit, this hypothesis remained unproven.
In this study we compared the actions of activin A (ßA-ßA), activin B (ßB-ßB), and activin C (ßC-ßC) homodimer proteins on human prostate tumor cell line growth and on the activation of AREs. This study also describes the development and validation of the first ELISA to detect the formation of activin AC protein in human prostate tumor cell line supernatants. In addition, we investigated whether the overexpression of the activin ßC-subunit in the PC3 human prostate tumor cell line, and thus the formation of activin AC heterodimers, reduced the levels of activin A homodimers and concomitantly decreased the stimulation of the activin intracellular signaling cascade.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Chinese hamster ovary (CHO) and pituitary cell lines
The CHO cell line was obtained from American Type Culture Collection. The LßT2 cell line was given by Pamela Mellon (University of California, San Diego, CA). This cell line was originally derived from pituitary tumors induced by the targeted expression of a transgene consisting of 1.8 kb of the rat LHß promoter linked to the oncogene simian virus 40 T antigen (25). Both cell lines were maintained in DMEM with 10% FCS. CHO cells were cultured in the presence of nonessential amino acids.
Expression constructs and reporter constructs
Human activin ßC-subunit cDNA was subcloned into the pRK5 expression vector as described previously (19). The pRK5 vector was supplied by Dr. Anthony Mason (Monash University, Melbourne, Australia). The reporter construct pGL3–5.5oFSHß (26) was a gift from Prof. William Miller (Department of Biochemistry, North Carolina State University, Raleigh, NC). The reporter construct 3XGRAS-PRL-lux consisting of three copies of the GnRH receptor-activating sequence, and the PRL promoter was a gift from Dr. Buffy S. Ellsworth (Colorado State University, Fort Collins, CO) (11). The reporter construct p3TP-lux, consisting of TPA response elements and a portion of the plasminogen activator inhibitor promoter, was a gift from Prof. J. Massagué (Memorial Sloan-Kettering Cancer Center, New York, NY) (27). The pSV-ß-galactosidase vector was purchased from Promega Corp. (Madison, WI). The pCMVß vector [consisting of a ß-galactosidase gene driven by the cytomegalovirus (CMV) promoter] was obtained from Clontech Laboratories (Palo Alto, CA). The pAR3-lux reporter construct consisted of a firefly luciferase gene driven by three tandem repeats of the Mix.2 ARE and a TATA box and was a gift from Prof. Yan Chen (Indiana University School of Medicine, Indianapolis, IN) (9). The plasmid pRL-CMV, which expresses Renilla luciferase under the control of the cytomegalovirus promoter, was obtained from Promega Corp.
Reagents
Human recombinant (hr-) activins A and B were purchased from R&D Systems, Inc. (Minneapolis, MN). hr-activin C was provided by Biopharm GmbH (Heidelberg, Germany). Bovine FS was provided by Prof. D. M. de Kretser (Monash Institute of Reproduction and Development, Melbourne, Australia). hr-FS288 (the short membrane-bound splice isoform of FS) was a gift from the NIDDK (Bethesda, MD). The bovine follicular fluid (bFF) used in the activin AC ELISA was aspirated from follicles from ovaries collected from a local abattoir. The ovaries were stored on ice until aspiration. The pooled bFF was spun for 10 min at 5000 x g, aliquoted, and stored at -80 C until use.
[3H]Thymidine incorporation assay for DNA synthesis
LNCaP cells were plated at a density of 5000 cells/well, in DMEM/5% FCS into 96-well plates (Falcon) for 72 h. Medium was removed and replaced with 40 ng/ml activin A, activin B, or activin C homodimers or vehicle buffer controls in DMEM/5% FCS and incubated for 2 d. Activin C homodimer protein at concentrations of 50, 100, 150, and 200 ng/ml with equivalent vehicle buffer controls were incubated with LNCaP cells in a similar manner. [3H]Thymidine (0.5 µCi/ml) was added to the cells for 20 h, after which the cells were harvested using a Micromate 196 Cell Harvester (Packard Instrument Co., Meriden, CT), and levels of [3H]thymidine incorporation were determined. Tritiated thymidine ([3H]thymidine) was obtained from NEN Life Science Products (Boston, MA).
Transient transfection of CHO and LßT2 cells
For transient transfection of the p3TP-lux or the pAR3-lux reporter constructs, 50,000 CHO cells/well were cultured in 24-well plates (70–80% confluence) for 24 h. The Fugene6 reagent (Roche, Castle Hill, Australia) was used for transfections at a ratio of 1:3 (micrograms of DNA to micrograms of Fugene6 reagent) according to the manufacturer’s instructions. Briefly, cells were transfected with 250 ng reporter construct and 25 ng pCMVß vector to monitor transfection efficiencies. The activin treatment was applied 24 h posttransfection; the cells were washed twice with PBS, and the medium was changed to DMEM plus nonessential amino acids and 0.2% FBS with or without 12.5 ng/ml activin A, activin B, and activin C homodimer protein. The cells were then incubated for an additional 24 h.
For transient transfection of the pGL3–5.5oFSHß or 3XGRAS-PRL-lux reporter constructs, 500,000 LßT2 cells/well were cultured in 24-well plates (70–80% confluence) for 24 h. The Fugene6 reagent (Roche) was then used for transfections at a ratio of 1:3 (micrograms of DNA to micrograms of Fugene6 reagent) according to the manufacturer’s instructions. Briefly, cells were transfected with 250 ng reporter construct and 25 ng pCMVß vector to monitor transfection efficiencies. The activin treatment was applied 24 h post transfection; the cells were prewashed with PBS, and the medium was changed to DMEM and 0.2% FCS with 12.5 ng/ml activin A, activin B, and activin C homodimer protein. The cells were then incubated for an additional 24 h before luciferase assays.
Luciferase and ß-galactosidase assays
Cells were washed twice with ice-cold PBS and then lysed in 200 µl lysis buffer (1% Triton X-100, 25 mM glycylglycine, 15 mM MgSO4, 4 mM EGTA, and 1 mM dithiothreitol). The cells were then incubated on ice for 30 min before collection of the cell lysate. For the luciferase assay, 50 µl cell lysate were mixed with 300 µl assay buffer [25 mM glycylglycine, 15 mM MgSO4, 4 mM EGTA, 15 mM potassium phosphate buffer (pH 7.8), 1 mM dithiothreitol, and 2 mM ATP]. The luciferase activity was measured for 2 sec using a luminometer (Berthold Australia, Bundoora, Australia) after injection of the luciferase substrate (Luciferin, Promega Corp.). For the ß-galactosidase assay, 10 µl supernatant were mixed with 50 µl Galacton-Star galactosidase substrate (PE Applied Biosystems, Foster City, CA), and the ß-galactosidase activity was counted after a 30-min incubation using a LumiCount 96-well plate reader (Packard Instrument Co.). Luciferase activity is expressed as relative activity calculated by dividing luciferase activity in relative light units by ß-galactosidase activity.
Activin AC ELISA
Plates were coated and blocked as previously described (28) with human activin ßC-subunit clone 1 monoclonal antibody (19) on 96-well ELISA plates (MaxiSorp, Nunc, Roskilde, Denmark). bFF was used as an interim standard. The top dose in the assay, equivalent to a 1:10 dilution, was assigned the arbitrary value of 10 U/ml. Standards and samples were diluted in DMEM/5% FCS, as used in the culture experiments. One hundred and twenty-five microliters of a 6% sodium dodecyl sulfate (SDS) solution in PBS was added (3% final concentration, wt/vol) to 125 µl sample or standard, mixed, boiled for 3 min, and allowed to cool. The addition of PBS to the SDS solution was found to improve the performance of the assay and the linearity of the dose-response curve of the standard and samples. Thereafter, 20 µl 30% H2O2 (2% final concentration, vol/vol) was added, and the tubes were incubated at room temperature for 30 min. To each well, were added 25 µl 20% BSA/0.1 M Tris/0.9% NaCl/5% Triton X-100/0.1% sodium azide before the addition of 100-µl duplicates of the treated samples. Plates were incubated overnight in a sealed humidified box. The next day, the plates were washed with 0.05 M Tris, 0.9% NaCl, 0.05% Tween 20, and 0.1% NaN3 before 50 µl biotinylated E4 monoclonal antibody directed at the activin ßA-subunit (29) in 5% BSA, 0.1 M Tris, 0.9% NaCl, 5% Triton X-100, and 0.1% sodium azide was added to each well and incubated for 2 h at room temperature. After washing, alkaline phosphatase linked to streptavidin (Invitrogen, Carlsbad, CA) was added to the wells and incubated at room temperature for 1 h. After further washes, the alkaline phosphatase activity was detected using an amplification kit (ELISA Amplification System, Invitrogen) in which the substrate was incubated for 1 h at room temperature, followed by the addition of an amplifying reagent. The reaction was stopped by the addition of 50 µl 0.4 M H2SO4. The plates were read at 492 nm with a 630-nm reference filter on a Multiskan RC plate reader (Labsystems, Helsinki, Finland), and data were processed using Genesis Lite EIA software (Labsystems). The assay was optimized and assessed for performance, specificity, accuracy, and precision.
Activin A ELISA
Activin A homodimer was measured using a specific ELISA (29) according to the manufacturer’s instructions (Oxford Bio-Innovations, Oxfordshire, UK) with some modifications. The standard used was human recombinant activin A as described previously (30), which was calibrated against human recombinant activin A (R&D Systems, Inc.). Serial dilutions of conditioned PC3 medium samples gave linear dose-response curves, which were parallel to the standard (data not shown). The average intraplate coefficient of variation (CV) was 7.8%, the interplate CV was 7.7% (n = 2), and the average limit of detection for the assays was 0.032 ng/ml.
Statistics
To determine linearity and parallelism of dose-response curves from serially diluted standard and test samples measured in the activin AC ELISA, the data were log-transformed, and linear regression analysis was performed. The curves were regarded as parallel if the 95% confidence limits of the slopes overlapped. Dose-response curves of the standard were compared using a two-tailed t test. Other data were analyzed after log transformation using one-way ANOVA with Dunnett’s multiple comparison or Bonferroni multiple comparison post hoc tests (PRISM, GraphPad, San Diego, CA).
Western blot and immunodetection
SDS-PAGE was performed under reducing conditions using a 15% polyacrylamide gel. Conditioned medium samples, hr-activin A, or hr-activin C proteins were diluted 1:2 in reducing buffer [7 mol/liter urea, 0.1% NaH2PO4.H2O, 1% SDS, and 0.01% bromophenol blue (pH 7.2)]. Samples were incubated at 100 C in a heat block for 10 min and centrifuged briefly. The gel was run at 200 V, constant milliamperes for 30 min with running buffer (0.19 mol/liter Tris, 0.29 mol/liter glycine, and 10% SDS). An Immobilon P polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA), that had been preincubated in methanol for 15 sec and with milliQ water for 2 min was equilibrated along with the gel in transfer buffer (0.7 mol/liter glycine, 0.3 mol/liter Tris, and 15.6% ethanol) for 5 min. The proteins in the gel were transferred to the membrane overnight at 30 V and 75 mAmps. After transfer, the membrane was soaked in milliQ water, then in methanol for 10 sec, followed by milliQ water for 2 min. The membrane was blocked (5% nonfat milk powder, 0.01% Tween 20 in PBS) for 60 min and washed (1% nonfat milk powder and 0.01% Tween 20 in PBS) for 5 min. All washes were repeated three times. Activin ßC-subunit clone 1 antibody was added at a 1:5000 dilution in 1% milk (0.3 µg/ml) in PBS overnight at 4 C. After washing, the membrane was incubated with goat antimouse HRP (Dako Corp., Carpinteria, CA; 1:10,000) in 1% milk in PBS for 2 h at room temperature. After subsequent washes, ECL Plus substrate (Amersham Pharmacia Biotech, Little Chalfont, UK) was added according to the manufacturer’s instructions. The membrane was placed in an x-ray cassette and exposed to X-OMAT film (Eastman Kodak, Rochester, NY).
Transient transfection of PC3 cells
PC3 cells were plated at 200,000 cells/well in DMEM/10% FCS into 12-well plates (70–80% confluence) for 24 h. Transient cotransfection combined pARE3-lux (1 µg), pRK5-ßC or pRK5 control (2.59 µg), and pRL-CMV (10 ng) control with Superfect (Qiagen, Valencia, CA), at a ratio of 1:1.7 (micrograms of DNA to microliters of Superfect reagent) according to the manufacturer’s instructions. Conditioned media and PC3 cells were collected at 24, 48, and 72 h and stored at -20 C.
Luciferase assay for PC3 cells
Cells were washed with PBS and then lysed with 300 µl passive lysis buffer (1x; Promega Corp.) while the culture plate was rocked at room temperature for 30 min. The luciferase assay was performed using the Dual-Luciferase Reporter Assay kit (Promega Corp.). Briefly, 20 µl PC3 cell lysate were added to 96-well luminescent solid assay plates (Corning, Acton, MA). Luciferase Assay Reagent (100 µl; Promega Corp.) was added, and firefly luciferase was measured on a LumiCount 96-well plate reader (Packard Instrument Co.). After the luciferase reading, 100 µl Stop and Glo reagent (Promega Corp.) were added to each well, and Renilla luciferase was measured as described above. The luciferase activities are represented as relative activities (firefly luciferase activity divided by Renilla luciferase activity).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The effects of activin A, B, and C homodimers on DNA synthesis in the LNCaP prostate tumor cell line were investigated (Fig. 1A
). Consistent with previous studies (1), activins A and B significantly (P < 0.01) inhibited DNA synthesis at a dose of 40 ng/ml compared with vehicle buffer only. In contrast, the same dose of activin C had no effect on LNCaP cell DNA synthesis, suggesting that activin C homodimer, in contrast to activins A and B, does not inhibit LNCaP cell proliferation. To discount the possibility that a higher concentration of activin C protein was needed to elicit a biological effect on LNCaP cells, a range of doses (50–200 ng/ml activin C) were added. Compared with the significant inhibition of DNA synthesis by activin A (P < 0.0001), doses of up to 200 ng/ml activin C had no obvious effect on LNCaP cell proliferation (Fig. 1B), confirming our previous observations that up to 200 ng/ml activin C had no effect on LNCaP or HepG2 cells (19).
|
). The ARE AR3-lux was stimulated by activin A approximately 4-fold (P < 0.01) and by activin B approximately 6-fold (P < 0.01) compared with vehicle alone (Fig. 2B). In the LßT2 cells, activin A activated the GnRH receptor-activating sequence element, 3XGRAS-PRL-lux, approximately 3.3-fold (P < 0.01), and activin B produced approximately 3.2-fold activation (P < 0.01) compared with vehicle alone (Fig. 2C). The FSHß promoter reporter construct, pGL3–5.5oFSHß, was stimulated by activin A approximately 1.6-fold (P < 0.05) and by activin B approximately 2.2-fold (P < 0.01) compared with vehicle alone (Fig. 2D). In contrast, activin C protein had no effect on any of the activin-responsive promoters tested (Fig. 2, A–D), suggesting that activin C homodimer does not induce the activin A and B signaling pathway.
|
). This assay is the first assay for measurement of activin AC heterodimer and was developed and validated for the specific purpose of measuring activin AC in culture medium.
|
Sample treatment
It has been shown previously that preassay sample denaturation and oxidation resulted in an increased response in similar immunoassays using the E4 monoclonal antibody directed at the activin ßA-subunit (29). The bFF standard (serially diluted in unconditioned culture medium) and conditioned medium samples from activin ßC-subunit transfected PC3 cells were assayed with and without pretreatment to assess the performance in this assay in order to achieve the best dose-response curve (Fig. 3A
). The bFF standard (
) did not display a dose-responsive effect without sample pretreatment. After a denaturation step with SDS/PBS and boiling, the bFF standard (
) dose response improved to small degree. An oxidation step improved the range of the bFF standard (
); however, the dose-response curve was not linear. The combination of both denaturation and oxidation steps resulted in the greatest range of bFF standard (
) and a linear dose-response curve. Similarly, the medium sample required pretreatment to establish a dose-response curve (
). Denaturation and boiling markedly improved the dose response of the medium sample (
); however, oxidation treatment had little effect (
). The combination of the pretreatments (denaturation, boiling, and oxidation) gave a linear dose-response of the medium sample (
). When this method was employed, both the bFF standard () and the activin ßC-subunit-transfected, PC3-conditioned medium sample () gave linear dose-response curves that were parallel to each other, as shown by a comparison of slopes with overlapping 95% confidence limits of log-transformed data (Fig. 3B). Subsequently, these pretreatment conditions were selected for use in the activin AC ELISA.
Accuracy and precision
The accuracy of the assay was determined by spiking medium samples with a known amount of bFF, equivalent to 1.0 U/ml, to determine the percent recovery of activin AC. Mean recoveries were 97.6 ± 7.3% from four samples on each of eight plates, indicating that quantitative recoveries were achieved from the test samples. The mean intraplate CV was 6.5%, and the interplate CV was 3.9%. The limit of detection, defined as the standard dose equivalent to the mean ± 2 SD of absorbance of the blank replicates (n = 6), was 0.04 U/ml. Therefore, both the accuracy and the precision of the novel activin AC ELISA were high, with quantitative recoveries of a known amount of bFF from spiked medium samples.
Specificity
The specificity of the assay was more difficult to quantify in the absence of purified activin AC protein of known mass. However, no interference or cross-reactivity was detected in the assay when high concentrations of activin B or C in the dose range of 15.63–500 ng/ml were added (data not shown). In addition, mean recoveries of a known amount of the bFF in the presence of either activin B or activin C (15.63–500 ng/ml) were 96.0 ± 3.1 and 100.9 ± 5.7 (n = 6), respectively. Therefore, high concentrations of activin B or C proteins did not interfere with the assay. As both the bFF used as the standard and the conditioned medium samples contain large amounts of activin A, the cross-reaction of activin A was assessed in several ways to determine the ability of the ELISA to accurately measure activin AC dimer. The bFF standard used in the activin AC ELISA contains the equivalent of 0.91–234 ng/ml (0.045–11.7 ng/well) activin A, as determined using the activin A ELISA. When a range of doses of activin A (0.313–20 and 6.25–400 ng/ml) was added alone in the activin AC ELISA, a small effect was seen only above a dose of 20 ng/well, which is equivalent to 100 ng/ml (Fig. 3C). Additionally, a dose of 50 ng/ml (2.5 ng/well) activin A was added to each of the doses of bFF in the standard curve. This is greater than the activin A concentration in the ßC-subunit-transfected, PC3-conditioned medium samples (<40 ng/ml; Fig. 5C). There was no displacement of the curve, nor was there any significant difference (P = 0.975) compared with the curve of the normal unspiked standard curve (Fig. 3C). These results suggest that the concentration of endogenous activin A was too low to interfere with measurement of the activin AC heterodimer concentration in the activin ßC-subunit transfected PC3 cell conditioned medium. This indicates that activin A does not have a significant effect or cross-reaction in the activin AC ELISA and that the activin AC heterodimer concentrations measured are not affected by the presence of activin A homodimer in the samples. To assess the potential interference of FS in the assay, 1 U/ml bFF was preincubated with a range of doses of hr-FS288 or bovine FS. Follistatin concentrations up to 1 µg/ml had no effect on the assay, demonstrating that the assay can measure total activin AC even in the presence of FS, bound or unbound (data not shown). Therefore, FS does not interfere with measurement of activin AC using the novel ELISA assay when added in excess. Follistatin binds to the activin ßA-subunit, but it is not known whether FS can bind to the human activin ßC-subunit. The recently published study by Hashimoto et al. (20) demonstrated the ability of bovine FS to bind to the newly cloned human activin ßE-subunit. As the activin ßE-subunit is closely related to the activin ßC-subunit (18), it is possible that FS also binds to the activin ßC-subunit, although the functional consequences of this possible interaction are not known. Nonetheless, the presence of excess FS did not affect the measurement of activin AC protein using the novel ELISA.
|
Activin AC heterodimer protein formation in vitro
To investigate the functional consequences of activin ßC-subunit overexpression in prostate tumor cells, the PC3 cell line was transfected with an expression vector consisting of the human activin ßC-subunit cDNA driven by the CMV promoter. The PC3 cell line was chosen for this series of experiments because it produces measurable amounts of activin A homodimer (1); therefore, the formation of the putative activin AC heterodimer could potentially be observed in this cell line upon overexpression of the activin ßC-subunit. Using a specific antibody to the activin ßC-subunit, Western blot analysis under reducing conditions showed that conditioned medium from PC3 cells transfected with the activin ßC-subunit contained a band of 13 kDa, similar in size to reduced hr-activin ßC protein (Fig. 4, lanes 1–3 and 6). No band was detected in PC3 medium from cells transfected with the vector alone (Fig. 4, lane 7). No cross-reaction was detected with 40 ng hr-activin A (Fig. 4, lane 4).
|
Endogenous activin AC was detected at low levels in cells transfected with the control vector, pRK5 alone (Fig. 5A, ). Overexpression of the activin ßC-subunit in PC3 cells resulted in increased levels of secreted activin AC protein (Fig. 5A, ) compared with cells transfected with the control vector (). Notably, at each individual time point, activin AC production was significantly higher in the supernatant from activin ßC-subunit-overexpressing cells PC3 cells than in the conditioned medium from control vector-transfected cells (P < 0.001). Measurable levels of the activin AC heterodimer were detected in conditioned medium from PC3 cells transfected with the control vector (Fig. 5A), indicating that these cells express low levels of endogenous activin ßC-subunit, consistent with our previous observation that the activin ßC mRNA subunit is expressed in this cell line (22), although this was below the sensitivity of the Western blot (Fig. 4).
To determine whether overexpression of the ßC-subunit affects the production of endogenous activin A dimer, activin A homodimeric protein was measured using the same conditioned medium samples (Fig. 5B) as those described above. Endogenously produced activin A was significantly lower at each individual time point (24, 48, and 72 h) in conditioned medium from activin ßC-subunit-overexpressing PC3 cells compared with corresponding control samples (P < 0.001). Although a component of the basal activin AC heterodimer measurement (Fig. 5A) may be explained by cross-reaction with activin A homodimers, we observed that when levels of activin AC heterodimer increased, those of activin A homodimers decreased (Fig. 5B). This observation together with the assay validation data argue strongly against considerable cross-reactivity with activin A homodimers in the activin AC heterodimer ELISA.
The decrease in activin A production associated with overproduction of activin AC suggests that the cells overexpressing the ßC-subunit may exhibit lower activin activity. To test this hypothesis, PC3 cells were cotransfected with the activin-responsive reporter construct, pAR3-lux, with or without the activin ßC-subunit expression vector (Fig. 5C). Relative luciferase activity in PC3 cells overexpressing the activin ßC-subunit was significantly lower at 48 h (P < 0.001) and 72 h (P < 0.001) compared with that in PC3 cells transfected with the control vector alone, but not at 24 h. Therefore, in the PC3 cells an increase in endogenously produced activin AC heterodimer was associated with the attenuation of endogenous production of activin A homodimer in addition to a significant decrease in pAR3-lux activation at 48 and 72 h. Significantly lower pAR3-lux activation was not observed at 24 h. Although activin A protein levels were significantly decreased at 24 h, this change was relatively small.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We have previously demonstrated that activin ßA- and ßC-subunits colocalize in the human prostate (19). Therefore, based on our previous results, we proposed that elevation of activin ßC-subunit levels in prostate tumor cells, relative to activin ßA-subunit levels, reduces the level of bioactive activin A due to the intracellular heterodimerization of activin AC. Activin A is a potent inhibitor of benign and malignant prostate cell growth in vitro (1, 22), and we proposed that a reduction in activin A would release cells from inhibitory constraints and promote cell proliferation.
The development of the novel activin AC ELISA enabled us to pursue investigations into the consequences of overexpressing the activin ßC-subunit in human prostate cancer cell lines. To investigate our hypothesis that the activin ßC-subunit is an antagonist of activin A homodimer formation, the activin ßC-subunit was overexpressed in an activin A-secreting human prostate tumor cell line, PC3. We initially established that activin ßA-ßC heterodimers could be synthesized intracellularly, and activin AC was measured in conditioned medium of activin ßC-transfected PC3 cells using the activin AC ELISA.
With the increase in activin AC heterodimer formation and the consequent reduction in activin A homodimer formation, a decrease in activation of the activin signaling pathway in activin ßC-transfected PC3 cells was observed, as measured by transactivation of a promoter containing an ARE (Fig. 5C). This is consistent with a lesser amount of available bioactive activin A acting via the activin receptors. These results are consistent with our hypothesis that activin ßC-subunit expression regulates the levels of activin A (and possibly activin AB) and therefore modulates the biological response to this ligand.
PC3 cells overexpressing the activin ßC-subunit display a reduction in endogenous activin A production as well as a decrease in activin signaling, which occurs even in the presence of FS. This is surprising, as exogenous activin A has no effect on activation of the ARE when transfected into the activin A- and FS288-secreting PC-3 cells (34). Furthermore, PC3 cell proliferation is not reduced by exposure to exogenous activin A unless FS is first removed with a neutralizing antibody (1). However, these data are strikingly similar to those previously reported with PA1 cells, where Delbaere and colleagues (35) were able to discriminate between exogenous and endogenous activin A responses by a human ovarian cancer cell line, PA1. PA1 cells, like PC3 cells, secrete FS and do not respond to exogenous activin A without prior removal of FS with heparin treatment. However, the induced production of endogenous activin A by PA1 cells leads to a reduction in the proliferation of these cells even in the presence of FS. They propose a model in which FS acts as a barrier to exogenous activin A, but not for endogenous activin A (35). It may be possible that the PC3 cells overexpressing the activin ßC-subunit provide another model system to test the effects of exogenous vs. endogenous activin action.
The heterodimerization of activin ßC and activin ßB-subunits would also result in a decrease in activin B homodimer formation and may contribute to the significant decrease in ARE activation in these experiments. Activin BC measurements are beyond the scope of these investigations and are currently not possible because an activin BC ELISA is not available. The possibility that activin C or activin AC (or even BC) is acting to block the activin receptor in these ßC-subunit transfected cells is an equally plausible explanation for the reduction in activin signaling. However, the role of activin heterodimers as activin receptor antagonists cannot be tested until purified activin AC is available. Previous studies from our laboratory imply that exogenous activin C protein does not disrupt the access of activin A to the activin receptor in LNCaP cells after coincubation of activin A with activin C (19).
The functional consequences of the proposed heterodimeric interaction are not unprecedented within the TGFß superfamily; members of the superfamily can act as antagonists of other members of this family through dimerization. For instance, a familiar example is heterodimerization of the activin ß-subunits with the inhibin 
-subunit to form inhibin proteins such as inhibin A or B (ßA and ßB), which reduces the formation of activin dimers. The resulting inhibin proteins themselves also have opposing actions to the activins. In addition, nodal antagonizes bone morphogenetic protein 7 signaling via the formation of an intracellular nodal bone morphogenetic protein 7 heterodimer (36). This is an example of mutual inhibition of signaling concurrent with dimeric ligand production (36). Our data suggest that the activin ßC-subunit may have a functional role via a heterodimerization mechanism analogous to those observed between other members of the TGFß superfamily. A recent study demonstrating other novel heterodimers of rat activin CE and BE in vitro using cotransfection techniques further demonstrates the potential for homodimer function to be modulated by dimerization with other activin ß-subunits (37).
The implication of the activin ßC-subunit acting as a potential antagonist of activin A homodimer formation is relevant to organs that express both activin ßA- and ßC-subunits, such as the human testis, ovary, liver, and prostate. Furthermore, the detection of activin AC heterodimers in a biological fluid such as bFF, which contains bioactive inhibins A and B and activins A and AB (30, 38), suggests that activin AC is functionally significant. We have previously demonstrated the immunolocalization of various combinations of activin ßA-, ßB-, and ßC-subunit proteins in human prostate basal, tumor, nerve, endothelial, and smooth muscle cells (19). Therefore, any cells that coexpress activin ßC- and ßA-subunit mRNA or protein may synthesize activin AC protein.
As these data support the hypothesis that activin ßC-subunit acts to reduce homo- and heterodimerization of other activin subunits, the lack of apparent phenotype of the activin ßC-subunit knockout mouse is not remarkable (23). Many of the TGFß superfamily members exhibit high homology and have overlapping functions. Therefore, other activin subunits may compensate for the loss of the activin ßC-subunit in these mice. This phenomenon has been observed in other knockout models, with the activin ßB-subunit able to replace some of the functions of the activin ßA-subunit (39). Furthermore, if our hypothesis is correct, other antagonists of activin formation or bioactivity, such as the inhibin -subunit and FS, may exhibit redundancy with the activin ßC-subunit. More elaborate cross-breeding programs between these different knockout models would address some of these issues. Furthermore, given the data presented here, mice overexpressing the activin ßC-subunit may be more informative than those in which the activin ßC-subunit has been knocked out.
We demonstrate for the first time that activin ßC-subunit overexpression drives the production of activin AC heterodimers, thereby reducing the levels of bioactive activin A. These data have important clinical implications, as elevated activin ßC-subunit in tumor cells may cause abnormal growth regulation and promote prostate tumor growth by disrupting activin A’s inhibitory ability. In conclusion, these data suggest that the activin ßC-subunit may act as a modulator of activin activity not only in the prostate, but in any organ that is regulated by activin A and that expresses the ßC-subunit. It is conceivable that activin AC itself may act as an activin receptor antagonist, be inactive, or have a novel function in addition to sequestration of the activin ßA-subunit; however, until the activin AC heterodimers are purified and appropriate bioassays developed, this possibility remains unknown. Certainly, the development of a novel ELISA for the activin AC heterodimer will prove invaluable in investigating the formation and the role of activin AC. Although this assay was developed and validated specifically for the quantification of activin AC heterodimers in conditioned medium, with further development the assay could be extended to measurement of activin AC in various biological fluids and in vivo. Development of assays and recombinant proteins for other heterodimers, such as activin BC and CE, will complement the activin AC assay and assist in understanding the roles of the activin ßC-subunit. The data reported here indicate that evaluation of expression of the activin ßC-subunit is warranted in tissues that synthesize or respond to dimers comprising activin ßA- or ßB-subunits.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Current address for J.-F.E.: Centre for Cancer Therapeutics, Ottawa Regional Cancer Center, 503 Smyth Road, Ottawa, Ontario, Canada K1H 1C4.
Abbreviations: ARE, Activin response element; bFF, bovine follicular fluid; CHO, Chinese hamster ovary; CMV, cytomegalovirus; CV, coefficient of variation; FCS, fetal calf serum; FS, follistatin; hr-, human recombinant; SDS, sodium dodecyl sulfate.
Received February 19, 2003.
Accepted for publication June 18, 2003.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  E. Gold, N. Jetly, M. K. O'Bryan, S. Meachem, D. Srinivasan, S. Behuria, L. G. Sanchez-Partida, T. Woodruff, S. Hedwards, H. Wang, et al. Activin C Antagonizes Activin A in Vitro and Overexpression Leads to Pathologies in Vivo Am. J. Pathol., January 1, 2009; 174(1): 184 - 195. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Barakat, A. E O'Connor, E. Gold, D. M de Kretser, and K. L Loveland Inhibin, activin, follistatin and FSH serum levels and testicular production are highly modulated during the first spermatogenic wave in mice Reproduction, September 1, 2008; 136(3): 345 - 359. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. M Bilezikjian, A. L Blount, C. J Donaldson, and W. W Vale Pituitary actions of ligands of the TGF-{beta} family: activins and inhibins. Reproduction, August 1, 2006; 132(2): 207 - 215. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 U. Muenster, C. A. Harrison, C. Donaldson, W. Vale, and W. H. Fischer An Activin-A/C Chimera Exhibits Activin and Myostatin Antagonistic Properties J. Biol. Chem., November 4, 2005; 280(44): 36626 - 36632. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E J Gold, X Zhang, A M Wheatley, S L Mellor, M Cranfield, G P Risbridger, N P Groome, and J S Fleming {beta}A- and {beta}C-activin, follistatin, activin receptor mRNA and {beta}C-activin peptide expression during rat liver regeneration J. Mol. Endocrinol., April 1, 2005; 34(2): 505 - 515. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Gaedeke, T. Boehler, K. Budde, H.-H. Neumayer, and H. Peters Glomerular activin A overexpression is linked to fibrosis in anti-Thy1 glomerulonephritis Nephrol. Dial. Transplant., February 1, 2005; 20(2): 319 - 328. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Wada, A. Maeshima, Y.-Q. Zhang, Y. Hasegawa, H. Kuwano, and I. Kojima Assessment of the function of the {beta}C-subunit of activin in cultured hepatocytes Am J Physiol Endocrinol Metab, August 1, 2004; 287(2): E247 - E254. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
